Figures
Abstract
Background
Elymus (Poaceae) is a large genus of polyploid species in the wheat tribe Triticeae. It is polyphyletic, exhibiting many distinct allopolyploid genome combinations, and its history might be further complicated by introgression and lineage sorting. We focus on a subset of Elymus species with a tetraploid genome complement derived from Pseudoroegneria (genome St) and Hordeum (H). We confirm the species' allopolyploidy, identify possible genome donors, and pinpoint instances of apparent introgression or incomplete lineage sorting.
Methodology/Principal Findings
We sequenced portions of three unlinked nuclear genes—phosphoenolpyruvate carboxylase, β-amylase, and granule-bound starch synthase I—from 27 individuals, representing 14 Eurasian and North American StStHH Elymus species. Elymus sequences were combined with existing data from monogenomic representatives of the tribe, and gene trees were estimated separately for each data set using maximum likelihood. Trees were examined for evidence of allopolyploidy and additional reticulate patterns. All trees confirm the StStHH genome configuration of the Elymus species. They suggest that the StStHH group originated in North America, and do not support separate North American and European origins. Our results point to North American Pseudoroegneria and Hordeum species as potential genome donors to Elymus. Diploid P. spicata is a prospective St-genome donor, though conflict among trees involving P. spicata and the Eurasian P. strigosa suggests either introgression of GBSSI sequences from P. strigosa into North American Elymus and Pseudoroegneria, or incomplete lineage sorting of ancestral GBSSI polymorphism. Diploid H. californicum and/or allotetraploid H. jubatum are possible H-genome donors; direct involvement of an allotetraploid Hordeum species would simultaneously introduce two distinct H genomes to Elymus, consistent with some of the relationships among H-genome sequences in Hordeum and Elymus.
Conclusions/Significance
Comparisons among molecular phylogenetic trees confirm allopolyploidy, identify potential genome donors, and highlight cases of apparent introgression or incomplete lineage sorting. The complicated history of this group emphasizes an inherent problem with interpreting conflicts among bifurcating trees—identifying introgression and determining its direction depend on which tree is chosen as a starting point of comparison. In spite of difficulties with interpretation, differences among gene trees allow us to identify reticulate species and develop hypotheses about underlying evolutionary processes.
Citation: Mason-Gamer RJ, Burns MM, Naum M (2010) Reticulate Evolutionary History of a Complex Group of Grasses: Phylogeny of Elymus StStHH Allotetraploids Based on Three Nuclear Genes. PLoS ONE 5(6): e10989. https://doi.org/10.1371/journal.pone.0010989
Editor: Simon Joly, Montreal Botanical Garden, Canada
Received: December 21, 2009; Accepted: May 10, 2010; Published: June 9, 2010
Copyright: © 2010 Mason-Gamer et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: The work was supported by National Science Foundation DEB-9974181 to RJMG (http://www.nsf.gov). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Introduction
Untangling reticulate relationships among species presents an interesting challenge to systematists, and an opportunity to uncover previously undetected evolutionary processes. Comparisons among gene trees can clarify historical relationships among species, and the examination of topological conflicts among trees can reveal complicating factors such as retention of ancestral genetic polymorphism, past or ongoing gene exchange, allopolyploidy, or a combination of these. Distinguishing among potential causes of phylogenetic conflict is often difficult, but careful comparisons among trees can help pinpoint the species involved, and allow specific hypotheses to be formulated. In the present study, we focus on species that are explicitly reticulate in that they are all allotetraploids, and potentially secondarily reticulate if they have arisen through multiple independent origins or undergone hybridization at the tetraploid level. We assess the role of reticulation at both levels in the genus Elymus L. of the wheat tribe Triticeae (Poaceae), using phylogenetic analyses of three unlinked, low-copy genetic markers.
Many inferences of reticulate evolution have been based on comparisons among gene trees; in plants, comparisons between chloroplast DNA (cpDNA) and nuclear ribosomal internal transcribed spacer (ITS) phylogenies are especially widely used (e.g., [1]–[7]). The reasons are partly historical – for technical reasons, these high-copy markers were among the first to be widely used for plant phylogenetic studies. They remain in frequent use in studies of reticulation (e.g., [8]–[12]), where they offer both methodological advantages and a kingdom-wide backdrop of published sequences within which new data can be interpreted. However, both cpDNA and ITS sequence data sets have some disadvantages. The chloroplast genome is maternally inherited in most angiosperms, and its ability to identify a maternal donor can be an advantage. However, its inability to provide information about other genetic donors is often a major limitation. The biparentally-inherited ITS does have the potential to reveal multiple genome donors, but its arrangement in long repetitive arrays promotes the confounding effects of concerted evolution, both within arrays [13]–[15] and among them [16], [17]. Thus, ITS copies can potentially convert toward one or the other parent, and the resulting sequence homogeneity can obscure a history of contributions from multiple distinct donors.
Low-copy nuclear genes can, like ITS, reveal multiple genome donors, and they are less subject to gene conversion. However, they do have some practical disadvantages. They can be more difficult to amplify because of their low copy number, and because online databases often contain fewer comparable sequences from which amplification primers can be designed. The smaller sequence database also narrows the phylogenetic context within which new data sets can be analyzed, and makes it more difficult to assemble the crucial copy-number information that would prevent misinterpretation of unsuspected variation among paralogs. In spite of the difficulties, a variety of single- and low-copy nuclear genes have been successfully used in many studies of reticulate relationships in plants over the last decade (e.g., [18]–[32]). Sequence data from some low-copy genes are now becoming plentiful across a broad range of angiosperms.
This study presents three low-copy nuclear gene trees from a group of tetraploid species in the wheat tribe, Triticeae. The wheat tribe is especially well known for its economically important members, including wheat, barley, and rye. The tribe's economic importance has driven an interest in its evolutionary history seemingly disproportionate to its size (about 300 species), yet a singular, straightforward phylogenetic estimate for the tribe remains elusive. One reason for this is that a history of incomplete lineage sorting and/or gene exchange has complicated relationships among the diploid lineages, so that sequence data from different genes yield conflicting trees [33], [34]. A second confounding issue is that the tribe includes a large number of genetically diverse allopolyploid lineages. The most well known of these are the tetraploid and hexaploid cultivated wheats (Triticum L.), but far more numerous are those that combine genomes from the wheatgrass genus Pseudoroegneria (Nevski) Á.Löve (genome designation St) with one or more genomes from other Triticeae genera (e.g., [35]). Under the genomic definition of genera [35], [36] widely applied to the Triticeae, most of the St-genome allopolyploids are classified as Elymus. Within Elymus, the St genome can be combined with a variety of other genomes, including that of Hordeum L. (genome designation H), Agropyron Gaertn. (P), Australopyrum (Tzvelev) Á.Löve (W), and an unknown donor (Y), and in many combinations including StStHH, StStYY, StStHHHH, StStStStHH, StStStStYY, StStYYYY, StStHHYY, StStYYWW, and StStYYPP [36]–[49]. Other St-containing allopolyploids include the autoallooctoploid Pascopyrum smithii (Rydb.) Á. Löve, which combines the Pseudoroegneria and Hordeum genomes with the Ns genome of Psathyrostachys Nevski an StStHHNsNsNsNs configuration [50]. Thinopyrum Á.Löve includes some octo- and decaploid species which are hypothesized to combine the St genome with the E and/or J genomes characteristic of Thinopyrum [51]–[53].
In this study, we focus on the StStHH Elymus tetraploids. This northern temperate group of about 50 species is distributed throughout much of North America, Europe, and western Asia. Numerous studies provide evidence that Pseudoroegneria and Hordeum were the genome donors to these tetraploids (e.g., [54] and references therein). Our results clearly confirm these studies, but we considered the StStHH genome configuration to be well-established before the start of the study, rather than a hypothesis in need of additional tests. Our main goals, therefore, were to: (1) determine whether the North American vs. Eurasian StStHH species arose from separate polyploidization events; (2) identify possible progenitor species from within Hordeum and Pseudoroegneria; and (3) find out whether reticulate patterns in StStHH Elymus extend beyond those clearly attributable to allopolyploidy alone.
Materials and Methods
Samples
The analyses include 27 individuals representing 14 StStHH tetraploid Elymus species: 11 individuals representing five Eurasian species (Table 1), and 16 individuals representing nine North American species (Table 2). Most of the accessions were obtained from the USDA and have associated chromosome counts (2n = 28), confirming their tetraploid nature. Nearly all of the sequences from the Eurasian Elymus samples were newly generated for this study; the few exceptions [55] are noted in Table 1. Many of the North American Elymus pepC and GBSSI sequences were drawn from earlier studies [29], [56], while the North American β-amylase sequences are new. The numbers of intraspecific samples of North American species differ among the three gene trees, but this does not affect our main conclusions.
Three single- or low-copy nuclear genes were amplified from each Elymus individual, and 8–24 clones per gene per individual were checked; the goal was to obtain copies from the St and H genomes from each. For each gene, the cloned sequences from Elymus were analyzed with previously-published sequences from a reasonably broad sample of the tribe's known genomic diversity, including representatives of 15 monogenomic genera (Table 3). These include the donor of the St genome (Pseudoroegneria), and of the genomes known to co-occur with St in allopolyploid nuclei: Hordeum (H), Agropyron (P), Australopyrum (W), Psathyrostachys (Ns; this genome is represented by tetraploid Leymus Hochst. in the pepC data set), and Thinopyrum (J and/or E). Additional monogenomic genera include Aegilops L., Crithopsis Jaub. & Spach (except for pepC), Dasypyrum (Coss. & Durieu) T.Durand, Eremopyrum (Ledeb.) Jaub. & Spach, Henrardia C.E.Hubb. (except for pepC), Heteranthelium Hochst. ex Jaub. & Spach, Peridictyon Seberg, Fred., & Baden, Secale L., Taeniatherum Nevski, and Triticum. The sample includes most of the monogenomic genera accepted in genome-based classifications of the Triticeae (e.g., [35], [36]). All three gene trees were rooted with a representative of Bromus L.; Bromeae and Triticeae have repeatedly been shown to form a single clade, with Bromeae as either sister or paraphyletic to a monophyletic Triticeae [57], [58].
Nearly all of the sequences from the monogenomic species were previously published in various sources [21], [29], [56], [59]–[64], with a few exceptions as noted in Table 3. Information about the data and taxa can be found therein, but the primary discussions about the characteristics of each marker and data set are: pepC [56]; β-amylase [61]; and GBSSI [29], [63]. Based on studies of grass genomes in crop species, the three nuclear markers appear to be on three different chromosomes (more below). This assumption is tentative because it is based on a small number of grass species, but for this study the three genes are assumed to be unlinked, and to represent independent phylogenetic estimates.
Molecular methods and alignment
Similar molecular methods were followed for each of the three nuclear gene fragments (detailed protocols are found in the works cited above for each marker). For each Elymus individual, three PCR replicates were run per individual, in order to counter the potential effects of PCR drift [65]. PCR products from replicate reactions were combined and cleaned on columns (Qiagen). Cleaned products were cloned into pGEM-T Easy vectors (Promega), and transformed into E. coli JM109 competent cells (Promega) following the manufacturer's protocol, except that all reactions were halved. Cloned fragments were amplified directly from white colonies using the same primers as were used for the original PCR, in 30–40 µl reactions with 0.5 units Taq polymerase (Sigma), a 1× concentration of the included Taq buffer, 45–60 nmol MgCl2, 6–8 nmol of each nucleotide, and 30–40 pmol of each primer. Amplified fragments were cleaned for sequencing using 1 unit shrimp alkaline phosphatase (USB) and 5 units exonuclease I (USB). Sequencing reaction included 1–3 µl of cleaned product, 2 µl BigDye Terminators v.3.1 (Applied Biosystems), and 3.2 pmol primer in a 10 µl volume. Reactions were run on an ABI Prism 377 (our lab) or ABI 3730 DNA Analyzer (Pritzker Lab, Field Museum of Natural History). Between 8 and 24 cloned PCR products from each individual were screened with a single sequencing primer, yielding a single-stranded partial sequence of about 600 basepairs. We examined these preliminary sequences to identify the two homoeologous sequence types (St and H) that we expected to find within each tetraploid individual. Representative clones of each were fully sequenced on both strands and added to the data set. If either homoeologous copy was missing from an initial sample of 8–12 clones, the corresponding gene from that individual was re-amplified and cloned, and 12 additional sequences were screened. We also included distinct, same-genome alleles from within individuals when they were encountered, although this was not our main goal. Based on the 600-basepair preliminary sequences, same-genome sequences that differed by more than three basepair substitutions were fully sequenced and added to the data set. The three-basepair threshold was arbitrary, but we reasoned that it was large enough to reveal distinct alleles rather than Taq errors.
PCR amplification of intra-individual variants can yield chimeric sequences [66]–[68]. A small number of recombinants were identified by inspection of alignments prior to phylogenetic analysis; such sequences are often visible because the St and H variants have length differences in some of the introns. A few more were identified following closer examination of sequences on long branches in preliminary maximum parsimony analyses. Such sequences were confirmed as recombinant by inspection or if, when they were divided at the presumed point of recombination and analyzed as separate sequences, one portion was phylogenetically St-like and the other was H-like. Chimeric sequences were interpreted as PCR artifacts and removed from the analyses.
The pepC gene is a member of a three-copy family in grasses [69]; the sequences used here appear to be homologous to the widely-expressed housekeeping copy. Based on the location of similar sequences in the rice genome (Genbank AP005781 and AP005802) and a comparative grass genome map [70], this gene copy is assumed to be on the Triticeae group 5 homoeologous chromosomes. The original Triticeae pepC data set [56] combined two fragments designated region 1 (approximately 1 kb; Genbank AY553236–AY553269) and region 2 (approximately 600 bp; Genbank AY548399–548432); the present data set includes just region 1 sequences. The 1100-bp PCR products obtained using primers 467F(1) and 1672R(2) [56] include partial exons 1 and 2, along with the intervening intron, which is approximately 1000 bp long. The intron exhibits considerable length variation, including insertion and excision of transposons [21]. Most length variation could be accommodated by manually adjusting the alignment (File S1). An ambiguous region of the alignment consisting mainly of short runs of C and/or T (positions 67–109), and two regions affected by transposon activity (690–771 and 1035–1119), were excluded from the analysis.
The β-amylase genes form a small family in the Triticeae, with several copies expressed in the endosperm and one that is ubiquitously expressed [71]. Based on sequence similarity, the sequences used here appear to represent the ubiquitously-expressed copy; this copy has been mapped to the Triticeae group 2 homoeologous chromosomes [72]. The 1400-bp β-amylase PCR products were obtained using primers 2a-for and 5a-bac [61], and include partial exons 2 and 5, complete exons 3 and 4, and introns 2–4, which are about 250, 100, and 400 bp in length, respectively. The β-amylase alignment (File S2) was generally straightforward; most length differences were easy to interpret. One ambiguous simple sequence region (positions 553–570) and two regions corresponding to Stowaway-like transposon activity in some sequences (positions 635–765 and 1475–1641) [73] were excluded from the analyses.
The GBSSI PCR products were obtained using the F-for and M-bac primers [63], which amplify an approximately 1300-bp fragment that includes partial exons 9 and 14, exons 10–13, and introns 9–13, which are about 100 bp each. The putatively single-copy GBSSI gene maps to the Triticeae group 7 homoeologous chromosomes [70], [74], or to a portion of chromosome 4 translocated from, and thus homoeologous to, the group 7 chromosomes [70], [75]. The GBSSI alignment (File S3) is generally straightforward in spite of numerous small insertions and deletions in the introns. Three ambiguously-aligned regions (positions 910–1020, 1138–1231, and 1383–1515) were excluded from the phylogenetic analyses.
Phylogenetic analyses
Prior to phylogenetic analyses, 16 nested models of sequence evolution [76]–[78] were examined for each data set using preliminary maximum parsimony trees, and the resulting maximum-likelihood (ML) scores were compared using a likelihood ratio test [78]–[81]. For each data set, the general time-reversible (GTR) substitution model [82], [83] led to a large and significant increase in score compared to the Jukes-Cantor [84], Kimura two-parameter [85], and Hasegawa-Kishino-Yano [86] models, as did the addition of a gamma (Γ) distribution with shape parameter α to model among-site rate variation [87]. Adding an invariable sites (I) parameter [86] to the GTR+Γ model led to a non-significant increase in the pepC and β-amylase scores, and a significant increase in the GBSSI score. Therefore, the GTR+Γ model was used for the ML analysis of the pepC and β-amylase data, and the GTR+I+Γ model was used for the analyses of the GBSSI data sets.
All ML analyses were run using the Mac OS X UNIX version of GARLI v. 0.95 [88]. Following the recommendations of the author, runs were set for an unlimited number of generations, and automatic termination following 10,000 generations without a significant (lnL increase of 0.01) topology change. For each data set, thirty analyses were run with random starting tree topologies, and the tree with the best score was used to display the gene tree. Branch support (BS) for each tree was estimated based on 100 ML bootstrap replicates in GARLI with searches as above, except that the stopping criterion was lowered to 5,000 generations without a significant topology change. Bootstrap values≥50% were recorded on the best ML trees.
Results
General
On all three gene trees (Figures 1–3) the Elymus sequences fall into two main groups, with Pseudoroegneria and with Hordeum. The two distinct Elymus groups are interpreted as St and H sequences, respectively, derived from Pseudoroegneria and Hordeum progenitors. Nearly all of the Elymus individuals yield sequences in both of the main clades, with the following exceptions: for pepC, only an H sequence was recovered from E. virginicus 4 and E. mutabilis 1; for β-amylase, only an H sequence was recovered from E. glaucus 6, E. hystrix 1, E. virginicus 4, and E. sibiricus 1; and for GBSSI, only an H sequence was recovered from E. riparius 1. No individuals lack a sequence type in all three data sets, and only Elymus virginicus 4 is missing a sequence type for two of the three genes. There are cases where individual plants show within-genome (presumably allelic) variation, and these cases are used where possible to shed light on the evolutionary history of the StStHH Elymus group. However, the study was designed primarily to capture intergenomic variation within individuals (i.e., St vs. H) rather than allelic variation, so patterns of intra-individual variation within genomes are probably more widespread than the data show.
The best-scoring ML tree was selected from 30 GARLI analyses of the pepC sequence data set under a GTR+Γ model of sequence evolution. Numbers above branches show ML bootstrap support ≥50%. “NA” and “Eu” distinguish North American and Eurasian Elymus species, respectively. Where applicable, numbers following taxon names distinguish individuals within species, and are consistent among Figures 1–3, S1. Letters following these numbers designate cloned sequences from within individuals, and are specific to each gene tree.
The best-scoring ML tree was selected from 30 GARLI analyses of the β-amylase sequence data set under a GTR+Γ model of sequence evolution. Numbers above branches show ML bootstrap support ≥50%. “NA” and “Eu” distinguish North American and Eurasian Elymus species, respectively. Where applicable, numbers following taxon names distinguish individuals within species, and are consistent among Figures 1–3, S1. Letters following these numbers designate cloned sequences from within individuals, and are specific to each gene tree.
The best-scoring ML tree was selected from 30 GARLI analyses of the GBSSI sequence data set under a GTR+I+Γ model of sequence evolution. Numbers above branches show ML bootstrap support ≥50%; the 28% value is mentioned in the text. “NA” and “Eu” distinguish North American and Eurasian Elymus species, respectively. Where applicable, numbers following taxon names distinguish individuals within species, and are consistent among Figures 1–3, S1. Letters following these numbers designate cloned sequences from within individuals, and are shared by Figures 3 and S1.
There are several cases of apparent β-amylase homeolog silencing in Elymus, inferred from exon sequences. Of the Eurasian species, two of the four E. caninus individuals have a stop codon in exon 2 of their St-genome copies (clones 2d and 4d). Silent copies are more widespread among the North American individuals, and all involve the H-genome copy. The E. hystrix 1a, E. riparius 1b, and E. virginicus 9e clones share a 2-basepair deletion in exon 2, and E. hystrix 1a has a second, single-basepair deletion in exon 4; the E. elymoides 1c and E. glaucus 6a clones each have a unique, single-basepair deletion in exon 2; and E. wawawaiensis 3b has a stop codon in exon 4. Note that for two individuals (E. glaucus 6 and E. hystrix 1), no functional β-amylase copy was recovered; in both cases, the St copy was not present among the sequenced clones, and the H copy includes a frame-shifting deletion.
The pepC phylogeny
The St-genome sequences of Pseudoroegneria and Elymus form a well-supported clade (100% BS; Figure 1). The Elymus St sequences show little diversity overall, and are most closely related to the only native North American Pseudoroegneria species, P. spicata (85% BS), from which they show very little divergence. The only phylogenetic structure within the clade groups the two E. sibiricus sequences with E. brachyaristatus (70% BS), and one of the two E. dentatus sequences with E. mutabilis (84% BS). There is no phylogenetic distinction between the North American and Eurasian species.
The remaining Elymus pepC sequences represent the H-genome, and form a clade with part of Hordeum (Figure 1). The Elymus H-genome sequences, like the St sequences, show little diversity. They form a clade (78% BS) with sequences from H. californicum, a native North American diploid species, and with H. jubatum, a North American allotetraploid whose ITS sequences were derived from H. californicum and H. roshevitzii [89]. This clade includes three subgroups: most of Elymus along with H. californicum and one of the H. jubatum genomes (77% BS); the four remaining Elymus sequences (74% BS); and the second H. jubatum genome (95% BS). The four individuals in the smaller Elymus clade (E. mutabilis 1 and 2, E. dentatus 1, and E. caninus 2) are also represented in the larger H clade; thus, the small clade appears to reveal allelic variation in the H genome. The three species in this clade are Eurasian natives, so the allele may be restricted to Europe, though more intensive sampling would be required to support this.
The β-amylase phylogeny
Within the St-genome clade on the β-amylase tree (Figure 2; 99% BS), the Elymus sequences are again most closely related to P. spicata, though the support for this relationship is only moderate (72% BS). Compared to the pepC tree, there is more phylogenetic structure among the β-amylase St sequences of Elymus and P. spicata, including weak (64% BS) support for a Eurasian species clade within a paraphyletic North American Pseudoroegneria/Elymus assemblage. Intraspecific sampling is limited, but within the Eurasian clade, the St sequences representing E. dentatus 1 and 2 form a monophyletic group (62% BS), while those from E. caninus and from E. mutabilis do not. The St sequences representing the North American species E. wawawaiensis are non-monophyletic, and those from E. lanceolatus 1 and 2 are unresolved; the remaining species only have a single representative on the tree.
In contrast to the pepC tree, the H sequences from Elymus do not group with the diploid H. californicum on the β-amylase tree, but only with one of the H. jubatum genomes (100% BS). The H. californicum sequence forms a clade with the other H. jubatum genome (100% BS) within a large, multi-species Hordeum clade (92% BS) sister to (97% BS) the Elymus/H. jubatum clade. There is much less resolution among the Elymus H sequences than among the St sequences. The few relationships with >50% BS support are within-continent groups, but there is no suggestion of a Eurasian species clade, as there is in the St-sequence group.
The GBSSI phylogeny
The structure of the St-genome clade on the GBSSI tree differs from those on the other two trees. The group includes a paraphyletic “core” assemblage on short branches, in which the Elymus sequences are similar to those from five P. spicata individuals (1, 2, 3, 4, and 6). Within this assemblage, the Eurasian Elymus sequences form a moderately supported clade (76% BS) derived from within a paraphyletic group that includes P. spicata and six sequences from four North American Elymus species. This pattern by itself is similar to that in the β-amylase St clade, and is suggestive of a single origin of Eurasian species from within a paraphyletic group of North American Elymus and Pseudoroegneria species. The GBSSI St clade, however, is unique in having a separate subclade (87% BS) on a relatively long branch nested within the core group. The subclade includes the second of two sequences from P. spicata 6, the remaining ten North American Elymus sequences, and, surprisingly, two Chinese accessions of P. strigosa. Like P. spicata 6, three Elymus individuals have gene copies in both the paraphyletic assemblage of short branches, and in the long-branched subclade: E. lanceolatus 1 (clones a and d) and 2 (d and c), and E. wawawaiensis 3 (b and a). At the species level, E. virginicus is also represented in both groups, with individual 4 separated from individual 9. Another unique feature of the GBSSI tree is the placement of P. libanotica and P. tauri far outside of the main St-sequence clade, though their position within the tree is not convincingly resolved. The placement of these sequences as a possible result of diploid-level introgression, and their contribution to some of the species in a different group of Elymus tetraploids (genomes StStYY) is discussed elsewhere [62]. Based on the present sample, however, these species play no role in the evolution of the StStHH Elymus tetraploids.
The GBSSI H-genome clade, like the St-genome clade, is more complex than its counterparts on the pepC and β-amylase trees. All but one of the North American Elymus sequences form a well-supported clade (100% BS), along with H. californicum, one of the tetraploid H. jubatum genomes, and two Eurasian species (E. dentatus 1 and 2, and E. mutabilis 2). The remaining, mostly-Eurasian Elymus sequences form a very weak (28% BS) group with much greater sequence diversity. This assemblage includes a small clade of sequences from E. sibiricus 1 and 3 and E. mutabilis 1 (100% BS), and a larger clade (93% BS) with the second H. jubatum genome, two Eurasian Elymus species (E. brachyaristatus and E. caninus), and one accession of the North American species E. lanceolatus. The high diversity among the GBSSI H-genome sequences, compared to those on the pepC and β-amylase trees, was unexpected enough to raise suspicion about previously undetected alignment artifacts in the introns. There is considerable length variation in the GBSSI introns, so we ruled out the possible effects of non-orthologous intron alignment by running a ML analysis of just the H-genome clade using the exons only, for which the alignment is unambiguous. The resulting tree (Figure S1) has a similar (though less well-resolved) topology, including the same clade of very similar, mostly North American sequences, with the remaining mostly-Eurasian sequences forming a weak group of long branches outside of the main Elymus clade.
Discussion
The StStHH genome configuration
Our initial assumption that the Elymus species included here are StStHH-genome tetraploids was based on the results of numerous studies of meiotic chromosome pairing (e.g., [54] and references therein). As expected, all three gene trees unequivocally support the Pseudoroegneria + Hordeum origin of the sampled Elymus species, in agreement with an earlier analysis of RPB2 gene sequences [90]. Nearly all of the Elymus individuals have two distinct copies of all three genes, in clades with Pseudoroegneria and Hordeum. No individuals are missing either copy of all three genes; thus, the occasional missing copies from a few individuals might represent sampling artifacts, copy loss, or unique changes in primer sites, but do not suggest that either genome is absent altogether.
The pepC and β-amylase trees
The pepC and β-amylase trees are in general agreement with regard to the relationships among the Elymus species. Both show the North American and Eurasian species to be very closely related, and neither support independent origins for the two geographic groups. The St- and H-genome clades on both trees support a North American origin for the StStHH Elymus tetraploids. Of the Pseudoroegneria and Hordeum species included in the analysis, those most closely related to Elymus are North American species. Furthermore, within the St-sequence clade on the β-amylase tree, the Eurasian Elymus species form a clade (albeit weakly supported) within a broader paraphyletic assemblage of North American Elymus and Pseudoroegneria sequences. Evidence for a single, North American origin of the StStHH tetraploids is at odds with the earlier suggestion that the North American and Eurasian species arose separately, based on limited karyotype [91] and isoenzyme data [92], [93]. More recently, separate origins were moderately well-supported in a phylogenetic analysis of a nuclear RNA polymerase II (RPB2) gene [90]; H-genome RPB2 sequences separated Elymus into largely American and Eurasian subclades with, respectively, American and Eurasian Hordeum species. The RPB2 St-genome sequences did not reveal a geographic pattern, and were ambiguous with regard to whether the closest Pseudoroegneria species was North American or Eurasian, placing one accession each of P. spicata and P. stipifolia within Elymus, and a second accession of P. spicata outside of the Elymus clade on a very long branch. (While all of our trees implicate P. spicata as a potential donor to the StStHH tetraploids, none point to P. stipifolia.)
The pepC and β-amylase trees do differ with respect to the relationships between Elymus and Hordeum; specifically, the possible roles of diploid H. californicum and allotetraploid H. jubatum in the origin of StStHH Elymus. On the pepC tree, the H-genome sequences of Elymus are grouped with, and very similar to, the North American diploid H. californicum, and with both genomes of the allotetraploid H. jubatum. This is consistent with the results of an earlier study of repetitive DNA sequences [94], and one based on starch synthase data [29], both of which suggested H. californicum as a possible H-genome donor to Elymus. However, in contrast to the pepC tree, the β-amylase tree does not place Elymus with H. californicum, but instead with one of the H. jubatum genomes, while H. californicum is grouped with the other H. jubatum genome in a separate, multi-species Hordeum clade. Together, these trees and the differences between them suggest that a tetraploid similar to H. jubatum might have been involved in the history of Elymus, whether through past introgression among the Elymus, H. californicum, and H. jubatum lineages, or through a direct contribution from a tetraploid H. jubatum-like species to Elymus. Direct involvement of an H. jubatum-like ancestor would have led to the simultaneous introduction of both of its homoelogous H genomes, and in a successfully diploidized StStHH tetraploid, they would then behave as homologous alleles. Thus, depending on changes in allele frequency through time, Elymus might exhibit one or the other, or both, of the H. jubatum-like homoeologs.
The relationships among H-genome sequences could, instead, reflect introgression following tetraploid Elymus formation. It is impossible to trace a precise sequence of events, but we can envision scenarios consistent with the data. For example, if H. californicum was, in fact, the H-genome donor to StStHH Elymus, as suggested by the pepC tree, then H. californicum could be “misplaced” on the β-amylase tree, having acquired its β-amylase gene copy through introgression after the formation of Elymus. Alternatively, Elymus's placement on the β-amylase tree, far from H. californicum, might indicate that it was Elymus's β-amylase gene that was acquired through introgression; its close relationship to the second genome of H. jubatum indicates that species as a potential source. Additional samples of H. californicum and H. jubatum might support or refute our hypotheses, or suggest other conceivable scenarios, but in any case, it appears that H. jubatum was involved at some stage in the history of StStHH Elymus.
The GBSSI tree suggests further introgression or lineage sorting
If Elymus relationships are in some ways similar on the pepC and β-amylase trees, and relatively straightforward to interpret, the GBSSI results complicate the interpretation. In the St sequence clade, the paraphyletic “core” group of very similar sequences is, by itself, reminiscent of the pattern on the β-amylase tree, with the Eurasian species forming a moderately-supported clade within a paraphyletic group of North American Elymus and Pseudoroegneria species. This group, when considered alone, supports a North American origin of the StStHH tetraploid group from a P. spicata-like ancestor. However, the long-branched clade that arises from within the core group is unique to the GBSSI tree. It includes sequences from several North American Elymus individuals, some of which are also represented in the core group, and two Chinese accessions of a Eurasian Pseudoroegneria species, P. strigosa. The clade also includes one P. spicata sequence; the same individual (#6) has a second sequence in the core group with the rest of P. spicata. The dual placement of several individuals in both the core group and in the derived clade could reveal gene duplication. However, such a duplication event (at the base of the “St” clade; Figure 3) should also be evident in P. strigosa and in the Eurasian Elymus species, unless we postulate at least two subsequent, independent paralog losses. Thus, a more parsimonious explanation is that the relationships among GBSSI sequences from P. spicata, P. strigosa, and Elymus result from either past gene exchange or from the maintenance of a shared ancestral polymorphism. Introgression of a P. strigosa-like GBSSI allele into North America could explain the close relationship between P. strigosa and some of the sequences from P. spicata and North American Elymus, though the exact sequence of events is not clear from the present sample. The P. strigosa allele might have been introduced to North America through hybridization between P. spicata and P. strigosa, and then passed from P. spicata to North American Elymus through hybridization, or through formation of new StStHH tetraploids and subsequent hybridization among tetraploid lineages. The allele might have been introduced directly into Elymus through hybridization with a tetraploid (StStStSt) accession of P. strigosa. Given the possible Eurasian origin of the allele, we could also postulate an introduction via Eurasian Elymus tetraploids, but so far, none of the sampled Eurasian StStHH Elymus species have alleles in this clade. Alternatively, the P. spicata GBSSI polymorphism, including the allele in a close relationship with the P. strigosa sequences, could reflect the maintenance of ancestral polymorphism as a result of incomplete lineage sorting. The subsequent introduction of both alleles from P. spicata into North American Elymus is consistent with the placement of Elymus alleles with both P. strigosa and P. spicata on the GBSSI tree. The Eurasian Elymus species lack the polymorphism; assuming this is not merely a sampling artifact, it appears that the P. strigosa-like allele was either never introduced into the Eurasian group, or that it was subsequently lost.
In the GBSSI H-clade, there is, again, a “core” group of very similar sequences (though monophyletic in this case) that includes most of the North American and a few Eurasian Elymus sequences with H. californicum and one genome of the tetraploid H. jubatum. The relationships among the Elymus sequences in the core clade once again suggest a North American origin, with a clade of Eurasian sequences nested within a paraphyletic North American Elymus and Hordeum group. However, the arrangement of Elymus sequences outside of the core clade bears no resemblance to either of the other trees, or to the St-genome clade on this tree. These species are primarily Eurasian, except for one of the two accessions of E. lanceolatus. The only Hordeum sequences loosely associated with these Elymus sequences represent the second genome from the tetraploid H. jubatum; this provides some additional support that H. jubatum was somehow involved in the history of the StStHH Elymus species. This interpretation is not entirely satisfying, however, because the pattern is so unlike that on either of the other trees; these Elymus species are unexpectedly divergent from H. jubatum and from one another. Thus, there is no clear indication of a donor-recipient relationship among the sequences outside the core clade, whether from introgression or otherwise, so it is difficult to speculate on processes that might explain the topology of H-genome clade on the GBSSI tree. Consideration of the earlier RPB2 analysis [90] further complicates the interpretation. The Eurasian and American H-genome clades uncovered in that study were grouped with American and Eurasian representatives of Hordeum – H. stenostachys and H. bogdanii, respectively – but these were the only two Hordeum species included in the RPB2 analysis, so a potential role of H. californicum and H. jubatum in the origin of Elymus was not assessed.
Summary and Conclusion
Our first goal was to test the hypothesis that the North American and Eurasian Elymus species arose independently; this was suggested by allozyme [92], [93] and cytological [91] studies, and more explicitly supported by a recent molecular phylogenetic analysis of RPB2 [90]. However, our data support a single, North American origin. In the studies using cytological and allozyme data, issues of small sample size or the interpretation of complicated results could be invoked to explain away the contradiction, but the RPB2 results are more difficult to square with ours. We stand by the conclusion of a single origin because none of our three trees support separate origins. However, there are clearly other processes in play beyond the number of StStHH origins, and we have only four nuclear gene trees (including RPB2) in consideration, so additional trees could tip the interpretation in another direction. Our second goal, the identification of possible progenitors within Pseudoroegneria and Hordeum, yielded reasonably consistent results. Pseudoroegneria spicata is supported as the most likely St-genome donor among the species included on all of our trees (albeit with complications involving introgression or incomplete lineage sorting on the GBSSI tree), and as a possible progenitor on the RPB2 tree. Hordeum californicum and/or its tetraploid derivative H. jubatum are suggested as H-genome donors to Elymus on all three of our trees; past interactions among these species remain to be clarified with more targeted sampling of these species and their closest relatives within Hordeum, and additional gene trees. Finally, our third goal was to identify reticulate patterns beyond allopolyploidy. Examination of the GBSSI tree relative to the others revealed a case of introgression or incomplete lineage sorting, revealed by the discovery of P. strigosa-like St-genome allele in North American Elymus and Pseudoroegneria, and a more confusing situation involving the GBSSI H-genome.
Our attempts to propose evolutionary scenarios to reconcile conflicting patterns among Elymus gene trees reveal a general problem with inferring reticulate events from multiple conflicting trees. The interpretation depends heavily on which tree is initially assumed to be closest to the “best” or “true” tree (with or without a clear explanation), after which the differences on the remaining trees are attributed to processes such as introgression or incomplete lineage sorting. In other words, the sequence of historical events leading to gene tree conflict is defined by which tree is selected as a reference tree, or the one with which other trees are viewed to be in conflict. (Perhaps the most familiar examples in plants involve conflicts between ITS and cpDNA trees, which are generally interpreted as cpDNA introgression. In that case, the ITS tree is being used as the reference tree, sometimes without explicit justification.) Furthermore, if there are numerous conflicts among trees, a different tree could potentially be selected as the reference tree for each group of conflicting taxa. If many gene trees are available, and one branching pattern is shared by a large majority of them, then it is probably reasonable to interpret the differences on the few conflicting trees with reference to the majority of trees. With just three trees presented here (or four, including RBP2 [90]), and without a clear majority pattern, the distinction between “real” and “conflicting” relationships is not always straightforward. On one hand, for example, the unexpected placement of some Pseudoroegneria and Elymus St sequences with P. strigosa on the GBSSI tree, in conflict with the pepC and β-amylase trees, seems like a fairly straightforward case in which the GBSSI gene was affected by introgression or maintenance of an ancestral polymorphism. On the other hand, an assumption that H. californicum is the diploid H-genome donor to StStHH Elymus based on the pepC and GBSSI trees, and consequent interpretation of the β-amylase and RPB2 trees in terms of H-genome introgression, is more arbitrary. A proposed evolutionary scenario would be quite different if either the β-amylase or the RPB2 tree were assumed to represent the true tree with respect to H-genome sequence relationships. In spite of the difficulties with interpreting gene tree conflict as a specific sequence of evolutionary events, such patterns can certainly highlight the importance of such events, pinpoint the taxa involved, and yield hypotheses to be tested with targeted sampling within and among conflicting taxa, and with data from additional nuclear loci.
Supporting Information
File S1.
Sequence data set - phosphoenolpyruvate carboxylase.
https://doi.org/10.1371/journal.pone.0010989.s001
(0.15 MB TXT)
File S2.
Sequence data set - beta-amylase.
https://doi.org/10.1371/journal.pone.0010989.s002
(0.21 MB TXT)
File S3.
Sequence data set - granule-bound starch synthase I.
https://doi.org/10.1371/journal.pone.0010989.s003
(0.19 MB TXT)
Figure S1.
Gene tree based on granule-bound starch synthase exon sequences. The best-scoring ML tree was selected from 30 GARLI analyses of GBSSI exons under a GTR+I+Γ model of sequence evolution. The taxa are the same as in the H-genome clade from Figure 3, but the analysis differs in that introns were excluded. Numbers above branches show ML bootstrap support ≥50%. “NA” and “Eu” distinguish North American and Eurasian Elymus species, respectively. Numbers following taxon names distinguish individuals within species where applicable, and are consistent among Figures 1–3, S1. Letters following these numbers designate cloned sequences from within individuals, and are shared between Figures 3 and S1.
https://doi.org/10.1371/journal.pone.0010989.s004
(0.52 MB TIF)
Acknowledgments
Thanks to the USDA-ARS National Germplasm System for seeds, and to Barbara Whitlock for input on the manuscript. Three anonymous reviewers provided thoughtful reviews; their suggestions substantially improved the manuscript. Academic Editor Simon Joly provided an additional review, and we especially acknowledge his input regarding the complicated relationship between Hordeum and Elymus.
Author Contributions
Conceived and designed the experiments: RMG. Performed the experiments: RMG MMB MN. Analyzed the data: RMG MMB MN. Contributed reagents/materials/analysis tools: RMG. Wrote the paper: RMG.
References
- 1. Guggisberg A, Bretagnolle F, Mansion G (2006) Allopolyploid origin of the Mediterranean endemic, Centaurium bianoris (Gentianaceae), inferred by molecular markers. Systematic Botany 31: 368–379.
- 2. Hughes CE, Bailey CD, Harris SA (2002) Divergent and reticulate species relationships in Leucaena (Fabaceae) inferred from multiple data sources: insights into polyploid origins and nrDNA polymorphism. American Journal of Botany 89: 1057–1073.
- 3. Mansion G, Zeltner L, Bretagnolle F (2005) Phylogenetic patterns and polyploid evolution within the Mediterranean genus Centaurium (Gentianaceae-Chironieae). Taxon 54: 931–950.
- 4. Mummenhoff K, Franzke A, Koch MA (1997) Molecular phylogenetics of Thlaspi s.l. (Brassicaceae) based on chloroplast DNA restriction site variation and sequences of the internal transcribed spacers of nuclear ribosomal DNA. Canadian Journal of Botany 75: 469–482.
- 5. Seelanan T, Schnabel A, Wendel JF (1997) Congruence and consensus in the cotton tribe (Malvaceae). Systematic Botany 22: 259–290.
- 6. Vander Stappen J, De Laet J, Gama-López S, Van Campenhout S, Volckaert G (2002) Phylogenetic analysis of Stylostanthes (Fabaceae) based on the internal transcribed spacer region of nuclear ribosomal DNA. Plant Systematics and Evolution 234: 27–51.
- 7. Widmer A, Baltisberger M (1999) Molecular evidence for allopolyploid speciation and a single origin of the narrow endemic Draba ladina (Brassicaceae). American Journal of Botany 86: 1282–1289.
- 8. Escobar García P, Schönswetter P, Fuertes Aquilar J, Nieto Feliner G, Schneeweiss GM (2009) Five molecular markers reveal extensive morphological homoplasy and reticulate evolution in the Malva alliance (Malvaceae). Molecular Phylogenetics and Evolution 50: 226–239.
- 9. Kim S-C, Mejias JA, Lubinsky P (2008) Molecular confirmation of the hybrid origin of the critically endangered western Mediterranean endemic Sonchus pustulatus (Asteraceae: Sonchinae). Journal of Plant Research 121: 357–364.
- 10. Kim S-T, Donoghue MJ (2008) Incongruence between cpDNA and nrITS trees indicates extensive hybridization within Eupersicaria (Polygonaceae). American Journal of Botany 95: 1122–1135.
- 11. Liu S-C, Lu C-T, Wang J-C (2009) Reticulate hybridization of Alpinia (Zingiberaceae) in Taiwan. Journal of Plant Research 122: 305–316.
- 12. Stefanović S, Costea M (2008) Reticulate evolution in the parasitic genus Cuscuta (Convolvulaceae): over and over again. Botany 86: 791–808.
- 13.
Arnheim N (1983) Concerted evolution of multigene families. In: Nei M, Koehn RK, editors. Evolution of Genes and Proteins. Sunderland, MA: Sinauer. pp. 38–61.
- 14.
Hamby RK, Zimmer EA (1992) Ribosomal RNA as a phylogenetic tool in plant systematics. In: Soltis PS, Soltis DE, Doyle JJ, editors. Molecular Systematics of Plants. New York: Chapman and Hall. pp. 50–91.
- 15. Jorgensen RA, Cluster PD (1988) Modes and tempos in the evolution of nuclear ribosomal DNA: new characters for evolutionary studies and new markers for genetic and population studies. Annals of the Missouri Botanical Garden 75: 1238–1247.
- 16. Álvarez I, Wendel JF (2003) Ribosomal ITS sequences and plant phylogenetic inference. Molecular Phylogenetics and Evolution 29: 417–434.
- 17. Wendel JF, Schnabel A, Seelanen T (1995) Bi-directional interlocus concerted evolution following allopolyploid speciation in cotton (Gossypium). Proceedings of the National Academy of Sciences of the United States of America 92: 280–284.
- 18. Brysting AK, Oxelman B, Huber KT, Moulton V, Brochmann C (2007) Untangling complex histories of genome mergings in high polyploids. Systematic Biology 56: 467–476.
- 19. Doust AN, Penly AM, Jacobs SWL, Kellogg EA (2007) Congruence, conflict, and polyploidization shown by nuclear and chloroplast markers in the monophyletic “bristle clade” (Paniceae, Panicoideae, Poaceae). Systematic Botany 32: 531–544.
- 20. Cronn RC, Small RL, Haselkorn T, Wendel JF (2003) Cryptic repeated genomic recombination during speciation in Gossypium gossypioides. Evolution 57: 2475–2489.
- 21. Mason-Gamer RJ (2008) Allohexaploidy, introgression, and the complex phylogenetic history on Elymus repens (Poaceae). Molecular Phylogenetics and Evolution 47: 598–611.
- 22. Doyle JJ, Doyle JL, Brown AHD, Palmer RG (2002) Genomes, multiple origins, and lineage recombination in the Glycine tomentella (Leguminosae) polyploid complex: histone H3-D gene sequences. Evolution 56: 1388–1402.
- 23. Emshwiller E, Doyle JJ (2002) Origins of domestication and polyploidy in oca (Oxalis tuberosa: Oxalidaceae). 2. Chloroplast-expressed glutamine synthase data. American Journal of Botany 89: 1042–1056.
- 24. Friar EA, Prince LM, Cruse-Sanders JM, McGlaughlin ME, Butterworth CA, et al. (2008) Hybrid origin and genomic mosaicism of Dubautia scabra (Hawaiian silversword alliance; Asteraceae, Madiinae). Systematic Botany 33: 589–597.
- 25. Ge S, Sang T, Lu B-R, Hong D-Y (1999) Phylogeny of rice genomes with emphasis on origins of allotetraploid species. Proceedings of the National Academy of Sciences of the United States of America 96: 14400–14405.
- 26. Järvinen P, Palme A, Morales LO, Lannenpaa M, Keinanen M, et al. (2004) Phylogenetic relationships of Betula species (Betulaceae) based on nuclear ADH and chloroplast matK sequences. American Journal of Botany 91: 1834–1845.
- 27. Joly S, Starr JR, Lewis WH, Bruneau A (2006) Polyploid and hybrid evolution in roses east of the Rocky Mountains. American Journal of Botany 93: 412–425.
- 28. Lihová J, Shimizu KK, Marhold K (2006) Allopolyploid origin of Cardamine asarifolia (Brassicaceae): Incongruence between plastid and nuclear ribosomal DNA sequences solved by a single-copy nuclear gene. Molecular Phylogenetics and Evolution 39: 759–786.
- 29. Mason-Gamer RJ (2001) Origin of North American species of Elymus (Poaceae: Triticeae) allotetraploids based on granule-bound starch synthase gene sequences. Systematic Botany 26: 757–768.
- 30. Popp M, Erixon P, Eggens F, Oxelman B (2005) Origin and evolution of a circumpolar polyploid species complex in Silene (Caryophyllaceae) inferred from low copy nuclear RNA polymerase introns, rDNA, and chloroplast DNA. Systematic Botany 30: 302–313.
- 31. Rodríguez F, Spooner DM (2009) Nitrate reductase phylogeny of potato (Solanum sect. Petota) genomes with emphasis on the origins of the polyploid species. Systematic Botany 34: 207–219.
- 32. Smedmark JEE, Eriksson T, Bremer B (2005) Allopolyploid evolution in Geinae (Colurieae: Rosaceae) - building reticulate species trees from bifurcating gene trees. Organisms Diversity & Evolution 5: 275–283.
- 33. Kellogg EA, Appels R, Mason-Gamer RJ (1996) When gene trees tell different stories: the diploid genera of Triticeae (Gramineae). Systematic Botany 21: 321–347.
- 34. Mason-Gamer RJ, Kellogg EA (1996) Testing for phylogenetic conflict among molecular data sets in the tribe Triticeae (Gramineae). Systematic Biology 45: 524–545.
- 35. Löve Á (1984) Conspectus of the Triticeae. Feddes Repertorium 95: 425–521.
- 36.
Dewey DR (1984) The genomic system of classification as a guide to intergeneric hybridization within the perennial Triticeae. In: Gustafson JP, editor. Gene manipulation in plant improvement, Proc 16th Stadler genetics symposium. New York, NY: Plenum Publishing Company. pp. 209–279.
- 37. Dewey DR (1967) Synthetic hybrids of new world and old world Agropyrons: III. Agropyron repens X tetraploid Agropyron spicatum. American Journal of Botany 54: 93–98.
- 38. Dewey DR (1968) Synthetic Agropyron-Elymus hybrids: III. Elymus canadensis X Agropyron caninum, A. trachycaulum, and A. striatum. American Journal of Botany 55: 1133–1139.
- 39. Dewey DR (1970) Genome relations among Elymus canadensis, Elymus triticoides, Elymus dasystachyus, and Agropyron smithii. American Journal of Botany 57: 861–866.
- 40. Dewey DR (1974) Cytogenetics of Elymus sibiricus and its hybrids with Agropyron tauri, Elymus canadensis, and Agropyron caninum. Botanical Gazette 135: 80–87.
- 41. Jensen KB (1990) Cytology and morphology of Elymus pendulinus, E. pendulinus ssp. multiculmis, and E. parviglume (Poaceae: Triticeae). Botanical Gazette 151: 245–251.
- 42. Jensen KB (1993) Cytogenetics of Elymus magellanicus and its intra-and inter-generic hybrids with Pseudoroegneria spicata, Hordeum violaceum, E. trachycaulus, E. lanceolatus, and E. glaucus (Poaceae: Triticeae). Genome 36: 72–76.
- 43. Jensen KB (1996) Genome analysis of Eurasian Elymus thoroldianus, E. melantherus, and E. kokonoricus (Triticeae: Poaceae). International Journal of Plant Sciences 157: 136–141.
- 44. Lu B-R, Salomon B, von Bothmer R (1995) Interspecific hybridizations with Elymus confusus and E. dolichatherus, and their genomic relationships (Poaceae: Triticeae). Plant Systematics and Evolution 197: 1–17.
- 45. Lu B-R, von Bothmer R (1991) Production and cytogenetic analysis of the intergeneric hybrids between nine Elymus species and common wheat (Triticum aestivum L.). Euphytica 58: 81–95.
- 46. Lu B-R, von Bothmer R (1993) Meiotic analysis of Elymus caucasicus, E. longearistatus, and their interspecific hybrids with twenty-three Elymus species (Triticeae, Poaceae). Plant Systematics and Evolution 185: 35–53.
- 47. Salomon B (1993) Interspecific hybridizations in the Elymus semicostatus group (Poaceae). Genome 36: 889–905.
- 48. Salomon B, Lu B-R (1992) Genomic groups, morphology, and sectional delimitation in Eurasian Elymus (Poaceae, Triticeae). Plant Systematics and Evolution 180: 1–13.
- 49. Salomon B, Lu B-R (1994) Genomic relationships between species of the Elymus semicostatus group and Elymus sensu lato (Poaceae). Plant Systematics and Evolution 191: 199–211.
- 50. Dewey DR (1975) The origin of Agropyron smithii. American Journal of Botany 62: 524–530.
- 51. Chen Q, Conner RL, Laroche A, Thomas JB (1998) Genome analysis of Thinopyrum intermedium and Thinopyrum ponticum using genomic in situ hybridization. Genome 41: 580–586.
- 52. Liu Z-W, Wang RR-C (1993) Genome analysis of Elytrigia caespitosa, Lophopyrum nodosum, Pseudoroegneria geniculata ssp. scythia, and Thinopyrum intermedium (Triticeae, Gramineae). Genome 36: 102–111.
- 53. Zhang XY, Dong YS, Wang RR-C (1996) Characterization of genomes and chromosomes in partial amphiploids of the hybrid Triticum aestivum x Thinopyrum ponticum by in situ hybridization, isozyme analysis, and RAPD. Genome 39: 1062–1071.
- 54.
Dewey DR (1982) Genomic and phylogenetic relationships among North American perennial Triticeae. In: Estes JR, editor. Grasses and grasslands. Norman: Oklahoma University Press. pp. 51–81.
- 55. Mason-Gamer RJ (2007) Allopolyploids of the genus Elymus (Triticeae, Poaceae): a phylogenetic perspective. Aliso 23: 372–379.
- 56. Helfgott DM, Mason-Gamer RJ (2004) The evolution of North American Elymus (Triticeae, Poaceae) allotetraploids: evidence from phosphoenolpyruvate carboxylase gene sequences. Systematic Botany 29: 850–861.
- 57. Davis JI, Soreng RJ (2007) A preliminary phylogenetic analysis of the grass subfamily Pooideae (Poaceae), with attention to structural features of the plastid and nuclear genomes, including an intron loss in GBSSI. Aliso 23: 335–348.
- 58. Grass Phylogeny Working Group (2001) Phylogeny and subfamilial classification of the grasses (Poaceae). Annals of the Missouri Botanical Garden 88: 372–457.
- 59. González MC, Echevarría C, Vidal J, Cejudo FJ (2002) Isolation and characterization of a wheat phosphoenolpyruvate carboxylase gene. Modelling of the encoded protein. Plant Science 162: 233–238.
- 60. Mason-Gamer RJ (2004) Reticulate evolution, introgression, and intertribal gene capture in an allohexaploid grass. Systematic Biology 53: 25–37.
- 61. Mason-Gamer RJ (2005) The β-amylase genes of grasses and a phylogenetic analysis of the Triticeae (Poaceae). American Journal of Botany 92: 1045–1058.
- 62. Mason-Gamer RJ, Burns MM, Naum M (2010) Phylogenetic relationships and reticulation among Asian Elymus (Poaceae) allotetraploids: analysis of three nuclear genes. Molecular Phylogenetics and Evolution 54: 10–22.
- 63. Mason-Gamer RJ, Weil CF, Kellogg EA (1998) Granule-bound starch synthase: structure, function, and phylogenetic utility. Molecular Biology and Evolution 15: 1658–1673.
- 64. Rohde W, Becker D, Salamini F (1988) Structural analysis of the waxy locus from Hordeum vulgare. Nucleic Acids Research 16: 7185–7186.
- 65. Wagner A, Blackstone N, Cartwright P, Dick M, Misof B, et al. (1994) Surveys of gene families using polymerase chain reaction: PCR selection and PCR drift. Systematic Biology 43: 250–261.
- 66. Bradley RD, Hillis DM (1997) Recombinant DNA sequences generated by PCR amplification. Molecular Biology and Evolution 14: 592–593.
- 67. Cronn R, Cedroni M, Haselkorn T, Grover C, Wendel JF (2002) PCR-mediated recombination in amplification products derived from polyploid cotton. Theoretical and Applied Genetics 104: 482–489.
- 68. Judo MSB, Wedel AB, Wilson C (1998) Stimulation and suppression of PCR-mediated recombination. Nucleic Acids Research 26: 1819–1825.
- 69. Lepiniec L, Keryer E, Philippe H, Gadal P, Crétin C (1993) Sorghum phosphoenolpyruvate carboxylase gene family: structure, function, and molecular evolution. Plant Molecular Biology 21: 487–502.
- 70. Devos KM, Gale MD (1997) Comparative genetics in the grasses. Plant Molecular Biology 35: 3–15.
- 71. Ziegler P (1999) Cereal beta-amylases. Journal of Cereal Science 29: 195–204.
- 72. Sharp PJ, Kreis M, Shewry PR, Gale MD (1988) Location of β-amylase sequences in wheat and its relatives. Theoretical and Applied Genetics 75: 286–290.
- 73. Mason-Gamer RJ (2007) Multiple homoplasious insertions and deletions of a Triticeae (Poaceae) DNA transposon: a phylogenetic perspective. BMC Evolutionary Biology 7: 92.
- 74. Kleinhofs A (1997) Integrating barley RFLP and classical marker maps. Barley Genetics Newsletter 27: 105–112.
- 75. Korzun V, Malyshev S, Voymokov A, Börner A (1997) RFLP-mapping of three mutant loci in rye (Secale cereale L.) and their relation to homoeologous loci within the Gramineae. Theoretical and Applied Genetics 95: 468–473.
- 76. Frati F, Simon C, Sullivan J, Swofford DL (1997) Evolution of the mitochondrial cytochrome oxidase II gene in Collembola. Journal of Molecular Evolution 44: 145–158.
- 77. Sullivan J, Markert JA, Kilpatrick CW (1997) Phylogeography and molecular systematics of the Peromyscus aztecus species group (Rodentia: Muridae) inferred using parsimony and likelihood. Systematic Biology 46: 426–440.
- 78.
Swofford DL, Olsen GJ, Waddell PJ, Hillis DM (1996) Phylogenetic inference. In: Hillis DM, Moritz C, Mable BK, editors. Molecular systematics, 2nd edition. Sunderland: Sinauer. pp. 407–514.
- 79. Felsenstein J (1981) Evolutionary trees from DNA sequences: a maximum likelihood approach. Journal of Molecular Evolution 17: 368–376.
- 80. Huelsenbeck JP, Crandall KA (1997) Phylogeny estimation and hypothesis testing using maximum likelihood. Annual Review of Ecology and Systematics 28: 437–466.
- 81. Huelsenbeck JP, Rannala B (1997) Phylogenetic methods come of age: testing hypotheses in an evolutionary context. Science 276: 227–232.
- 82. Rodríguez F, Oliver JL, Marín A, Meduna JR (1990) The general stochastic model of nucleotide substitution. Journal of Theoretical Biology 142: 485–501.
- 83. Tavaré S (1986) Some probabilistic and statistical problems in the analysis of DNA sequences. Lectures on Mathematics in the Life Sciences 17: 57–86.
- 84.
Jukes TH, Cantor CR (1969) Evolution of protein molecules. In: Munro NH, editor. Mammalian protein metabolism. New York: Academic Press. pp. 21–132.
- 85. Kimura M (1980) A simple method for estimating evolutionary rate of base substitutions through comparative studies of nucleotide sequences. Journal of Molecular Evolution 16: 111–120.
- 86. Hasegawa M, Kishino H, Yano T (1985) Dating the human-ape split by a molecular clock of mitochondrial DNA. Journal of Molecular Evolution 22: 160–170.
- 87. Yang Z (1993) Maximum-likelihood estimation of phylogeny from DNA sequences when substitution rates differ over sites. Molecular Biology and Evolution 10: 1396–1401.
- 88.
Zwickl DJ (2006) Genetic algorithm approaches for the phylogenetic analysis of large biological sequence datasets under the maximum likelihood criterion.PhD Thesis. Austin: The University of Texas.
- 89. Blattner FR (2004) Phylogenetic analysis of Hordeum (Poaceae) as inferred by nuclear rDNA ITS sequences. Molecular Phylogenetics and Evolution 33: 289–299.
- 90. Sun G-L, Ni Y, Daley T (2008) Molecular phylogeny of RPB2 gene reveals multiple origin, geographic distribution of H genome, and the relationship of the Y genome to other genomes of Elymus species. Molecular Phylogenetics and Evolution 46: 897–907.
- 91. Linde-Laursen I, Seberg O, Salomon B (1994) Comparison of the Giemsa C-banded and N-banded karyotypes of two Elymus species, E. dentatus and E. glaucescens (Poaceae: Triticeae). Plant Systematics and Evolution 192: 165–176.
- 92.
Jaaska V (1998) Isoenzyme data on the origin of North American allotetraploid Elymus species. In: Jaradat AA, editor. Triticeae III. Enfield, New Hampshire: Science Publishers, Inc. pp. 209–216.
- 93. Jaaska V (1992) Isoenzyme variation in the grass genus Elymus (Poaceae). Hereditas 117: 11–22.
- 94. Tsujimoto H, Gill BS (1991) Repetitive DNA sequences from polyploid Elymus trachycaulus and the diploid progenitor species: detection and genomic affinity of Elymus chromatin added to wheat. Genome 34: 782–789.