Molecular genetic variability of Australian isolates of five cereal rust pathogens

Mycol. Res. 107 (5): 545–556 (May 2003). f The British Mycological Society

545

DOI: 10.1017/S0953756203007809 Printed in the United Kingdom.

Molecular genetic variability of Australian isolates of five cereal rust pathogens

Felicity J. KEIPER1*, Matthew J. HAYDEN1,2, Robert F. PARK1 and Colin R. WELLINGS1# 1

Plant Breeding Institute, University of Sydney, Cobbitty, PMB 11, Camden NSW 2570, Australia. Value Added Wheat CRC Ltd, Locked Bag No. 1345, North Ryde NSW 1670, Australia. E-mail : [email protected] 2

Received 7 August 2002; accepted 29 March 2003.

Rust fungi cause economically important diseases of cereals, and their ability to rapidly evolve new virulent races has hindered attempts to control them by genetic resistance. PCR-based molecular tools may assist in understanding the genetic structure of pathogen populations. The high multiplex DNA fingerprinting techniques, amplified fragment length polymorphisms (AFLP), selectively amplified microsatellites (SAM) and sequence-specific amplification polymorphisms (S-SAP) were assessed for their potential in investigations of the genetic relationships among isolates of the wheat rust pathogens, Puccinia graminis f. sp. tritici (Pgt), Puccinia triticina (Pt), and P. striiformis f. sp. tritici (Pst), the oat stem rust pathogen P. graminis f. sp. avenae (Pga), and a putative new P. striiformis special form tentatively designated Barley grass yellow rust (Bgyr). Marker information content, as indicated by the number of species-specific fragments, polymorphic fragments among pathotypes, percentage of polymorphic loci, and the marker index, was highest for the SAM assay, followed by the AFLP and S-SAP assays. UPGMA analysis revealed that all marker types efficiently discriminated the five different taxa and Mantel tests revealed significant correlations between the marker types. Within pathogen groups, the marker types differed in the amount of variation detected among isolates; however, the major differences were consistent and polymorphism was generally low. This was reflected by the AMOVA analysis that significantly partitioned 90 % of the genetic variation between taxa. Of the three marker types, SAMS were the most informative, and have the potential for the development of locus-specific microsatellites.

INTRODUCTION The Uredinales are obligate parasites that cause rust diseases in many plant species. The rust diseases of cereals are among the most economically important plant diseases. Australian populations of the wheat rust pathogens Puccinia graminis f. sp. tritici (Pgt) (causing stem or black rust) and Puccinia triticina (Pt) (causing leaf or brown rust) have been monitored for pathogenic variability continuously for over 70 yr, and the pathogen causing stripe rust (yellow rust), P. striiformis f. sp. tritici (Pst), since it was first detected in Australia in 1979. Pathogenic variability in P. graminis f. sp. avenae (Pga), the stem rust pathogen of oats, has also been monitored in annual surveys for many years. The development of rust resistant cereal varieties is the most economical means of sustained disease control, however the rust pathogens often evolve new virulent pathotypes on previously resistant cultivars. Ongoing surveys

* Corresponding author. # On secondment from NSW Department of Agriculture.

monitoring pathogenic variability have provided strong evidence that the structure of Australian pathogen populations is determined by periodic exotic introductions, mutation selection and somatic hydridisation (Wellings & McIntosh 1990, Park, Burdon & McIntosh 1995, Park, Burdon & Jahoor 1999). Genetic analysis of pathogen populations is fundamental to understanding the mechanisms generating genetic variation, host-pathogen coevolution, and in the management of resistance (Aradhya, Chan & Parfitt 2001). Molecular markers have been used widely to characterise fungal plant pathogen populations, in particular for the assessment of genetic diversity, phylogenetic relationships, and the characterisation of pathotypes (Majer et al. 1996, Arenal et al. 1999). One of the most powerful assays is amplified fragment length polymorphism (AFLP) analysis (Vos et al. 1995). AFLPs are detected as dominant multi-locus DNA fingerprints that are based on the selective amplification of adapter-ligated restriction fragments from total digests of genomic DNA. This assay has a high multiplex ratio and is applicable to DNA of any origin or

Genetic variability of cereal rust pathogens complexity without prior investment in sequence knowledge, primer synthesis, or characterisation of DNA probes (Vos et al. 1995). AFLPs have several advantages, including marker neutrality and the assessment of large numbers of independent loci covering the entire genome, that make them ideal for the analysis of genetic variation among fungal isolates (Majer et al. 1996). AFLPs were used to study genetic variation among Pst pathotypes collected from Australia and New Zealand (Steele et al. 2001), and to map avirulence genes in Pgt (Zambino, Kubelik & Szabo 2000), however polymorphism has generally been reported as low. Several PCR-based assays have been developed based on modifications to the AFLP technique, to exploit different sources of DNA polymorphism. Microsatellites, or simple sequence repeats (SSRs), consist of tandem repeats of short (1–6 bp) nucleotide sequences that are highly polymorphic, abundant, and randomly distributed throughout most eukaryotic genomes (Weber & May 1989). These markers are used widely in studies of genetic diversity (Jarne & Lagoda 1996), however their routine application is restricted by high development costs (Ro¨der et al. 1998). The selectively amplified microsatellites (SAM) assay is a modification of the AFLP technique that employs a 5k anchored SSR primer in combination with an AFLP primer to promote the amplification of DNA fragments containing microsatellite sequences (Hayden & Sharp 2001). This SSR profiling technique was originally developed to enable the rapid identification and development of informative SSR markers, however, like AFLP, the SAM assay has a high multiplex ratio. The use of SAM analysis as a dominant multi-locus marker system has not been reported previously. Insertion of transposable elements constitutes a major source of spontaneous mutation in a variety of organisms (Kinsey & Helber 1989). Retrotransposons are the most common class of transposable elements in eukaryotes and are subdivided into two classes (Waugh et al. 1997). Class I retrotransposons move via an RNA intermediate and are further subdivided into elements that comprise long terminal repeats (LTRs), of which Ty1-copia elements are the major group, and those without long terminal repeats, such as the LINE-like group. Class II retrotransposons are subdivided into elements with short inverted terminal repeats (ITRs) and those with ITRs of variable length (Daboussi & Langin 1994). Retrotransposons and related sequences identified in fungal genomes reflect the entire spectrum of eukaryotic transposable elements (Daboussi 1996). A modification of the AFLP assay, known as sequence-specific amplification polymorphism (S-SAP, Waugh et al. 1997), is a high multiplex-ratio marker system that produces fragments consisting of retrotransposon-like sequences that exhibit widespread genome coverage (Waugh et al. 1997). The horizontal transfer and proliferation of transposable elements has been implicated in the genetic variation and pathogenic specialisation observed in asexual fungal pathogens including Fusarium

546 oxysporum (Daboussi & Langin 1994), Magnaporthe grisea (Dobinson, Harris & Hamer 1993), and Cladosporium fulvum (McHale et al. 1992). In the present work, the AFLP technique and two modifications of this procedure, SAM analysis and the S-SAP assay, were used to investigate genetic variation among collections of pathotypes representing the wheat rust pathogens Pgt, Pt, Pst, the oat stem rust pathogen Pga, and a yet to be classified pathogen causing a disease known as Barley grass yellow rust (Bgyr). Bgyr was first detected in Australia in 1998 and exhibited pathogenicity not only on naturalised weedy barley grass species (Hordeum glaucum and H. leporinum) but also on some genotypes of wheat and barley (H. vulgare), indicating that it may be a new special form of P. striiformis (Wellings et al. 2000). The purpose of this study was to identify which of the three multilocus procedures provided the greatest ability to differentiate between, and most importantly, within species and special forms of Puccinia, with a view to perform larger studies to address evolutionary and diagnostic questions.

MATERIALS AND METHODS Isolates Isolates of 47 rust pathogens were chosen from collections maintained at the Plant Breeding Institute Cobbitty, University of Sydney (Table 1). Isolates representing the wheat rust pathogens were selected based on variation in pathogenicity. Pathotypes were determined through assessment of virulence patterns on sets of differential cultivars (McIntosh et al. 1995). Given the high level of pathogenic variation within Pga in Australia (data not presented), isolates for this study were selected to represent the geographical range of the disease in Australia. The set of Bgyr isolates tested were collected throughout the south-eastern states of Australia.

DNA extraction and template preparation DNA was extracted from 50 mg of urediniospores as described by Brake, Irwin & Park (2001). For the AFLP assay, 100 ng of genomic DNA was incubated for three hours at 37 xC, then at room temperature (22 x) for three hours with 5 U each of PstI and MseI in 60 ml of 1r New England Biotechnologies (NEB) buffer II containing 100 ng mlx1 BSA, 1 mM dATP, 1 U T4 ligase, and 5 and 50 pmol of PstI and MseI adapters (Table 2), respectively. DNA templates for the SAM and S-SAP assays were prepared in an identical manner, except that the PstI restriction enzyme and adapter were not included in the reaction mixtures. For all three assays, 12 primer combinations were tested in the selective amplification, and all PCR amplifications were performed using a Corbett PC-960C thermocycler.

F. J. Keiper and others

547

Table 1. List of Puccinia isolates studied indicating species, accession number, pathotype, site, and year of collection. Pga, P. graminis f. sp. avenae; Pgt, P. graminis f. sp. tritici; Pt, P. triticina; Pst, P. striiformis f. sp. tritici; Bgyr, P. striiformis ‘Barley grass yellow rust ’. Rust

Accession no.

Pathotypea

Site of collection

Year

Pga

1 2 3 4 5 6 7 8 9 10

992579 992678 992655 992664 992629 992630 992584 992573 992585 992728

94 94 94 94 41 30 94 94 94 94

Condobolin NSW Mendooran NSW Young NSW Yeoval NSW Stawell VIC Winchelsea VIC Moona WA Mindarabin WA Moona WA Roseworthy SA

1999 1999 1999 1999 1999 1999 1999 1999 1999 1999

Pgt

1 2 3 4 5 6 7 8 9 10

840552 540129 690924 691042 334 781219 730879 751188 690822 730560

34-2,12,13 21-0 34-2 194-1,2,3,5,6 126-5,6,7,11 98-1,2,3,5,6 343-1,2,3,5,6 21-1,2,10 326-1,2,3,5,6 34-2,11

Kingaroy QLD Southern NSW Wallbundrie NSW Tichbourne NSW ?b Southern NSW Kerang VIC Ceduna SA Clinton SA Cobar NSW

1984 1954 1969 1969 ?b 1978 1973 1975 1969 1973

Pt

1 710208 2 810043 3 720469 4 890163 5 670028 6 760583 7 840045 8 200347 9 630550 10 900053

64-1,3 53-1,(6),(7),10,11 10-1,2,3,4 76-1,3,10,12 26-1,3 76-2,3,6,(7) 104-2,3,(6),(7),11 104-1,2,3,(6),(7),11,13 135-2 64-(6),(7),(10),11

Castle Hill NSW Palmerston North NZ Narrabri NSW Manawatu NZ Takapara NZ Narrabri NSW Mt Derimut VIC Urania SA Ravensworth NSW Narrabri NSW

1971 1981 1972 1989 1967 1976 1984 2000 1967 1991

Pst

1 2 3 4 5 6 7 8 9

861725 821559 821552 881510 821589 861773 831917 881584 911582

110 E143 A+ 104 E137 Ax 104 E137 A+ 104 E9 Ax 106 E139 Ax 109 E141 Ax 108 E141 A+ 104 E137 Ax, Yr17+ 106 E139 Ax, YrSk+

Richmond TAS Grafton NSW Forbes NSW Trundle NSW Springston NZ Christchurch NZ Rutherglen VIC Cobbitty NSW Lincoln NZ

1986 1982 1982 1988 1982 1986 1983 1988 1991

Bgyr

1 2 3 4 5 6 7 8

991660 991717 002110 981570 002135 991527 991551 991539

Bgyr Bgyr Bgyr Bgyr Bgyr Bgyr Bgyr Bgyr

Penshurst VIC Blackwell TAS Horsham VIC Darlington Point NSW Jugiong NSW Moree NSW Horsham VIC Temora NSW

1999 1999 2000 1998 2000 1999 1999 1999

a Pathotype nomenclatures for Pgt, Pt and Pst were described by McIntosh, Wellings & Park (1995), and for Pga by Stewart & Roberts (1970). b From W. L. Waterhouse, 1926.

AFLP analysis AFLP procedures were performed essentially as described by Vos et al. (1995), except that the methylation sensitive restriction enzyme PstI replaced EcoRI because it is known to cleave low-copy DNA preferentially (Burr et al. 1988). Preamplification reactions were performed in 20 ml reaction mixtures containing 5 pmol each of PstI+0 and MseI+0 primers (Table 2), 0.2 mM dNTP, 1rPCR buffer (Advanced Biotechnologies Integrated Sciences, Sydney), 1.5 mM MgCl2, 1 U Taq

DNA polymerase (Advanced Biotechnologies), and 2 ml of digested/ligated PstI-MseI template DNA. PCR was performed for 25 cycles of denaturation at 92 x, annealing at 56 x and extension at 72 x for 60 s each. Preamplification products were diluted 1 :20 with TE buffer (1 mM EDTA, 10 mM Tris-HCl, pH 8). Selective amplifications were performed in a similar reaction mix using 5 pmol each of PstI+1 and MseI+3 primers (Table 2) and 2 ml of diluted preamplified template. PCR was performed with 35 cycles of the following profile : denaturation at 92 x, annealing

Genetic variability of cereal rust pathogens Table 2. Sequences (5k-3k) of adapters and primers used in the AFLP, SAM and S-SAP assays. Adapters PstI MseI

CTCGTAGACTGCGTACATGCA CATCTGACGCATGT GACGATGAGTCCTGAG TACTCAGGACTCAT

Preamplification primers PstI+0 AGACTGCGTACATGCAG MseI+0 GATGAGTCCTGAGTAA Selective amplification primers PstI+1 AGACTGCGTACATGCAG+A/C/G/T MseI+3 GATGAGTCCTGAGTAA+CAA GATGAGTCCTGAGTAA+CCA GATGAGTCCTGAGTAA+CCT GATGAGTCCTGAGTAA+GAA GATGAGTCCTGAGTAA+GAG GATGAGTCCTGAGTAA+GAT GATGAGTCCTGAGTAA+GCA GATGAGTCCTGAGTAA+GCC GATGAGTCCTGAGTAA+GCG GATGAGTCCTGAGTAA+GCT GATGAGTCCTGAGTAA+GGA GATGAGTCCTGAGTAA+GGC GATGAGTCCTGAGTAA+GGT GATGAGTCCTGAGTAA+GTA GATGAGTCCTGAGTAA+GTC GATGAGTCCTGAGTAA+GTG GATGAGTCCTGAGTAA+GTT Bare-1a CTAGGGCATAATTCCAACA PCT6b KKVRVRVCTCTCTCTCTCT KKHVHVHTGTGTGTGTGTG PGT6b a

Waugh et al. (1997). K=G or T ; V=A, C or G ; R=A or G; H=A, C or T; each primer consisted of equimolar amounts of each oligomer. b

(see below), and extension at 72 x for 60 s each over 10 cycles. The annealing temperature for the first cycle was 65 x, reducing by 1 x/cycle. For the next 25 cycles, the denaturation and annealing steps were reduced to 30 s with annealing at 56 x. SAM analysis Preamplification reactions were performed in 25 ml reaction mixtures containing 20 pmol 5k-anchored SSR primer mix (Table 2), 0.2 mM dNTP, 1rPCR buffer, 1.5 mM MgCl2, 5 pmol MseI+0 primer, and 2 ml of digested/ligated template DNA. Hot-start PCR was performed by heating the reactions for 5 min at 94 x, followed by the addition of 1 U Taq DNA polymerase, and 25 cycles of denaturation at 92 x, annealing at 55 x and extension at 72 x for 60 s each. Preamplification products were diluted 1 :20 with TE buffer. SAM amplification was performed essentially as described by Hayden & Sharp (2001). Reaction mixtures contained 5 pmol MseI+3 primer, 20 pmol 5k-anchored SSR primer mix, 0.2 mM dNTP, 1rPCR buffer, 1.5 mM MgCl2, 1 U Taq DNA polymerase, and 2 ml of diluted preamplification product in a total volume of 20 ml. PCR was performed for 33 cycles with the following profile : 60 s denaturation at 92 x, 60 s

548 annealing (see below) and 30 s extension at 72 x. The annealing temperature for the first cycle was 65 x, reducing by 1 x/cycle for seven consecutive cycles, followed by five cycles at 57 x. The denaturation and annealing steps were reduced to 30 s with annealing at 55 x for the next 20 cycles.

S-SAP analysis Preamplification reactions were performed in 25 ml reaction mixtures containing 5 pmol each of MseI+0 and Bare-1 primers (Table 2), 0.2 mM dNTP, 1rPCR buffer, 1.5 mM MgCl2, and 2 ml of digested/ligated template DNA. Hot-start PCR amplification and product dilution was performed as described for SAM analysis. Selective S-SAP amplification was performed as described by Waugh et al. (1997). Reaction mixtures contained 5 pmol each of MseI+3 and Bare-1 primers, 0.2 mM dNTP, 1rPCR buffer, 1.5 mM MgCl2, 1 U Taq DNA polymerase, and 2 ml of diluted preamplified DNA template in a total volume of 20 ml. The selective amplification PCR profile used was that described for AFLP analysis. Marker scoring Following selective amplification, an equal volume of gel loading buffer (98 % formamide, 10 mM EDTA, 0.25 % xylene cyanol) was added to each reaction. The PCR products were denatured by heating at 94 x for 3 min, chilled on ice, and separated on a 6 % sequencing gel under standard conditions (Sambrook, Fritsch & Maniatis 1989). PCR products were visualised by silver staining. For all three marker types, each amplified fragment was treated as a separate character and scored as either present (1) or absent (0) across the 47 pathogen isolates. All 47 isolates were scored manually on a single gel for each primer combination tested. Fragments that were poorly resolved, or of low intensity, were not scored. The size range scored for all three marker types was 80–500 bp.

Statistical analyses For each of the three marker types, a marker index was calculated as described by Powell et al. (1996). Percentage polymorphic loci were calculated for each species and marker type using the presence/absence data matrix and the program Tools for Population Genetic Analysis (TFPGA, Miller 1997a). The percentage of polymorphic loci was based on the percentage of loci not fixed for one allele. Similarity matrices representing all pairwise comparisons of rust isolates were constructed using the presence/absence data for each marker type and Jaccard’s similarity coefficient : GS(ij)=a/(a+b+c)

F. J. Keiper and others (Jaccard 1908), where GS(ij) is the measure of genetic similarity between individuals i and j, a is the number of polymorphic fragments that are shared by i and j, b is the number of fragments present in i but absent in j, and c is the number of fragments present in j but absent in i. The Jaccard similarity matrices were used to perform cluster analyses using the unweighted pair-group method using arithmetic means (UPGMA) procedure. Support for the clusters was evaluated using bootstrapping analyses with 1000 iterations (Felsenstein 1985). The goodness of fit of the UPGMA clustering analysis with the original Jaccard’s similarity coefficient matrix was determined through calculation of the cophenetic correlation values. The correlations between the similarity matrices generated for each of the three marker types were determined using the Mantel test (Mantel 1967), with statistical significance determined by 1000 random permutations for each test. Construction of the Jaccard matrices, cluster analyses, cophenetic values and Mantel tests were performed using the program NTSYSpc version 2.1 (Exeter Software), and the bootstrapping analyses were performed using WinBoot (Yap & Nelson 1996). Analysis of Molecular Variance (AMOVA, Excoffier 1995) was used to assess genetic variation. A pairwise Euclidean distance matrix was constructed using the presence/absence data matrices and the program AMOVA-PREP (Miller 1997b) in preparation for WINAMOVA. AMOVA performs a classical analysis of variance on the distance matrix, and in this study, partitioned the genetic variation between two levels: among species/special forms, and among isolates within taxa. The significance of the AMOVA variance components and fixation indices were tested using nonparametric permutation procedures (Excoffier, Smouse & Quattro 1992). The fixation index, FST, was calculated by the AMOVA analysis and provided a measure of genetic differentiation of groups. Values of FST greater than 0.25 indicate significant genetic differentiation (Hartl & Clarke 1997).

RESULTS Optimisation of marker assay conditions Preliminary investigations were undertaken to determine the optimal assay conditions required to generate reproducible and scorable DNA fingerprints for each type of marker system. For the AFLP assay, optimal fingerprints were achieved using four selective nucleotides on the adapter primers during selective amplification (PstI+1, MseI+3). Optimal SAM fingerprints were generated using MseI+3 adapter primers in combination with a 5k-anchored SSR primer targeting (CT)n or (GT)n dinucleotide repeats. The S-SAP assay also required MseI+3 adapter primers in combination with the Bare-1 primer for optimal DNA fingerprints. In preliminary S-SAP experiments, primers targeting different types of retrotransposon elements were tested,

549 Table 3. Comparison of the information revealed by AFLP, SAM and S-SAP markers using 12 primer combinations across 47 cereal rust isolates representing five Puccinia species and special forms. Pga, P. graminis f. sp. avenae; Pgt, P. graminis f. sp. tritici; Pt, P. triticina; Pst, P. striiformis f. sp. tritici; Bgyr, P. striiformis ‘Barley grass yellow rust’. AFLP Total band number Mean (¡S.D.) band number per primer combination Species specific bands Bands polymorphic within species Pga Pgt Pt Pst Bgyr Marker indexa Percentage polymorphic loci Pga Pgt Pt Pst Bgyr a

SAM

S-SAP

1484 953 827 120.33¡24.80 79.42¡14.08 68.92¡10.86

972 (67.3%) 226 (15.6%)

788 (82.7%) 384 (40.3%)

555 (67.1%) 199 (24.1%)

67 62 23 10 61 25.32

128 99 95 27 197 39.97

43 68 14 3 86 21.30

6.67 5.05 0.94 0.74 6.33

10.81 9.44 6.30 2.83 17.10

5.44 8.71 2.66 0.36 14.99

Powell et al. (1996).

including Class I (Ty1-copia group and SINE) and Class II (Stowaway and Barfly) retrotransposons (Waugh et al. 1997, Chang, O’Donoughue & Bureau 2001). All retrotransposon primers tested generated scorable and reproducible DNA profiles. However, Ty1-copia-like retrotransposons amplified by the Bare1 primer (Waugh et al. 1997) revealed the highest frequency of polymorphic fragments and was used in subsequent experiments.

Marker polymorphism All three marker assays generated highly informative and reproducible fingerprints. Using only 12 primer combinations in each type of marker analysis, a total of 1484 AFLP fragments, 953 SAM fragments, and 827 S-SAP fragments were amplified across the 47 pathogen isolates (Table 3). For every primer combination in each of the three marker types, no fragments were monomorphic across all five taxa, i.e. each scored fragment was absent in at least one of the species/special forms. For all three marker types, a large proportion of the total number of fragments amplified were species/ special form-specific. The SAM assay revealed the greatest number of species-specific fragments (82.7 %), and the AFLP and S-SAP assays revealed similar proportions (y67 %). The SAM assay also revealed the greatest variation among isolates of the same species/special form with 40.3 % of scored fragments polymorphic within taxa (Table 3), but the least number of fragments per primer combination (mean¡S.D. 68.92¡10.86, Table 3). In contrast, the

Genetic variability of cereal rust pathogens

550

(A)

100%

100%

57.3%

100% 99.9% 100%

100%

56.4% 100% 70.5% 97.2% 82.3%

100% 99.8%

100%

100% 100%

0

0.15 0.50

0.75

100%

(B)

43.1%

56.4%

63.1%

99.3%

98.4% 100% 89.9%

100% 60.4% 100% 67.4% 55.7% 97.1% 100%

100%

100%

0

0.35

0.50

0.75

100%

75.6% 100%

69.3%

84.1%

0

0.15 0.50

0.75

Fig. 1. For caption, see facing page.

Pgt

Pt

Pst

Bgyr

1 2 10 6 9 7 8 3 4 5 5 1 8 2 3 6 9 7 4 10 1 3 4 8 9 2 10 5 6 7 3 1 2 4 5 7 6 8 9 1 2 3 4 5 6 7 8

Pga

Pgt

Pt

Pst

Bgyr

1.00

(C)

100%

Pga

1.00

100%

99.9%

1 7 8 9 10 6 4 2 3 5 1 3 8 9 7 2 4 6 10 5 1 2 3 4 6 5 9 7 8 10 1 2 3 9 5 4 6 7 8 2 3 5 1 4 6 8 7

1 2 7 8 4 10 99.7% 6 99.9% 9 3 5 1 74.6% 8 2 3 100% 6 7 4 9 46.7% 10 5 1 2 3 4 43.3% 5 6 100% 2 8 9 57% 10 1 4 77.1% 5 6 100% 78 9 43.6% 23 90.9% 12 90.8% 3 4 100% 5 100% 6 7 8

1.00

Pga

Pgt

Pt

Pst

Bgyr

F. J. Keiper and others AFLP assay produced the greatest band number per primer combination (mean¡S.D. 120.33¡24.80, Table 3) but the lowest within species/special forms polymorphism with only 15.6 % of the scored fragments being informative (Table 3), with the S-SAP results intermediate. Within the SAM assay, the number of fragments produced by the two different microsatellite repeats used per primer combination were similar (CT mean¡S.D. 81.4¡18.81 ; GT mean¡S.D. 78.00¡11.05). The highest marker index value of 39.97 was obtained for the SAM assay, and the lowest for the S-SAP assay at 21.30 (Table 3). The estimates of percentage polymorphic loci (P) revealed different levels of genetic variation among isolates within species/special forms for each of the three multi-locus marker types (Table 3). The highest P values were obtained for the SAM markers, with the greatest polymorphism among isolates detected for the Bgyr group (P=17.10). This group was also the most polymorphic for the S-SAP assay (P=14.99), whereas Pga isolates exhibited the greatest variation with AFLPs (P=6.67). For all three marker types, the least genetic polymorphism was exhibited amongst the Pst isolates.

Genetic diversity Cluster analyses for each marker type revealed five distinct clusters representing each species/special form (Fig. 1). All analyses revealed differentiation of the five taxa, with the exception of the stripe rust pathogens (Pst and Bgyr), at y10% (or less) genetic similarity. With AFLPs, Pst and Bgyr exhibited 80% genetic similarity, and within Bgyr, there was a distinct division into two groups at 86% genetic similarity (Fig. 1A). The two groups consisted of isolates 1–5 (see Table 1), all within 98% genetic similarity, and isolates 6–8 all within 99% similarity. Isolates with identical AFLP genotypes were revealed in Pgt (isolates 3, 8 and 9), Pt (isolates 3 and 4), and Pst (isolates 4–6). SAMs and S-SAPs separated Pst and Bgyr differently, with Bgyr isolates 6–8 grouped more closely to the Pst isolates than Bgyr isolates 1–5. With SAMs (Fig. 1B), the five Bgyr isolates and all Pst isolates were separated from Bgyr isolates 6–8 at 37 % genetic similarity, and with S-SAPs (Fig. 1C), this division occurred at 67 % genetic similarity. Among the stripe rust pathogens, Pst and the five Bgyr isolates exhibited up to 72 % genetic similarity with SAMs, and 77% with S-SAPs. Identical SAM genotypes were detected for Bgyr isolates 7 and 8 only. With S-SAPs, Pga isolates 1 and 2 were genetically

551 Table 4. Mantel test results for correlation (r) between similarity matrices constructed from AFLP, SAM and S-SAP markers within and among isolates representing five Puccinia species and special forms, and the probability of statistical significance based on 1000 random permutations. Pga, P. graminis f. sp. avenae; Pgt, P. graminis f. sp. tritici; Pt, P. triticina; Pst, P. striiformis f. sp. tritici; Bgyr, P. striiformis ‘Barley grass yellow rust’. Comparison species

AFLP/SAM r

P

AFLP/S-SAP r

SAM/S-SAP

P

r

P

Pga Pgt Pt Pst Bgyr

0.809 x0.121 0.114 0.588 0.984

0.018 0.544 0.325 0.010 0.018

0.939 x0.014 0.248 0.585 0.988

0.004 0.364 0.175 0.037 0.008

0.812 0.932 0.154 0.719 0.983

0.020 0.001 0.211 0.012 0.007

Total

0.977

0.001

0.995

0.001

0.989

0.001

identical, as were Pst isolates 1 and 4–9. The bootstrapping analysis generated high confidence values for the major nodes differentiating taxa and the major groups among isolates within species/special forms of all three dendrograms. Bootstrapping values among the isolates comprising the smaller clusters within taxa were generally less than 50 %. Comparison of the dendrograms constructed from the original Jaccard similarity coefficient and the UPGMA cluster analysis for each of the three marker types indicated very high goodness of fit with high co-phenetic correlation values : AFLP r=0.999 ; SAM r=0.999 ; S-SAP r=0.998. Mantel tests for all pairwise comparisons of the complete similarity matrices for each of the three marker types revealed significant correlations (AFLP/SAM r=0.977 ; AFLP/S-SAP r=0.995 ; SAM/S-SAP r=0.989 ; P<0.01) (Table 4). At the species/special form level, significant correlations were detected between the three marker types for Pga and the stripe rust pathogens. For Pgt, only the SAM/ S-SAP comparison was significant (P<0.01), and no significant correlations were detected between the three marker types for Pt. The distribution of genetic variation detected by each of the three marker types was tested statistically by AMOVA (Table 5). For all marker types, more than 90% of the detected genetic variation was partitioned between the different pathogen species/special forms, with less than 10 % distributed among the isolates within each taxon. Significance tests of variance components indicated that the among-taxon classification was significantly better (P<0.05) than random classification for all three marker types. High FST values for all three marker

Fig. 1. UPGMA dendrograms depicting patterns of genetic similarity among Australian isolates of five cereal rust pathogens estimated by AFLPs (A), SAMs (B), and S-SAPs (C). Pga, Puccinia graminis f. sp. avenae ; Pgt, P. graminis f. sp. tritici ; Pt, P. triticina ; Pst, P. striiformis f. sp. tritici ; Bgyr, P. striiformis ‘Barley grass yellow rust’. Numbers given for each species/special form represent isolates listed in Table 1. Values given on dendrogram branches are percent of bootstrap data sets exhibiting the cluster.

Genetic variability of cereal rust pathogens

552

Table 5. AMOVA for 47 cereal rust isolates representing five Puccinia species and special forms using AFLP, SAM and S-SAP markers. Statistics include the degrees of freedom (D.F.), sums of squared deviations (SSDs), mean squared deviations (MSDs), variance component estimates, the percentages of total variance ( % total) contributed by each component, the probability (P) of obtaining a more extreme component estimate by chance alone (estimated from 1000 sampling realisations), and FST values.

Source of variation

D.F.

SSD

MSD

Variance compo- % nent total P

FST

AFLP Species 4 Isolates 42 Total 46

12345.08 3086.27 327.94 388.53 9.25 9.25 12733.62

97.26 <0.001 0.973 2.74

SAM Species 4 Isolates 42 Total 46

6312.29 1578.07 166.33 729.03 17.36 17.36 7041.32

90.55 <0.001 0.906 9.45

S-SAP Species 4 Isolates 42 Total 46

6438.17 1609.54 170.63 357.83 8.52 8.52 6795.00

95.24 <0.001 0.952 4.76

types (>0.9) indicated very high genetic differentiation among taxa. DISCUSSION Marker polymorphism The utility of three multi-locus DNA fingerprinting methods in discriminating among a group of diverse cereal rust isolates representing five species and special forms was assessed. All three assays detected polymorphisms among isolates within each taxon, however they differed in the amount of variability detected (Table 3). For each marker system, the polymorphism detected was the result of the presence or absence of a restriction site. In addition, the two AFLP modifications, SAM and S-SAP, detected polymorphism associated with variation in microsatellite repeat length, and the insertion or deletion of Ty1 copia-like retrotransposon elements, respectively, adjacent to restriction sites. The AFLP assay produced the greatest total number of polymorphic fragments, and in terms of polymorphism per primer combination, the most informative fingerprints. However, the SAM assay provided the greatest discrimination within taxa due to the detection of a greater number of polymorphic fragments within each group, and the highest marker index. The SAM assay also detected the largest proportion of speciesspecific fragments. The amount of diversity revealed by the S-SAP assay was intermediate of that revealed by the AFLP and SAM assays. The level of polymorphism revealed by a marker system is dependant on the number of markers used, the degree of genome coverage, and the type of DNA sequence variation being assayed (Powell et al. 1996). All three assays revealed a large number of markers, indicating a high rate of single nucleotide substitution

or insertion/deletion mutations adjacent to restriction sites (Powell et al. 1996). The higher information content of SAM markers compared to AFLPs and S-SAPs suggested a relatively higher rate of microsatellite evolution, and concurs with reports that multi-locus SSR fingerprint analysis is more suited to studies of genetic variability than other multi-locus assays (Singh et al. 2002), especially when genetic variation is expected to be low (Witsenboer, Vogel & Michelmore 1997). The main disadvantage of multi-locus SSR fingerprinting is the prevalence of dominant markers, which hinders the identification of allelic fragments, particularly in complex fingerprints (Hayden & Sharp 2001). In the present work, putative allelic fragments were identified with each primer combination for at least one SSR locus among isolates of a species/special form (data not shown). These species-specific SAMs occurred predominantly among isolates of Pga and Pgt. In future work, these SAMs will be converted into locus-specific, co-dominant SSR markers. Retrotransposons have also been reported to have a high rate of evolution, and greater levels of polymorphism are often detected by the S-SAP assay compared to AFLP analysis (Waugh et al. 1997). Retro-elements are dispersed throughout fungal genomes (Daboussi 1996), and have been identified in Fusarium oxysporum (Daboussi, Langin & Brygoo 1992, Di Petro, Anaya & Roncero 1994) and Colletotrichum gloeosporioides (He et al. 1996), but not yet for Puccinia. The S-SAP assay results presented here, and those of the preliminary investigation, indicated that various types of retrotransposon-like elements occur at a high copy number in the Puccinia genomes examined (data not shown). Of the elements assayed, the Ty1 copia-like elements revealed the greatest level of polymorphism and proved more informative than AFLPs, based on the proportion of within-taxon polymorphic fragments (Table 3). Various studies have demonstrated different transposition activities for different types of retrotransposon elements. It is possible that Ty1-copia-like elements are relatively inactive elements in the Puccinia genome, and that a greater level of polymorphism may be achieved when transposon elements characterised in other fungal genomes are assayed with S-SAP. For example, the Fusarium genome contains at least six different families of elements, of which the Floret and Palm elements, with characteristics of LTR-retrotransposons, and the active Fot1, Fot2, Impala, and Hop families that resemble bacterial transposons, have been characterised (Daboussi et al. 1992). A survey of these different elements in the Puccinia genome might identify retrotransposons that reveal higher levels of polymorphism. Therefore, the S-SAP technique has considerable potential for revealing greater levels of polymorphism in Puccinia. Of the three DNA marker assays, only the AFLP technique has been previously used in studies of genetic variation in a rust pathogen. In the study by Steele et al. (2001), no AFLP polymorphisms were detected among

F. J. Keiper and others 14 Pst pathotypes collected from Australia and New Zealand. In the present work, 10 polymorphic AFLPs were found among nine Pst pathotypes collected from Australia and New Zealand, of which four (isolates 1, 3, 7 and 9, see Table 1) were used in the study by Steele et al. (2001). In the present work, 12 primer combinations were used to generate 1484 AFLPs, compared to four primer combinations and 100 AFLPs in the study by Steele et al. (2001). This demonstrates the low polymorphism in this species in Australia and the need to either screen many primer combinations to detect polymorphism, or to exploit alternative sources of polymorphism to determine genetic variation. In contrast, AFLPs have detected polymorphism in other fungal pathogen species including Cladosporium fulvum and Pyrenopeziza brassicae, where marker systems such as RFLP detected very little variation (Majer et al. 1996). The two modifications to the AFLP technique tested here, SAM and S-SAP analyses, have not been reported previously in DNA fingerprinting studies of cereal rust pathogens. A major difficulty in interpreting multi-locus fingerprinting data is the inability to distinguish polymorphism arising from contaminant DNA, particularly with organisms such as the rust fungi that cannot be readily grown in axenic culture. The potential to amplify contaminant DNA was taken into consideration in scoring all multi-locus fingerprints in the present work by using a highly conservative approach. Therefore, it is possible that true polymorphisms, particularly among isolates of the same species/special forms, were overlooked. This problem may be circumvented with the use of locus-specific markers (Enkerli et al. 2001). The conversion of AFLPs to locus-specific markers can be problematic because it is difficult to capture the source of the polymorphism in the PCR amplicon (Shan, Blake & Talbert 1999, Reamon-Buttner & Jung 2000). In contrast, SAMs can easily be converted into conventional SSRs (Hayden & Sharp 2001). As markers for cereal rust pathogens, locus-specific multiallelic SSRs may have greater application in biotype/ pathotype discrimination and identification. The requirement for the preparation of preamplified DNA template in each of these assays may be considered labourious and a major disadvantage, however in this instance it was considered highly advantageous. Large-scale DNA extractions require large amounts of fungal urediniospores. For cereal rust pathogens, increasing pure cultures is a lengthy process and impractical where large numbers of isolates are required for genetic studies. Therefore, the ability to use only a small amount of DNA for all three marker systems, obtained from a single extraction from 50 mg of urediniospores, was considered highly beneficial. Pathotype variation The cluster analyses revealed that all three marker types efficiently discriminated among the five taxa

553 (Fig. 1), and the Mantel tests proved that all three similarity matrices were significantly correlated for all taxa except for Pt (Table 4). In the case of Pt, the poor correlation between marker types was probably related to the vast difference in the number of polymorphic fragments detected by each assay (e.g. 23 AFLPs, 95 SAMs, 14 S-SAPs ; Table 3). The significant separation of Pga and Pgt is consistent with their different host ranges, and the separation of Bgyr as a distinct group with AFLPs is of particular interest. Within species/special form clusters, major differences of interest were revealed (Fig. 1). The clustering patterns consistent among the three marker types were the major clusters involving the separation of Pga isolates 3 and 5, and Pgt isolates 5 and 10 from the remaining eight isolates of those groups, and Bgyr isolates 6–8 from the other five isolates of that group. The within-taxon clusters reflected the major polymorphisms found within that group. For example, the absence of an AFLP band in Pgt isolate 5 (standard race 126) and its presence in every other Pgt isolate was responsible for 43 of a total of 62 polymorphic fragments observed for that group, and the separation of this isolate from the rest of the group. Standard race 126 was first detected in Western Australia in 1925 and predominated until 1954, after which it declined in frequency and has not been detected in annual pathogenicity surveys in Australia since 1963 (Luig 1977). Due to the pathogenic distinctness of race 126, it was concluded to have been of exotic origin (Luig 1977), and this is supported by the results presented here. Pgt isolate 10 (Pathotype 34-2,11, Table 1) was detected in 1957 and is believed to be a somatic hybrid between an isolate of standard races 126 and 21 that was detected in 1954 (Luig 1977). Pathotype 34-2,11 generally clustered separately from, and between, the main group of eight isolates and race 126. Within Pga, isolate 5 differed from the rest of the group with all three marker systems, suggesting that it may have a different origin. This is of interest because the 10 isolates of Pga examined, whilst representing pathotypes 30, 41 and 94, were similar in their pathogenicity (data not shown). The high genetic similarity between Pst isolates detected in the present study is consistent with observations of Pst in Australia that imply the introduction of a single isolate in 1979, and subsequent single-step mutations at loci governing pathogenicity (Wellings & McIntosh 1990). Based on the pathogenicity of each Bgyr isolate on differential genotypes used to characterise Pst, and its recent detection in Australia, this group is believed to have arisen from a single introduction in 1998 (Wellings et al. 2000). The separation of the eight isolates into two distinct groups was therefore not expected. Additionally, the two groups do not reflect any apparent geographic trends. The genetic separation of the two groups was caused primarily by the presence of fragments in isolates 6–8, and the absence of these fragments in the other five isolates. For example, of the

Genetic variability of cereal rust pathogens 61 AFLP fragments detected for Bgyr, 46 (75 %) were present in isolates 6–8 only. Similarly, 76 % of the polymorphic S-SAPs were present in isolates 6–8 but absent in the other five isolates, however only 39% of the SAMs followed this trend. There are several possible explanations for this result, with perhaps the simplest being the introduction of genetically dissimilar Bgyr isolates. Considering the relatively high level of within group genetic variation, compared to the other four taxa, in a collection spanning only three years, this hypothesis seems highly probable. It is also possible that Bgyr and Pst have undergone somatic hybridisation on the mutual host barley grass. Support for this hypothesis may be the greater genetic similarity between Bgyr isolates 1–5 and the group of Pst isolates compared to the genetic similarity between the two groups within Bgyr revealed by the SAM (Fig. 1B) and S-SAP (Fig. 1C) cluster analyses. Alternatively, the presence of fragments in isolates 6–8 and their absence in the other five Bgyr isolates could indicate genomic instability. The absence of fragments could be evidence for the loss or movement of genomic elements. These possibilities will be tested in future studies, once sufficient isolates of Bgyr are available. In general, the clustering of isolates within taxa did not follow expectations based on the pathogenicity of the isolates or their geographical origin (see Table 1). This finding concurs with low correlations between pathogenicity and molecular variation for Pt and Pst using dsRNA (Dickinson, Wellings & Pryor 1990), Pst using RAPDs (Chen, Line & Leung 1993), and Puccinia coronata f. sp. avenae (causal agent of oat crown rust) using DNA amplification fingerprinting (Brake et al. 2001). In addition, different pathotypes have at times shown little or no molecular variation. Newton, Caten & Johnson (1985) found within species uniformity in isozyme phenotypes among P. striiformis, P. recondita and P. hordei isolates that contrasted in pathogenicity, and Steele et al. (2001) found no AFLP variation among Pst isolates. These findings were interpreted as indicating that DNA polymorphism is independent of pathogenicity, and that the genome as a whole evolves at a much faster rate than genes governing pathogenicity (Chen et al. 1993). It has been proposed that a direct relationship between pathogenicity and molecular data would be observed if pathogenicity was controlled by many genes distributed throughout the genome (Koch et al. 1991). A lack of correlation between classifications based on molecular and geographical or ecological origin has been reported previously for other phytopathogens, including 36 Epicoccum nigrum isolates collected from Europe, America and Africa assayed with five different types of PCR-based molecular markers (Arenal et al. 1999). However, major differences aside, there was not enough polymorphism among the isolates in the present work to determine with any degree of accuracy the genetic relationships within species/special forms with the three sets of markers. This was supported by low bootstrapping

554 values of generally less than 50%, for smaller isolate clusters within taxa (Fig. 1), and the AMOVA analysis that significantly partitioned 90 % of the genetic variation between taxa (Table 5). Two of the three marker types revealed identical phenotypes (putative clones) within species/special forms, however these were not the same for each marker assay (Fig. 1). The AFLP assay revealed the greatest number of putative clones (Fig. 1A), and these occurred within Pgt, Pt and Pst, followed by the S-SAP assay with identical isolates revealed within Pga and Pst (Fig. 1C). This reflects the low level of polymorphism within species/special forms detected by these two assays, compared to the SAM assay where identical isolates were absent. This finding illustrates differences in ability to detect variation between closely related individuals among different multi-locus marker types, and highlights the need for care when interpreting such data. In agreement with the present work, Arenal et al. (1999) found that five different PCRbased molecular marker types revealed different levels of similarity among isolates of Epicoccum nigrum. In addition, codominant RFLP markers were found to be poor in measuring genetic similarity among clones within populations of Septoria tritici (McDonald & Martinez 1991). The levels of similarity detected will depend on the number of markers, and the results here suggest that a very large number of markers is needed to clearly obtain an accurate indication of the genetic relationships among isolates within different taxa. In conclusion, all three multi-locus marker types presented here provided similar information in terms of differentiation among Puccinia species/special forms, and the major clusters within taxa. The markers differed primarily in the amount of variability they detected among isolates within taxa. AFLPs proved to be the best marker type for discriminating species, as well as differentiating the putative new special form tentatively designated Bgyr. The higher information content of SAMs compared to S-SAPs indicated that microsatellites undergo a higher rate of evolution than Ty1 copia-like retroelements in the Puccinia genome, and this assay will be used in future investigations of genetic relatedness among isolates and pathotypes, and for the development of locus-specific SSRs. To better characterise genetic relationships between and within pathotypes, it is clear that a greater number of isolates will need to be examined. The genetic variability present in a given pathotype will be under-represented by the use of a single isolate, especially when genetic variation is expected to occur within pathotypes. In addition, with all three marker systems presented, a greater number of loci require sampling to better discriminate among isolates of low genetic differentiation. More powerful discriminatory markers such as codominant, locus-specific SSRs may better address evolutionary and diagnostic questions in working with these species.

F. J. Keiper and others ACKNOWLEDGEMENTS The authors thank Shahidul Haque for DNA samples. Financial support provided by the Grains Research and Development Corporation of Australia is gratefully acknowledged.

REFERENCES Aradhya, M. K., Chan, H. M. & Parfitt, D. E. (2001) Genetic variability in the pistachio late blight fungus, Alternaria alternata. Mycological Research 105: 300–306. Arenal, F., Platas, G., Martin, J., Salazar, O. & Pelaez, F. (1999) Evaluation of different PCR-based DNA fingerprinting techniques for assessing the genetic variability of isolates of the fungus Epicoccum nigrum. Journal of Applied Microbiology 87: 898–906. Brake, V. M., Irwin, J. A. G. & Park, R. F. (2001) Genetic variability in Australian isolates of Puccinia coronata f. sp. avenae assessed with molecular markers and pathogenicity markers. Australian Plant Pathology 30 : 259–266. Burr, B., Burr, F. A., Thompson, K. H., Albertson, M. C. & Stuber, C. W. (1988) Gene mapping with recombinant inbreds in maize. Genetics 118: 519–526. Chang, R.-Y., O’Donoughue, L. S. & Bureau, T. E. (2001) InterMITE polymorphisms (IMP): a high throughput transposonbased genome mapping and fingerprinting approach. Theoretical and Applied Genetics 102: 773–781. Chen, X., Line, R. F. & Leung, H. (1993) Relationship between virulence variation and DNA polymorphism in Puccinia striiformis. Phytopathology 83: 1498–1497. Daboussi, M.-J. (1996) Fungal transposable elements: generators of diversity and genetic tools. Journal of Genetics 75: 325–339. Daboussi, M.-J. & Langin, T. (1994) Transposable elements in the fungal plant pathogen Fusarium oxysporum. Genetica 93: 49–59. Daboussi, M.-J., Langin, T. & Brygoo, Y. (1992) Fot1, a new family of fungal transposable elements. Molecular and General Genetics 232: 12–16. Dickinson, M. J., Wellings, C. R. & Pryor, A. (1990) Variation in the double-stranded RNA phenotype between and within different rust species. Canadian Journal of Botany 68: 599–604. Di Petro, A., Anaya, N. & Roncero, M. I. G. (1994) Occurrence of a retrotransposon-like sequence among different formae speciales and races of Fusarium oxysporum. Mycological Research 98: 993–996. Dobinson, K. F., Harris, R. E. & Hamer, J. E. (1993) Grasshopper, a long terminal repeat (LTR) retroelement in the phytopathogenic fungus Magnaporthe grisea. Molecular Plant-Microbe Interactions 6: 114–126. Enkerli, J., Widmer, F., Gessler, C. & Keller, S. (2001) Strain-specific microsatellite markers in the entomopathogenic fungus Beauveria brongniartii. Mycological Research 105: 1079–1087. Excoffier, L. (1995) AMOVA 1.55 (Analysis of Molecular Variance). Genetics and Biometry Laboratory. University of Geneva, Switzerland. Excoffier, L., Smouse, P. & Quattro, J. (1992) Analysis of molecular variance inferred from metric distances among DNA haplotypes: application to human mitochondrial DNA restriction sites. Genetics 131: 479–491. Felsenstein, J. (1985) Confidence limits on phylogenies: an approach using the bootstrap. Evolution 39: 783–791. Hartl, D. L. & Clarke, A. G. (1997) Principles of Population Genetics. 3rd edn. Sinauer Associates, Sunderland, MA. Hayden, M. J. & Sharp, P. J. (2001) Targeted development of informative microsatellite (SSR) markers. Nucleic Acids Research 29: e44. He, C., Nourse, J. P., Kelemu, S., Irwin, J. A. G. & Manners, J. M. (1996) CgT1: a non-LTR retrotransposon with restricted distribution in the fungal phytopathogen Colletotrichum gloeosporioides. Molecular and General Genetics 252: 320–331.

555 Jaccard, P. (1908) Nouvelles recherches sur la distribution florale. Bulletin Socie´te´ Vaudoise des Sciences Naturelles 44: 223–270. Jarne, P. & Lagoda, P. J. L. (1996) Microsatellites, from molecules to populations and back. Trends in Ecology and Evolution 11: 424–429. Kinsey, J. A. & Helber, J. (1989) Isolation of a transposable element from Neurospora crassa. Proceedings of the National Academy of Sciences, USA 86: 1929–1933. Koch, E., Song, K., Osborn, T. C. & Williams, P. H. (1991) Relationship between pathogenicity and phylogeny based on restriction fragment length polymorphism in Leptosphaeria maculans. Molecular Plant-Microbe Interactions 4 : 341–349. Luig, N. H. (1977) The establishment and success of exotic strains of Puccinia graminis tritici in Australia. Proceedings of the Ecological Society of Australia 10 : 89–96. McDonald, B. A. & Martinez, J. P. (1991) DNA fingerprintng of the plant pathogenic fungus Mycosphaerella graminicola (Anamorph Septoria tritici). Experimental Mycology 15 : 146–158. McHale, M. T., Roberts, I. N., Noble, S. M., Beaumont, C., Whitehead, M. P., Seth, D. & Oliver, R. P. (1992) CfT1: an LTRretrotransposon in Cladosporium fulvum, a fungal pathogen of tomato. Molecular and General Genetics 233: 337–347. McIntosh, R. A., Wellings, C. R. & Park, R. F. (1995) Wheat Rusts An Atlas of Resistance Genes. CSIRO Publications, Melbourne. Majer, D., Mithen, R., Lewis, B. G., Vos, P. & Oliver, R. P. (1996) The use of AFLP fingerprinting for the detection of genetic variation in fungi. Mycological Research 100 : 1107–1111. Mantel, N. A. (1967) The detection of disease clustering and a generalised regression approach. Cancer Research 27: 209–220. Miller, M. P. (1997a) Tools for Population Genetic Analysis (TFPGA). Version 1.3. Department of Biological Sciences, Northern Arizona University, Flagstaff AZ. Miller, M. P. (1997b) Amova-Prep. Department of Biological Sciences, Northern Arizona University, Flagstaff AZ. Newton, A. C., Caten, C. E. & Johnson, R. (1985) Variation for isozymes and double-stranded RNA among isolates of Puccinia striiformis and two other cereal rusts. Plant Pathology 34: 235–247. Park, R. F., Burdon, J. J. & Jahoor, A. (1999) Evidence for somatic hybridization in nature in Puccinia recondita f. sp. tritici, the leaf rust pathogen of wheat. Mycological Research 103: 715–723. Park, R. F., Burdon, J. J. & McIntosh, R. A. (1995) Studies on the origin, spread, and evolution of an important group of Puccinia recondita f. sp. tritici pathotypes in Australasia. European Journal of Plant Pathology 101 : 613–622. Powell, W., Morgante, M., Andre, C., Hanafey, M., Vogel, J., Tingey, S. & Rafalski, A. (1996) The comparison of RFLP, RAPD, AFLP and SSR (microsatellite) markers for germplasm analysis. Molecular Breeding 2: 225–238. Reamon-Buttner, S. M. & Jung, C. (2000) AFLP-derived STS markers for the identification of sex in Asparagus officinalis L. Theoretical and Applied Genetics 100 : 432–438. Ro¨der, M. S., Korzun, V., Wendehake, K., Plaschke, J., Tixier, M. H., Leroy, P. & Ganal, M. W. (1998) A microsatellite map of wheat. Genetics 149: 2007–2023. Sambrook, J., Fritsch, E. F. & Maniatis, T. (1989) Molecular Cloning : a Laboratory Manual. 2nd edn. Cold Spring Harbor Laboratory Press, New York. Shan, X., Blake, T. K. & Talbert, L. E. (1999) Conversion of AFLP markers to sequence-specific PCR markers in barley and wheat. Theoretical and Applied Genetics 98 : 1072–1078. Singh, A., Chaudhury, A., Srivastava, P. S. & Lakshmikumaran, M. (2002) Comparison of AFLP and SAMPL markers for assessment of intra-population genetic variation in Azadirachta indica A. Juss. Plant Science 162: 17–25. Steele, K. A., Humpreys, E., Wellings, C. R. & Dickinson, M. J. (2001) Support for a stepwise mutation model for pathogen evolution in Australasian Puccinia striiformis f. sp. tritici by use of molecular markers. Plant Pathology 50: 174–180.

Genetic variability of cereal rust pathogens Stewart, D. M. & Roberts, B. J. (1970) Identifying races of Puccinia graminis f. sp. avenae: a modified international system. USDA Technical Bulletin 1416. Vos, P., Hogers, R., Bleeker, M., Reijans, M., van de Lee, T., Hornes, M., Frijter, B. A., Pot, J., Peleman, J., Kuiper, M. & Zabeau, M. (1995) AFLP: a new technique for DNA fingerprinting. Nucleic Acids Research 23 : 4407–4414. Waugh, R., McLean, K., Flavell, A. J., Pearce, S. R., Kumar, A., Thomas, B. B. T. & Powell, W. (1997) Genetic distribution of Bare-1-like retrotransponsonable elements in the barley genome revealed by sequence-specific amplification polymorphisms (S-SAP). Molecular and General Genetics 253: 687–694. Weber, J. L. & May, P. E. (1989) Abundant class of human DNA polymorphisms which can be typed using the polymerase chain reaction. American Journal of Human Genetics 44: 388–396. Wellings, C. R., Burdon, J. J., McIntosh, R. A., Wallwork, H., Raman, H. & Murray, G. M. (2000) A new variant of Puccinia

556 striiformis causing stripe rust on barley and wild Hordeum species in Australia. Plant Pathology 49: 803. Wellings, C. R. & McIntosh, R. A. (1990) Puccinia striiformis f. sp. tritici in Australasia: pathogenic changes during the first 10 years. Plant Pathology 39: 316–325. Witsenboer, H., Vogel, J. M. & Michelmore, R. W. (1997) Identification, genetic localization and allelic diversity of selectively amplified polymorphic loci in lettuce and wild relatives (Lactuca spp.) Genome 40 : 923–936. Yap, V. & Nelson, R. J. (1996) WinBoot: a program for performing bootstrap analysis of binary data to determine the confidence limits of UPGMA-based dendrograms. [IRRI Discussion Paper Series No. 14.] International Rice Institute, Manilla. Zambino, P. J., Kubelik, A. R. & Szabo, L. J. (2000) Gene action and linkage of avirulence genes to DNA markers in the rust fungus Puccinia graminis. Phytopathology 90: 819–826. Corresponding Editor: D. S. Hibbett