Clonal populations of Leptosphaeria maculans contaminating cabbage in Mexico

Authors


E-mail: rouxel@versailles.inra.fr

Abstract

The race structure and genotypic diversity of four Leptosphaeria maculans populations isolated from Brassica oleracea (broccoli, cauliflower, cabbage, etc.) in central Mexico (Aguascalientes, Guanajuato and Zacatecas states) were analysed. Race structure was characterized by an unusually low diversity at three locations out of four. Fourteen minisatellite markers revealed a high proportion of repeated multilocus genotypes in these populations, combined with a significant linkage disequilibrium and strong clonal fraction (65–87%). The occurrence of the mating-type idiomorphs always significantly departed from the 1:1 proportion expected in the case of random mating. Each population thus consists of a few (four to nine) multilocus genotypes which are specific to each location. These data strongly support the hypothesis of exclusive, or a high rate of, clonal multiplication. Comparison of cropping practices between B. oleracea and B. napus indicate that the shift in reproductive behaviour of the fungus is chiefly man-mediated.

Introduction

Fungal phytopathogens often exhibit a wide range of mating behaviours, ranging from obligate sexual to fully asexual cycle. Many fungal species thus have a reproduction cycle comprising a mixture of sexual and asexual reproduction. The mating system is expected to affect strongly the adaptation response of a parasite population, as it modulates the level and structure of genetic diversity within and among individuals. Predominantly outcrossing populations should present more genotype diversity than asexual ones (McDonald & Linde, 2002), and produce more rapidly advantageous gene combinations, thus allowing the parasites to circumvent host resistance (Milgroom, 1996; Taylor et al., 1999). Short-term changes in the environment could provide an advantage to sex when mutations at several loci are required for adaptation to the new situation (Brown, 1999). Conversely, clonality allows the fast amplification of a highly fit individual, and a shift of the population to a new optimal gene combination (McDonald & Linde, 2002). In many plant pathogenic fungi, reproduction is mainly clonal (e.g. all Fusarium oxysporum formae speciales, Magnaporthe oryzae) (Milgroom, 1996; Taylor et al., 1999). These pathogens exist as a series of discrete clones or clonal lineages with little evidence for recombination among clonal lineages (McDonald & Linde, 2002). However, this may be a man-mediated bias, because truly asexual lineages are believed not to persist for evolutionary significant lengths of times (Taylor et al., 1999). Whereas sexual reproduction favours diversification of genotypes and better adaptation to changing environments, clonal populations are also able to generate pathogenic diversity by mechanisms including parasexuality (Tinline & MacNeil, 1969), transposition events (Hua-Van et al., 2011), transfer of nonconventional genetic material, such as ‘B’ chromosomes, via lateral gene transfer (Zolan, 1995; Ma et al., 2010), mitosis-driven chromosomal rearrangements (Zolan, 1995), spontaneous losses caused by location of genes in dispensable regions of the genome (e.g. telomeres; Farman, 2007), or generation of point mutations as a result of large population sizes.

Leptosphaeria maculans is an important pathogen of crucifers, with major economic incidence on oilseed rape (Brassica napus) worldwide. It develops gene-for-gene interactions with its brassica hosts (B. napus, B. rapa and B. juncea) whereby major resistance genes in the plant (termed Rlm) ‘recognize’ avirulence genes (AvrLm) in the fungus and induce plant defence responses (Rouxel & Balesdent, 2005). Population surveys to identify the combination of AvrLm alleles present in individual isolates of the fungus (termed races) are of major agronomical importance in choosing resistance genes that can be used by farmers at a given location (Dilmaghani et al., 2009). Leptosphaeria maculans has a very complex life and parasitic cycle, with sexual mating occurring on stem residues, followed by large-scale dispersion of ascospores initiating a short necrotrophic stage on leaves, in turn followed by a lengthy period of endophytic symptomless systemic growth within plant tissues, leading eventually to necrosis at the base of the stem many months after the initiation of the leaf lesions (Rouxel & Balesdent, 2005). Stem canker is of agronomic significance because it can cause lodging of the plant before harvest. Like many other ascomycetes, L. maculans is heterothallic and has a single mating type (MAT) locus with two alternate forms (termed idiomorphs): haploid isolates must carry opposite idiomorphs to be able to mate. In all parts of the world, the disease is considered as monocyclic, i.e. one annual cycle takes place between primary infection and the next sexual reproduction (West et al., 2001). The asexual multiplication that takes place mainly on leaf spots is well known to be of very low incidence in the disease cycle because of the low infection efficiency of conidia, their limited germination potential, and their small-scale dispersion (West et al., 2001; Travadon et al., 2007). However, recent findings indicate that in some specific climate and cropping conditions such as those in western Canada, conidia can be dispersed over greater distances than expected (Guo & Fernando, 2005) and asexual multiplication may not be negligible in the fungus life cycle (Dilmaghani et al., 2012). In addition, the low infectivity of conidia can be enhanced if ascospore contamination occurs concomitantly, suggesting cryptic importance of conidia in the epidemic cycle (Li et al., 2006, 2007).

Leptosphaeria maculans is also an important pathogen of B. oleracea (broccoli, cauliflower, cabbages, etc.), and the first epidemics were recorded at the end of the 19th and the beginning of the 20th centuries on B. oleracea (Wilcox, 1913; Henderson, 1918). However, it is now mostly considered to be of decreasing importance to B. oleracea because of improved seed quality, including use of treated seeds, management of seedbeds and use of rotations in the field (Sherf & MacNab, 1986; Babadoost, 1989). In contrast to what is described on oilseed rape, the sexual stage is rarely found and is believed to play little or no part in the disease cycle on B. oleracea (Babadoost, 1989). Recent phylogeography studies indicated that L. maculans may have been primarily a pathogen of vegetable brassicas, but became adapted to oilseed rape and widely disseminated following the success of oilseed rape as a major crop (Dilmaghani et al., 2012). However, only L. maculans populations originating from oilseed rape have been analysed to date and there is no information on how isolates from cabbage may be different in terms of race or population structures (Gout et al., 2006a; Dilmaghani et al., 2012). Preliminary small-scale analysis of L. maculans isolates obtained from cabbages in Mexico indicated a very low diversity of isolates and a disequilibrium in mating-type distribution within the populations (Moreno-Rico et al., 2001; Dilmaghani et al., 2009).

In this study, B. oleracea trap cultivars were set up in three locations in central Mexico and isolates naturally infecting leaves were collected. Race and population structures were analysed and the populations compared to previously analysed populations originating from oilseed rape. The factors that may have caused changes in reproduction regimes in these specific populations are discussed.

Materials and methods

Brassica oleracea core collection

To evaluate the occurrence of known resistance genes in B. oleracea and the existence of novel resistance sources, a core collection of 392 accessions of B. oleracea was constructed and structured at ISA Lisbon. The choice of the accessions was made by request to the main European and international genebanks, and included a few genotypes from two Japanese seed companies (Table S1). The accessions originated from 28 different countries from Europe, Asia, America and Africa (Table S1). The collection encompassed eight varieties representing the main B. oleracea morphotypes (var. acephala, kale; var. alboglabra, Chinese broccoli; var. botrytis, cauliflower and romanesco; var. italica, broccoli; var. capitata, cabbage; var. costata, Portuguese cabbage; var. gemmifera, Brussels sprouts and var. gongylodes, Italian turnip) along with one ‘rapid-cycling’ accession developed at the University of Madison, Wisconsin (Table S1).

Fungal isolates and DNA extraction

Four European reference isolates were used for the characterization of the B. oleracea collection (Table S1): three were obtained from oilseed rape in Germany (T12aD34) (Balesdent et al., 2005), France (v11.1.1) (Rouxel et al., 2003b) and Portugal (Lm501) (J. S. Dias, unpublished data). All three isolates have the same complement of Avr genes and belong to the most common race of L. maculans found on oilseed rape in Europe, race Av5-6-7-8 (Balesdent et al., 2006; Stachowiak et al., 2006) (for more details on race terminology, see footnote in Table 4). The fourth isolate, BBA62908, one of the oldest available in European collections, is of uncertain date of isolation and host origin, but is thought to have been collected from Beta vulgaris in Germany in 1966 (Balesdent et al., 2005). It harbours most of the currently known avirulence genes and belongs to race Av1-2-4-5-6-7-8 (Rouxel et al., 2003b; Balesdent et al., 2006), similarly to currently known isolates obtained from cabbage (Dilmaghani et al., 2009).

The Mexican isolates were collected from two different states of Mexico: Aguascalientes (Asientos, 71 isolates and Aguascalientes, 58 isolates) and Guanajuato (Salamanca, 71 isolates) in 2008 (Fig. 1). To these were added 28 isolates obtain in 2005 from a third state, Zacatecas (Fig. 1), the race structure of which was described by Dilmaghani et al. (2009). Different broccoli, cauliflower and white cabbage cultivars were used as trap cultivars according to the protocol established by Balesdent et al. (2006) for oilseed rape. Whenever possible these were first confirmed to harbour no resistance genes known in oilseed rape following a cotyledon inoculation test with a set of differential isolates. Cultivars Iron Man, Neptuno, Monaco and Marathon (Table 1), for example, lacked resistance genes Rlm1, Rlm2, Rlm4, Rlm6 and Rlm7, and, when this could be evaluated, Rlm5 and Rlm8 (M. H. Balesdent, unpublished data). Such a lack of Rlm genes is consistent with the common assumption of the high level of susceptibility of B. oleracea to L. maculans (Ferreira et al., 1993) and the data presented here on the B. oleracea core collection.

Figure 1.

 Locations of sampling of Mexican isolates of Leptosphaeria maculans and schematic representation of the genotypes (MLGs) and their frequency at each location. The genotypes correspond to the combination of alleles at 14 minisatellite loci. For a given location each distinct genotype is represented by a different shading. Note that, as no MLG is shared between two locations, similar colour codes at two sampling locations are not indicative of shared MLGs. As, Asientos, Aguascalientes state; Ag, Aguascalientes, Aguascalientes state; Oj, Ojocaliente, Zacatecas state; Sa, Salamanca, Guanajuato state.

Table 1. Origins and characteristics of Leptosphaeria maculans isolates analysed in this study
Sampling locationStateSublocation B. oleracea var.aCultivarTotal no. of isolatesNo. of selected isolatesbNo. of distinct genotypes
  1. aVarieties of B. oleracea: var. italica, broccoli; var. botrytis, cauliflower; var. capitata, white cabbage.

  2. bNumber of isolates randomly selected in each location from the total sample for population genetics studies.

  3. cUnknown genotype of white cabbage.

OjocalienteZacatecas‘El Diamente’ ranch botrytis Apex 8 81
OjocalienteZacatecasVitar ranch italica Marathon 7 73
OjocalienteZacatecasOld ranch italica 104 7 72
OjocalienteZacatecasOld ranch italica Neptuno 7 62
AguascalientesAguascalientes‘Medio kilo’ ranch italica Iron Man 58308
AsientosAguascalientesEl Borrego italica Patriot 44167
AsientosAguascalientesEl Borrego capitata Unknownc 38103
SalamancaGuanajuato‘El Fuerte’ ranch italica Domador103304

The different locations had specific contrasting traits: presence of a unique plant genotype at a single location (Aguascalientes and Salamanca), presence of two unrelated B. oleracea morphotypes at a single location (Asientos), and presence of four B. oleracea genotypes in three separated fields (Ojocaliente; Table 1). In all cases except at the Ojocaliente location, one plot of the highly susceptible oilseed rape (B. napus var. oleifera) cv. Westar was placed alongside the B. oleracea plots to be used as a conventional B. napus trap cultivar (Balesdent et al., 2006), but none of these showed the presence of leaf spots typical of primary symptoms caused by L. maculans in the course of the experiment.

In the fields, leaf lesions were never found on B. oleracea plants recently transplanted from the seedbed and leaves with symptoms were only found 1–3 months after transplanting. One (and only one) leaf showing symptoms and the presence of pycnidia was collected from each individual plant, and care was taken to collect diseased material from distantly located plants. Small pieces of leaves with lesions were placed in Petri dishes containing wetted tissue paper and incubated for 2–3 days to induce pycnidia to exude conidia. The conidia were picked up with toothpicks under a binocular microscope and streaked onto V8 agar medium containing antibiotics (West et al., 2002) and hyphal tip isolates were obtained from single pycnidia. All the cultures were maintained on V8 juice agar and conidia produced on V8 juice agar as previously described (Ansan-Melayah et al., 1995). Conidia were collected from 12- to 15-day-old sporulating cultures for pathogenicity tests and DNA extraction (Balesdent et al., 1998). Genomic DNA was extracted from conidial suspensions using the DNeasy 96 Plant Kit and the QIAGEN BioRobot 3000 in accordance with the manufacturer’s recommendations. In vitro crosses between isolates harbouring opposite mating types were performed according to Ansan-Melayah et al. (1995).

Pathogenicity tests

For large-scale screening of resistance sources in B. oleracea, 392 accessions of the core collection were inoculated with the four reference European isolates (Table S1). For identification of the presence of known Avr genes in Mexican isolates, a differential plant set comprising genetically fixed cultivars/lines of B. napus that possess one or two major resistance (Rlm) genes was used for pathogenicity tests (Table 2). Isolates were inoculated onto cotyledons according to established protocols (Balesdent et al., 2006). Lesion phenotypes were scored on a 1–6 scale at 14 and at 21 days after inoculation according to the IMASCORE rating scale (Balesdent et al., 2006). Scores of 1–3 indicated resistance and isolates were avirulent (i.e. harbouring the avirulence allele matching the corresponding Rlm gene), whereas scores of >3 to 6 indicated susceptibility and isolates were virulent (i.e. harbouring the virulence allele matching the corresponding Rlm gene). The Avr allele combination for each isolate, termed a race, was encoded as Avx-y-z according to Balesdent et al. (2005).

Table 2. List and Rlm gene complement of Brassica napus cultivars used to characterize the collection of isolates of Leptosphaeria maculans (Balesdent et al., 2005; M. H. Balesdent, unpublished data)
Cultivar/lineResistance gene(s) present
WestarNone
Columbus Rlm1, Rlm3
Koster Rlm2, Rlm9
Darmor-MX Rlm6, Rlm9
Pixel Rlm4
01-23-2-1 Rlm7
03-22-3-1 Rlm3
Goéland Rlm9
Grizzly Rlm2, Rlm3
Heaven Rlm2, Rlm7
00-136-2-1 Rlm1, Rlm7

Isolate genotyping and data collection

Fourteen genetically and physically independent minisatellite (MS) markers were used for population genetics analyses (Table S2). PCRs were performed in an Eppendorf Mastercycler EP Gradient thermocycler (Eppendorf) as described previously (Gout et al., 2006b). The PCR products were separated by electrophoresis in 3% agarose gels supplemented with ethidium bromide. The allele sizes were determined using quantity one 1-D analysis software (Bio-Rad), by comparison with band sizes of the 1-kb+ ladder (Invitrogen) and that of an internal control with alleles of known size and number of repeat units (the sequenced L. maculans isolate, v23.1.3). Data were scored as number of repeat units for each minisatellite.

A multiplex PCR, developed to rapidly characterize the mating type of L. maculans isolates, was used to determine the distribution of the Mat1-1 and Mat1-2 mating types within populations (Cozijnsen & Howlett, 2003).

Data analyses

Race structure

The frequency of avirulence alleles and race complexity of isolates (number of virulences per isolate) within populations were analysed. Population diversity was analysed using the Margalef index, which measures species (here race) richness of each population, and the Simpson index of richness, which takes into account the evenness of races within populations (Pinon & Frey, 1997; Balesdent et al., 2006).

Genetic variation within samples

Nei’s gene diversity (H) was estimated at each locus and within each subpopulation (Nei, 1973) as implemented in fstat v. 2.9.3 (Goudet, 1995). fstat was also used to determine the number of alleles per locus, the numbers of private alleles in the populations, and to estimate allelic richness based on a minimum sample size of 26 individuals. The proportion of polymorphic loci was calculated for each population by averaging the number of loci with frequencies of the most common allele below 0·99. The number of different multilocus genotypes (MLG) and the level of genotypic diversity, defined as the probability that two individuals taken at random have different genotypes, were calculated using the multilocus program (v. Win 1.3b; Agapow & Burt, 2001). These data were used to quantify the proportion of repeated multilocus genotypes, calculated as 1 – ((number of different genotypes)/(total number of isolates)). Given the low diversity at the 14 MS loci, multilocus was also used to estimate their power to identify all genetically unique isolates within samples by resampling loci and recalculating the number of unique MLGs and the level of genotypic diversity for all locus combinations from 1 to 14. The mean level of genotypic diversity reached an asymptote over all combinations of loci with a subset of nine of the 14 markers, indicating that a reliable identification of clones could be expected in the sampled populations.

Two analyses were performed to test for the presence of true clones in the populations. First, overall tests for clonal population structure were carried out using the genodive program (v. 2.0b13; Meirmans & Van Tienderen, 2004). This analysis compares the observed diversity of MLGs to that of permuted data sets which approximate expectations under random sexual reproduction. If clonal structure exists, then the mean number of unique MLGs simulated is greater than the number of unique MLGs observed. Secondly, the probability that repeated MLGs were indeed clones and not just frequent genotypes resulting from the combinations of frequent alleles was determined using the genclone program (v. 2.0b13; Meirmans & Van Tienderen, 2004). The probability of observing n copies of an MLG in a sample (Psex) was estimated for each multicopy MLG. If the Psex value of a repeated MLG at n = 2 is significant (significance determined at Psex < 0·05), then all copies of that MLG can be considered to result from asexual reproduction. To examine the mutational relationships among the repeated genotypes, a clustering analysis of the 26 MLGs was performed using a minimum spanning tree approach (‘minimum.spanning.tree’) implemented in the R package igraph (Csardi & Nepusz, 2006). The genetic distances of Bruvo et al. (2004), which incorporate MS repeat number, were calculated for all pairs of MLGs and were used to compute a minimum spanning tree of haplotypes.

The hypothesis of random mating was tested using three approaches. First, the distribution of mating types was compared to the 1:1 ratio expected under random mating for a haploid fungus, using a χ2 test on the data set without or after clone correction. Secondly, associations of alleles among different loci were calculated with the standardized version of the index of association rd, using multilocus. The index of association is a measure of multilocus linkage disequilibrium (Brown et al., 1980), which gives information on whether two different individuals sharing the same allele at one locus are more likely to share the same allele at another locus. rd varies between 0 (complete panmixia) and 1 (no recombination). The significance of rd was established by comparing the observed value with the distribution obtained from 1000 randomizations for which alleles at each locus are resampled without replacement to simulate the effect of random mating. Thirdly, linkage disequilibrium between all pairs of loci was calculated in genepop v. 3.4 (Raymond & Rousset, 1995). The null hypothesis that genotypes at one locus are independent from genotypes at another locus was tested using Fisher’s exact test and the Markov chain method using default settings.

Population differentiation and clustering methods

For these analyses, a clone-corrected data set was used in which only a single individual for each MLG was included per population. Genetic differentiation between clone-corrected populations was examined using two different methods. First, non-parametric methods were used: a principal coordinate analysis (PCoA) was performed on a matrix of chord distances among samples (Cavalli-Sforza & Edwards, 1967). The chord-distance matrix was based upon the allele present at each locus for each isolate and was built using the populations software (Langella, 1997) and the PCoA was performed with GenAlEx (Peakall & Smouse, 2006). This analysis included previously investigated worldwide populations of L. maculans (Dilmaghani et al., 2012). Secondly, hierarchical analysis of molecular variance (amova) was estimated using arlequin v. 3.1 (Zoological Institute, Department of Biology, University of Bern), whereby the partitioning of genetic variability was examined among/within populations. The significance was tested by 10 000 randomizations of the haplotype data. Thirdly, arlequin was used to calculate FST to examine pairwise differences between populations separated spatially and temporally, and between the types of site sampled.

Results

Susceptibility of B. oleracea to L. maculans

All 392 accessions of B. oleracea were extremely susceptible to the L. maculans isolate with the highest number of avirulence alleles, BBA62908, while only six accessions showed moderate susceptibility to two or three of the isolates obtained from oilseed rape. Similarly, Apex, Marathon and Iron Man were shown in previous studies to be highly to moderately susceptible to local populations of L. maculans or to BBA62908 (Moreno-Rico et al., 2005b; M. H. Balesdent, unpublished data). Hence, all accessions analysed here, whatever their country of origin, the morphotype to which they belonged or the isolate used, were moderately to highly susceptible to L. maculans (Table S1).

Frequency of avirulence alleles and race structure of Mexican populations

Similar to what was observed following a sampling limited to one location and only a few isolates in Ojocaliente (Dilmaghani et al., 2009), the Mexican populations showed remarkably low race diversity relative to that known for populations originating from B. napus worldwide. All isolates were highly virulent on the control B. napus cv. Westar (also used, unsuccessfully, as a trap cultivar alongside the B. oleracea fields in Mexico) and produced abundant pycnidia on cotyledon lesions (data not shown). Avirulence towards Rlm6 was fixed in all populations, as were avirulence towards Rlm2, Rlm4 and Rlm7, and virulence towards Rlm3 and Rlm9 in three populations out of four (Table 3). Accordingly, race structure was identical in Salamanca to that previously found in Ojocaliente (Dilmaghani et al., 2009) and only one race was represented, race Av1-2-4-(5)-6-7-(8). This race has rarely been found anywhere other than Mexico to date, the only two exceptions being isolate BBA62908 and an isolate obtained from cabbage in France in the 1970s (Balesdent et al., 2005). Only two races were found in Aguascalientes, of which the main one was Av1-2-4-(5)-6-7-(8) and the second one only differed by the absence of AvrLm1 (Tables 3 and 4). Only the Asientos location departed from this extremely simple race structure (Tables 3 and 4). AvrLm6 was fixed in this population, and AvrLm2 and AvrLm1 were prevalent in it, but all other alleles were polymorphic, and AvrLm3 and AvrLm9, which were absent from other populations, were found in the Asientos population (Table 3). Nine different races were found at this location, but many of them were only represented by a limited number of isolates (Table 4). In Asientos, the prevalent race was Av1-2-4-(5)-6-7-(8), as at all other Mexican locations (Table 4).

Table 3. Number of races and frequencies of avirulence (AvrLm) alleles in Mexican isolates of Leptosphaeria maculans
Sample locationNo. of isolatesNo. of racesMargalef indexSimpson diversity indexFrequency of avr alleles (%)
AvrLm1 AvrLm2 AvrLm3 AvrLm4 AvrLm6 AvrLm7 AvrLm9
  1. aData for Ojocaliente are from Dilmaghani et al. (2009).

Aguascalientes5820·250·24286·21000100100100 0
Salamanca711001001000100100100 0
Asientos7191·880·38195·790·115·578·910078·911.3
Ojocaliente281001001000100100100 0
Table 4. Leptosphaeria maculans races identified in Mexican populations
RaceAguascalientes (%)Salamanca (%)Asientos (%)Ojocaliente (%)No. of isolates
  1. aRace terminology according to Balesdent et al. (2005): each L. maculans race is identified by the successive numbers of Avr loci for which the isolate has been characterized and is avirulent, preceded by the letters ‘Av’, to show that only Avr alleles are indicated. Avr loci for which the genotyping has not been possible are indicated in parentheses. Note that presence of AvrLm9 in the isolate does not allow identification of the genotype at locus AvrLm6 with the differential set used here.

Av1-2-4-(5)-6-7-(8)a86·210078·6100204
Av2-4-(5)-6-7-(8)13·80 008
Av2-3-(5)-6-(8)00 1·401
Av3-(5)-(6)-(8)-900 2·802
Av1-2-(5)-6-(8)00 4·303
Av1-2-3-(5)-6-(8)00 1·401
Av1-3-(5)-6-(8)00 2·802
Av1-2-3-(5)-(6)-(8)-900 2·802
Av1-3-(5)-(6)-(8)-900 4·303
Av1-2-(5)-(6)-(8)-900 1·401

Genetic diversity

Population genetics studies were performed on a subset of 26–30 randomly chosen isolates per population (including all 28 isolates from the Ojocaliente population for which races were previously determined; Dilmaghani et al., 2009. A total of 114 isolates was tested (Table 1). Of the MS markers used, one, MinLm935.2, showed a complete absence of polymorphism in all populations, as observed previously in some Canadian populations in which the same allele as the one found here was prevalent (Dilmaghani, 2010). The other 13 MS markers also showed a very limited range of variation compared to populations obtained worldwide from B. napus (Dilmaghani et al., 2012), with only two to four alleles, depending on the population (Table 5). The range of variation was even more limited within single locations, with only three cases where more than two alleles were present at one given MS locus (MinLm1377 and MinLm8 at Ojocaliente, and MinLm3 at Asientos; Table 5). In total, 34 alleles were found using the 14 MS loci (Table 5). In accordance with these data, fixed alleles were found at all locations, but mainly at Aguascalientes, with seven fixed alleles out of 13 polymorphic loci (Table 5). As previously found in western Canadian populations, some MS showed a remarkable lack of diversity in Mexican populations, whereas they were very diverse in populations from other areas (Gout et al., 2006a; Dilmaghani et al., 2012). Gene diversity was low (H ≤ 0·40), as was allelic richness (overall allelic richness around 1·71) at all four locations (Table 5).

Table 5. Frequency of minisatellite (MS) alleles across four populations of Leptosphaeria maculans from Mexico
LocusMS alleleaOjocalienteAguascalientesAsientosSalamanca Hb
  1. aAlleles are defined as n-x by the number of repeats of the core motif of the minisatellite (see Table S2 for more details on minisatellites analysed here).

  2. bH, Nei’s index of diversity at each locus across each population.

  3. cPrivate alleles, defined here as alleles found in a single population throughout the study, are underlined.

  4. dAllelic richness averaged across loci (±standard error).

MinLm5554-x0·180·530·771·000·47
5-x0·820·470·230·00
MinLm6323-x1·001·001·000·600·19
5-x0·000·000·00 0·40c
MinLm935.23-x1·001·001·001·000·00
MinLm9676-x0·960·470·230·000·66
7-x0·040·530·000·50
12-x0·000·000·770·50
MinLm13343-x0·000·000·310·400·29
4-x1·001·000·690·60
MinLm13683-x0·820·470·270·600·50
3·5-x0·180·530·730·40
MinLm13772-x0·680·530·000·500·67
4-x0·000·000·770·50
8-x0·140·470·230·00
9-x 0·18 0·000·000·00
MinLm17214-x0·180·000·850·500·47
5-x0·821·000·150·50
MinLm18381-x0·860·130·350·500·50
2-x0·140·870·650·50
MinLm24513-x 1·00 0·000·000·000·37
4-x0·001·001·001·00
MinLm311-x0·000·000·770·600·58
13-x0·000·000·040·40
15-x1·001·000·190·00
MinLm53-x0·961·000·770·500·32
4-x 0·04 0·000·000·00
5-x0·000·000·230·50
MinLm63-x0·870·370·040·500·61
4-x0·000·000·00 0·50
5-x0·130·630·960·00
MinLm89-x0·110·000·230·400·32
10-x0·851·000·770·60
11-x 0·04 0·000·000·00
Total gene diversity H 0·170·200·280·400·26
Allelic richnessd 1·78 ± 0·191·43 ± 0·141·86 ± 0·141·79 ± 0·431·71 ± 0·43

Genotypic diversity

Genotypic diversity was very low, with only 26 MLG found among the 114 individuals analysed and four to nine MLGs found per location (Table 6). As a result, the G index of genotypic diversity was among the lowest found to date for L. maculans populations (Gout et al., 2006a; Dilmaghani et al., 2012; Table 6). Most MLGs were detected many times in each population (up to 15 times for a MLG present in Ojocaliente; Fig. 1, Table 7) and seven MLGs were represented by a single isolate. Each MLG was only found at a single location, and none of the MLGs were shared between the Mexican populations. genodive analysis indicated strong probabilities of clonality in all the populations (< 0·001). Furthermore, nine of the 19 repeated MLGs had Psex < 0·05 at n = 2 and seven additional multicopy MLGs had Psex < 0·01 at n = 3 (Table 7), indicating evidence of clonal reproduction in all four populations.

Table 6. Distribution of mating types and measures of genotypic diversity in Mexican populations of Leptosphaeria maculans
LocationSample sizeMat1-1/Mat1-2aNo. of MLGsPercentage of polymorphic lociClonal fractionGenotypic diversityrdb Pc
  1. aMat1-1/Mat1-2 is the ratio of the number of isolates carrying mating type 1 (Mat1-1) to isolates carrying mating type 2 (Mat1-2); the Mat1-1/Mat1-2 ratio in the clone-corrected data set is indicated in brackets; the probability associated with the χ2 test for the 1:1 distribution ratio is indicated; *< 0·05; ***< 0·001.

  2. brd, index of association.

  3. cProbability associated with the index of association following 1000 randomizations of the haplotype data.

Ojocaliente2819/9* (3/2)564·290·820·670·20<0·001
Aguascalientes3019/11* (4/4)842·860·730·850·32<0·001
Asientos2612/14* (6/3)978·570·650·810·19<0·001
Salamanca3027/3*** (3/1)478·570·870·680·72<0·001
Table 7. List of multilocus genotypes (MLGs) identified in the Mexican populations of Leptosphaeria maculans
MLG idOriginNo. of isolatesMat1-1/Mat1-2a Psexb (n = 2) Psex (n = 3)Varieties of B. oleracea sampled (no. of MLGs detected)
  1. aMat1-1/Mat1-2 is the ratio of the number of isolates sharing a repeated MLG and carrying mating type 1 (Mat1-1) to isolates carrying mating type 2 (Mat1-2).

  2. bPsex is the probability that any pair of identical MLGs originated from distinct events of sexual reproduction.

MLG1Ojocaliente50/50·003<0·001 B. oleracea var. italica cv. 104 (5)
MLG2Ojocaliente33/00·009<0·001 B. oleracea var. italica cv. Marathon (3)
MLG3Ojocaliente1515/00·0390·001 B. oleracea var. botrytis cv. Apex (7)
B. oleracea var. italica cv. 104 (2)
B. oleracea var. italica cv. Marathon (1)
B. oleracea var. italica cv. Neptuno (5)
MLG4Ojocaliente40/40·021<0·001 B. oleracea var. italica cv. Marathon (4)
MLG5Ojocaliente11/0 B. oleracea var. italica cv. Neptuno (1)
MLG6Aguascalientes33/00·1050·006 B. oleracea var. italica cv. Iron Man (3)
MLG7Aguascalientes99/00·0540·001 B. oleracea var. italica cv. Iron Man (9)
MLG8Aguascalientes44/00·0640·002 B. oleracea var. italica cv. Iron Man (4)
MLG9Aguascalientes10/1 B. oleracea var. italica cv. Iron Man (1)
MLG10Aguascalientes30/30·1130·007 B. oleracea var. italica cv. Iron Man (3)
MLG11Aguascalientes33/00·1090·006 B. oleracea var. italica cv. Iron Man (3)
MLG12Aguascalientes60/60·1340·009 B. oleracea var. italica cv. Iron Man (6)
MLG13Aguascalientes10/1 B. oleracea var. italica cv. Iron Man (1)
MLG14Asientos100/100·029<0·001 B. oleracea var. capitata (unknown) (4)
B. oleracea var. italica cv. Patriot (6)
MLG15Asientos30/30·0490·001 B. oleracea var. italica cv. Patriot (3)
MLG16Asientos11/0 B. oleracea var. italica cv. Patriot (1)
MLG17Asientos55/0<0·001<0·001 B. oleracea var. capitata (unknown) (5)
MLG18Asientos11/0 B. oleracea var. capitata (unknown) (1)
MLG19Asientos22/00·003 B. oleracea var. italica cv. Patriot (2)
MLG20Asientos11/0 B. oleracea var. italica cv. Patriot (1)
MLG21Asientos22/00·054 B. oleracea var. italica cv. Patriot (2)
MLG22Asientos10/1 B. oleracea var. italica cv. Patriot (1)
MLG23Salamanca1212/00·022<0·001 B. oleracea var. italica cv. Domador (12)
MLG24Salamanca33/00·0810·003 B. oleracea var. italica cv. Domador (3)
MLG25Salamanca30/30·002<0·001 B. oleracea var. italica cv. Domador (3)
MLG26Salamanca1212/0<0·001<0·001 B. oleracea var. italica cv. Domador (12)

The low level of genotypic diversity within Mexican populations is also well illustrated by the minimum spanning tree showing the mutational relationships among the 26 MLGs (Fig. 2). The tree indicates that rare or singleton MLGs tend to group by region of origin, thereby supporting the epidemiological history of these isolates. Their distribution across the tree also suggests that novel MLGs evolved by accumulating mutational differences from one or a few founder genotypes in each population, which is exemplified by MLG3 from the Asientos population or MLG7 and MLG12 from the Aguascalientes population (Fig. 2). The local evolution of genotypes in a stepwise fashion further corroborates the hypothesis of a high rate of clonal multiplication of L. maculans in these populations.

Figure 2.

 Minimum spanning tree of 114 Leptosphaeria maculans isolates inferred from 14 minisatellite loci. Each circle represents a unique genotype. Node size is proportional to MLG frequency and distances between nodes are proportional to the distance between MLGs according to Bruvo et al. (2004). The code next to each circle identifies each MLG, and circle shading indicates the region of origin of the isolates.

Mating-type distribution and gametic equilibrium

Both mating types were present in each population, as shown following PCR amplification, but only one mating type could be detected in each repeated MLG (Tables 6 and 7). In addition, tests were performed on two Mexican isolates, OMR19 (Mat1-1) and OMR21 (Mat1-2), to determine if they could be crossed in vitro with one of the two tester isolates obtained from oilseed rape, v23.1.2 (Mat1-1) and v23.1.3 (Mat1-2). Pseudothecia produced for each of the compatible crosses produced abundant and viable ascospores (data not shown). However, in all locations investigated here the mating-type distribution significantly departed from the 1:1 distribution observed in most other locations investigated to date (Gout et al., 2006a; Dilmaghani et al., 2012; Table 6).

The rd index of association was high in all populations, and as high as 0·72 (< 0·001) for the Salamanca population, suggesting essentially clonal reproductive behaviour (Table 6). Accordingly, high clonal fractions (65–87%) were observed for all the populations (Table 6). Tests for pairwise linkage disequilibrium over all populations indicated that 90·8% of the pairs of loci showed a statistical association after Bonferroni correction (N = 65, adjusted alpha = 0·00076). The proportion of pairs of loci in linkage disequilibrium in each population ranged from 20% in Asientos to 100% in Salamanca.

Population differentiation

Hierarchical amova indicated that 40·6% of the total variance was distributed among the populations, compared with 59·4% within populations (FST values ranging from 0·26 to 0·40, all significantly different from 0). PCoA analyses separated the clustered Mexican populations from all previously characterized populations (Fig. 3).

Figure 3.

 Principal component analysis of a matrix of chord distance among multilocus minisatellite haplotypes of Leptosphaeria maculans. Symbol code as follows: triangles, EU; star, Western Australia; dots, Georgia (USA); cross, Ontario (Canada); square, Chile; diamonds, Saskatchewan and Alberta (western Canada); hyphens, Manitoba (western Canada); pale crosses, Mexican (Mx) locations investigated here. Data for all locations except Mexico are from Dilmaghani et al. (2012).

Clonal lineages obtained from different hosts in the same location

Individuals sharing the same MLG had the same mating type and belonged to the same race. In addition, a series of identical genotypes could be obtained from different host plants in the same location and the same putative clonal lineage was present on the two B. oleracea morphotypes grown in Asientos or on the four B. oleracea morphotypes grown in Ojocaliente (Table 7).

Discussion

Following extensive analyses of races of L. maculans present on oilseed rape (Balesdent et al., 2006; Stachowiak et al., 2006; Dilmaghani et al., 2009; Kutcher et al., 2010) and population structure (Gout et al., 2006a; Dilmaghani et al., 2012), this study is the first to investigate populations of L. maculans occurring on B. oleracea in a part of the world where oilseed rape is a very minor crop. The data obtained here indicated some very unusual features of the populations investigated, compared with what is known for populations attacking oilseed rape elsewhere: (i) the Mexican populations are drastically separated from all other populations in terms of structure; (ii) they show an unusually low diversity and complexity of races; and (iii) sexual recombination seems to be dispensable in their pathogenic cycle.

While L. maculans infections were common on B. oleracea at the end of the 19th century and in the first half of the 20th century, they were only recently reported in Mexico. Broccoli and cauliflower have been grown in central states of Mexico since 1970 (Moreno-Rico et al., 2001), but the first symptoms of ‘blackleg’ on broccoli were only observed in 1988 in Zacatecas and Aguascalientes, and subsequently expanded to other B. oleracea crops to become a limiting factor in cauliflower production (Moreno-Rico et al., 2001). This would suggest that the fungus was recently introduced in Mexico from existing population(s) in other parts of the world, and that population genetic tools could help to trace the origin of the founder population. However, race and genotype structure indicate the extreme isolation of Mexican populations from all currently known populations obtained from B. napus. Seed transmission of L. maculans to initiate disease in B. oleracea has been reported multiple times and one of the main control recommendations is the use of certified disease-free or treated seeds (Wilcox, 1913; Henderson, 1918; Sherf & MacNab, 1986; Babadoost, 1989). Accordingly, L. maculans is suggested to have been introduced in Mexico via contaminated seeds produced in the USA (California and Washington states) (Moreno-Rico et al., 2001). Unfortunately, there is currently no information on populations of L. maculans in these American states to help evaluate how different they may be from other populations of the American continent (Dilmaghani et al., 2012). One may hypothesize that specificity towards B. oleracea has led to a host-mediated isolation of B. oleracea populations versus oilseed rape populations. However, this hypothesis is unlikely because, in controlled conditions, cross-infectivity was shown to be possible and common: all B. oleracea genotypes from the core collection analysed here were readily attacked (and usually showed very high levels of susceptibility) by three isolates obtained from B. napus in Europe, and all Mexican isolates attacking B. oleracea in the field, even though often harbouring an unusually high number of avirulence alleles, were able to attack an oilseed rape genotype without the corresponding resistance genes (such as those grown in Canada). Sampling of more isolates from B. oleracea crops from other parts of the world would be needed to evaluate further whether population isolation is a host-mediated factor or the result of other causes.

The Mexican populations showed an unusually low race richness and diversity, and maintenance of rare avirulence alleles. The small number of virulence alleles harboured and the maintenance of avirulence alleles such as AvrLm2 are reminiscent of the situation recently described for some Canadian populations of L. maculans attacking B. napus (Dilmaghani et al., 2009), in which these features were attributed to the long-time use of highly susceptible oilseed rape genotypes harbouring no major resistance genes, whereas in other parts of the world the use of such genes sequentially eradicated most of the avirulence alleles matching them (Rouxel et al., 2003a). The high susceptibility of B. oleracea to L. maculans (Ferreira et al., 1993; this study) is consistent with the hypothesis that Mexican populations were only submitted to limited R-gene selection, and thus maintained many avirulence alleles that were preserved in populations because of their probable role as effectors involved in pathogenicity (Rouxel et al., 2011). The extremely low race diversity is also relevant if the populations have essentially asexual multiplication, with no meiotic recombination of characters. The fact that the main (and unique) race obtained in 2005 in Ojocaliente is still the main (and unique) race 3 years later in two locations in Aguascalientes and Guanajuato also indicates a lack of selection pressure.

The epidemiology of L. maculans on oilseed rape is well documented and indicates the importance of ascospores in the epidemics (West et al., 2001). Ascospores have major advantages over conidia in initiating and disseminating the disease: (i) they are much more efficient at germinating, penetrating the plant tissues and expressing primary symptoms on leaves (up to 2 days before conidia; Li et al., 2004); (ii) their infectivity is much higher: whereas only a few ascospores are enough to infect unwounded leaves, conidia are unlikely to infect unwounded leaves unless applied at very high concentrations (West et al., 2001); ascospore infection, however, has been shown to promote conidia infectivity and to render them more likely to play a role in disease epidemiology, at least under Australian conditions (Li et al., 2006); (iii) ascospores are known to be dispersed by wind over several kilometres (up to 8–10 km) (West et al., 2001); and (iv) sexual reproduction is instrumental in the generation of variation at avirulence loci to rapidly adapt the fungus to new resistance sources (Rouxel et al., 2011). By contrast, conidia are small hyaline spores produced in a mucilage, and can only be dispersed by rain-splash within a few centimetres (Travadon et al., 2007), although one study reported trapping of wind-dispersed conidia up to 45 m from the inoculation site in Canada (Guo & Fernando, 2005). In accordance with the importance of sexual reproduction in the life and pathogenic cycle, most of the L. maculans populations obtained from oilseed rape analysed to date showed a high genotypic diversity and low clonal fraction indicative of regular sexual mating (Gout et al., 2006a; Dilmaghani et al., 2012). The present study showed a very high clonal fraction and low gene diversity in Mexican populations and identified a limited and discrete number of clonal lineages, seemingly specific to each location but unrelated to the host genotype on which they were trapped. These data are strongly indicative of a lack of sexual reproduction and of ascospore-mediated initiation of epidemics, consistent with the lack of leaf spots observed on the susceptible oilseed rape cv. Westar plants grown alongside B. oleracea fields and the reported rarity of the sexual stage and its low incidence in the disease cycle on cabbage (Babadoost, 1989). The reason why the fungus does not undergo sexual crossing under Mexican conditions could be linked to loss of one of the mating-type alleles following the introduction of a limited number of isolates. This phenomenon was observed for Phytophthora infestans, the potato blight pathogen, when it moved out of its native region, Mexico; this resulted in essentially asexual populations in North America and Europe for more than 100 years (Smart & Fry, 2001). However, both L. maculans mating types were present in all Mexican populations and were balanced if clone-corrected data sets were considered. Another possibility is the inactivation of one of the mating-type idiomorphs by mutation, but this is extremely unlikely, because Mexican isolates of both mating types can readily mate in vitro whenever crossed with other L. maculans isolates, including isolates obtained from oilseed rape. In addition, Moreno-Rico et al. (2005a) reported that pseudothecia could develop on stubble with symptoms left under natural conditions for 5–6 months. Because B. oleracea crops are only grown in the field for 3 months and leftover residues are incorporated into the soil when crop rotation occurs, it is likely that cultural practices are responsible for preventing differentiation and maturation of pseudothecia. Interestingly, short cultivation times of oilseed rape were also associated with low development of pseudothecia and high clonal fractions in three western Canada populations showing clonal fractions of up to 67% (one field in Manitoba) (Dilmaghani et al., 2012). The L. maculans life cycle is unusual in that leaf symptom expression is followed by a very long symptomless period of growth within plant tissues until the fungus reaches the stem base, the most likely place to meet mating partners (Rouxel & Balesdent, 2005). This stage of the plant colonization can last many months in cropping conditions where oilseed rape is grown in the field for up to 9 months (Australia) or 11 months (Europe). In Mexican and some Canadian conditions, where plants are only grown for 3 months in the field, and in the case of Mexico where only a few MLGs are found in each population, it is probable that systemic colonization will only be partial and not conducive to the presence of numerous mating partners at the stem lesion stage. While sexual reproduction does not seem to be a part of the epidemiology of the disease in Mexican populations, it remains to be clarified how conidia can disseminate genotypes so widely in the field while rain-splash only ensures small-scale dissemination (Travadon et al., 2007). The most probable explanation lies in the very different modes of culture between oilseed rape and B. oleracea. Oilseed rape is sown directly in the field, while B. oleracea is first grown in greenhouses or seedbeds for a month before being transplanted to the field. One of the main recommendations of extension bodies to farmers is to carefully monitor seedbeds before transplanting (Sherf & MacNab, 1986; Babadoost, 1989; FAO, 2000). In infected seedbeds, symptoms are described as generally appearing 2–3 weeks before transplanting time. Symptoms on cotyledons usually cause seedlings to fall over. These losses remain unnoticed as they may be confused with damping-off diseases caused by other fungi. However, diseased plantlets will promote abundant production of conidia on prematurely killed seedlings, resulting in many secondary infections in the seedbeds (favoured by the artificial rain-splash effect of sprinkler irrigation), promoting the dissemination of a limited number of genotypes on many plants, eventually throughout a field (FAO, 2000). Similarly, insufficient rotations will allow survival of the fungus in the soil, with the possibility of soilborne conidia initiating disease (Li et al., 2007) and disseminating a single genotype widely in seedbeds or in the field. For these reasons, 3- to 4-year rotations are recommended between cruciferous crops in both fields and seedbeds (Wilcox, 1913; Babadoost, 1989; FAO, 2000).

The origin of clonality in fungal or oomycete phytopathogens is unclear and probably variable from one case to the other. Moving a fungus from its native biogeographic range can result in establishment of only one mating type (see above). Other possibilities include the loss by mutation of one of the mating types in a founding and isolated population, loss of female fertility, as documented for Magnaporthe grisea (Zeigler, 1998), or eradication of an alternative host on which sexual reproduction takes place, as reported for the rust Puccinia graminis (Taylor et al., 1999). The present work suggests another possible explanation: culture conditions conducive to the large-scale dispersal of conidia over small distances, i.e. numerous individual plants available as a result of high plantlet density, along with short cultivation times which do not favour sexual mating. Alternatively, L. maculans may exemplify the rise of meiotic recombination, departing from a life cycle characterized by essentially asexual multiplication. Although the cabbage-growing industry, well established decades before oilseed rape became a major crop (Wilcox, 1913), has short cropping periods which do not favour a sexual cycle in L. maculans, the rise of oilseed rape, with its long growing season (at least in Europe and Australia) and the use of susceptible cultivars over large acreages, has provided a new epidemic cycle for the fungus in which the sexual stage has become prevalent.

Acknowledgements

A. Dilmaghani was funded by grants from INRA Département ‘Santé des Plantes et Environnement’ and INRA ‘Direction des Relations Internationales’. Part of this work was supported by EU contract FAIR3CT96-1669 (IMASCORE). Thanks are due to Peter Gladders and anonymous reviewers for excellent suggestions to improve the manuscript.

Ancillary