Evolutionary history of Phakopsora pachyrhizi ( the Asian soybean rust ) in Brazil based on nucleotide sequences of the internal transcribed spacer region of the nuclear ribosomal DNA

Phakopsora pachyrhizi has dispersed globally and brought severe economic losses to soybean growers. The fungus has been established in Brazil since 2002 and is found nationwide. To gather information on the temporal and spatial patterns of genetic variation in P. pachyrhizi, we sequenced the nuclear internal transcribed spacer regions (ITS1 and ITS2). Total genomic DNA was extracted using either lyophilized urediniospores or lesions removed from infected leaves sampled from 26 soybean fields in Brazil and one field in South Africa. Cloning prior to sequencing was necessary because direct sequencing of PCR amplicons gave partially unreadable electrophoretograms with peak displacements suggestive of multiple sequences with length polymorphism. Sequences were determined from four clones per field. ITS sequences from African or Asian isolates available from the GenBank were included in the analyses. Independent sequence alignments of the ITS1 and ITS2 datasets identified 27 and 19 ribotypes, respectively. Molecular phylogeographic analyses revealed that ribotypes of widespread distribution in Brazil displayed characteristics of ancestrality and were shared with Africa and Asia, while ribotypes of rare occurrence in Brazil were indigenous. The results suggest P. pachyrhizi found in Brazil as originating from multiple, independent long-distance dispersal events.


Introduction
Soybean [Glycine max (L.) Merrill] holds a prominent position in the Brazilian economy because of the dimension of the cultivated area and the volume of grain production.To achieve the desirable level of productivity, the Brazilian breeding programs have been actively involved in the release of new varieties with improved agronomic traits, particularly high yielding capabilities.Until some years ago, diseases that imposed important economic losses to Brazilian soybean growers were regarded as of minor importance and were restricted to certain geographic regions and climatic conditions (Almeida et al., 2005).With the global spread of the Asian soybean rust, caused by the fungus Phakopsora pachyrhizi Sydow & Sydow, the national soybean production system is facing a new challenge that requires a new set of management skills from the soybean researchers and growers.The challenges are enormous since no geographic region or state is free from the occurrence of the Asian soybean rust and no resistant variety has yet been released to soybean growers.
In addition to the classical Rpp1, Rpp2, Rpp3 and Rpp4 resistance genes (Bromfield and Hartwig, 1980;Hartwig, 1986), several major resistance genes have been identified in new plant introductions or cultivars (Monteros et al., 2007;Pierozzi et al., 2008;Garcia et al., 2008).However, even with the availability of these new sources to Brazilian breeders, the durability of the disease resistance genes cannot be predicted because of our lack of understanding about fundamental aspects of the fungus biology, such as the level of genetic diversity present in the Brazilian cropping regions.
The Asian soybean rust pathogen was first identified in 1902 in Japan (Bromfield, 1984).During the following 60 years, the disease was identified in the Philippines, China, Australia and several other countries in Southeast Asia.Later, the disease spread to the African continent and was reported in Zimbabwe (1997) andSouth Africa (2001).Until the 2000/2001 season, the American continent was free of the Asian soybean rust.However, the fungus was reported in Par-aguay (2001) and was established in Brazil and Argentina (2002), Bolivia (2003) as well as the United States (2004) (Schneider et al., 2005).Long-distance dispersal of the spores is attributed mainly to wind storms (Isard et al., 2005).
While P. pachyrhizi causes an infestation that results in large economic losses to soybean growers, the disease caused by a closely related species naturally found in the Americas, Phakopsora meibomiae (Arthur) Arthur, results in low or no significant crop losses (Ono et al., 1992).A PCR-based protocol (Frederick et al., 2002) allowed the distinction between P. pachyrhizi and P. meibomiae isolates.Use of this protocol led to the detection of dual infection in some soybean fields located in southern Brazil (Piuga et al., 2003).
Phylogeography studies have been intensively employed to investigate the genetic diversity of natural populations of both plants and animals.It deals with the principles and processes that govern the geographic distributions of genealogical lineages, especially those within and among closely related species (Avise, 2000).A phylogeographic approach generally aims to understand genetic diversity in connection with dispersal history (Avise, 2000).The potential for the use of molecular phylogeography in the investigation of native or introduced species is enormous, mostly because it provides information about genetic diversity not only in the spatial dimension but also in the temporal dimension (Avise, 2000).Molecular phylogeography techniques have been more frequently used in the areas of ecology, conservation genetics and biogeography.
The utilization of molecular phylogeographic approaches to uncover patterns of genetic diversity in species of agricultural importance is in its infancy.Recently, this approach was used to reveal patterns of genetic diversity in the plant pathogens Fusarium graminearum (O'Donnell 2000), Alternaria alternata (Peever et al., 2002), Phaeosphaeria nodorum (Stukenbrock et al., 2006) and Mycosphaerella graminicola (Banke et al., 2004).
In the present study, we used a phylogeographic approach through the molecular characterization of the two internal transcribed spacer regions (ITS1 and ITS2) of the 18S-26S ribosomal RNA genes of P. pachyrhizi and P. meibomiae to address the following questions: (1) What levels of ITS variation are present in P. pachyrhizi and P. meibomiae in Brazil?(2) Is there any association between patterns of ITS sequence variation and geographic location of soybean fields in Brazil?(3) What are the genealogical relationships among P. pachyrhizi found in Brazil and that of African or Asian origins?

Sampling strategy
Phakopsora-infected soybean leaves were collected from 26 soybean fields in Brazil and one field in South Africa.
In Brazil, sampling sites were selected such that together they would represent the area in which soybeans were cultivated during the 2005/2006 growing season.In sampling, we had no particular preference for a given soybean variety, crop system, soil type or field size.Infected leaves were randomly sampled from several soybean plants throughout the field, pooled together according to the field of origin, bagged and shipped to our laboratory, where uredinospores were harvested using a fine paint brush.
Soybean leaves sampled in many of the fields showed symptoms of a mild infection only.In these cases, the leaves were not brushed; instead, leaf segments containing the lesions were removed with the help of a small pair of scissors.Urediniospores or leaf segments were pooled together according to field of origin, lyophilized and maintained at -80 °C until further use.Although inoculation of soybean plants under greenhouse conditions with fieldderived urediniospores could increase fungal tissue availability for DNA extraction, we choose not to use this procedure because the inoculation process might favor some genotypes in detriment of others in case our field samples contained a pool of Phakopsora strains.
In South Africa, infected leaves were collected from naturally infected highly susceptible local cultivars on "Cedara Agricultural Research Station", near Pietermaritzburg, during the 2005/2006 growing season.In this case, leaves were sampled four times during the growing season and samples from each stage were pooled and handled separately.Genomic DNA was extracted at the ARC-Grain Crops Institute, Potchefstroom, following the protocol described below.After extraction, the genomic DNA was lyophilized and sent to our laboratory in Brazil for further analyses.

DNA extraction
Total genomic DNA was extracted using the CTAB method of Doyle and Doyle (1987), with modifications.Each extraction included 30 mg of lyophilized urediniospores or leaf fragments.Crushing was done within a microtube in the presence of 600 μL of a prewarmed (to 65 °C) CTAB extraction buffer (1.4 M NaCl; 100 mM Tris-HCl, pH 8.0; 20 mM EDTA; 5% CTAB; 2% β-mercaptoetanol).The homogenate was incubated for 15 min at 65 °C in the same buffer.Subsequently, 400 μL of a chloroform/isoamilic alcohol mixture (24:1) was added and the suspension was centrifuged for 10 min at 13.000 rpm.The DNA was precipitated with equal volume of isopropanol and washed twice with cold 70% ethanol.The DNA was solubilized in 30 μL TE and stored at -20 °C for subsequent use.

PCR amplification, cloning and sequence editing
Preliminary amplifications of the target gene regions were performed via the polymerase chain reaction (PCR) in a GeneAmp PCR System 9700 (Applied Biosystems).Initially, the universal primers ITS4 and ITS5 of White et al. (1990) were used and an amplicon of about 650 bp was obtained.ITS4 and ITS5 are primers that target the internal transcribed spacer regions (ITS1, intervenient 5.8S sequence, and ITS2) of the 18S-26S ribosomal RNA gene cluster.When soybean genomic DNA was used as the sole source of DNA template in control experiments, PCR amplification with ITS4 and ITS5 also yielded an amplicon of similar length (data not shown).We regarded this soybean amplicon as potentially misleading because soybean genomic DNA was present as a contaminant in the fungal DNA samples.Since elimination of the contaminant DNA was operationally unfeasible, we designed new primers for the specific amplification of the ITS regions of P. pachyrhizi.To design the new primers, alignments of ITS sequences of P. pachyrhizi and Glycine max available in the GenBank (accession numbers: DQ354536 and X02623, respectively) were performed.Alignment analyses indicated sequences in the P. pachyrhizi ITS region with very low homology to correspondent sequences in the Glycine max ITS regions.These low-homology sequences of the P. pachyrhizi ITS region were fed into the software Primer3 (Rozen and Skaletsky, 2000) to design two new primers: ITSPP3 (5'-GTTCAGTGGGTAGTCTCACCTGATT-3') and ITSPP5 (5'-GCAACGGCACTTTACTGGCTC-3').PCR with ITSPP3 and ITSPP5 showed no detectable amplicon when the source of the DNA template was Phakopsora-free soybean leaves, but yielded a single amplicon of the expected size when our fungal DNA samples were used as a DNA template (data not shown).
The fungal tissue samples yielded a very limited amount of DNA upon extraction.Moreover, the DNA was of poor quality in most of the samples as the source of DNA was field-grown soybean leaves.Consequently, amplifications of the whole ITS region with primers ITSPP5 and ITSPP3 were feasible in few samples only.To circumvent this problem, we choose to amplify the ITS1 and ITS2 regions separately using our two new primers and the primers ITS6 and ITS3 of White et al. (1990).PCR with the primer pairs ITSPP5/ITS6 (aimed at the ITS1 region) or ITS3/ITSPP3 (aimed at the ITS2 region) each consistently gave a single amplicon.PCR with any of the above-mentioned primer pairs showed no detectable amplicon when the source of the DNA template was Phakopsora-free soybean leaves (data not shown).
For final PCR amplification, 60 ng of genomic DNA was used in a 25 μL final reaction volume.The concentrations of the reagents were: 10 mM Tris-HCl (pH 8.3), 50 mM KCl, 0.1% Tween 20, 1.5 mM MgCl 2 , 0.2 mM for each deoxynucleoside triphosphate (dNTP), 1.25 μL DMSO, 0.5 μM for each of the primers, 0.25 U Taq DNA polymerase (Phoneutria).The PCR program was 5 min at 94 °C followed by 35 cycles of 1 min at 94 °C for template denaturation, 1 min at 58 °C for primer annealing, and 1 min at 72 °C for primer extension, plus a final extension step of 5 min at 72 °C.A sample of 5 μL of each amplifica-tion reaction was visualized in a 0.8% agarose gel stained with a 0.2 μg/mL ethidium bromide solution, and a single fragment was identified.The PCR product was cloned into pGEM T-Easy vector (Promega), using the manufacture's recommendations.Samples were loaded in a MegaBace DNA Analysis System 500 (Amersham Biosciences Corp.) for sequence detection.
Direct sequencing of most of the Phakopsora-specific amplicons gave partially unreadable electrophoretograms, with overlapping peaks.The peak displacements were suggestive that sequences differing in length could be present within the amplicons.Cloning of the amplicons prior to sequencing resulted in electrophoretograms free of ambiguities.Four clones from each of the 26 Brazilian fields (a total of 104 clones) and four clones from each of the four growing stages of the South-African field (a total of 16 clones) were sequenced for each ITS region.
The 120 sequences for each of the two ITS regions were imported into Sequencher version 4.8 (Gene Codes Corp.) and editing was accomplished through manual corrections.Sequences derived from cloning were easily aligned and adjustments due to the presence of insertion/deletions (indels) were required.
Thirteen sequences of the ITS region of Phakopsora pachyrhizi (GenBank Accessions AF333488 to AF333500) and two sequences of the ITS region of Phakopsora meibomiae (GenBank Accessions AF333501 and AF333502) were available at the time this study was undertaken.The sequences had been obtained by Frederick et al. (2002) from isolates sampled in Africa (Zimbabwe), Asia (Australia, India, Indonesia, the Philippines, Taiwan, and Thailand) and Hawaii.Sampling dates ranged from 1972 to 2000.We included these sequences in our analyses, but excluded the intervening 5.8s RNA gene.

Data analyses
Population structure was estimated by analysis of molecular variance (AMOVA) using Arlequin 3.11 (Excoffier et al., 2005).The total genetic diversity was partitioned into components of two hierarchical levels: between soybean fields and among soybean fields.The significance of the genetic differentiation was tested with 1000 permutations, where P denotes the probability of having a more extreme variance component than would have been obtained for the observed values by chance alone.Arlequin 3.01 also estimated the gene diversity and nucleotide diversity (Nei, 1987) indexes.
We used the computer program TCS version 1.13 (Clement et al. 2000) to obtain an unrooted ribotype cladogram that would reveal genealogical relationships among the ITS sequences.This program applies statistical parsimony by implementing the networking algorithm developed by Templeton et al. (1992).The program collapses the original sequences into ribotypes and calculates a distance matrix (number of mutational differences) for all pairwise comparisons of ribotypes until the probability exceeds 95%.The number of mutational differences just above this cutoff point is the maximum number of mutational connections between pairs of sequences justified by the parsimony criterion (Clement et al., 2000).The program was run considering indels as a 5 th character state and coded such that each indel, regardless of its size, was considered a single state.We estimated two ribotype networks, for ITS1 and ITS2 regions, respectively.

Results
Datasets for ITS1 and ITS2 regions were assembled independently.Each contained 133 sequences from P. pachyrhizi and two sequences from P. meibomiae.Among the sequences of P. pachyrhizi in each dataset, 104 were obtained in Brazil, 16 from South Africa, and 13 from GenBank.Overall, the length of the ITS1 region ranged from 251 to 252 bp, the 5.8S was 153 bp, and the ITS2 region ranged from 229 to 240 bp.
Independent sequence alignments of the ITS1 and ITS2 datasets uncovered 27 and 19 ribotypes, respectively.Twenty four base substitutions and four 1-bp indels distinguished the 27 ribotypes of the ITS1 region (Figure 1).For the ITS2 region, 17 polymorphic sites (ten base substitutions and seven indels) defined the 19 ribotypes.Among the seven indels found in the ITS2 ribotypes, five were 1 bp long, one was 3 bp long, and one was 6 bp long (Figure 2).The sequences have been deposited in GenBank (EU930070 to EU930096 for ITS1 ribotypes and EU930097 to EU930115 for ITS2 ribotypes).
The genealogical relationships among the ribotypes of P. pachyrhizi and P. meibomiae were revealed by assembling two ribotype networks, one for each ITS datasets.In each case, species-specific networks were obtained.Consistently, one of the networks contained the sequences of P. meibomiae, while the other network contained the sequences of P. pachyrhizi (Figures 3 and 4).The maximum number of mutational connections between pairs of sequences, justified by the 'parsimony' criterion with 95% confidence, was estimated as six.The estimation procedure detected many interior nodes (represented by a small circle) to which none of the sequences could be assigned.These nodes represent inferred intermediate ribotypes between two nearest-neighbor ribotypes in the network that differed by two or more mutations (Templeton, 1998).These missing intermediates are either ribotypes that have vanished, or ribotypes with such a low frequency in the population that they were not sampled.
According to the predictions of the coalescent theory (Templeton et al., 1992), ancestral ribotypes characteristically: (a) are located at the inner part of the network, (b) occur at high frequencies, (c) show a large number of connections to low frequency ribotypes, and additionally (d) are widespread at large geographic areas.In contrast, derived ribotypes are: (a) located in tips of the network, (b) occur at low frequency, (c) are connected to a ribotype that is located in a central part of the network, and (d) are circumscribed to limited geographic areas, such as a single population or adjacent populations.
For the ITS1 region of P. pachyrhizi, the most common ribotypes were ribotypes A and B (Table 1), which occupy the inner parts of the network and are connected to a number of low frequency ribotypes (Figure 3).Moreover, these two ribotypes showed the most widespread geographic distribution among the ITS1 ribotypes (Figure 5) and may have an ancient origin because they were identified in GenBank accessions collected in Australia (1972and 1979), Hawaii (1995and 1998), India (1973), Indonesia (1972), the Philippines (1977), Taiwan (1972and 1980), Thailand (1976) and Zimbabwe (2000).Ribotypes A and B also showed the most widespread occurrence within the Brazilian territory.Ribotype D also appeared with high frequency (although not as high as that shown by ribotypes A and B) and occupies a central position in the ITS1 network (Figure 3).It was collected from Cedara (South Africa) and from five soybean fields in Brazil (Table 1 and Figure 5).Ribotypes A, B, and D should, therefore, be treated as having an ancestral relationship to the remaining ribotypes in the network.Additional evidence of the ancestrality of ribotypes A, B, and D emerges when the network configuration data are evaluated with respect to the geographic distribution of their ribotype components (Figure5).Most of the remaining ITS1 ribotypes occurred at very low frequencies, 22 of them being singletons, that is, they were each identified within a single soybean field.Three of these singletons (ribotypes α, T and Z) were exclusive to Cedara (South Africa) (Figure 5).For the ITS2 region of P. pachyrhizi, the most common ribotypes were ribotypes A' and B' (Table 2).Although these two ribotypes occupy central positions and are connected to a number of tip ribotypes in the network recovered for ITS2 region (Figure 4), they were found mostly at Cedara (South Africa) and in Brazil (Figure 6).Ribotype A' was uncovered from a single GenBank accession (AF333500) collected in Zimbabwe ( 2000    cate that these two ribotypes, although frequently found in Brazil, may be of recent origin.Two ribotypes of rare occurrence (D' and E') showed interesting placements in the network.Among the GenBank accessions, they were those of higher frequencies, appearing seven and five times, respectively.Ribotype E' occupies a tip position and was not found in Brazil, but was identified in samples collected in Asia since 1972.Ribotype D', which is internally connected to E' was collected from only two Brazilian fields, but was identified in samples collected in Asia between 1972 and 1980, and in Zimbabwe, Africa in 2000.Ribotypes D' and E' were both absent from the South African samples.The majority of the remaining ITS2 ribotypes of P. pachyrhizi were singletons (Figure 6).Overall, most of the Brazilian fields exhibited from two to three ribotypes for each ITS region (Tables 1-2 and Figure 5-6).In South Africa, DNA samples from infected leaves that were sampled four times during the growing season gave distinct ribotypes (data not shown).These findings are consistent with the lack of readability of elec-trophoretograms derived from direct sequencing of PCR products and is an indication that P. pachyrhizi may possess variation for the ITS sequences evaluated in this study.
Analysis of molecular variance (AMOVA) was performed with data sets taken from each ITS region and only included data collected from the Brazilian fields (Table 3).For the ITS1 dataset, results of the AMOVA revealed that 90.77% of the total molecular variance can be attributed to within field variation.A similar result (84.3%) was obtained when AMOVA was performed using the ITS2 dataset.These high levels of within field diversity are consistent with the high number of singletons uncovered during sequence analysis.Genetic differences among the Brazilian fields were low.AMOVA revealed that only 9.23% of the variation, for the ITS1 dataset, and 15.7% of the variation, for the ITS2 dataset, can be attributed to differences among fields.As shown in Tables 2 and 3, the most frequent ribotypes were shared by most of the Brazilian fields, which is consistent with differences among fields being low.(Nei, 1987), **Nucleotide diversity (Nei and Li, 1979).
Nucleotide diversity and gene diversity are presented in Table 4.Nucleotide diversity estimates the probability that two randomly chosen homologous nucleotides will be different and is equivalent to the level of polymorphism within population (Nei, 1987).Gene diversity is defined as the probability that two randomly chosen ribotypes will be different in the sample, and is used as a measure of the genetic variability within a population (Nei and Li, 1979).Gene diversity for ITS1 reached the lowest value (0.0) in Chapadão do Sul (MS).For ITS2, gene diversity reached the lowest value (0.0) in Itiquira (MT), Rio Verde (GO), Mineiros (GO), and Taquarituba (SP).In each of these five fields, only a single ribotype was uncovered, and therefore no diversity was found.On the other hand, gene diversity for ITS1 reached the highest possible value (1.0) in Alto Taquari (MT), Rio Verde (GO), and Pitanga (PR).For ITS2, gene diversity was the highest in Alto Taquari (MG), Palmas (TO), and Pitanga (PR).Each of the four clones sequenced in each of these fields yielded a distinct ribotype.Two fields with the same number of ribotypes, such as Cerejeiras (RO) and Palmas (TO) with two ribotypes each, did not necessarily yield the same estimate for nucleotide diversity.The reason for different estimates between Cerejeiras (RO) and Palmas (TO) is that nucleotide diversity takes into consideration the number of mutations that distinguishes each of the ribotypes.In our investigation, nucleotide diversity ranged from zero (in those fields where a single ribotype was identified), to 0.014 at Pitanga (PR) (with four ribotypes for ITS2) and 0.011 at Alto Taquari (MT) (with four ribotypes for ITS1).

Discussion
One of the objectives of this investigation was to determine the levels of ITS variation in P. pachyrhizi and P. meibomiae present in soybean fields of Brazil.A previous study already had revealed that nucleotide sequence similarity among isolates of either P. pachyrhizi or P. meibomiae were greater than nucleotide sequence similarity between the two species (Frederick et al., 2002) and reference sequences for comparative purposes were available at the GenBank.Sequence alignments and data analyses revealed that each of the 208 sequences of Brazilian origin identified in this study were either identical to, or diverged by a few mutation steps from the sequences of P. pachyrhizi deposited in the GenBank.None of the sequences were either identical to or had any ancestral-descendent relationship to the sequences of P. meibomiae.Species-specific networks obtained with our datasets confirmed this trend.Despite being a weaker pathogen that does not cause economic damage to soybeans, P. meibomiae is endemic to the Americas and infects a range of host species in addition to soybeans (Ono et al., 1992).Until 2000, this was the only soybean rust species present in Brazil (Carvalho Jr and Figueiredo, 2000).It was therefore plausible to expect that our sequence analyses might confirm the incidence of P. meibomiae in soybean leaves harvested in at least some of the Brazilian fields.Such double incidence had been detected previously (Piuga et al., 2003).However, none of the sequences we obtained could be attributed to P. meibomiae.Our results do not allow us to exclude the presence of P. meibomiae from soybean fields in Brazil, but they indicate that the current incidence of this species may be extremely low compared to the occurrence of the P. pachyrhizi.Levels of nucleotide diversity in both the ITS regions of P. pachyrhizi were low.However, for a species that was first reported in Brazil in 2002 (Yorinori and Paiva, 2002), and that appears to be maintained asexually (the sexual stage has not yet been reported), the level of intraspecific ITS sequence variation, as revealed by the number of ribotypes uncovered, was surprisingly high.Although species that are maintained via clonal propagation are expected to display low levels of genetic diversity, there are many instances in which strictly asexual plant pathogens possess unusually high levels of genetic diversity, as in Puccinia triticina (Kolmer, 2001;Goyeau et al., 2007).Several events are invoked to explain such high levels of genetic diversity in imperfect fungi, including parasexuality and heterokaryosis (Kuhn et al., 1995;Taylor et al., 1999), as well as high mutation rate (Bentley et al., 1998;O'Donnell et al., 1999).The occurrence of any of these processes in Phakopsora pachyrhizi is, however, yet to be investigated.
The phylogeographic framework proposed by Templeton et al. (1992) and co-workers allowed us to investigate the association between ITS sequence variation, geographic location of soybean fields in Brazil, and the genealogical relationships among Brazilian, African, and Asian ribotypes.The overall pattern we identified in this investigation is that those ribotypes with the most widespread occurrence in Brazil also displayed characteristics of ancestrality and were shared with Africa in particular, and to a lesser degree with Asia.On the other hand, ribotypes of rare occurrence in Brazil are endemic and were limited to a single field or to fields in the same vicinity in Brazil.
These findings support an African origin for the populations of P. pachyrhizi found in Brazil as ribotypes D, A' and B', which occurred in both Africanand Brazilian samples, but not in those of Asian-Australian origin.Likewise, ribotype E' occurred in Asian, but not in the African or Brazilian samples.Ribotype B occurred in the Brazilian and South African samples, and in the Australian sample, but not in any of the Asian samples, indicating that the African and Brazilian samples may be more closely related to the Australian sample.More recent samples from Asia and Australia are needed to confirm these indications.
One possible scenario that may account for the natural introduction of the Asian soybean rust into Brazil would involve a transatlantic dispersal of spores from Africa mediated by air currents.Atmospheric pathways have been suggested to contribute to long-distance dispersal of rust pathogens (Nagarajan and Singh, 1990), and recently this possibility has been applied to investigate the origin of P. pachyrhizi found in the United States (Isard et al., 2005).
Regardless of the mechanism of introduction, a single long-distance dispersal event is not compatible with the pattern of distribution of genetic diversity we uncovered for P. pachyrhizi in Brazil.Shared ribotypes with Africa that have distinct patterns of geographic distribution in Brazil are suggestive of multiple, independent introduction events.Considering that the DNA regions being investigated are not under selection pressure, it is plausible to expect that early-arriving ribotypes would have originated in Africa/Asia and would display a wider geographic distribution in Brazil.Similarly, ribotypes with an African/Asian origin, but being of rare occurrence in Brazil, would have arrived later.This rationale can be illustrated with the ITS2 ribotypes as an example.Ribotype D' is expected to represent a recent introduction because it was present in several African and Asian isolates sampled from 1972 to 2000, but was found in only two Brazilian fields.Conversely, ribotypes A' and B' were probably introduced earlier since they are of widespread occurrence in Brazil and are shared with Africa.

Freire
et al.

Freire et al. 923 Figure 1 -
Figure 1 -Sequence alignment of the variable sites for the ITS1 region defining the 27 ribotypes uncovered in 133 cloned sequences of Phakopsora pachyrhizi.Each fragment spans 254 bases.Dots indicate similarity to ribotype A and hyphens indicate gaps.Numbers indicate the nucleotide position having ribotype A as a reference sequence during alignment.

Figure 2 -
Figure 2 -Sequence alignment of the variable sites for the ITS2 region defining the 19 ribotypes uncovered in 133 cloned sequences of Phakopsora pachyrhizi.Each fragment spans 241 bases.Dots indicate similarity to ribotype A' and hyphens indicate gaps.Numbers indicate the nucleotide position having ribotype A' as a reference sequence during alignment.

Figure 3 -
Figure 3 -Networks obtained with TCS 1.13 for Phakopsora pachyrhizi (I) and Phakopsora meibomiae (II) using ITS1 ribotypes.Indels were considered as 5 th character state and coded such that each one, regardless of its size, was considered as only one character.Letters inside circles identify the detected ribotype.Each line in the network represents one mutational change, and small, open circles indicate extinct or unsampled ribotypes.

Figure 4 -
Figure 4 -Networks obtained with TCS 1.13 for Phakopsora pachyrhizi (I) and Phakopsora meibomiae (II) using ITS2 ribotypes.Indels were considered as 5 th character state and coded such that each one, regardless of its size, was considered as only one character.Letters inside circles identify the detected ribotype.Each line in the network represents one mutational change, and small, open circles indicate extinct or unsampled ribotypes.
Asian soybean rust in BrazilTable 2 -Distribution of the 19 ITS2 ribotypes among 133 cloned sequences of Phakopsora pachyrhizi.The total number of cloned sequences analyzed per site (#), the total number of ribotypes found in each site (R SH ), and the number of sites that share a given ribotype (R SITE ) are indicated.Sites were numbered from 1 to 27 as indicated in parenthesis after the site name.

Freire et al. 927 Figure 5 -
Figure 5 -Geographic distribution of the 27 ribotypes for the ITS1 region of Phakopsora pachyrhizi.(I) Distribution of ribotypes found in Africa and Asia.(II) Distribution of the ribotypes found in Brazil.Numbers on the map correspond to the same population codes given in Table 1.Areas of pie charts represent composite haplotype frequency.Ribotype colors and codes are as in the network shown in Figure 3.

Figure 6 -
Figure 6 -Geographic distribution of the 19 ribotypes for the ITS2 region of Phakopsora pachyrhizi.(I) Distribution of ribotypes found in Africa and Asia.(II) Distribution of the ribotypes found in Brazil.Numbers on the map correspond to the same population codes given in Table 2. Areas of pie charts represent composite haplotype frequency.Ribotype colors and codes are as in the network shown in Figure 4.

Table 3 -
Summary of the AMOVA based on ITS1 and ITS2 datasets for Phakopsora pachyrhizi sampled in 26 soybean fields in Brazil p-values are the probabilities of having a more extreme variance component than would have been obtained for the observed values by chance alone.Probabilities were calculated by 1000 random permutations. *

Table 4 -
Gene diversity* and nucleotide diversity** of Phakopsora pachyrhizi in 26 soybean fields in Brazil *Gene diversity