850

Systematic Botany (2004) 29(4) pp 850ndash861q Copyright 2004 by the American Society of Plant Taxonomists

The Evolution of North American Elymus (Triticeae Poaceae)Allotetraploids Evidence from Phosphoenolpyruvate Carboxylase

Gene Sequences

D MEGAN HELFGOTT1 and ROBERTA J MASON-GAMER

University of Illinois at Chicago Department of Biological Sciences (MC 066) 845 West Taylor Street ChicagoIllinois 60607

1Author for Correspondence Current address Department of Biological Sciences 2325 North CliftonMcGowan Rm 118 DePaul University Chicago Illinois 60614 (dhelfgotdepauledu)

Communicating Editor Gregory M Plunkett

ABSTRACT Cytogenetic studies of North American Elymus suggest that the genus is an allopolyploid derivative of Pseu-doroegneria (St) and Hordeum (H) To test this we conducted a phylogenetic analysis of North American Elymus species withina broad sample of diploid Triticeae taxa using cloned sequences from one member of the nuclear gene family encodingphosphoenolpyruvate carboxylase The phylogeny supports the hypothesis that Pseudoroegneria and Hordeum are the diploidprogenitors of the North American Elymus tetraploids Each tetraploid Elymus individual has two distinct forms of the geneand each form is in a strongly supported clade with sequences from either Pseudoroegneria or Hordeum suggesting that theseElymus species have an St 1 H genomic content This pattern is consistent with a single (or multiple very similar) polyploidancestor(s) of the North American tetraploids confirming earlier results based on granule-bound starch synthase I genesequence data We also examined the utility of the phosphoenolpyruvate carboxylase gene to reconstruct the evolutionaryhistory of the Triticeae by comparing it to starch synthase gene sequence data Both nuclear data sets are phylogeneticallyinformative but suggest somewhat different evolutionary histories among genera within the tribe

Allopolyploid taxa can be difficult to interpret with-in phylogenies of their diploid relatives The problemswith placing hybrid taxa within the bifurcating treesthat are obtained from the most common methods ofphylogenetic analysis have been discussed (eg Hull1979 Cronquist 1987) and examined experimentally(eg McDade 1992 1997) Analyses of individual genetrees however allow these problems to be circum-vented as long as their tree-like history has not beendisrupted by recombination Thus molecular phylo-genetic data have recently revealed the reticulate his-tories of several polyploid species or groups includingfor example Gossypium L (eg Cronn et al 1996 2003Seelanan et al 1997 Small and Wendel 2000) Geum L(Smedmark et al 2003) Glycine Willd (eg Doyle etal 2002 Rauscher et al 2002) Oxalis L (Emshwillerand Doyle 1998 2002) Oryza L (Ge et al 1999) andPaeonia L (eg Sang and Zhang 1999)

In the Triticeae data from both chloroplast (Mason-Gamer and Kellogg 1996a Petersen and Seberg 1998Mason-Gamer et al 2002) and nuclear genes (Hsiao etal 1995 Kellogg and Appels 1995 Petersen and Seberg2002) have greatly increased overall understanding ofthe evolution of the diploid members of the tribe re-vealing surprising patterns of reticulate evolution (egKellogg et al 1996) The polyploids however havebeen the focus of far fewer molecular phylogeneticstudies yet they may hold the key to understandingthe history of the tribe as a whole Of the approxi-mately 350 Triticeae species whose chromosome num-ber is known 75 are of polyploid origin (eg Love1984) The present study focuses on North Americanmembers of Elymus L an entirely polyploid genus of

about 150 species found in temperate regions throughmuch of the world

The circumscription of Elymus varies greatly amongtreatments In North America one widely used defi-nition follows the Manual of Grasses of the United States(Hitchcock 1951) and is based on morphological char-acteristics including multiple spikelets per node and anon-disarticulating rachis The genomic definition ofElymus (eg Dewey 1984 Love 1984 Barkworth andDewey 1985) on the other hand is based on presumedoverall genome similarity deduced from the degree ofchromosome pairing in meiotic cells of interspecifichybrids Under this system the circumscription ofNorth American Elymus is very different from the mor-phology-based circumscription (Barkworth and Dewey1985 Fig 1 of Mason-Gamer 2001) Within the geno-mic classification system Elymus is defined as thoseallopolyploid species with at least one set of Pseudo-roegneria (Nevski) ALove (St) genomes When consid-ered worldwide Elymus may be tetraploid hexaploidor octoploid and may combine the St genome withthe H genome from Hordeum L P from AgropyronGaertner W from Australopyrum (Tzvelev) ALove andY from an unknown donor (eg Dewey 1984) TheNorth American native Elymus species which are thefocus of this study are nearly all tetraploid with a pre-sumed StStHH genomic complement Data from allo-zymes (Jaaska 1995 1998) and from the nuclear waxygene encoding granule-bound starch synthase I(GBSSI) were in agreement that Elymus is derivedfrom Pseudoroegneria and Hordeum (Mason-Gamer2001) but the demonstrated gene tree conflict withinthe Triticeae (Kellogg et al 1996 Mason-Gamer and

2004] 851HELFGOTT AND MASON-GAMER ELYMUS PHYLOGENY

FIG 1 A map of the PepC region (exons 1ndash4) used in thisstudy

Kellogg 1996b) precludes drawing strong phylogeneticconclusions from the analysis of a single gene

Data from an additional nuclear marker are neededto confirm the genomic content of North American Ely-mus We have turned to sequence data from the geneencoding phosphoenolpyruvate carboxylase (PepC EC41131) The PepC genes form a small multigene fam-ily in the plants studied so far (eg Cushman and Boh-nert 1989a 1989b Cretin et al 1991 Kawamura et al1992 Lepiniec et al 1992 1993 Ernst and Westhoff1997) Sorghum vulgare Pers was the first species forwhich all three genes in the PepC gene family werecharacterized (Lepiniec et al 1993) Members of thefamily are approximately 6 kb in length and includeten exons (Lepiniec et al 1993) Each copy codes for afunctionally distinct PepC isoform including a widely-expressed housekeeping form an inducible form ex-pressed in the root and a light-regulated form in-volved in C4 photosynthesis (eg Lepiniec et al 1994)Several PepC isoforms have been described in othergrasses as well including maize and wheat (Izui et al1986 Hudspeth and Grula 1989 Kawamura et al 1992Gonzalez et al 1998)

Sequence data from the PepC gene have alreadybeen shown to be phylogenetically useful at a numberof taxonomic levels Gehrig et al (2001) constructedphylogenies using PepC sequence data spanning theplant kingdom to gain insights into the evolution ofthe PepC gene family itself The phylogenetic analysesof all available PepC sequences from seed plants bySvensson et al (2003) revealed that grass sequencesform three distinct clades relative to the other seedplant sequences suggesting that the grass isoformsrepresent gene duplications that occurred prior to theorigin of grasses Besnard et al (2002) used sequencesfrom C4-specific isoforms of PepC to confirm the rela-tionships among grass tribes that had been previouslyproposed based on other molecular markers Malcom-ber (2002) employed PepC sequence data to examineevolutionary relationships within the genus GaertneraLam (Rubiaceae)

The utility of single- and low-copy genes for phy-logenetic analyses is increased if the genes are wellcharacterized and if there are enough available se-quence data for designing amplification primers Theavailability of sequences representing multiple PepCisoforms from several diverse grasses allowed us todevelop primers specific to a single PepC gene copythe apparent ortholog of the housekeeping form in Sor-

ghum (GenBank accession X59925 Lepiniec et al 1991)and Zea (X61489 Kawamura et al 1992) The presentphylogenetic analysis based on a portion of this geneaddresses two objectives 1) to confirm the genomicconstitution of North American Elymus species and 2)to examine the phylogenetic utility of the housekeep-ing isoform of the PepC gene by comparing PepC andwaxy gene sequence data from the Triticeae

MATERIALS AND METHODS

Taxon Sampling Sequences from the PepC gene were sampledfrom 27 species (Table 1) Ingroup members included seven speciesof Elymus multiple accessions of the putative parent species inHordeum and Pseudoroegneria and a broad sample of other mono-genomic genera in the Triticeae Bromus tectorum L (Bromeae) wasincluded as an outgroup (Kellogg 1992 Davis and Soreng 1993)Sequences were submitted to GenBank under the accession num-bers AY553236-AY553269 and AY548399-AY548432 and the datamatrixes were deposited at TreeBASE (study accession numberS1053 matrix accession numbers M1794-M1795)

Molecular Data Using three grass sequences from GenBankwe designed primers to amplify and sequence a 17-kb portion ofthe housekeeping PepC gene (Table 2) The data set consists of twonon-contiguous regions corresponding to base pairs 526ndash1632 and1849ndash2441 of Triticum aestivum (Genbank accession AJ007705) (Fig1) This copy of PepC is similar to sequences mapped to chromo-some 9 of the rice genome (AP005802 and AP005781) and usingGale and Devosrsquos (1998) map of twelve grass genomes we pos-tulate that it is on the long arm of the group 5 Triticeae chromo-somes

DNA Extraction Amplificationand Sequencing Total genomicDNA was isolated from individual plants following Doyle andDoyle (1987) Gene amplification was performed following Mason-Gamer et al (1998) with the addition of 5 DMSO In all casesPepC regions 1 and 2 were amplified separately We used PepCprimers 219F or 467F(1) and 1672R(1) or 1672R(2) to amplify re-gion 1 and PepC 1827F and 2443R for region 2 (Table 2) BecauseElymus species are presumed allopolyploids amplification reac-tions were run in triplicate and combined before cloning to coun-ter the potential effects of PCR drift (Wagner et al 1994 Mason-Gamer 2001)

The PCR products were purified using Geneclean (Bio101) ac-cording to the manufacturerrsquos protocol and the cloning reactionswere carried out using Promegarsquos pGem-T Easy cloning kit ac-cording to instructions except that the final volumes of the ligationand transformation reactions were halved Target colonies wereisolated using QIAGEN or Promega miniprep kits In a few casescloned inserts were amplified directly from colonies using theoriginal PCR primers and cleaned using exonuclease I (Exo) andshrimp alkaline phosphatase (SAP) as in Mason-Gamer (2004) Forthree diploid taxa (Bromus tectorum Secale cereale S montanumsubsp montanum) PepC region 2 was sequenced directly from apooled set of PCR products cleaned using the Exo-SAP protocol

Individuals were sequenced using ABI Big Dye Terminators ac-cording to the manufacturerrsquos instructions except that sequencingreagent volumes were quartered and the final reaction volumewas 10 mL Both strands were sequenced for all taxa The sequenc-ing reactions were visualized using an ABI model Prism-377 orMJ Research Basestation automated sequencer then assembled us-ing Sequencher vers 30 or 41 (Gene Codes Corp)

In Elymus we initially screened clones for sequence variationusing a single sequencing primer (PepC-467F for region 1 or PepC-1827F for region 2) until both St and H sequence types werefound or until all clones generated for a specific individual wereexamined Once the St and H clones were found within an indi-vidual both forms were completely sequenced using both internaland external primers Then all sequences were carefully inspectedfor mosaic sequence patterns combining the St and H genome inone recombinant molecule

852 [Volume 29SYSTEMATIC BOTANY

TABLE 1 List of taxa collection information and GenBank numbers Voucher numbers containing names are collectorsrsquo accessionnumbers Voucher H 5555 is from the Swedish-Danish Triticeae Consortium sample provided by Dr O Seberg The remaining lsquolsquoPIrsquorsquovoucher numbers are Plant Introduction accessions from the National Plant Germplasm System (US Dept Agriculture httpwwwars-gringovnpgs) samples were provided by the USDA-ARS Data are presented in the following sequence species name PepCvoucher Clone Region-I Clone Region-II Elymus genome GenBank No Region-I Region-II Used in PepC-waxy comparison (yesno) waxy voucher waxy GenBank No

Elymus E californicus (Bolander) Gould Barkworth sn (UTC) 1645B 637C NA AY553241 AY548404 Yes Same AY011012E canadensis L PI 531568 (ID) 973A 534A St AY553248 AY548411 Yes Barkworth 97-86 (UTC) AY556481 E canadensis PI531568 (ID) 973B 534F H AY553242 AY548405 No mdash mdash E canadensis PI 578675 (ID) 215C 48C H AY553243 AY548406No mdash mdash E elymoides (Rafin) Swezey PI 531606 (ID) 648C 632C St AY553249 AY548412 Yes Same AY010992 E elymoidesPI 531606 (ID) 957A 632A H AY553244 AY548407 Yes Same AY010965 E glaucus Buckley W6 10215 (ID) 219A 51A StAY553250 AY548413 Yes MasonGamer 130 (ID) AY010979 E hystrix L Barkworth 97-87 (UTC) 981C 622C St AY553251AY548414 Yes Same AY010982 E hystrix Barkworth 97-87 (UTC) 981B 622A H AY553245 AY548408 No mdash mdash E lanceolatus(Scribn amp Smith) Gould PI 531623 (ID) 1049B 604B St AY553252 AY548415 Yes W6 14220 (ID) AY010984 E lanceolatus PI531623 (ID) 1049C 604J H AY553246 AY548409 Yes Same AY010969 E wawawaiensis J Carlson ex Barkworth PI 598812(ID) 229A 74A St AY553253 AY548416 Yes Same AY010990 E wawawaiensis PI 598812 (ID) 229B 74P H AY553247AY548410 Yes Same AY010978

Triticeae Aegilops comosa Sm G 602 (GH) 1305A 1296A N A AY553236 AY548399 Yes Same AF079263 Agropyroncristatum (L) Gaertn PI 279802 (GH) 1242B 540A N A AY553237 AY548400 Yes Same AF079271 Australopyrum velutinum(Nees) B Simon D 2873-2878 (GH) 1209B 1128A N A AY553238 AY548401 Yes Same AY011004 Dasypyrum villosum (L)Candargy D 2990 (GH) 1641A 1134A N A AY553240 AY548403 Yes PI 470279 (GH) AY556480 Eremopyrum orientale (L)Jaub amp Spach H 5555 (GH) 884D 543E N A AY553254 AY548417 Yes Same AY011007 Heteranthelium piliferum (Banks ampSol) Hochst PI 402352 (GH) 1301B 1292A N A AY553255 AY548418 Yes Same AF079277 Hordeum brachyantherum subspcalifornicum (Cov amp Steb) Bothmer MA-138-1-40 (GH) 339B 312B N A AY553256 AY548419 Yes Same AF079273 H jubatumL MasonGamer 106 (ID) 940C 316A N A AY553257 AY548420 Yes Same AY010963 H marinum Hudson PI 304346 (ID)1246C 301A N A AY553258 AY548421 Yes Same AY010959 H murinum L Ciho 15683 (ID) 989A 297A N A AY553259AY548422 Yes Same AY010960 H vulgare L MasonGamer 107 (ID) 1346BB 1314A N A AY553260 AY548423 Yes Rohdeet al 1988 X07931 Leymus racemosus subsp sabulosus (MBieb) Tzvelev R-20-21ndash25 (GH) 1592B 1718A N A AY553261AY548424 No mdash mdash Peridictyon sanctum (Janka) Seberg Fred amp Baden Jensen 248 (GH) 1341A 1157A N A AY553262AY548425 Yes Same AF079278 Pseudoroegneria spicata (Pursh) A Love PI 610986 (ID) 352A 325A N A AY553263 AY548426Yes Same AY010999 P spicata D 2844 (GH) 804B 610A N A AY553264 AY548427 Yes Same AY011000 Secale cereale LKellogg sn (GH) 2729B 2766 N A AY553266 AY548429 Yes Same AY011009 S montanum subsp montanum Guss T 36554(GH) 2734C 2770 N A AY553267 AY548430 No mdash mdash S montanum subsp anatolicum (Boiss) Tzelev PI 206992 (GH) 2739A1145A N A AY553265 AY548428 Yes PI 206991 (GH) AY011008 Taeniatherum caput-medusae (L) Nevski MasonGamer 189d(ID) 1588A 1141A N A AY553268 AY548431 Yes PI 208075 (ID) AY011010 Thinopyrum elongatum (Host) DRDeweyMasonGamer 113 (ID) 1460A 1116F N A AY553269 AY548432 Yes PI 531719 (GH) AF079284 Triticum aestivum L mdash mdashmdash N A AJ007705 AJ007705 No mdash mdash

Bromeae Bromus tectorum L Kellogg sn (GH) 1662C 2774 N A AY553239 AY548402 Yes Same AY362757

TABLE 2 List of primers

Primer NameRelationship to Triticum

aestivum AJ007705 Primer (59ndash39)

Region 1 (1ndash1562) 526ndash1632PepC-291FPepC-467F(1)PepC-796RPepC-952R

219ndash240467ndash489796ndash813952ndash973

ACTCCTGCCATCCGCCTTCTATGCTGCTCGTCCCCGCCAAGGTGTAAAAACACCAGATTACGCACAATGACTGACACGATTTGAGATTC

PepC-1387RPepC-1555RPepC-1672R(1)PepC-1672R(2)

1387ndash14081555ndash15751672ndash16961672ndash1696

GTCAAGGCATAGTCGTTTCAAGAATCCATCAATCAACATAGAGAGCTTGTTATCATCTTCCCGAGTTCAGCTTGTTATCATCTCTCCGAGTTCA

Region II (1566ndash2185) 1849ndash2441PepC-1827FPepC-2443R

1827ndash18482443ndash2464

ARAYTCRGCAATCACAGAATCTRATCCCAATGTTCTTCAATGC

Sequence Analysis Boundaries between the PepC introns andexons were determined by comparison with published Triticumaestivum (AJ007705) and Sorghum vulgare (X59925) sequences Se-quences were initially aligned using the default setting in ClustalW ver 15 (Thompson et al 1994) then adjusted manually usingMacClade ver 4 (Maddison and Maddison 2002) Three regions(35 characters alignment positions 393ndash403 996ndash972 and 1280ndash

1296) were judged to have an ambiguous alignment and were ex-cluded prior to phylogenetic analyses Percent of gap charactersmissing characters and nucleotide composition for each specieswere determined using MacClade

Phylogeny Reconstruction Maximum likelihood (ML) meth-ods were used to generate hypotheses about relationships amongtaxa using PAUP ver 40b10 (Swofford 2002) Initially maximum

2004] 851HELFGOTT AND MASON-GAMER ELYMUS PHYLOGENY

FIG 1 A map of the PepC region (exons 1ndash4) used in thisstudy

Kellogg 1996b) precludes drawing strong phylogeneticconclusions from the analysis of a single gene

Data from an additional nuclear marker are neededto confirm the genomic content of North American Ely-mus We have turned to sequence data from the geneencoding phosphoenolpyruvate carboxylase (PepC EC41131) The PepC genes form a small multigene fam-ily in the plants studied so far (eg Cushman and Boh-nert 1989a 1989b Cretin et al 1991 Kawamura et al1992 Lepiniec et al 1992 1993 Ernst and Westhoff1997) Sorghum vulgare Pers was the first species forwhich all three genes in the PepC gene family werecharacterized (Lepiniec et al 1993) Members of thefamily are approximately 6 kb in length and includeten exons (Lepiniec et al 1993) Each copy codes for afunctionally distinct PepC isoform including a widely-expressed housekeeping form an inducible form ex-pressed in the root and a light-regulated form in-volved in C4 photosynthesis (eg Lepiniec et al 1994)Several PepC isoforms have been described in othergrasses as well including maize and wheat (Izui et al1986 Hudspeth and Grula 1989 Kawamura et al 1992Gonzalez et al 1998)

Sequence data from the PepC gene have alreadybeen shown to be phylogenetically useful at a numberof taxonomic levels Gehrig et al (2001) constructedphylogenies using PepC sequence data spanning theplant kingdom to gain insights into the evolution ofthe PepC gene family itself The phylogenetic analysesof all available PepC sequences from seed plants bySvensson et al (2003) revealed that grass sequencesform three distinct clades relative to the other seedplant sequences suggesting that the grass isoformsrepresent gene duplications that occurred prior to theorigin of grasses Besnard et al (2002) used sequencesfrom C4-specific isoforms of PepC to confirm the rela-tionships among grass tribes that had been previouslyproposed based on other molecular markers Malcom-ber (2002) employed PepC sequence data to examineevolutionary relationships within the genus GaertneraLam (Rubiaceae)

The utility of single- and low-copy genes for phy-logenetic analyses is increased if the genes are wellcharacterized and if there are enough available se-quence data for designing amplification primers Theavailability of sequences representing multiple PepCisoforms from several diverse grasses allowed us todevelop primers specific to a single PepC gene copythe apparent ortholog of the housekeeping form in Sor-

ghum (GenBank accession X59925 Lepiniec et al 1991)and Zea (X61489 Kawamura et al 1992) The presentphylogenetic analysis based on a portion of this geneaddresses two objectives 1) to confirm the genomicconstitution of North American Elymus species and 2)to examine the phylogenetic utility of the housekeep-ing isoform of the PepC gene by comparing PepC andwaxy gene sequence data from the Triticeae

MATERIALS AND METHODS

Taxon Sampling Sequences from the PepC gene were sampledfrom 27 species (Table 1) Ingroup members included seven speciesof Elymus multiple accessions of the putative parent species inHordeum and Pseudoroegneria and a broad sample of other mono-genomic genera in the Triticeae Bromus tectorum L (Bromeae) wasincluded as an outgroup (Kellogg 1992 Davis and Soreng 1993)Sequences were submitted to GenBank under the accession num-bers AY553236-AY553269 and AY548399-AY548432 and the datamatrixes were deposited at TreeBASE (study accession numberS1053 matrix accession numbers M1794-M1795)

Molecular Data Using three grass sequences from GenBankwe designed primers to amplify and sequence a 17-kb portion ofthe housekeeping PepC gene (Table 2) The data set consists of twonon-contiguous regions corresponding to base pairs 526ndash1632 and1849ndash2441 of Triticum aestivum (Genbank accession AJ007705) (Fig1) This copy of PepC is similar to sequences mapped to chromo-some 9 of the rice genome (AP005802 and AP005781) and usingGale and Devosrsquos (1998) map of twelve grass genomes we pos-tulate that it is on the long arm of the group 5 Triticeae chromo-somes

DNA Extraction Amplificationand Sequencing Total genomicDNA was isolated from individual plants following Doyle andDoyle (1987) Gene amplification was performed following Mason-Gamer et al (1998) with the addition of 5 DMSO In all casesPepC regions 1 and 2 were amplified separately We used PepCprimers 219F or 467F(1) and 1672R(1) or 1672R(2) to amplify re-gion 1 and PepC 1827F and 2443R for region 2 (Table 2) BecauseElymus species are presumed allopolyploids amplification reac-tions were run in triplicate and combined before cloning to coun-ter the potential effects of PCR drift (Wagner et al 1994 Mason-Gamer 2001)

The PCR products were purified using Geneclean (Bio101) ac-cording to the manufacturerrsquos protocol and the cloning reactionswere carried out using Promegarsquos pGem-T Easy cloning kit ac-cording to instructions except that the final volumes of the ligationand transformation reactions were halved Target colonies wereisolated using QIAGEN or Promega miniprep kits In a few casescloned inserts were amplified directly from colonies using theoriginal PCR primers and cleaned using exonuclease I (Exo) andshrimp alkaline phosphatase (SAP) as in Mason-Gamer (2004) Forthree diploid taxa (Bromus tectorum Secale cereale S montanumsubsp montanum) PepC region 2 was sequenced directly from apooled set of PCR products cleaned using the Exo-SAP protocol

Individuals were sequenced using ABI Big Dye Terminators ac-cording to the manufacturerrsquos instructions except that sequencingreagent volumes were quartered and the final reaction volumewas 10 mL Both strands were sequenced for all taxa The sequenc-ing reactions were visualized using an ABI model Prism-377 orMJ Research Basestation automated sequencer then assembled us-ing Sequencher vers 30 or 41 (Gene Codes Corp)

In Elymus we initially screened clones for sequence variationusing a single sequencing primer (PepC-467F for region 1 or PepC-1827F for region 2) until both St and H sequence types werefound or until all clones generated for a specific individual wereexamined Once the St and H clones were found within an indi-vidual both forms were completely sequenced using both internaland external primers Then all sequences were carefully inspectedfor mosaic sequence patterns combining the St and H genome inone recombinant molecule

852 [Volume 29SYSTEMATIC BOTANY

TABLE 1 List of taxa collection information and GenBank numbers Voucher numbers containing names are collectorsrsquo accessionnumbers Voucher H 5555 is from the Swedish-Danish Triticeae Consortium sample provided by Dr O Seberg The remaining lsquolsquoPIrsquorsquovoucher numbers are Plant Introduction accessions from the National Plant Germplasm System (US Dept Agriculture httpwwwars-gringovnpgs) samples were provided by the USDA-ARS Data are presented in the following sequence species name PepCvoucher Clone Region-I Clone Region-II Elymus genome GenBank No Region-I Region-II Used in PepC-waxy comparison (yesno) waxy voucher waxy GenBank No

Elymus E californicus (Bolander) Gould Barkworth sn (UTC) 1645B 637C NA AY553241 AY548404 Yes Same AY011012E canadensis L PI 531568 (ID) 973A 534A St AY553248 AY548411 Yes Barkworth 97-86 (UTC) AY556481 E canadensis PI531568 (ID) 973B 534F H AY553242 AY548405 No mdash mdash E canadensis PI 578675 (ID) 215C 48C H AY553243 AY548406No mdash mdash E elymoides (Rafin) Swezey PI 531606 (ID) 648C 632C St AY553249 AY548412 Yes Same AY010992 E elymoidesPI 531606 (ID) 957A 632A H AY553244 AY548407 Yes Same AY010965 E glaucus Buckley W6 10215 (ID) 219A 51A StAY553250 AY548413 Yes MasonGamer 130 (ID) AY010979 E hystrix L Barkworth 97-87 (UTC) 981C 622C St AY553251AY548414 Yes Same AY010982 E hystrix Barkworth 97-87 (UTC) 981B 622A H AY553245 AY548408 No mdash mdash E lanceolatus(Scribn amp Smith) Gould PI 531623 (ID) 1049B 604B St AY553252 AY548415 Yes W6 14220 (ID) AY010984 E lanceolatus PI531623 (ID) 1049C 604J H AY553246 AY548409 Yes Same AY010969 E wawawaiensis J Carlson ex Barkworth PI 598812(ID) 229A 74A St AY553253 AY548416 Yes Same AY010990 E wawawaiensis PI 598812 (ID) 229B 74P H AY553247AY548410 Yes Same AY010978

Triticeae Aegilops comosa Sm G 602 (GH) 1305A 1296A N A AY553236 AY548399 Yes Same AF079263 Agropyroncristatum (L) Gaertn PI 279802 (GH) 1242B 540A N A AY553237 AY548400 Yes Same AF079271 Australopyrum velutinum(Nees) B Simon D 2873-2878 (GH) 1209B 1128A N A AY553238 AY548401 Yes Same AY011004 Dasypyrum villosum (L)Candargy D 2990 (GH) 1641A 1134A N A AY553240 AY548403 Yes PI 470279 (GH) AY556480 Eremopyrum orientale (L)Jaub amp Spach H 5555 (GH) 884D 543E N A AY553254 AY548417 Yes Same AY011007 Heteranthelium piliferum (Banks ampSol) Hochst PI 402352 (GH) 1301B 1292A N A AY553255 AY548418 Yes Same AF079277 Hordeum brachyantherum subspcalifornicum (Cov amp Steb) Bothmer MA-138-1-40 (GH) 339B 312B N A AY553256 AY548419 Yes Same AF079273 H jubatumL MasonGamer 106 (ID) 940C 316A N A AY553257 AY548420 Yes Same AY010963 H marinum Hudson PI 304346 (ID)1246C 301A N A AY553258 AY548421 Yes Same AY010959 H murinum L Ciho 15683 (ID) 989A 297A N A AY553259AY548422 Yes Same AY010960 H vulgare L MasonGamer 107 (ID) 1346BB 1314A N A AY553260 AY548423 Yes Rohdeet al 1988 X07931 Leymus racemosus subsp sabulosus (MBieb) Tzvelev R-20-21ndash25 (GH) 1592B 1718A N A AY553261AY548424 No mdash mdash Peridictyon sanctum (Janka) Seberg Fred amp Baden Jensen 248 (GH) 1341A 1157A N A AY553262AY548425 Yes Same AF079278 Pseudoroegneria spicata (Pursh) A Love PI 610986 (ID) 352A 325A N A AY553263 AY548426Yes Same AY010999 P spicata D 2844 (GH) 804B 610A N A AY553264 AY548427 Yes Same AY011000 Secale cereale LKellogg sn (GH) 2729B 2766 N A AY553266 AY548429 Yes Same AY011009 S montanum subsp montanum Guss T 36554(GH) 2734C 2770 N A AY553267 AY548430 No mdash mdash S montanum subsp anatolicum (Boiss) Tzelev PI 206992 (GH) 2739A1145A N A AY553265 AY548428 Yes PI 206991 (GH) AY011008 Taeniatherum caput-medusae (L) Nevski MasonGamer 189d(ID) 1588A 1141A N A AY553268 AY548431 Yes PI 208075 (ID) AY011010 Thinopyrum elongatum (Host) DRDeweyMasonGamer 113 (ID) 1460A 1116F N A AY553269 AY548432 Yes PI 531719 (GH) AF079284 Triticum aestivum L mdash mdashmdash N A AJ007705 AJ007705 No mdash mdash

Bromeae Bromus tectorum L Kellogg sn (GH) 1662C 2774 N A AY553239 AY548402 Yes Same AY362757

TABLE 2 List of primers

Primer NameRelationship to Triticum

aestivum AJ007705 Primer (59ndash39)

Region 1 (1ndash1562) 526ndash1632PepC-291FPepC-467F(1)PepC-796RPepC-952R

219ndash240467ndash489796ndash813952ndash973

ACTCCTGCCATCCGCCTTCTATGCTGCTCGTCCCCGCCAAGGTGTAAAAACACCAGATTACGCACAATGACTGACACGATTTGAGATTC

PepC-1387RPepC-1555RPepC-1672R(1)PepC-1672R(2)

1387ndash14081555ndash15751672ndash16961672ndash1696

GTCAAGGCATAGTCGTTTCAAGAATCCATCAATCAACATAGAGAGCTTGTTATCATCTTCCCGAGTTCAGCTTGTTATCATCTCTCCGAGTTCA

Region II (1566ndash2185) 1849ndash2441PepC-1827FPepC-2443R

1827ndash18482443ndash2464

ARAYTCRGCAATCACAGAATCTRATCCCAATGTTCTTCAATGC

Sequence Analysis Boundaries between the PepC introns andexons were determined by comparison with published Triticumaestivum (AJ007705) and Sorghum vulgare (X59925) sequences Se-quences were initially aligned using the default setting in ClustalW ver 15 (Thompson et al 1994) then adjusted manually usingMacClade ver 4 (Maddison and Maddison 2002) Three regions(35 characters alignment positions 393ndash403 996ndash972 and 1280ndash

1296) were judged to have an ambiguous alignment and were ex-cluded prior to phylogenetic analyses Percent of gap charactersmissing characters and nucleotide composition for each specieswere determined using MacClade

Phylogeny Reconstruction Maximum likelihood (ML) meth-ods were used to generate hypotheses about relationships amongtaxa using PAUP ver 40b10 (Swofford 2002) Initially maximum

852 [Volume 29SYSTEMATIC BOTANY

TABLE 1 List of taxa collection information and GenBank numbers Voucher numbers containing names are collectorsrsquo accessionnumbers Voucher H 5555 is from the Swedish-Danish Triticeae Consortium sample provided by Dr O Seberg The remaining lsquolsquoPIrsquorsquovoucher numbers are Plant Introduction accessions from the National Plant Germplasm System (US Dept Agriculture httpwwwars-gringovnpgs) samples were provided by the USDA-ARS Data are presented in the following sequence species name PepCvoucher Clone Region-I Clone Region-II Elymus genome GenBank No Region-I Region-II Used in PepC-waxy comparison (yesno) waxy voucher waxy GenBank No

Elymus E californicus (Bolander) Gould Barkworth sn (UTC) 1645B 637C NA AY553241 AY548404 Yes Same AY011012E canadensis L PI 531568 (ID) 973A 534A St AY553248 AY548411 Yes Barkworth 97-86 (UTC) AY556481 E canadensis PI531568 (ID) 973B 534F H AY553242 AY548405 No mdash mdash E canadensis PI 578675 (ID) 215C 48C H AY553243 AY548406No mdash mdash E elymoides (Rafin) Swezey PI 531606 (ID) 648C 632C St AY553249 AY548412 Yes Same AY010992 E elymoidesPI 531606 (ID) 957A 632A H AY553244 AY548407 Yes Same AY010965 E glaucus Buckley W6 10215 (ID) 219A 51A StAY553250 AY548413 Yes MasonGamer 130 (ID) AY010979 E hystrix L Barkworth 97-87 (UTC) 981C 622C St AY553251AY548414 Yes Same AY010982 E hystrix Barkworth 97-87 (UTC) 981B 622A H AY553245 AY548408 No mdash mdash E lanceolatus(Scribn amp Smith) Gould PI 531623 (ID) 1049B 604B St AY553252 AY548415 Yes W6 14220 (ID) AY010984 E lanceolatus PI531623 (ID) 1049C 604J H AY553246 AY548409 Yes Same AY010969 E wawawaiensis J Carlson ex Barkworth PI 598812(ID) 229A 74A St AY553253 AY548416 Yes Same AY010990 E wawawaiensis PI 598812 (ID) 229B 74P H AY553247AY548410 Yes Same AY010978

Triticeae Aegilops comosa Sm G 602 (GH) 1305A 1296A N A AY553236 AY548399 Yes Same AF079263 Agropyroncristatum (L) Gaertn PI 279802 (GH) 1242B 540A N A AY553237 AY548400 Yes Same AF079271 Australopyrum velutinum(Nees) B Simon D 2873-2878 (GH) 1209B 1128A N A AY553238 AY548401 Yes Same AY011004 Dasypyrum villosum (L)Candargy D 2990 (GH) 1641A 1134A N A AY553240 AY548403 Yes PI 470279 (GH) AY556480 Eremopyrum orientale (L)Jaub amp Spach H 5555 (GH) 884D 543E N A AY553254 AY548417 Yes Same AY011007 Heteranthelium piliferum (Banks ampSol) Hochst PI 402352 (GH) 1301B 1292A N A AY553255 AY548418 Yes Same AF079277 Hordeum brachyantherum subspcalifornicum (Cov amp Steb) Bothmer MA-138-1-40 (GH) 339B 312B N A AY553256 AY548419 Yes Same AF079273 H jubatumL MasonGamer 106 (ID) 940C 316A N A AY553257 AY548420 Yes Same AY010963 H marinum Hudson PI 304346 (ID)1246C 301A N A AY553258 AY548421 Yes Same AY010959 H murinum L Ciho 15683 (ID) 989A 297A N A AY553259AY548422 Yes Same AY010960 H vulgare L MasonGamer 107 (ID) 1346BB 1314A N A AY553260 AY548423 Yes Rohdeet al 1988 X07931 Leymus racemosus subsp sabulosus (MBieb) Tzvelev R-20-21ndash25 (GH) 1592B 1718A N A AY553261AY548424 No mdash mdash Peridictyon sanctum (Janka) Seberg Fred amp Baden Jensen 248 (GH) 1341A 1157A N A AY553262AY548425 Yes Same AF079278 Pseudoroegneria spicata (Pursh) A Love PI 610986 (ID) 352A 325A N A AY553263 AY548426Yes Same AY010999 P spicata D 2844 (GH) 804B 610A N A AY553264 AY548427 Yes Same AY011000 Secale cereale LKellogg sn (GH) 2729B 2766 N A AY553266 AY548429 Yes Same AY011009 S montanum subsp montanum Guss T 36554(GH) 2734C 2770 N A AY553267 AY548430 No mdash mdash S montanum subsp anatolicum (Boiss) Tzelev PI 206992 (GH) 2739A1145A N A AY553265 AY548428 Yes PI 206991 (GH) AY011008 Taeniatherum caput-medusae (L) Nevski MasonGamer 189d(ID) 1588A 1141A N A AY553268 AY548431 Yes PI 208075 (ID) AY011010 Thinopyrum elongatum (Host) DRDeweyMasonGamer 113 (ID) 1460A 1116F N A AY553269 AY548432 Yes PI 531719 (GH) AF079284 Triticum aestivum L mdash mdashmdash N A AJ007705 AJ007705 No mdash mdash

Bromeae Bromus tectorum L Kellogg sn (GH) 1662C 2774 N A AY553239 AY548402 Yes Same AY362757

TABLE 2 List of primers

Primer NameRelationship to Triticum

aestivum AJ007705 Primer (59ndash39)

Region 1 (1ndash1562) 526ndash1632PepC-291FPepC-467F(1)PepC-796RPepC-952R

219ndash240467ndash489796ndash813952ndash973

ACTCCTGCCATCCGCCTTCTATGCTGCTCGTCCCCGCCAAGGTGTAAAAACACCAGATTACGCACAATGACTGACACGATTTGAGATTC

PepC-1387RPepC-1555RPepC-1672R(1)PepC-1672R(2)

1387ndash14081555ndash15751672ndash16961672ndash1696

GTCAAGGCATAGTCGTTTCAAGAATCCATCAATCAACATAGAGAGCTTGTTATCATCTTCCCGAGTTCAGCTTGTTATCATCTCTCCGAGTTCA

Region II (1566ndash2185) 1849ndash2441PepC-1827FPepC-2443R

1827ndash18482443ndash2464

ARAYTCRGCAATCACAGAATCTRATCCCAATGTTCTTCAATGC

Sequence Analysis Boundaries between the PepC introns andexons were determined by comparison with published Triticumaestivum (AJ007705) and Sorghum vulgare (X59925) sequences Se-quences were initially aligned using the default setting in ClustalW ver 15 (Thompson et al 1994) then adjusted manually usingMacClade ver 4 (Maddison and Maddison 2002) Three regions(35 characters alignment positions 393ndash403 996ndash972 and 1280ndash

1296) were judged to have an ambiguous alignment and were ex-cluded prior to phylogenetic analyses Percent of gap charactersmissing characters and nucleotide composition for each specieswere determined using MacClade

Phylogeny Reconstruction Maximum likelihood (ML) meth-ods were used to generate hypotheses about relationships amongtaxa using PAUP ver 40b10 (Swofford 2002) Initially maximum

2004] 853HELFGOTT AND MASON-GAMER ELYMUS PHYLOGENY

parsimony (MP) analyses were performed to obtain trees on whichstarting parameters for ML analyses would be estimated Parsi-mony analyses assumed equal weights for characters and charac-ter state changes and gaps were treated as missing data An initialtree was obtained via random taxon stepwise addition and a heu-ristic tree search was performed using MULTREES ACCTRANoptimization and tree-bisection-reconnection (TBR) branch swap-ping (Hendy and Penny 1982) The possibility of multiple tree is-lands was explored by running 10000 random addition replicateswith twenty trees held at each step

Parameters for 16 models of sequence evolution were estimatedfor all most-parsimonious trees (eg Swofford et al 1996 Frati etal 1997 Sullivan et al 1997) Four models of nucleotide substitu-tionmdashJukes-Cantor (Jukes and Cantor 1969) Kimura two-param-eter (Kimura 1980) Hasegawa-Kishino-Yano (Hasegawa et al1985) and general time reversible (GTR Yang 1993)mdashwere ex-amined Each substitution model was paired with each of fourmodels of among-site rate variation 1) no rate heterogeneity 2)some sites invariable (I Hasegawa et al 1985) with equal rates ofchange among the remaining sites 3) rate heterogeneity amongsites following a gamma distribution (G Yang 1994) and 4) somesites invariable with gamma-distributed variation among the re-maining sites (I1G Gu et al 1995 Waddell and Penny 1996) Us-ing the tree with the highest ML scores across all 16 models ex-amined we compared the models with the three highest scoresemploying a likelihood ratio test and compared the results to aX2 distribution (Felsenstein 1981 Huelsenbeck and Crandall 1997Huelsenbeck and Rannala 1997 Sanderson 1998 but see Goldman1993)

Among the models of molecular evolution tested the one in-corporating the most parameters (GTR1I1G) produced the high-est log likelihood value (2744450) However this value was notsignificantly different from the one obtained using GTR1G alone(2744508) so the GTR1G model was chosen for all subsequentML analyses Two sequential ML analyses were conducted Thefollowing estimated model parameters were used as initial set-tings in the first ML search [R (AC) 5 108 R (AG) 5 297 R (AT)5 042 R (CG) 5 138 R (CT) 5 362 R (GT) 5 100 pi (A) 5025 pi (C) 5 020 pi (G) 5 021 pi (T) 5 033 a 5 049] usinga heuristic tree searching strategy (the initial tree obtained viastepwise random addition with twenty trees held at each step andTBR branch swapping) This search resulted in a single tree(2744462) with the following estimated ML parameters [R (AC)5 108 R (AG) 5 304 R (AT) 5 044 R (CG) 5 141 R (CT) 5367 R (GT) 5 100 pi (A) 5 025 pi (C) 5 020 pi (G) 5 021pi (T) 5 033 a 5 048] These estimates were used as the startingpoint for the second ML analysis which produced a tree with thesame score as the first

Branch support for the parsimony tree was estimated using thenon-parametric bootstrap (Felsenstein 1985 Sanderson and Don-oghue 1989) Bootstrap values were calculated from 10000 datasets using the lsquolsquofast bootstraprsquorsquo tree search strategy implementedin PAUP Posterior probability values for clades on the ML treewere inferred with Mr Bayes (ver 20 Huelsenbeck and Ronquist2001) using the GTR 1 G (Yang 1993) model a random startingtree and a uniform prior probability The analysis included threechains (two hot and one cold) which ran for a million generationswith trees saved every 1000 generations We determined the pla-teau point of the chains at 4000 generations by graphing theweighted maximum likelihood scores obtained and discarding thefour trees saved before the plateau A 95 majority-rule consensustree was calculated from the set of remaining trees and the groupfrequencies were used as posterior probability estimates

Congruence Testing Between PepC and waxy Previous studiescomparing molecular data sets of the Triticeae uncovered incon-gruence among some of the gene trees suggesting that portionsof the nuclear genome have different histories (Kellogg et al 1996Mason-Gamer and Kellogg 1996b) We tested whether the PepCdata were congruent with sequence data from a 13-kb portion ofthe single-copy nuclear waxy gene previously obtained for thesame members of the Triticeae (Mason-Gamer and Kellogg 2000Mason-Gamer 2001) The waxy and PepC loci are believed to be on

different chromosomes The waxy gene is found on the group 7chromosomes in the Triticeae (Devos et al 1995 Kleinhofs 1997)or on a portion of chromosome 4 translocated from and thus ho-moeologous to the group 7 chromosomes (Devos et al 1995 Kor-zun et al 1997) The PepC gene is hypothesized here to be on thelong arm of chromosome 5 (see Materials and Methods lsquolsquoMolec-ular Datarsquorsquo)

We tested congruence between the PepC and waxy data sets us-ing the Shimodaira-Hasegawa (SH Shimodaira and Hasegawa1999) test with a specific model of evolution as implemented inPAUP ver 40b10 (Swofford 2002) Prior to analyses the waxy andPepC data sets were reduced to 29 taxa which matched exactly tothe level of species (Table 1) Two tree topologies were evaluated(PepC and waxy) obtained from MP analyses of each data partitionfollowing the same protocol used to analyze the full PepC data setexcept that only 100 random addition replicates were run We de-termined the appropriate model of sequence evolution for both thewaxy and PepC data partitions and calculated which parsimonytree had the highest ML scores using the same procedure em-ployed to analyze the complete PepC data set The two highest-scoring trees (one from each data partition) were used as the twotest trees

Two separate SH tests were conducted The ML scores for eachtest tree were estimated first under the model of sequence evolu-tion best fitting the PepC data and then under the model that bestfit the waxy data In both cases the log-likelihood differences be-tween the trees were determined The first of these was comparedto the RELL bootstrap distribution of log-likelihood differencesgenerated using the PepC data and ML parameters while the sec-ond was compared to the distribution generated using the waxydata and ML parameters If the score difference between two treesfell within 95 of the appropriate RELL-generated scores in a one-tailed comparison the estimated p-value was not considered sig-nificant at the 005 level If both SH tests resulted in significant p-values the trees were assumed to be incongruent

Next we used constraint analyses to examine the source of con-flict between the PepC and waxy trees Initially branch support forthe parsimony trees was estimated following the methods used inthe analysis of the full PepC data set Then we constructed single-node constraint trees for each clade with bootstrap support above50 in the reduced PepC and waxy trees This yielded eight con-straint trees based on the PepC tree and ten based on the waxytree Finally we completed a series of constrained MP heuristicsearches using stepwise addition MULTREES ACCTRAN opti-mization and TBR branch swapping We imposed each of the eightPepC constraint trees in MP searches of the waxy data and eachof the ten waxy constraint trees in MP searches of the PepC dataIf the constrained trees were longer it suggested that the corre-sponding nodes are sources of conflict between the data sets

RESULTS

Sequence Length and Variation Complete DNA se-quences were generated for 27 species The 17-kb por-tion of the PepC gene examined spans exons 1ndash4 how-ever the majority of the sequence data (80) are de-rived from introns 1ndash3 Compared to the portion ofthe waxy gene used by Mason-Gamer (2001) for thesame set of taxa (Table 3) the PepC gene is approxi-mately 350 bp longer has considerably more introncharacters and provides more parsimony-informativecharacters (although the percentage of parsimony-in-formative characters is somewhat lower) There is noevidence that either the H or the St copy of the PepCgene is non-functional there are no length changes orstop codons in the exons examined

All of the Elymus individuals examined have twodistinct PepC gene variants One (the H form) groups

854 [Volume 29SYSTEMATIC BOTANY

TABLE 3 Data characteristics 1 Approximately half of region 2 is missing for Secale montanum subsp montanum

PepC PepC Reduced waxy

Aligned LengthSequence Lengthmdashaverage (minimummaximum) Gap Charactersmdashaverage (minimummaximum) Missing Charactersmdashaverage (minimummaximum) Exon

21851573 (1418ndash1693)267 (211ndash339)14 (0ndash3311)

195

21261587 (1429ndash1710)254 (196ndash328)06 (00ndash41)

201

15301232 (1191ndash1299)195 (152ndash222)001 (00ndash01)499

GC ContentParsimony Informative Charactersmdash ()Number of Parsimony Trees (steps)Consistency IndexRetention Index

430 (421ndash444)97 (211)

3 (750)06100882

430 (421ndash439)90 (192)

6 (674)06310814

583 (568ndash595)122 (187)

30 (763)04780656

with members of Hordeum sect Critesion and the sec-ond (the St form) groups within Pseudoroegneria How-ever for two Elymus individuals we only have a par-tial sequence of one of the two forms for E canadensiswe have St sequences from region 2 but not from re-gion 1 and for E glaucus we have an H copy fromregion 2 but not from region 1 Because approximately60 of the nucleotides for these sequences are absentthey were not included in the phylogenetic analysesonly the H copy is included for E canadensis and onlythe St copy is included for E glaucus

Phylogenetic Analyses of PepC Data Parsimonyanalysis of the PepC data yielded three most-parsi-monious trees (750 steps CI excluding uninformativecharacters 5 0610 RI 5 0822) The only topologicaldifference between the three trees was the placementof two taxa within the Elymus 1 H clade (E lanceolatusand H jubatum) However once nodes with zero-lengthbranches were collapsed there was only one tree inwhich E lanceolatus and H jubatum were part of a po-lytomy within the Elymus 1 H clade The final MLsearch resulted in a single tree (2744462 Fig 2 seeMaterials and Methods lsquolsquoPhylogeny Reconstructionrsquorsquofor estimated parameters)

The ML and parsimony trees have the same generaltopology (Fig 2) In each E californicus is the first todiverge followed by Leymus racemosus Next Hordeumforms a weakly supported clade sister to a secondweakly supported clade containing the remainingmembers of the tribe Many of the other nodes resolvedin both the parsimony and ML trees are either incon-gruent or weakly supported but every clade with highbootstrap (BS) or posterior probabilities (PP) was re-covered by both methods of analysis

The PepC data suggest that the Elymus tetraploidscombine the H and St genomes There is a stronglysupported clade consisting of Hordeum jubatum Hbrachyantherum and Elymus (BS 5 98 and PP 5 100)A second strongly supported group includes Elymusand Pseudoroegneria spicata (BS and PP 5 100) Elymuscalifornicus is the only Elymus species that does notgroup with Hordeum or Pseudoroegneria but is foundnear the base of the tree There are four other lineageswith high bootstrap support and posterior probabili-

ties 1) an Agropyron cristatum and Eremopyrum orientaleclade (BS and PP 5 100) 2) an Aegilops comosa Triticumaestivum and Taeniatherum caput-medusae clade (BS 586 and PP 5 100) 3) a Secale L clade (BS 5 97 and PP5 100) and 4) a Hordeum murinum and H vulgare Lclade sister to the Elymus 1 H lineage (BS 5 83 andPP 5 99)

Comparison of PepC and waxy Sequence charac-teristics for the reduced PepC and waxy data sets areshown in Table 3 Parsimony analysis of the PepC datapartition yielded six most-parsimonious trees (674steps CI excluding uninformative characters 5 0631RI 5 0814) including two tree islands of three treeseach The strict consensus of all six trees is less re-solved than the strict consensus from the entire PepCdata set but has the same general topology and well-supported nodes (Fig 3) Among the six most-parsi-monious trees tree 6 had the highest log likelihoodscore and was chosen as the PepC tree for use in theSH test Once again the log likelihood value obtainedunder the (GTR1I1G) model was not significantly dif-ferent than the one obtained using (GTR1G) alone(2688734 vs2688770) so we used the simpler modelFigure 3 depicts parsimony tree 6 under the ML pa-rameters best fitting the PepC data [R (AC) 5 098 R(AG) 5 283 R (AT) 5 043 R (CG) 5 133 R (CT) 5342 R (GT) 5 100 pi (A) 5 025 pi (C) 5 020 pi(G) 5 021 pi (T) 5 033 a 5 052]

Parsimony analysis of the reduced waxy data yielded30 shortest trees (763 steps CI excluding uninforma-tive characters 5 0478 RI 5 0656) The strict consen-sus of these trees (Fig 4) is similar to the strict con-sensus trees of tree islands 3 and 4 in Mason-Gamerrsquos(2001) analysis of waxy data All most-parsimoniouswaxy trees include a single well-supported Hordeum 1Elymus clade (BS 5 89) Within this group the ElymusH sequences form a well-supported clade with Hbrachyantherum (BS 5 100) while H marinum H mu-rinum H vulgare and H jubatum form a weakly-sup-ported clade The Elymus and Pseudoroegneria sequenc-es are grouped into two distinct well-supportedclades (St1 and St2) but the St sequences do not forma monophyletic group The St1 clade (BS 5 100) is sis-ter to a weakly supported clade including Aegilops L

2004] 855HELFGOTT AND MASON-GAMER ELYMUS PHYLOGENY

FIG 2 The maximum likelihood tree based on PepC sequence data generated under a GTR1 G model of evolution ln (L52744462) Bold lines indicate the Elymus 1 Pseudoroegneria and Elymus 1 Hordeum clades Genome designations are providedfor Elymus Pseudoroegneria and Hordeum taxa The posterior probabilities are above the branches and bootstrap values ($ 75)are given in parentheses Gray lines represent portions of the tree that are inconsistent with the parsimony results PepC clonenumbers are provided next to taxon names

856 [Volume 29SYSTEMATIC BOTANY

FIG 3 The reduced PepC parsimony topology used as thetest tree in the SH test depicted using ML estimated param-eters best fitting the PepC data under GTR1G Bold lines in-dicate the Elymus 1 Pseudoroegneria and Elymus 1 Hordeumclades Bootstrap values greater than 49 are above the nodesNodes that are sources of conflict in the constrained analysesare marked with asterisks and the resulting increase in treelength in each of the associated constrained analyses is givenGray lines represent nodes that collapse in the strict consensusof all most parsimonious trees Taxon labels as in Fig 2

FIG 4 The reduced waxy parsimony tree used as the testtree in the SH test depicted using ML estimated parametersbest fitting the waxy data under GTR1I1G Bold lines indicatethe Elymus 1 Pseudoroegneria and Elymus 1 Hordeum cladesBootstrap values above 49 are above the nodes Nodes thatare sources of conflict in the constrained analyses are markedwith asterisks and the resulting increase in tree length in eachof the associated constrained analyses is given Gray lines rep-resent nodes that collapse in the strict consensus of all mostparsimonious trees Taxon labels as in Fig 2

Dasypyrum (Coss amp Durieu) P Candargy and Thino-pyrum ALove and St2 (BS 5 98) is sister to Australo-pyrum velutinum As in the PepC tree E californicusdoes not group with other Elymus species The twoSecale species form a strongly supported clade (BS 5100) sister to Heteranthelium piliferum (BS 5 68) Thisgroup is in turn sister to a clade including Agropyroncristatum Eremopyrum orientale and Taeniatherum caput-medusae (BS 5 68) Among the 30 most-parsimonioustrees tree 1 had the highest log likelihood score andwas chosen as the waxy tree for use in the SH test Themost parameter rich model (GTR1I1G) had a signif-icantly higher log likelihood value than the other mod-els tested and was used in the SH test Figure 4 de-picts parsimony tree 1 under the ML parameters bestfitting the waxy data [R (AC) 5 150 R (AG) 5 353R (AT) 5 136 R (CG) 5 166 R (CT) 5 487 R (GT)5 100 pi (A) 5 022 pi (C) 5 026 pi (G) 5 030 pi(T) 5 021 P-inv 5 035 a 5 071]

Both SH tests yielded significant p-values The SHtest using the PepC data and associated ML parametersresulted in a difference in tree scores of 55194 P 0001 (PepC score 52688770 waxy score 52743964)The SH test using the waxy data and associated MLparameters resulted in a difference in tree scores of

39726 P 0001 (waxy score 52598225 PepC score52637950)

In constrained analyses of the PepC data five of theten nodes tested from the waxy tree (Fig 4 asterisks)resulted in an increase in length relative to the uncon-strained PepC trees (674 steps) In constrained analysesof the waxy data five of the eight PepC tree nodes (Fig3 asterisks) resulted in increases in length over theunconstrained waxy trees (763 steps) The length in-creases associated with each node are shown in Figs3 and 4

DISCUSSION

Evidence of St 1 H Genome Content All tetraploidElymus individuals examined here have two copies ofthe PepC isoform sequenced in the study One copyforms a clade with Hordeum and the second with Pseu-doroegneria confirming that North American Elymusspecies do have an St 1 H genomic content and there-fore are allotetraploid derivatives of Hordeum and Pseu-doroegneria (Fig 2) The data thus confirm Deweyrsquos(1984) and Loversquos (1984) genomic concept of NorthAmerican Elymus in which all species (with the excep-tion of the octoploid E californicus) are allotetraploidscontaining the St and H genomes (Dewey 1984 Love1984) Some systematists have raised justifiable con-cerns regarding the use of genomic pairing data as the

2004] 857HELFGOTT AND MASON-GAMER ELYMUS PHYLOGENY

primary criterion for both grouping and ranking oftaxa in the Triticeae (eg Baum et al 1987 Kellogg1989 Seberg and Petersen 1998) However the PepCphylogeny corroborates the cytogenetic findings andis in agreement with previous studies based on bothisozyme (Jaaska 1995 1998) and waxy gene sequencedata (Mason-Gamer 2001) A classification systembased on chromosome pairing provides a reasonablestarting point in that it groups taxa based on a char-acter that appears to reflect evolutionary relatednessHowever its utility is limited because it cannot provideinsights into the hierarchical relationships among thetaxa examined (eg Kellogg 1989)

Elymus californicusmdashis it a Member of ElymusThe octoploid Elymus californicus is the only Elymusspecies in the present study in which the St and Hgenomes have not been detected This is not surpris-ing because the classification of E californicus has beenunstable since it was first described in 1880 and itsgenomic makeup remains uncertain (Love 1984 MaryBarkworth pers comm) In an attempt to gain in-sights into the genomic complement of this enigmaticspecies Jensen and Wang (1997) screened for the Stand Ns genomes using genome-specific RAPD mark-ers Their study found no evidence of the St genomebut did detect a marker specific to the Ns genomewhich is found in the diploid genus PsathyrostachysNevski and in the allopolyploid genus Leymus HochstFrom this information Jensen and Wang (1997) sug-gested that E californicus should be transferred to Ley-mus The PepC data neither refute nor confirm thesefindings directly placing Elymus californicus at the baseof the tree Nevertheless the PepC phylogeny clearlydemonstrates that E californicus does not belong withinElymus The waxy gene phylogeny by Mason-Gamer etal (2001) was also consistent with Jensen and Wangrsquos(1997) findings revealing a weak association betweenPsathyrostachys and E californicus in most of the most-parsimonious trees Within the chloroplast DNA phy-logeny by Mason-Gamer et al (2002) the placement ofE californicus is equivocal E californicus is weaklygrouped with Hordeum in some of the shortest treesThe PepC data add to the mounting evidence suggest-ing that E californicus does not belong within Elymusbut additional sampling and phylogenetic analyses areneeded to develop a firm understanding of how thisspecies is related to other members of the Triticeae

Elymus H and St Genomes The PepC data indicatethat all North American tetraploid species of Elymushave St 1 H genome content but there is little reso-lution among Elymus individuals in either the St or theH clade This pattern is consistent with the hypothesisthat the Elymus tetraploids are formed from a singleset of genome donors or multiple closely related setsof donors but the specific progenitor species involvedremain a mystery The Elymus H genome sequences

form a strongly supported clade with the two NorthAmerican Hordeum taxa included in the study (Hbrachyantherum and H jubatum) while the Elymus Stsequences form an unresolved polytomy with P spicatathe only North American Pseudoroegneria species (Fig2) The lack of resolution among the Elymus St copiesis similar to that previously observed among chloro-plast DNA sequences (Mason-Gamer et al 2002) In astudy based on the waxy gene (Mason-Gamer et al2001) St sequences formed two distinct clades whichdid not correspond to geographical or morphologicaldifferentiation It was unclear whether the pattern wasthe result of allelic variation or recent gene duplicationprior to the divergence of the St group The PepC andwaxy studies have provided hints on which Hordeumspecies may be most closely related to Elymus but ad-ditional sampling of both Hordeum and Pseudoroegneriais needed before we can develop a better picture of theorigin of the H and St genomes

Distinction Among Hordeum Genomes A detaileddiscussion of Hordeum evolution in light of the PepCphylogeny is impossible given the limited number oftaxa sampled However the data do resolve the fourHordeum genomes (H Xa I and Xu) and providesome insights into how these groups are related (Fig2) Taxa with genomes I (Hordeum vulgare) or Xu (Hor-deum murinum) form a strongly supported monophy-letic group which in turn is sister to a second well-supported clade that includes species with the H or Xa(Hordeum marinum) genomes This corroborates thestudy by Komatsuda et al (1999) of Hordeum diploidtaxa using Vrs1 nuclear intron sequence data The PepCresults are also in accordance with Petersen and Se-bergrsquos (2003) analysis of diploid Hordeum taxa inwhich they combined the chloroplast data of Doebleyet al (1992) and Komatsuda et al (1999) with datafrom the nuclear marker DMC1 (Petersen and Seberg2002 2003) and the chloroplast rbcL gene (Petersen andSeberg 1998) Nuclear and chloroplast DNA data werepartly contradictory but in the combined data set theXu and I genomes formed a sister group to the H andXa taxa although one H genome taxon (H brevisubu-latum Link) was sister to H marinum (Xa) Data fromthe nuclear ribosomal internal transcribed spacers(Frank Blattner pers comm) also recovered four ma-jor clades corresponding to the four Hordeum genomesbut the arrangement of the clades was somewhat dif-ferent Genomes H 1 Xa formed a clade sister to Xuand this clade is sister to I though the placement ofthe Xu clade is weakly supported It is clear from theseHordeum studies that the genome groups are in factphylogenetically informative

Conflict Between PepC and waxy Results from theSH test suggest that the PepC and waxy trees are sig-nificantly incongruent Constrained analyses confirmthese findings suggesting that there is disagreement

858 [Volume 29SYSTEMATIC BOTANY

involving some of the clades in the PepC and waxy phy-logenies (Figs 3 4) There are two cases within Hor-deum where taxa have different positions in the twotrees but are members of clades with moderate to highbootstrap support The first is Hordeum marinum in thePepC tree it is sister to the H clade while in the waxytree it is sister to H murinum (Xu) In the second casethe same accession of H jubatum has a different posi-tion in the PepC and waxy trees In the PepC phylogenyit is nested within the well-supported H clade whilein the waxy phylogeny it is found at the base of a gradeincluding H marinum (Xa) H murinum (Xu) and Hvulgare (I) While the reason for the conflicting place-ment of H jubatum is not known it is an allotetraploid(eg Joslashrgensen 1986 Bothmer et al 1987) and the twodata sets may have sampled gene copies representingthe two different homoeologous genomes

The placement of Taeniatherum caput-medusae is alsoat odds between the PepC and waxy phylogenies (Figs3 4) In the PepC tree T caput-medusae forms a well-supported clade (BS 5 95) with Aegilops comosa whilein the waxy tree it is sister to Eremopyrum orientale (BS5 50) which in turn is sister to Agropyron cristatum (BS5 70) The association seen in the waxy topology hasnot been seen in other gene trees but a close relation-ship between T caput-medusae and Aegilops was seen inthe chloroplast DNA phylogeny (Mason-Gamer et al2002) The PepC data strongly conflict with the waxydata on the placement of Taeniatherum with 36 addi-tional steps required in the constrained analysis Thewaxy data show far less conflict with Taeniatherumrsquosplacement in the PepC tree requiring only two addi-tional steps in the constrained analysis There may besome underlying signal within the waxy data set thatsupports a more congruent placement of TaeniatherumIn the PepC phylogeny E orientale forms a stronglysupported clade with A cristatum (BS 5 100) ratherthan Taeniatherum in agreement with the chloroplastDNA phylogeny (Mason-Gamer et al 2002) and theDMC1 phylogeny (Petersen and Seberg 2002) in whichAgropyron cristatum was grouped with two other Ere-mopyrum species E distans (KKoch) Nevski and E tri-ticeum (Gaertn) Nevski

Secale is well-supported on both trees (BS 5 100)which is not surprising A correlation between mono-genomic groups and clades on molecular phylogenetictrees has been previously reported (Hsiao et al 1995Kellogg and Appels 1995 Kellogg et al 1996 Mason-Gamer and Kellogg 1996a 1996b Petersen and Seberg1997) In the waxy tree the Secale clade is sister to Het-eranthelium piliferum forming a weakly supportedclade (BS 5 68) this association was previously sup-ported in Mason-Gamerrsquos chloroplast DNA phylogeny

Phylogenetic conflict was previously demonstratedamong Triticeae gene trees with the most extensivedifferences documented in comparisons between chlo-

roplast and nuclear data (Kellogg et al 1996 Mason-Gamer and Kellogg 1996b) There were a few instancesof statistically significant incongruence among the nu-clear data sets and the present comparison betweenPepC and waxy shows conflict involving several taxa Ifthe conflict is genuine the nuclear gene trees may re-flect differences among phylogenies of the chromo-somes or sections of chromosomes on which thegenes reside The discordance between PepC and waxygene tree topologies (as well as many others) high-lights the need for continued research on chromosomeevolution in the Triticeae in the hopes of better under-standing mechanisms behind the complex historicalpatterns that are emerging from phylogenetic studiesof this group

Two different historical scenarios provide possibleexplanations for the reticulate history of the tribe (Kel-logg et al 1996) First the pattern may result fromperiods of cladogenesis and dispersal interspersedwith hybridization and introgression between speciesThe level of hybridization slowed or stopped at somepoint in the past most genera in the tribe are inter-sterile and hybrids among them generally exhibit alack of chromosome pairing (eg Dewey 1984 and ref-erences therein) However some degree gene exchangeamong genera is possible if a small percentage of vi-able hybrid seeds are produced through chance aloneIt is also possible that present-day gene flow betweendiploid taxa is taking place through a polyploid inter-mediate Gene flow between diploids via a tetraploidbridge requires the formation of partially fertile trip-loids as has been reported for species of Aegilops byVardi (1973) Feldman et al (1979) and Kimber andFeldman (1987) There is also phylogenetic evidence ofhybridization or introgression Mason-Gamer and Kel-logg (2000) reported that an accession of Pseudoroeg-neria strigosa (M Bieb) ALove which was expected tobe a monogenomic St species has both St and H cop-ies of the waxy gene and an St chloroplast The resultwas consistent with both the allotetraploid-bridge anddirect hybridization hypotheses

A second possible explanation for the reticulate his-tory of the Triticeae is that it reflects a burst of diver-sification with rapid lineage sorting and fixation ofdifferent alleles into lineages which quickly becameintersterile (Kellogg et al 1996) Given the need forextensive ancestral polymorphism in this hypothesisit seems unlikely that rapid diversification and lineagesorting was the major driving force behind the retic-ulate history of the Triticeae However it is not clearhow we could determine which scenario (lineage sort-ing and hybridization-introgression) has had the big-ger impact on the history of the Triticeae In fact thesehypotheses are not mutually exclusive and it may bethat both processes have played some role in shapingthe history of the Triticeae

2004] 859HELFGOTT AND MASON-GAMER ELYMUS PHYLOGENY

In conclusion this study verifies Deweyrsquos (19821983) and Loversquos (1982 1984) insightful chromosome-based revisions of the Triticeae The PepC results clear-ly show that North American tetraploid members ofElymus have an St 1 H genomic content and are there-fore derived from Hordeum and Pseudoroegneria How-ever the current data reveal little genetic differentia-tion within North American Elymus Consequently therelationships among Elymus species and their relation-ships to specific progenitor species await a better-re-solved phylogeny Additional studies which integratemorphological cytological and phylogenetic informa-tion are necessary to understand better the historicalrelationships within Elymus The PepC-based analysisof North American Elymus is a small step forward inthis process

ACKNOWLEDGEMENTS The authors wish to thank Mary Bark-worth for sharing plant material the USDArsquos Germplasm Resourc-es Information Network for seeds Gregory Plunkett and twoanonymous reviewers for helpful comments on the manuscriptThe National Science Foundation (DEB-9974181) funded this work

LITERATURE CITED

BARKWORTH M E and D R DEWEY 1985 Genomically basedgenera in the perennial Triticeae of North America identifi-cation and membership American Journal of Botany 72 767ndash776

BAUM B R J R ESTES and P K GUPTA 1987 Assessment of thegenomic system of classification in the Triticeae AmericanJournal of Botany 74 1388ndash1395

BESNARD G B OFFMANN C ROBERT and C ROUCH 2002 As-sessment of the C4 phosphoenolpyruvate carboxylase genediversity in grasses (Poaceae) Theoretical and Applied Genetics105 404ndash412

BOTHMER R VON J FLINK and T LANDSTROM 1987 Meiosis inHordeum interspecific hybrids II Triploid hybrids Evolution-ary Trends in Plants 1 41ndash50

CRETIN C S SANTI E KERYER L LEPINIEC D TAGU J VIDALand P GADAL 1991 The phosphoenolpyruvate carboxylasegene family in Sorghum promoter structure amino acid se-quences and expression of genes Gene 99 87ndash94

CRONN R C R L SMALL T HASELKORN and J F WENDEL 2003Cryptic repeated genomic recombination during speciation inGossypium gossypioides Evolution 57 2475ndash2489

mdashmdashmdash X ZHAO A H PATERSON and J F WENDEL 1996 Poly-morphism and concerted evolution in a tandemly repeatedgene family 5S ribosomal DNA in diploid and allopolyploidcottons Journal of Molecular Evolution 42 685ndash705

CRONQUIST A 1987 A botanical critique of cladism Botanical Re-view 53 1ndash52

CUSHMAN J C and H J BOHNERT 1989a Nucleotide sequence ofthe Ppc2 gene encoding a housekeeping isoform of PEPCfrom Mesembryanthemum crystallinum Nucleic Acids Research17 6743ndash6744

mdashmdashmdash and mdashmdashmdash 1989b Nucleotide sequence of the gene en-coding a CAM specific isoform of PEPC from Mesembryan-themum crystallinum Nucleic Acids Research 17 6745ndash6746

DAVIS J I and R J SORENG 1993 Phylogenetic structure in thegrass family (Poaceae) as determined from chloroplast DNArestriction site variation American Journal of Botany 80 1444ndash1454

DEVOS K M J DUBCOVSKY J DVORAK C N CHINOY and M DGALE 1995 Structural evolution of wheat chromosomes 4A

5A and 7B and its impact on recombination Theoretical andApplied Genetics 91 282ndash288

DEWEY D R 1982 Genomic and phylogenetic relationshipsamong North American perennial Triticeae Pp 51ndash88 inGrasses and grasslands eds J R Estes R J Tyrl and J NBrunken Norman Oklahoma University Press

mdashmdashmdash 1983 Historical and current taxonomic perspectives of Ag-ropyron Elymus and related genera Crop Science 23 637ndash642

mdashmdashmdash 1984 The genomic system of classification as a guide tointergeneric hybridization within the perennial Triticeae Pp209ndash279 in Gene manipulation in plant improvement Proceedingsof the 6th Stadler genetics symposium ed J P Gustafson NewYork Columbia University Press

DOEBLEY J R VON BOTHMER and S LARSON 1992 ChloroplastDNA variation and the phylogeny of Hordeum (Poaceae)American Journal of Botany 79 576ndash584

DOYLE J J and J L DOYLE 1987 A rapid DNA isolation proce-dure for small quantities of fresh leaf tissue PhytochemicalBulletin 19 11ndash15

mdashmdashmdash mdashmdashmdash A H D BROWN and R G PALMER 2002 Ge-nomes multiple origins and lineage recombination in theGlycine tomentella complex histone H3-D gene sequences Evo-lution 56 1388ndash1402

EMSHWILLER E and J J DOYLE 1998 Origins of domesticationand polyploidy in oca (Oxalis tuberosa Oxalidaceae) nrDNAITS data American Journal of Botany 85 975ndash985

mdashmdashmdash and mdashmdashmdash 2002 Origins of domestication and polyploidyin oca (Oxalis tuberosa Oxalidaceae) 2 Chloroplast-expressedglutamine synthetase data American Journal of Botany 891042ndash1056

ERNST K and P WESTHOFF 1997 The phosphoenolpyruvate car-boxylase (ppc) gene family of Flaveria trinervia (C4) and Fpringlei (C3) molecular characterization and expression anal-ysis of the ppcB and ppcC genes Plant Molecular Biology 34427ndash443

FELDMAN M I STRAUSS and A VARDI 1979 Chromosome pair-ing and fertility of F1 hybrids of Ae longissima and Ae searsiiCanadian Journal of Genetics and Cytology 21 261ndash272

FELSENSTEIN J 1981 Evolutionary trees from DNA sequences amaximum likelihood approach Journal of Molecular Evolution17 368ndash376

mdashmdashmdash 1985 Confidence limits on phylogenies an approach usingthe bootstrap Evolution 39 783ndash791

FRATI F C SIMON J SULLIVAN and D L SWOFFORD 1997 Evo-lution of the mitochondrial cytochrome oxidase II gene inCollembola Journal of Molecular Evolution 44 145ndash158

GALE M D and K M DEVOS 1998 Plant comparative geneticsafter ten years Genome 282 656ndash659

GE S T SANG B-R LU and D-Y HONG 1999 Phylogeny of ricegenomes with emphasis on origins of allotetraploid speciesProceedings of the National Academy of Sciences USA 96 14400ndash14405

GEHRIG H H V HEUTE and M KLUGE 2001 New partial se-quences of phosphoenolpyruvate carboxylase as molecularphylogenetic markers Molecular Phylogenetics and Evolution 20262ndash274

GOLDMAN N 1993 Statistical tests of models of DNA substitutionJournal of Molecular Evolution 36 182ndash198

GONZALEZ M C L OSUNA C ECHEVARRIA J VIDAL and F JCEJUDO 1998 Expression and localization of phosphoenol-pyruvate carboxylase in developing and germinating wheatgrains Plant Physiology 116 1249ndash1258

GU X Y-X Fu and W-H Li 1995 Maximum likelihood esti-mation of the heterogeneity of substitution rate among nu-cleotide sites Molecular Biology and Evolution 12 546ndash557

HASEGAWA M H KISHINO and T YANO 1985 Dating of thehuman-ape split by a molecular clock of mitochondrial DNAJournal of Molecular Evolution 22 160ndash174

854 [Volume 29SYSTEMATIC BOTANY

TABLE 3 Data characteristics 1 Approximately half of region 2 is missing for Secale montanum subsp montanum

PepC PepC Reduced waxy

Aligned LengthSequence Lengthmdashaverage (minimummaximum) Gap Charactersmdashaverage (minimummaximum) Missing Charactersmdashaverage (minimummaximum) Exon

21851573 (1418ndash1693)267 (211ndash339)14 (0ndash3311)

195

21261587 (1429ndash1710)254 (196ndash328)06 (00ndash41)

201

15301232 (1191ndash1299)195 (152ndash222)001 (00ndash01)499

GC ContentParsimony Informative Charactersmdash ()Number of Parsimony Trees (steps)Consistency IndexRetention Index

430 (421ndash444)97 (211)

3 (750)06100882

430 (421ndash439)90 (192)

6 (674)06310814

583 (568ndash595)122 (187)

30 (763)04780656

with members of Hordeum sect Critesion and the sec-ond (the St form) groups within Pseudoroegneria How-ever for two Elymus individuals we only have a par-tial sequence of one of the two forms for E canadensiswe have St sequences from region 2 but not from re-gion 1 and for E glaucus we have an H copy fromregion 2 but not from region 1 Because approximately60 of the nucleotides for these sequences are absentthey were not included in the phylogenetic analysesonly the H copy is included for E canadensis and onlythe St copy is included for E glaucus

Phylogenetic Analyses of PepC Data Parsimonyanalysis of the PepC data yielded three most-parsi-monious trees (750 steps CI excluding uninformativecharacters 5 0610 RI 5 0822) The only topologicaldifference between the three trees was the placementof two taxa within the Elymus 1 H clade (E lanceolatusand H jubatum) However once nodes with zero-lengthbranches were collapsed there was only one tree inwhich E lanceolatus and H jubatum were part of a po-lytomy within the Elymus 1 H clade The final MLsearch resulted in a single tree (2744462 Fig 2 seeMaterials and Methods lsquolsquoPhylogeny Reconstructionrsquorsquofor estimated parameters)

The ML and parsimony trees have the same generaltopology (Fig 2) In each E californicus is the first todiverge followed by Leymus racemosus Next Hordeumforms a weakly supported clade sister to a secondweakly supported clade containing the remainingmembers of the tribe Many of the other nodes resolvedin both the parsimony and ML trees are either incon-gruent or weakly supported but every clade with highbootstrap (BS) or posterior probabilities (PP) was re-covered by both methods of analysis

The PepC data suggest that the Elymus tetraploidscombine the H and St genomes There is a stronglysupported clade consisting of Hordeum jubatum Hbrachyantherum and Elymus (BS 5 98 and PP 5 100)A second strongly supported group includes Elymusand Pseudoroegneria spicata (BS and PP 5 100) Elymuscalifornicus is the only Elymus species that does notgroup with Hordeum or Pseudoroegneria but is foundnear the base of the tree There are four other lineageswith high bootstrap support and posterior probabili-

ties 1) an Agropyron cristatum and Eremopyrum orientaleclade (BS and PP 5 100) 2) an Aegilops comosa Triticumaestivum and Taeniatherum caput-medusae clade (BS 586 and PP 5 100) 3) a Secale L clade (BS 5 97 and PP5 100) and 4) a Hordeum murinum and H vulgare Lclade sister to the Elymus 1 H lineage (BS 5 83 andPP 5 99)

Comparison of PepC and waxy Sequence charac-teristics for the reduced PepC and waxy data sets areshown in Table 3 Parsimony analysis of the PepC datapartition yielded six most-parsimonious trees (674steps CI excluding uninformative characters 5 0631RI 5 0814) including two tree islands of three treeseach The strict consensus of all six trees is less re-solved than the strict consensus from the entire PepCdata set but has the same general topology and well-supported nodes (Fig 3) Among the six most-parsi-monious trees tree 6 had the highest log likelihoodscore and was chosen as the PepC tree for use in theSH test Once again the log likelihood value obtainedunder the (GTR1I1G) model was not significantly dif-ferent than the one obtained using (GTR1G) alone(2688734 vs2688770) so we used the simpler modelFigure 3 depicts parsimony tree 6 under the ML pa-rameters best fitting the PepC data [R (AC) 5 098 R(AG) 5 283 R (AT) 5 043 R (CG) 5 133 R (CT) 5342 R (GT) 5 100 pi (A) 5 025 pi (C) 5 020 pi(G) 5 021 pi (T) 5 033 a 5 052]

Parsimony analysis of the reduced waxy data yielded30 shortest trees (763 steps CI excluding uninforma-tive characters 5 0478 RI 5 0656) The strict consen-sus of these trees (Fig 4) is similar to the strict con-sensus trees of tree islands 3 and 4 in Mason-Gamerrsquos(2001) analysis of waxy data All most-parsimoniouswaxy trees include a single well-supported Hordeum 1Elymus clade (BS 5 89) Within this group the ElymusH sequences form a well-supported clade with Hbrachyantherum (BS 5 100) while H marinum H mu-rinum H vulgare and H jubatum form a weakly-sup-ported clade The Elymus and Pseudoroegneria sequenc-es are grouped into two distinct well-supportedclades (St1 and St2) but the St sequences do not forma monophyletic group The St1 clade (BS 5 100) is sis-ter to a weakly supported clade including Aegilops L

2004] 855HELFGOTT AND MASON-GAMER ELYMUS PHYLOGENY

FIG 2 The maximum likelihood tree based on PepC sequence data generated under a GTR1 G model of evolution ln (L52744462) Bold lines indicate the Elymus 1 Pseudoroegneria and Elymus 1 Hordeum clades Genome designations are providedfor Elymus Pseudoroegneria and Hordeum taxa The posterior probabilities are above the branches and bootstrap values ($ 75)are given in parentheses Gray lines represent portions of the tree that are inconsistent with the parsimony results PepC clonenumbers are provided next to taxon names

856 [Volume 29SYSTEMATIC BOTANY

FIG 3 The reduced PepC parsimony topology used as thetest tree in the SH test depicted using ML estimated param-eters best fitting the PepC data under GTR1G Bold lines in-dicate the Elymus 1 Pseudoroegneria and Elymus 1 Hordeumclades Bootstrap values greater than 49 are above the nodesNodes that are sources of conflict in the constrained analysesare marked with asterisks and the resulting increase in treelength in each of the associated constrained analyses is givenGray lines represent nodes that collapse in the strict consensusof all most parsimonious trees Taxon labels as in Fig 2

FIG 4 The reduced waxy parsimony tree used as the testtree in the SH test depicted using ML estimated parametersbest fitting the waxy data under GTR1I1G Bold lines indicatethe Elymus 1 Pseudoroegneria and Elymus 1 Hordeum cladesBootstrap values above 49 are above the nodes Nodes thatare sources of conflict in the constrained analyses are markedwith asterisks and the resulting increase in tree length in eachof the associated constrained analyses is given Gray lines rep-resent nodes that collapse in the strict consensus of all mostparsimonious trees Taxon labels as in Fig 2

Dasypyrum (Coss amp Durieu) P Candargy and Thino-pyrum ALove and St2 (BS 5 98) is sister to Australo-pyrum velutinum As in the PepC tree E californicusdoes not group with other Elymus species The twoSecale species form a strongly supported clade (BS 5100) sister to Heteranthelium piliferum (BS 5 68) Thisgroup is in turn sister to a clade including Agropyroncristatum Eremopyrum orientale and Taeniatherum caput-medusae (BS 5 68) Among the 30 most-parsimonioustrees tree 1 had the highest log likelihood score andwas chosen as the waxy tree for use in the SH test Themost parameter rich model (GTR1I1G) had a signif-icantly higher log likelihood value than the other mod-els tested and was used in the SH test Figure 4 de-picts parsimony tree 1 under the ML parameters bestfitting the waxy data [R (AC) 5 150 R (AG) 5 353R (AT) 5 136 R (CG) 5 166 R (CT) 5 487 R (GT)5 100 pi (A) 5 022 pi (C) 5 026 pi (G) 5 030 pi(T) 5 021 P-inv 5 035 a 5 071]

Both SH tests yielded significant p-values The SHtest using the PepC data and associated ML parametersresulted in a difference in tree scores of 55194 P 0001 (PepC score 52688770 waxy score 52743964)The SH test using the waxy data and associated MLparameters resulted in a difference in tree scores of

39726 P 0001 (waxy score 52598225 PepC score52637950)

In constrained analyses of the PepC data five of theten nodes tested from the waxy tree (Fig 4 asterisks)resulted in an increase in length relative to the uncon-strained PepC trees (674 steps) In constrained analysesof the waxy data five of the eight PepC tree nodes (Fig3 asterisks) resulted in increases in length over theunconstrained waxy trees (763 steps) The length in-creases associated with each node are shown in Figs3 and 4

DISCUSSION

Evidence of St 1 H Genome Content All tetraploidElymus individuals examined here have two copies ofthe PepC isoform sequenced in the study One copyforms a clade with Hordeum and the second with Pseu-doroegneria confirming that North American Elymusspecies do have an St 1 H genomic content and there-fore are allotetraploid derivatives of Hordeum and Pseu-doroegneria (Fig 2) The data thus confirm Deweyrsquos(1984) and Loversquos (1984) genomic concept of NorthAmerican Elymus in which all species (with the excep-tion of the octoploid E californicus) are allotetraploidscontaining the St and H genomes (Dewey 1984 Love1984) Some systematists have raised justifiable con-cerns regarding the use of genomic pairing data as the

2004] 857HELFGOTT AND MASON-GAMER ELYMUS PHYLOGENY

primary criterion for both grouping and ranking oftaxa in the Triticeae (eg Baum et al 1987 Kellogg1989 Seberg and Petersen 1998) However the PepCphylogeny corroborates the cytogenetic findings andis in agreement with previous studies based on bothisozyme (Jaaska 1995 1998) and waxy gene sequencedata (Mason-Gamer 2001) A classification systembased on chromosome pairing provides a reasonablestarting point in that it groups taxa based on a char-acter that appears to reflect evolutionary relatednessHowever its utility is limited because it cannot provideinsights into the hierarchical relationships among thetaxa examined (eg Kellogg 1989)

Elymus californicusmdashis it a Member of ElymusThe octoploid Elymus californicus is the only Elymusspecies in the present study in which the St and Hgenomes have not been detected This is not surpris-ing because the classification of E californicus has beenunstable since it was first described in 1880 and itsgenomic makeup remains uncertain (Love 1984 MaryBarkworth pers comm) In an attempt to gain in-sights into the genomic complement of this enigmaticspecies Jensen and Wang (1997) screened for the Stand Ns genomes using genome-specific RAPD mark-ers Their study found no evidence of the St genomebut did detect a marker specific to the Ns genomewhich is found in the diploid genus PsathyrostachysNevski and in the allopolyploid genus Leymus HochstFrom this information Jensen and Wang (1997) sug-gested that E californicus should be transferred to Ley-mus The PepC data neither refute nor confirm thesefindings directly placing Elymus californicus at the baseof the tree Nevertheless the PepC phylogeny clearlydemonstrates that E californicus does not belong withinElymus The waxy gene phylogeny by Mason-Gamer etal (2001) was also consistent with Jensen and Wangrsquos(1997) findings revealing a weak association betweenPsathyrostachys and E californicus in most of the most-parsimonious trees Within the chloroplast DNA phy-logeny by Mason-Gamer et al (2002) the placement ofE californicus is equivocal E californicus is weaklygrouped with Hordeum in some of the shortest treesThe PepC data add to the mounting evidence suggest-ing that E californicus does not belong within Elymusbut additional sampling and phylogenetic analyses areneeded to develop a firm understanding of how thisspecies is related to other members of the Triticeae

Elymus H and St Genomes The PepC data indicatethat all North American tetraploid species of Elymushave St 1 H genome content but there is little reso-lution among Elymus individuals in either the St or theH clade This pattern is consistent with the hypothesisthat the Elymus tetraploids are formed from a singleset of genome donors or multiple closely related setsof donors but the specific progenitor species involvedremain a mystery The Elymus H genome sequences

form a strongly supported clade with the two NorthAmerican Hordeum taxa included in the study (Hbrachyantherum and H jubatum) while the Elymus Stsequences form an unresolved polytomy with P spicatathe only North American Pseudoroegneria species (Fig2) The lack of resolution among the Elymus St copiesis similar to that previously observed among chloro-plast DNA sequences (Mason-Gamer et al 2002) In astudy based on the waxy gene (Mason-Gamer et al2001) St sequences formed two distinct clades whichdid not correspond to geographical or morphologicaldifferentiation It was unclear whether the pattern wasthe result of allelic variation or recent gene duplicationprior to the divergence of the St group The PepC andwaxy studies have provided hints on which Hordeumspecies may be most closely related to Elymus but ad-ditional sampling of both Hordeum and Pseudoroegneriais needed before we can develop a better picture of theorigin of the H and St genomes

Distinction Among Hordeum Genomes A detaileddiscussion of Hordeum evolution in light of the PepCphylogeny is impossible given the limited number oftaxa sampled However the data do resolve the fourHordeum genomes (H Xa I and Xu) and providesome insights into how these groups are related (Fig2) Taxa with genomes I (Hordeum vulgare) or Xu (Hor-deum murinum) form a strongly supported monophy-letic group which in turn is sister to a second well-supported clade that includes species with the H or Xa(Hordeum marinum) genomes This corroborates thestudy by Komatsuda et al (1999) of Hordeum diploidtaxa using Vrs1 nuclear intron sequence data The PepCresults are also in accordance with Petersen and Se-bergrsquos (2003) analysis of diploid Hordeum taxa inwhich they combined the chloroplast data of Doebleyet al (1992) and Komatsuda et al (1999) with datafrom the nuclear marker DMC1 (Petersen and Seberg2002 2003) and the chloroplast rbcL gene (Petersen andSeberg 1998) Nuclear and chloroplast DNA data werepartly contradictory but in the combined data set theXu and I genomes formed a sister group to the H andXa taxa although one H genome taxon (H brevisubu-latum Link) was sister to H marinum (Xa) Data fromthe nuclear ribosomal internal transcribed spacers(Frank Blattner pers comm) also recovered four ma-jor clades corresponding to the four Hordeum genomesbut the arrangement of the clades was somewhat dif-ferent Genomes H 1 Xa formed a clade sister to Xuand this clade is sister to I though the placement ofthe Xu clade is weakly supported It is clear from theseHordeum studies that the genome groups are in factphylogenetically informative

Conflict Between PepC and waxy Results from theSH test suggest that the PepC and waxy trees are sig-nificantly incongruent Constrained analyses confirmthese findings suggesting that there is disagreement

858 [Volume 29SYSTEMATIC BOTANY

involving some of the clades in the PepC and waxy phy-logenies (Figs 3 4) There are two cases within Hor-deum where taxa have different positions in the twotrees but are members of clades with moderate to highbootstrap support The first is Hordeum marinum in thePepC tree it is sister to the H clade while in the waxytree it is sister to H murinum (Xu) In the second casethe same accession of H jubatum has a different posi-tion in the PepC and waxy trees In the PepC phylogenyit is nested within the well-supported H clade whilein the waxy phylogeny it is found at the base of a gradeincluding H marinum (Xa) H murinum (Xu) and Hvulgare (I) While the reason for the conflicting place-ment of H jubatum is not known it is an allotetraploid(eg Joslashrgensen 1986 Bothmer et al 1987) and the twodata sets may have sampled gene copies representingthe two different homoeologous genomes

The placement of Taeniatherum caput-medusae is alsoat odds between the PepC and waxy phylogenies (Figs3 4) In the PepC tree T caput-medusae forms a well-supported clade (BS 5 95) with Aegilops comosa whilein the waxy tree it is sister to Eremopyrum orientale (BS5 50) which in turn is sister to Agropyron cristatum (BS5 70) The association seen in the waxy topology hasnot been seen in other gene trees but a close relation-ship between T caput-medusae and Aegilops was seen inthe chloroplast DNA phylogeny (Mason-Gamer et al2002) The PepC data strongly conflict with the waxydata on the placement of Taeniatherum with 36 addi-tional steps required in the constrained analysis Thewaxy data show far less conflict with Taeniatherumrsquosplacement in the PepC tree requiring only two addi-tional steps in the constrained analysis There may besome underlying signal within the waxy data set thatsupports a more congruent placement of TaeniatherumIn the PepC phylogeny E orientale forms a stronglysupported clade with A cristatum (BS 5 100) ratherthan Taeniatherum in agreement with the chloroplastDNA phylogeny (Mason-Gamer et al 2002) and theDMC1 phylogeny (Petersen and Seberg 2002) in whichAgropyron cristatum was grouped with two other Ere-mopyrum species E distans (KKoch) Nevski and E tri-ticeum (Gaertn) Nevski

Secale is well-supported on both trees (BS 5 100)which is not surprising A correlation between mono-genomic groups and clades on molecular phylogenetictrees has been previously reported (Hsiao et al 1995Kellogg and Appels 1995 Kellogg et al 1996 Mason-Gamer and Kellogg 1996a 1996b Petersen and Seberg1997) In the waxy tree the Secale clade is sister to Het-eranthelium piliferum forming a weakly supportedclade (BS 5 68) this association was previously sup-ported in Mason-Gamerrsquos chloroplast DNA phylogeny

Phylogenetic conflict was previously demonstratedamong Triticeae gene trees with the most extensivedifferences documented in comparisons between chlo-

roplast and nuclear data (Kellogg et al 1996 Mason-Gamer and Kellogg 1996b) There were a few instancesof statistically significant incongruence among the nu-clear data sets and the present comparison betweenPepC and waxy shows conflict involving several taxa Ifthe conflict is genuine the nuclear gene trees may re-flect differences among phylogenies of the chromo-somes or sections of chromosomes on which thegenes reside The discordance between PepC and waxygene tree topologies (as well as many others) high-lights the need for continued research on chromosomeevolution in the Triticeae in the hopes of better under-standing mechanisms behind the complex historicalpatterns that are emerging from phylogenetic studiesof this group

Two different historical scenarios provide possibleexplanations for the reticulate history of the tribe (Kel-logg et al 1996) First the pattern may result fromperiods of cladogenesis and dispersal interspersedwith hybridization and introgression between speciesThe level of hybridization slowed or stopped at somepoint in the past most genera in the tribe are inter-sterile and hybrids among them generally exhibit alack of chromosome pairing (eg Dewey 1984 and ref-erences therein) However some degree gene exchangeamong genera is possible if a small percentage of vi-able hybrid seeds are produced through chance aloneIt is also possible that present-day gene flow betweendiploid taxa is taking place through a polyploid inter-mediate Gene flow between diploids via a tetraploidbridge requires the formation of partially fertile trip-loids as has been reported for species of Aegilops byVardi (1973) Feldman et al (1979) and Kimber andFeldman (1987) There is also phylogenetic evidence ofhybridization or introgression Mason-Gamer and Kel-logg (2000) reported that an accession of Pseudoroeg-neria strigosa (M Bieb) ALove which was expected tobe a monogenomic St species has both St and H cop-ies of the waxy gene and an St chloroplast The resultwas consistent with both the allotetraploid-bridge anddirect hybridization hypotheses

A second possible explanation for the reticulate his-tory of the Triticeae is that it reflects a burst of diver-sification with rapid lineage sorting and fixation ofdifferent alleles into lineages which quickly becameintersterile (Kellogg et al 1996) Given the need forextensive ancestral polymorphism in this hypothesisit seems unlikely that rapid diversification and lineagesorting was the major driving force behind the retic-ulate history of the Triticeae However it is not clearhow we could determine which scenario (lineage sort-ing and hybridization-introgression) has had the big-ger impact on the history of the Triticeae In fact thesehypotheses are not mutually exclusive and it may bethat both processes have played some role in shapingthe history of the Triticeae

2004] 859HELFGOTT AND MASON-GAMER ELYMUS PHYLOGENY

In conclusion this study verifies Deweyrsquos (19821983) and Loversquos (1982 1984) insightful chromosome-based revisions of the Triticeae The PepC results clear-ly show that North American tetraploid members ofElymus have an St 1 H genomic content and are there-fore derived from Hordeum and Pseudoroegneria How-ever the current data reveal little genetic differentia-tion within North American Elymus Consequently therelationships among Elymus species and their relation-ships to specific progenitor species await a better-re-solved phylogeny Additional studies which integratemorphological cytological and phylogenetic informa-tion are necessary to understand better the historicalrelationships within Elymus The PepC-based analysisof North American Elymus is a small step forward inthis process

ACKNOWLEDGEMENTS The authors wish to thank Mary Bark-worth for sharing plant material the USDArsquos Germplasm Resourc-es Information Network for seeds Gregory Plunkett and twoanonymous reviewers for helpful comments on the manuscriptThe National Science Foundation (DEB-9974181) funded this work

LITERATURE CITED

BARKWORTH M E and D R DEWEY 1985 Genomically basedgenera in the perennial Triticeae of North America identifi-cation and membership American Journal of Botany 72 767ndash776

BAUM B R J R ESTES and P K GUPTA 1987 Assessment of thegenomic system of classification in the Triticeae AmericanJournal of Botany 74 1388ndash1395

BESNARD G B OFFMANN C ROBERT and C ROUCH 2002 As-sessment of the C4 phosphoenolpyruvate carboxylase genediversity in grasses (Poaceae) Theoretical and Applied Genetics105 404ndash412

BOTHMER R VON J FLINK and T LANDSTROM 1987 Meiosis inHordeum interspecific hybrids II Triploid hybrids Evolution-ary Trends in Plants 1 41ndash50

CRETIN C S SANTI E KERYER L LEPINIEC D TAGU J VIDALand P GADAL 1991 The phosphoenolpyruvate carboxylasegene family in Sorghum promoter structure amino acid se-quences and expression of genes Gene 99 87ndash94

CRONN R C R L SMALL T HASELKORN and J F WENDEL 2003Cryptic repeated genomic recombination during speciation inGossypium gossypioides Evolution 57 2475ndash2489

mdashmdashmdash X ZHAO A H PATERSON and J F WENDEL 1996 Poly-morphism and concerted evolution in a tandemly repeatedgene family 5S ribosomal DNA in diploid and allopolyploidcottons Journal of Molecular Evolution 42 685ndash705

CRONQUIST A 1987 A botanical critique of cladism Botanical Re-view 53 1ndash52

CUSHMAN J C and H J BOHNERT 1989a Nucleotide sequence ofthe Ppc2 gene encoding a housekeeping isoform of PEPCfrom Mesembryanthemum crystallinum Nucleic Acids Research17 6743ndash6744

mdashmdashmdash and mdashmdashmdash 1989b Nucleotide sequence of the gene en-coding a CAM specific isoform of PEPC from Mesembryan-themum crystallinum Nucleic Acids Research 17 6745ndash6746

DAVIS J I and R J SORENG 1993 Phylogenetic structure in thegrass family (Poaceae) as determined from chloroplast DNArestriction site variation American Journal of Botany 80 1444ndash1454

DEVOS K M J DUBCOVSKY J DVORAK C N CHINOY and M DGALE 1995 Structural evolution of wheat chromosomes 4A

5A and 7B and its impact on recombination Theoretical andApplied Genetics 91 282ndash288

DEWEY D R 1982 Genomic and phylogenetic relationshipsamong North American perennial Triticeae Pp 51ndash88 inGrasses and grasslands eds J R Estes R J Tyrl and J NBrunken Norman Oklahoma University Press

mdashmdashmdash 1983 Historical and current taxonomic perspectives of Ag-ropyron Elymus and related genera Crop Science 23 637ndash642

mdashmdashmdash 1984 The genomic system of classification as a guide tointergeneric hybridization within the perennial Triticeae Pp209ndash279 in Gene manipulation in plant improvement Proceedingsof the 6th Stadler genetics symposium ed J P Gustafson NewYork Columbia University Press

DOEBLEY J R VON BOTHMER and S LARSON 1992 ChloroplastDNA variation and the phylogeny of Hordeum (Poaceae)American Journal of Botany 79 576ndash584

DOYLE J J and J L DOYLE 1987 A rapid DNA isolation proce-dure for small quantities of fresh leaf tissue PhytochemicalBulletin 19 11ndash15

mdashmdashmdash mdashmdashmdash A H D BROWN and R G PALMER 2002 Ge-nomes multiple origins and lineage recombination in theGlycine tomentella complex histone H3-D gene sequences Evo-lution 56 1388ndash1402

EMSHWILLER E and J J DOYLE 1998 Origins of domesticationand polyploidy in oca (Oxalis tuberosa Oxalidaceae) nrDNAITS data American Journal of Botany 85 975ndash985

mdashmdashmdash and mdashmdashmdash 2002 Origins of domestication and polyploidyin oca (Oxalis tuberosa Oxalidaceae) 2 Chloroplast-expressedglutamine synthetase data American Journal of Botany 891042ndash1056

ERNST K and P WESTHOFF 1997 The phosphoenolpyruvate car-boxylase (ppc) gene family of Flaveria trinervia (C4) and Fpringlei (C3) molecular characterization and expression anal-ysis of the ppcB and ppcC genes Plant Molecular Biology 34427ndash443

FELDMAN M I STRAUSS and A VARDI 1979 Chromosome pair-ing and fertility of F1 hybrids of Ae longissima and Ae searsiiCanadian Journal of Genetics and Cytology 21 261ndash272

FELSENSTEIN J 1981 Evolutionary trees from DNA sequences amaximum likelihood approach Journal of Molecular Evolution17 368ndash376

mdashmdashmdash 1985 Confidence limits on phylogenies an approach usingthe bootstrap Evolution 39 783ndash791

FRATI F C SIMON J SULLIVAN and D L SWOFFORD 1997 Evo-lution of the mitochondrial cytochrome oxidase II gene inCollembola Journal of Molecular Evolution 44 145ndash158

GALE M D and K M DEVOS 1998 Plant comparative geneticsafter ten years Genome 282 656ndash659

GE S T SANG B-R LU and D-Y HONG 1999 Phylogeny of ricegenomes with emphasis on origins of allotetraploid speciesProceedings of the National Academy of Sciences USA 96 14400ndash14405

GEHRIG H H V HEUTE and M KLUGE 2001 New partial se-quences of phosphoenolpyruvate carboxylase as molecularphylogenetic markers Molecular Phylogenetics and Evolution 20262ndash274

GOLDMAN N 1993 Statistical tests of models of DNA substitutionJournal of Molecular Evolution 36 182ndash198

GONZALEZ M C L OSUNA C ECHEVARRIA J VIDAL and F JCEJUDO 1998 Expression and localization of phosphoenol-pyruvate carboxylase in developing and germinating wheatgrains Plant Physiology 116 1249ndash1258

GU X Y-X Fu and W-H Li 1995 Maximum likelihood esti-mation of the heterogeneity of substitution rate among nu-cleotide sites Molecular Biology and Evolution 12 546ndash557

HASEGAWA M H KISHINO and T YANO 1985 Dating of thehuman-ape split by a molecular clock of mitochondrial DNAJournal of Molecular Evolution 22 160ndash174

2004] 855HELFGOTT AND MASON-GAMER ELYMUS PHYLOGENY

FIG 2 The maximum likelihood tree based on PepC sequence data generated under a GTR1 G model of evolution ln (L52744462) Bold lines indicate the Elymus 1 Pseudoroegneria and Elymus 1 Hordeum clades Genome designations are providedfor Elymus Pseudoroegneria and Hordeum taxa The posterior probabilities are above the branches and bootstrap values ($ 75)are given in parentheses Gray lines represent portions of the tree that are inconsistent with the parsimony results PepC clonenumbers are provided next to taxon names

856 [Volume 29SYSTEMATIC BOTANY

FIG 3 The reduced PepC parsimony topology used as thetest tree in the SH test depicted using ML estimated param-eters best fitting the PepC data under GTR1G Bold lines in-dicate the Elymus 1 Pseudoroegneria and Elymus 1 Hordeumclades Bootstrap values greater than 49 are above the nodesNodes that are sources of conflict in the constrained analysesare marked with asterisks and the resulting increase in treelength in each of the associated constrained analyses is givenGray lines represent nodes that collapse in the strict consensusof all most parsimonious trees Taxon labels as in Fig 2

FIG 4 The reduced waxy parsimony tree used as the testtree in the SH test depicted using ML estimated parametersbest fitting the waxy data under GTR1I1G Bold lines indicatethe Elymus 1 Pseudoroegneria and Elymus 1 Hordeum cladesBootstrap values above 49 are above the nodes Nodes thatare sources of conflict in the constrained analyses are markedwith asterisks and the resulting increase in tree length in eachof the associated constrained analyses is given Gray lines rep-resent nodes that collapse in the strict consensus of all mostparsimonious trees Taxon labels as in Fig 2

Dasypyrum (Coss amp Durieu) P Candargy and Thino-pyrum ALove and St2 (BS 5 98) is sister to Australo-pyrum velutinum As in the PepC tree E californicusdoes not group with other Elymus species The twoSecale species form a strongly supported clade (BS 5100) sister to Heteranthelium piliferum (BS 5 68) Thisgroup is in turn sister to a clade including Agropyroncristatum Eremopyrum orientale and Taeniatherum caput-medusae (BS 5 68) Among the 30 most-parsimonioustrees tree 1 had the highest log likelihood score andwas chosen as the waxy tree for use in the SH test Themost parameter rich model (GTR1I1G) had a signif-icantly higher log likelihood value than the other mod-els tested and was used in the SH test Figure 4 de-picts parsimony tree 1 under the ML parameters bestfitting the waxy data [R (AC) 5 150 R (AG) 5 353R (AT) 5 136 R (CG) 5 166 R (CT) 5 487 R (GT)5 100 pi (A) 5 022 pi (C) 5 026 pi (G) 5 030 pi(T) 5 021 P-inv 5 035 a 5 071]

Both SH tests yielded significant p-values The SHtest using the PepC data and associated ML parametersresulted in a difference in tree scores of 55194 P 0001 (PepC score 52688770 waxy score 52743964)The SH test using the waxy data and associated MLparameters resulted in a difference in tree scores of

39726 P 0001 (waxy score 52598225 PepC score52637950)

In constrained analyses of the PepC data five of theten nodes tested from the waxy tree (Fig 4 asterisks)resulted in an increase in length relative to the uncon-strained PepC trees (674 steps) In constrained analysesof the waxy data five of the eight PepC tree nodes (Fig3 asterisks) resulted in increases in length over theunconstrained waxy trees (763 steps) The length in-creases associated with each node are shown in Figs3 and 4

DISCUSSION

Evidence of St 1 H Genome Content All tetraploidElymus individuals examined here have two copies ofthe PepC isoform sequenced in the study One copyforms a clade with Hordeum and the second with Pseu-doroegneria confirming that North American Elymusspecies do have an St 1 H genomic content and there-fore are allotetraploid derivatives of Hordeum and Pseu-doroegneria (Fig 2) The data thus confirm Deweyrsquos(1984) and Loversquos (1984) genomic concept of NorthAmerican Elymus in which all species (with the excep-tion of the octoploid E californicus) are allotetraploidscontaining the St and H genomes (Dewey 1984 Love1984) Some systematists have raised justifiable con-cerns regarding the use of genomic pairing data as the

2004] 857HELFGOTT AND MASON-GAMER ELYMUS PHYLOGENY

primary criterion for both grouping and ranking oftaxa in the Triticeae (eg Baum et al 1987 Kellogg1989 Seberg and Petersen 1998) However the PepCphylogeny corroborates the cytogenetic findings andis in agreement with previous studies based on bothisozyme (Jaaska 1995 1998) and waxy gene sequencedata (Mason-Gamer 2001) A classification systembased on chromosome pairing provides a reasonablestarting point in that it groups taxa based on a char-acter that appears to reflect evolutionary relatednessHowever its utility is limited because it cannot provideinsights into the hierarchical relationships among thetaxa examined (eg Kellogg 1989)

Elymus californicusmdashis it a Member of ElymusThe octoploid Elymus californicus is the only Elymusspecies in the present study in which the St and Hgenomes have not been detected This is not surpris-ing because the classification of E californicus has beenunstable since it was first described in 1880 and itsgenomic makeup remains uncertain (Love 1984 MaryBarkworth pers comm) In an attempt to gain in-sights into the genomic complement of this enigmaticspecies Jensen and Wang (1997) screened for the Stand Ns genomes using genome-specific RAPD mark-ers Their study found no evidence of the St genomebut did detect a marker specific to the Ns genomewhich is found in the diploid genus PsathyrostachysNevski and in the allopolyploid genus Leymus HochstFrom this information Jensen and Wang (1997) sug-gested that E californicus should be transferred to Ley-mus The PepC data neither refute nor confirm thesefindings directly placing Elymus californicus at the baseof the tree Nevertheless the PepC phylogeny clearlydemonstrates that E californicus does not belong withinElymus The waxy gene phylogeny by Mason-Gamer etal (2001) was also consistent with Jensen and Wangrsquos(1997) findings revealing a weak association betweenPsathyrostachys and E californicus in most of the most-parsimonious trees Within the chloroplast DNA phy-logeny by Mason-Gamer et al (2002) the placement ofE californicus is equivocal E californicus is weaklygrouped with Hordeum in some of the shortest treesThe PepC data add to the mounting evidence suggest-ing that E californicus does not belong within Elymusbut additional sampling and phylogenetic analyses areneeded to develop a firm understanding of how thisspecies is related to other members of the Triticeae

Elymus H and St Genomes The PepC data indicatethat all North American tetraploid species of Elymushave St 1 H genome content but there is little reso-lution among Elymus individuals in either the St or theH clade This pattern is consistent with the hypothesisthat the Elymus tetraploids are formed from a singleset of genome donors or multiple closely related setsof donors but the specific progenitor species involvedremain a mystery The Elymus H genome sequences

form a strongly supported clade with the two NorthAmerican Hordeum taxa included in the study (Hbrachyantherum and H jubatum) while the Elymus Stsequences form an unresolved polytomy with P spicatathe only North American Pseudoroegneria species (Fig2) The lack of resolution among the Elymus St copiesis similar to that previously observed among chloro-plast DNA sequences (Mason-Gamer et al 2002) In astudy based on the waxy gene (Mason-Gamer et al2001) St sequences formed two distinct clades whichdid not correspond to geographical or morphologicaldifferentiation It was unclear whether the pattern wasthe result of allelic variation or recent gene duplicationprior to the divergence of the St group The PepC andwaxy studies have provided hints on which Hordeumspecies may be most closely related to Elymus but ad-ditional sampling of both Hordeum and Pseudoroegneriais needed before we can develop a better picture of theorigin of the H and St genomes

Distinction Among Hordeum Genomes A detaileddiscussion of Hordeum evolution in light of the PepCphylogeny is impossible given the limited number oftaxa sampled However the data do resolve the fourHordeum genomes (H Xa I and Xu) and providesome insights into how these groups are related (Fig2) Taxa with genomes I (Hordeum vulgare) or Xu (Hor-deum murinum) form a strongly supported monophy-letic group which in turn is sister to a second well-supported clade that includes species with the H or Xa(Hordeum marinum) genomes This corroborates thestudy by Komatsuda et al (1999) of Hordeum diploidtaxa using Vrs1 nuclear intron sequence data The PepCresults are also in accordance with Petersen and Se-bergrsquos (2003) analysis of diploid Hordeum taxa inwhich they combined the chloroplast data of Doebleyet al (1992) and Komatsuda et al (1999) with datafrom the nuclear marker DMC1 (Petersen and Seberg2002 2003) and the chloroplast rbcL gene (Petersen andSeberg 1998) Nuclear and chloroplast DNA data werepartly contradictory but in the combined data set theXu and I genomes formed a sister group to the H andXa taxa although one H genome taxon (H brevisubu-latum Link) was sister to H marinum (Xa) Data fromthe nuclear ribosomal internal transcribed spacers(Frank Blattner pers comm) also recovered four ma-jor clades corresponding to the four Hordeum genomesbut the arrangement of the clades was somewhat dif-ferent Genomes H 1 Xa formed a clade sister to Xuand this clade is sister to I though the placement ofthe Xu clade is weakly supported It is clear from theseHordeum studies that the genome groups are in factphylogenetically informative

Conflict Between PepC and waxy Results from theSH test suggest that the PepC and waxy trees are sig-nificantly incongruent Constrained analyses confirmthese findings suggesting that there is disagreement

858 [Volume 29SYSTEMATIC BOTANY

involving some of the clades in the PepC and waxy phy-logenies (Figs 3 4) There are two cases within Hor-deum where taxa have different positions in the twotrees but are members of clades with moderate to highbootstrap support The first is Hordeum marinum in thePepC tree it is sister to the H clade while in the waxytree it is sister to H murinum (Xu) In the second casethe same accession of H jubatum has a different posi-tion in the PepC and waxy trees In the PepC phylogenyit is nested within the well-supported H clade whilein the waxy phylogeny it is found at the base of a gradeincluding H marinum (Xa) H murinum (Xu) and Hvulgare (I) While the reason for the conflicting place-ment of H jubatum is not known it is an allotetraploid(eg Joslashrgensen 1986 Bothmer et al 1987) and the twodata sets may have sampled gene copies representingthe two different homoeologous genomes

The placement of Taeniatherum caput-medusae is alsoat odds between the PepC and waxy phylogenies (Figs3 4) In the PepC tree T caput-medusae forms a well-supported clade (BS 5 95) with Aegilops comosa whilein the waxy tree it is sister to Eremopyrum orientale (BS5 50) which in turn is sister to Agropyron cristatum (BS5 70) The association seen in the waxy topology hasnot been seen in other gene trees but a close relation-ship between T caput-medusae and Aegilops was seen inthe chloroplast DNA phylogeny (Mason-Gamer et al2002) The PepC data strongly conflict with the waxydata on the placement of Taeniatherum with 36 addi-tional steps required in the constrained analysis Thewaxy data show far less conflict with Taeniatherumrsquosplacement in the PepC tree requiring only two addi-tional steps in the constrained analysis There may besome underlying signal within the waxy data set thatsupports a more congruent placement of TaeniatherumIn the PepC phylogeny E orientale forms a stronglysupported clade with A cristatum (BS 5 100) ratherthan Taeniatherum in agreement with the chloroplastDNA phylogeny (Mason-Gamer et al 2002) and theDMC1 phylogeny (Petersen and Seberg 2002) in whichAgropyron cristatum was grouped with two other Ere-mopyrum species E distans (KKoch) Nevski and E tri-ticeum (Gaertn) Nevski

Secale is well-supported on both trees (BS 5 100)which is not surprising A correlation between mono-genomic groups and clades on molecular phylogenetictrees has been previously reported (Hsiao et al 1995Kellogg and Appels 1995 Kellogg et al 1996 Mason-Gamer and Kellogg 1996a 1996b Petersen and Seberg1997) In the waxy tree the Secale clade is sister to Het-eranthelium piliferum forming a weakly supportedclade (BS 5 68) this association was previously sup-ported in Mason-Gamerrsquos chloroplast DNA phylogeny

Phylogenetic conflict was previously demonstratedamong Triticeae gene trees with the most extensivedifferences documented in comparisons between chlo-

roplast and nuclear data (Kellogg et al 1996 Mason-Gamer and Kellogg 1996b) There were a few instancesof statistically significant incongruence among the nu-clear data sets and the present comparison betweenPepC and waxy shows conflict involving several taxa Ifthe conflict is genuine the nuclear gene trees may re-flect differences among phylogenies of the chromo-somes or sections of chromosomes on which thegenes reside The discordance between PepC and waxygene tree topologies (as well as many others) high-lights the need for continued research on chromosomeevolution in the Triticeae in the hopes of better under-standing mechanisms behind the complex historicalpatterns that are emerging from phylogenetic studiesof this group

Two different historical scenarios provide possibleexplanations for the reticulate history of the tribe (Kel-logg et al 1996) First the pattern may result fromperiods of cladogenesis and dispersal interspersedwith hybridization and introgression between speciesThe level of hybridization slowed or stopped at somepoint in the past most genera in the tribe are inter-sterile and hybrids among them generally exhibit alack of chromosome pairing (eg Dewey 1984 and ref-erences therein) However some degree gene exchangeamong genera is possible if a small percentage of vi-able hybrid seeds are produced through chance aloneIt is also possible that present-day gene flow betweendiploid taxa is taking place through a polyploid inter-mediate Gene flow between diploids via a tetraploidbridge requires the formation of partially fertile trip-loids as has been reported for species of Aegilops byVardi (1973) Feldman et al (1979) and Kimber andFeldman (1987) There is also phylogenetic evidence ofhybridization or introgression Mason-Gamer and Kel-logg (2000) reported that an accession of Pseudoroeg-neria strigosa (M Bieb) ALove which was expected tobe a monogenomic St species has both St and H cop-ies of the waxy gene and an St chloroplast The resultwas consistent with both the allotetraploid-bridge anddirect hybridization hypotheses

A second possible explanation for the reticulate his-tory of the Triticeae is that it reflects a burst of diver-sification with rapid lineage sorting and fixation ofdifferent alleles into lineages which quickly becameintersterile (Kellogg et al 1996) Given the need forextensive ancestral polymorphism in this hypothesisit seems unlikely that rapid diversification and lineagesorting was the major driving force behind the retic-ulate history of the Triticeae However it is not clearhow we could determine which scenario (lineage sort-ing and hybridization-introgression) has had the big-ger impact on the history of the Triticeae In fact thesehypotheses are not mutually exclusive and it may bethat both processes have played some role in shapingthe history of the Triticeae

2004] 859HELFGOTT AND MASON-GAMER ELYMUS PHYLOGENY

In conclusion this study verifies Deweyrsquos (19821983) and Loversquos (1982 1984) insightful chromosome-based revisions of the Triticeae The PepC results clear-ly show that North American tetraploid members ofElymus have an St 1 H genomic content and are there-fore derived from Hordeum and Pseudoroegneria How-ever the current data reveal little genetic differentia-tion within North American Elymus Consequently therelationships among Elymus species and their relation-ships to specific progenitor species await a better-re-solved phylogeny Additional studies which integratemorphological cytological and phylogenetic informa-tion are necessary to understand better the historicalrelationships within Elymus The PepC-based analysisof North American Elymus is a small step forward inthis process

ACKNOWLEDGEMENTS The authors wish to thank Mary Bark-worth for sharing plant material the USDArsquos Germplasm Resourc-es Information Network for seeds Gregory Plunkett and twoanonymous reviewers for helpful comments on the manuscriptThe National Science Foundation (DEB-9974181) funded this work

LITERATURE CITED

BARKWORTH M E and D R DEWEY 1985 Genomically basedgenera in the perennial Triticeae of North America identifi-cation and membership American Journal of Botany 72 767ndash776

BAUM B R J R ESTES and P K GUPTA 1987 Assessment of thegenomic system of classification in the Triticeae AmericanJournal of Botany 74 1388ndash1395

BESNARD G B OFFMANN C ROBERT and C ROUCH 2002 As-sessment of the C4 phosphoenolpyruvate carboxylase genediversity in grasses (Poaceae) Theoretical and Applied Genetics105 404ndash412

BOTHMER R VON J FLINK and T LANDSTROM 1987 Meiosis inHordeum interspecific hybrids II Triploid hybrids Evolution-ary Trends in Plants 1 41ndash50

CRETIN C S SANTI E KERYER L LEPINIEC D TAGU J VIDALand P GADAL 1991 The phosphoenolpyruvate carboxylasegene family in Sorghum promoter structure amino acid se-quences and expression of genes Gene 99 87ndash94

CRONN R C R L SMALL T HASELKORN and J F WENDEL 2003Cryptic repeated genomic recombination during speciation inGossypium gossypioides Evolution 57 2475ndash2489

mdashmdashmdash X ZHAO A H PATERSON and J F WENDEL 1996 Poly-morphism and concerted evolution in a tandemly repeatedgene family 5S ribosomal DNA in diploid and allopolyploidcottons Journal of Molecular Evolution 42 685ndash705

CRONQUIST A 1987 A botanical critique of cladism Botanical Re-view 53 1ndash52

CUSHMAN J C and H J BOHNERT 1989a Nucleotide sequence ofthe Ppc2 gene encoding a housekeeping isoform of PEPCfrom Mesembryanthemum crystallinum Nucleic Acids Research17 6743ndash6744

mdashmdashmdash and mdashmdashmdash 1989b Nucleotide sequence of the gene en-coding a CAM specific isoform of PEPC from Mesembryan-themum crystallinum Nucleic Acids Research 17 6745ndash6746

DAVIS J I and R J SORENG 1993 Phylogenetic structure in thegrass family (Poaceae) as determined from chloroplast DNArestriction site variation American Journal of Botany 80 1444ndash1454

DEVOS K M J DUBCOVSKY J DVORAK C N CHINOY and M DGALE 1995 Structural evolution of wheat chromosomes 4A

5A and 7B and its impact on recombination Theoretical andApplied Genetics 91 282ndash288

DEWEY D R 1982 Genomic and phylogenetic relationshipsamong North American perennial Triticeae Pp 51ndash88 inGrasses and grasslands eds J R Estes R J Tyrl and J NBrunken Norman Oklahoma University Press

mdashmdashmdash 1983 Historical and current taxonomic perspectives of Ag-ropyron Elymus and related genera Crop Science 23 637ndash642

mdashmdashmdash 1984 The genomic system of classification as a guide tointergeneric hybridization within the perennial Triticeae Pp209ndash279 in Gene manipulation in plant improvement Proceedingsof the 6th Stadler genetics symposium ed J P Gustafson NewYork Columbia University Press

DOEBLEY J R VON BOTHMER and S LARSON 1992 ChloroplastDNA variation and the phylogeny of Hordeum (Poaceae)American Journal of Botany 79 576ndash584

DOYLE J J and J L DOYLE 1987 A rapid DNA isolation proce-dure for small quantities of fresh leaf tissue PhytochemicalBulletin 19 11ndash15

mdashmdashmdash mdashmdashmdash A H D BROWN and R G PALMER 2002 Ge-nomes multiple origins and lineage recombination in theGlycine tomentella complex histone H3-D gene sequences Evo-lution 56 1388ndash1402

EMSHWILLER E and J J DOYLE 1998 Origins of domesticationand polyploidy in oca (Oxalis tuberosa Oxalidaceae) nrDNAITS data American Journal of Botany 85 975ndash985

mdashmdashmdash and mdashmdashmdash 2002 Origins of domestication and polyploidyin oca (Oxalis tuberosa Oxalidaceae) 2 Chloroplast-expressedglutamine synthetase data American Journal of Botany 891042ndash1056

ERNST K and P WESTHOFF 1997 The phosphoenolpyruvate car-boxylase (ppc) gene family of Flaveria trinervia (C4) and Fpringlei (C3) molecular characterization and expression anal-ysis of the ppcB and ppcC genes Plant Molecular Biology 34427ndash443

FELDMAN M I STRAUSS and A VARDI 1979 Chromosome pair-ing and fertility of F1 hybrids of Ae longissima and Ae searsiiCanadian Journal of Genetics and Cytology 21 261ndash272

FELSENSTEIN J 1981 Evolutionary trees from DNA sequences amaximum likelihood approach Journal of Molecular Evolution17 368ndash376

mdashmdashmdash 1985 Confidence limits on phylogenies an approach usingthe bootstrap Evolution 39 783ndash791

FRATI F C SIMON J SULLIVAN and D L SWOFFORD 1997 Evo-lution of the mitochondrial cytochrome oxidase II gene inCollembola Journal of Molecular Evolution 44 145ndash158

GALE M D and K M DEVOS 1998 Plant comparative geneticsafter ten years Genome 282 656ndash659

GE S T SANG B-R LU and D-Y HONG 1999 Phylogeny of ricegenomes with emphasis on origins of allotetraploid speciesProceedings of the National Academy of Sciences USA 96 14400ndash14405

GEHRIG H H V HEUTE and M KLUGE 2001 New partial se-quences of phosphoenolpyruvate carboxylase as molecularphylogenetic markers Molecular Phylogenetics and Evolution 20262ndash274

GOLDMAN N 1993 Statistical tests of models of DNA substitutionJournal of Molecular Evolution 36 182ndash198

GONZALEZ M C L OSUNA C ECHEVARRIA J VIDAL and F JCEJUDO 1998 Expression and localization of phosphoenol-pyruvate carboxylase in developing and germinating wheatgrains Plant Physiology 116 1249ndash1258

GU X Y-X Fu and W-H Li 1995 Maximum likelihood esti-mation of the heterogeneity of substitution rate among nu-cleotide sites Molecular Biology and Evolution 12 546ndash557

HASEGAWA M H KISHINO and T YANO 1985 Dating of thehuman-ape split by a molecular clock of mitochondrial DNAJournal of Molecular Evolution 22 160ndash174

856 [Volume 29SYSTEMATIC BOTANY

FIG 3 The reduced PepC parsimony topology used as thetest tree in the SH test depicted using ML estimated param-eters best fitting the PepC data under GTR1G Bold lines in-dicate the Elymus 1 Pseudoroegneria and Elymus 1 Hordeumclades Bootstrap values greater than 49 are above the nodesNodes that are sources of conflict in the constrained analysesare marked with asterisks and the resulting increase in treelength in each of the associated constrained analyses is givenGray lines represent nodes that collapse in the strict consensusof all most parsimonious trees Taxon labels as in Fig 2

FIG 4 The reduced waxy parsimony tree used as the testtree in the SH test depicted using ML estimated parametersbest fitting the waxy data under GTR1I1G Bold lines indicatethe Elymus 1 Pseudoroegneria and Elymus 1 Hordeum cladesBootstrap values above 49 are above the nodes Nodes thatare sources of conflict in the constrained analyses are markedwith asterisks and the resulting increase in tree length in eachof the associated constrained analyses is given Gray lines rep-resent nodes that collapse in the strict consensus of all mostparsimonious trees Taxon labels as in Fig 2

Dasypyrum (Coss amp Durieu) P Candargy and Thino-pyrum ALove and St2 (BS 5 98) is sister to Australo-pyrum velutinum As in the PepC tree E californicusdoes not group with other Elymus species The twoSecale species form a strongly supported clade (BS 5100) sister to Heteranthelium piliferum (BS 5 68) Thisgroup is in turn sister to a clade including Agropyroncristatum Eremopyrum orientale and Taeniatherum caput-medusae (BS 5 68) Among the 30 most-parsimonioustrees tree 1 had the highest log likelihood score andwas chosen as the waxy tree for use in the SH test Themost parameter rich model (GTR1I1G) had a signif-icantly higher log likelihood value than the other mod-els tested and was used in the SH test Figure 4 de-picts parsimony tree 1 under the ML parameters bestfitting the waxy data [R (AC) 5 150 R (AG) 5 353R (AT) 5 136 R (CG) 5 166 R (CT) 5 487 R (GT)5 100 pi (A) 5 022 pi (C) 5 026 pi (G) 5 030 pi(T) 5 021 P-inv 5 035 a 5 071]

Both SH tests yielded significant p-values The SHtest using the PepC data and associated ML parametersresulted in a difference in tree scores of 55194 P 0001 (PepC score 52688770 waxy score 52743964)The SH test using the waxy data and associated MLparameters resulted in a difference in tree scores of

39726 P 0001 (waxy score 52598225 PepC score52637950)

In constrained analyses of the PepC data five of theten nodes tested from the waxy tree (Fig 4 asterisks)resulted in an increase in length relative to the uncon-strained PepC trees (674 steps) In constrained analysesof the waxy data five of the eight PepC tree nodes (Fig3 asterisks) resulted in increases in length over theunconstrained waxy trees (763 steps) The length in-creases associated with each node are shown in Figs3 and 4

DISCUSSION

Evidence of St 1 H Genome Content All tetraploidElymus individuals examined here have two copies ofthe PepC isoform sequenced in the study One copyforms a clade with Hordeum and the second with Pseu-doroegneria confirming that North American Elymusspecies do have an St 1 H genomic content and there-fore are allotetraploid derivatives of Hordeum and Pseu-doroegneria (Fig 2) The data thus confirm Deweyrsquos(1984) and Loversquos (1984) genomic concept of NorthAmerican Elymus in which all species (with the excep-tion of the octoploid E californicus) are allotetraploidscontaining the St and H genomes (Dewey 1984 Love1984) Some systematists have raised justifiable con-cerns regarding the use of genomic pairing data as the

2004] 857HELFGOTT AND MASON-GAMER ELYMUS PHYLOGENY

primary criterion for both grouping and ranking oftaxa in the Triticeae (eg Baum et al 1987 Kellogg1989 Seberg and Petersen 1998) However the PepCphylogeny corroborates the cytogenetic findings andis in agreement with previous studies based on bothisozyme (Jaaska 1995 1998) and waxy gene sequencedata (Mason-Gamer 2001) A classification systembased on chromosome pairing provides a reasonablestarting point in that it groups taxa based on a char-acter that appears to reflect evolutionary relatednessHowever its utility is limited because it cannot provideinsights into the hierarchical relationships among thetaxa examined (eg Kellogg 1989)

Elymus californicusmdashis it a Member of ElymusThe octoploid Elymus californicus is the only Elymusspecies in the present study in which the St and Hgenomes have not been detected This is not surpris-ing because the classification of E californicus has beenunstable since it was first described in 1880 and itsgenomic makeup remains uncertain (Love 1984 MaryBarkworth pers comm) In an attempt to gain in-sights into the genomic complement of this enigmaticspecies Jensen and Wang (1997) screened for the Stand Ns genomes using genome-specific RAPD mark-ers Their study found no evidence of the St genomebut did detect a marker specific to the Ns genomewhich is found in the diploid genus PsathyrostachysNevski and in the allopolyploid genus Leymus HochstFrom this information Jensen and Wang (1997) sug-gested that E californicus should be transferred to Ley-mus The PepC data neither refute nor confirm thesefindings directly placing Elymus californicus at the baseof the tree Nevertheless the PepC phylogeny clearlydemonstrates that E californicus does not belong withinElymus The waxy gene phylogeny by Mason-Gamer etal (2001) was also consistent with Jensen and Wangrsquos(1997) findings revealing a weak association betweenPsathyrostachys and E californicus in most of the most-parsimonious trees Within the chloroplast DNA phy-logeny by Mason-Gamer et al (2002) the placement ofE californicus is equivocal E californicus is weaklygrouped with Hordeum in some of the shortest treesThe PepC data add to the mounting evidence suggest-ing that E californicus does not belong within Elymusbut additional sampling and phylogenetic analyses areneeded to develop a firm understanding of how thisspecies is related to other members of the Triticeae

Elymus H and St Genomes The PepC data indicatethat all North American tetraploid species of Elymushave St 1 H genome content but there is little reso-lution among Elymus individuals in either the St or theH clade This pattern is consistent with the hypothesisthat the Elymus tetraploids are formed from a singleset of genome donors or multiple closely related setsof donors but the specific progenitor species involvedremain a mystery The Elymus H genome sequences

form a strongly supported clade with the two NorthAmerican Hordeum taxa included in the study (Hbrachyantherum and H jubatum) while the Elymus Stsequences form an unresolved polytomy with P spicatathe only North American Pseudoroegneria species (Fig2) The lack of resolution among the Elymus St copiesis similar to that previously observed among chloro-plast DNA sequences (Mason-Gamer et al 2002) In astudy based on the waxy gene (Mason-Gamer et al2001) St sequences formed two distinct clades whichdid not correspond to geographical or morphologicaldifferentiation It was unclear whether the pattern wasthe result of allelic variation or recent gene duplicationprior to the divergence of the St group The PepC andwaxy studies have provided hints on which Hordeumspecies may be most closely related to Elymus but ad-ditional sampling of both Hordeum and Pseudoroegneriais needed before we can develop a better picture of theorigin of the H and St genomes

Distinction Among Hordeum Genomes A detaileddiscussion of Hordeum evolution in light of the PepCphylogeny is impossible given the limited number oftaxa sampled However the data do resolve the fourHordeum genomes (H Xa I and Xu) and providesome insights into how these groups are related (Fig2) Taxa with genomes I (Hordeum vulgare) or Xu (Hor-deum murinum) form a strongly supported monophy-letic group which in turn is sister to a second well-supported clade that includes species with the H or Xa(Hordeum marinum) genomes This corroborates thestudy by Komatsuda et al (1999) of Hordeum diploidtaxa using Vrs1 nuclear intron sequence data The PepCresults are also in accordance with Petersen and Se-bergrsquos (2003) analysis of diploid Hordeum taxa inwhich they combined the chloroplast data of Doebleyet al (1992) and Komatsuda et al (1999) with datafrom the nuclear marker DMC1 (Petersen and Seberg2002 2003) and the chloroplast rbcL gene (Petersen andSeberg 1998) Nuclear and chloroplast DNA data werepartly contradictory but in the combined data set theXu and I genomes formed a sister group to the H andXa taxa although one H genome taxon (H brevisubu-latum Link) was sister to H marinum (Xa) Data fromthe nuclear ribosomal internal transcribed spacers(Frank Blattner pers comm) also recovered four ma-jor clades corresponding to the four Hordeum genomesbut the arrangement of the clades was somewhat dif-ferent Genomes H 1 Xa formed a clade sister to Xuand this clade is sister to I though the placement ofthe Xu clade is weakly supported It is clear from theseHordeum studies that the genome groups are in factphylogenetically informative

Conflict Between PepC and waxy Results from theSH test suggest that the PepC and waxy trees are sig-nificantly incongruent Constrained analyses confirmthese findings suggesting that there is disagreement

858 [Volume 29SYSTEMATIC BOTANY

involving some of the clades in the PepC and waxy phy-logenies (Figs 3 4) There are two cases within Hor-deum where taxa have different positions in the twotrees but are members of clades with moderate to highbootstrap support The first is Hordeum marinum in thePepC tree it is sister to the H clade while in the waxytree it is sister to H murinum (Xu) In the second casethe same accession of H jubatum has a different posi-tion in the PepC and waxy trees In the PepC phylogenyit is nested within the well-supported H clade whilein the waxy phylogeny it is found at the base of a gradeincluding H marinum (Xa) H murinum (Xu) and Hvulgare (I) While the reason for the conflicting place-ment of H jubatum is not known it is an allotetraploid(eg Joslashrgensen 1986 Bothmer et al 1987) and the twodata sets may have sampled gene copies representingthe two different homoeologous genomes

The placement of Taeniatherum caput-medusae is alsoat odds between the PepC and waxy phylogenies (Figs3 4) In the PepC tree T caput-medusae forms a well-supported clade (BS 5 95) with Aegilops comosa whilein the waxy tree it is sister to Eremopyrum orientale (BS5 50) which in turn is sister to Agropyron cristatum (BS5 70) The association seen in the waxy topology hasnot been seen in other gene trees but a close relation-ship between T caput-medusae and Aegilops was seen inthe chloroplast DNA phylogeny (Mason-Gamer et al2002) The PepC data strongly conflict with the waxydata on the placement of Taeniatherum with 36 addi-tional steps required in the constrained analysis Thewaxy data show far less conflict with Taeniatherumrsquosplacement in the PepC tree requiring only two addi-tional steps in the constrained analysis There may besome underlying signal within the waxy data set thatsupports a more congruent placement of TaeniatherumIn the PepC phylogeny E orientale forms a stronglysupported clade with A cristatum (BS 5 100) ratherthan Taeniatherum in agreement with the chloroplastDNA phylogeny (Mason-Gamer et al 2002) and theDMC1 phylogeny (Petersen and Seberg 2002) in whichAgropyron cristatum was grouped with two other Ere-mopyrum species E distans (KKoch) Nevski and E tri-ticeum (Gaertn) Nevski

Secale is well-supported on both trees (BS 5 100)which is not surprising A correlation between mono-genomic groups and clades on molecular phylogenetictrees has been previously reported (Hsiao et al 1995Kellogg and Appels 1995 Kellogg et al 1996 Mason-Gamer and Kellogg 1996a 1996b Petersen and Seberg1997) In the waxy tree the Secale clade is sister to Het-eranthelium piliferum forming a weakly supportedclade (BS 5 68) this association was previously sup-ported in Mason-Gamerrsquos chloroplast DNA phylogeny

Phylogenetic conflict was previously demonstratedamong Triticeae gene trees with the most extensivedifferences documented in comparisons between chlo-

roplast and nuclear data (Kellogg et al 1996 Mason-Gamer and Kellogg 1996b) There were a few instancesof statistically significant incongruence among the nu-clear data sets and the present comparison betweenPepC and waxy shows conflict involving several taxa Ifthe conflict is genuine the nuclear gene trees may re-flect differences among phylogenies of the chromo-somes or sections of chromosomes on which thegenes reside The discordance between PepC and waxygene tree topologies (as well as many others) high-lights the need for continued research on chromosomeevolution in the Triticeae in the hopes of better under-standing mechanisms behind the complex historicalpatterns that are emerging from phylogenetic studiesof this group

Two different historical scenarios provide possibleexplanations for the reticulate history of the tribe (Kel-logg et al 1996) First the pattern may result fromperiods of cladogenesis and dispersal interspersedwith hybridization and introgression between speciesThe level of hybridization slowed or stopped at somepoint in the past most genera in the tribe are inter-sterile and hybrids among them generally exhibit alack of chromosome pairing (eg Dewey 1984 and ref-erences therein) However some degree gene exchangeamong genera is possible if a small percentage of vi-able hybrid seeds are produced through chance aloneIt is also possible that present-day gene flow betweendiploid taxa is taking place through a polyploid inter-mediate Gene flow between diploids via a tetraploidbridge requires the formation of partially fertile trip-loids as has been reported for species of Aegilops byVardi (1973) Feldman et al (1979) and Kimber andFeldman (1987) There is also phylogenetic evidence ofhybridization or introgression Mason-Gamer and Kel-logg (2000) reported that an accession of Pseudoroeg-neria strigosa (M Bieb) ALove which was expected tobe a monogenomic St species has both St and H cop-ies of the waxy gene and an St chloroplast The resultwas consistent with both the allotetraploid-bridge anddirect hybridization hypotheses

A second possible explanation for the reticulate his-tory of the Triticeae is that it reflects a burst of diver-sification with rapid lineage sorting and fixation ofdifferent alleles into lineages which quickly becameintersterile (Kellogg et al 1996) Given the need forextensive ancestral polymorphism in this hypothesisit seems unlikely that rapid diversification and lineagesorting was the major driving force behind the retic-ulate history of the Triticeae However it is not clearhow we could determine which scenario (lineage sort-ing and hybridization-introgression) has had the big-ger impact on the history of the Triticeae In fact thesehypotheses are not mutually exclusive and it may bethat both processes have played some role in shapingthe history of the Triticeae

2004] 859HELFGOTT AND MASON-GAMER ELYMUS PHYLOGENY

In conclusion this study verifies Deweyrsquos (19821983) and Loversquos (1982 1984) insightful chromosome-based revisions of the Triticeae The PepC results clear-ly show that North American tetraploid members ofElymus have an St 1 H genomic content and are there-fore derived from Hordeum and Pseudoroegneria How-ever the current data reveal little genetic differentia-tion within North American Elymus Consequently therelationships among Elymus species and their relation-ships to specific progenitor species await a better-re-solved phylogeny Additional studies which integratemorphological cytological and phylogenetic informa-tion are necessary to understand better the historicalrelationships within Elymus The PepC-based analysisof North American Elymus is a small step forward inthis process

ACKNOWLEDGEMENTS The authors wish to thank Mary Bark-worth for sharing plant material the USDArsquos Germplasm Resourc-es Information Network for seeds Gregory Plunkett and twoanonymous reviewers for helpful comments on the manuscriptThe National Science Foundation (DEB-9974181) funded this work

LITERATURE CITED

BARKWORTH M E and D R DEWEY 1985 Genomically basedgenera in the perennial Triticeae of North America identifi-cation and membership American Journal of Botany 72 767ndash776

BAUM B R J R ESTES and P K GUPTA 1987 Assessment of thegenomic system of classification in the Triticeae AmericanJournal of Botany 74 1388ndash1395

BESNARD G B OFFMANN C ROBERT and C ROUCH 2002 As-sessment of the C4 phosphoenolpyruvate carboxylase genediversity in grasses (Poaceae) Theoretical and Applied Genetics105 404ndash412

BOTHMER R VON J FLINK and T LANDSTROM 1987 Meiosis inHordeum interspecific hybrids II Triploid hybrids Evolution-ary Trends in Plants 1 41ndash50

CRETIN C S SANTI E KERYER L LEPINIEC D TAGU J VIDALand P GADAL 1991 The phosphoenolpyruvate carboxylasegene family in Sorghum promoter structure amino acid se-quences and expression of genes Gene 99 87ndash94

CRONN R C R L SMALL T HASELKORN and J F WENDEL 2003Cryptic repeated genomic recombination during speciation inGossypium gossypioides Evolution 57 2475ndash2489

mdashmdashmdash X ZHAO A H PATERSON and J F WENDEL 1996 Poly-morphism and concerted evolution in a tandemly repeatedgene family 5S ribosomal DNA in diploid and allopolyploidcottons Journal of Molecular Evolution 42 685ndash705

CRONQUIST A 1987 A botanical critique of cladism Botanical Re-view 53 1ndash52

CUSHMAN J C and H J BOHNERT 1989a Nucleotide sequence ofthe Ppc2 gene encoding a housekeeping isoform of PEPCfrom Mesembryanthemum crystallinum Nucleic Acids Research17 6743ndash6744

mdashmdashmdash and mdashmdashmdash 1989b Nucleotide sequence of the gene en-coding a CAM specific isoform of PEPC from Mesembryan-themum crystallinum Nucleic Acids Research 17 6745ndash6746

DAVIS J I and R J SORENG 1993 Phylogenetic structure in thegrass family (Poaceae) as determined from chloroplast DNArestriction site variation American Journal of Botany 80 1444ndash1454

DEVOS K M J DUBCOVSKY J DVORAK C N CHINOY and M DGALE 1995 Structural evolution of wheat chromosomes 4A

5A and 7B and its impact on recombination Theoretical andApplied Genetics 91 282ndash288

DEWEY D R 1982 Genomic and phylogenetic relationshipsamong North American perennial Triticeae Pp 51ndash88 inGrasses and grasslands eds J R Estes R J Tyrl and J NBrunken Norman Oklahoma University Press

mdashmdashmdash 1983 Historical and current taxonomic perspectives of Ag-ropyron Elymus and related genera Crop Science 23 637ndash642

mdashmdashmdash 1984 The genomic system of classification as a guide tointergeneric hybridization within the perennial Triticeae Pp209ndash279 in Gene manipulation in plant improvement Proceedingsof the 6th Stadler genetics symposium ed J P Gustafson NewYork Columbia University Press

DOEBLEY J R VON BOTHMER and S LARSON 1992 ChloroplastDNA variation and the phylogeny of Hordeum (Poaceae)American Journal of Botany 79 576ndash584

DOYLE J J and J L DOYLE 1987 A rapid DNA isolation proce-dure for small quantities of fresh leaf tissue PhytochemicalBulletin 19 11ndash15

mdashmdashmdash mdashmdashmdash A H D BROWN and R G PALMER 2002 Ge-nomes multiple origins and lineage recombination in theGlycine tomentella complex histone H3-D gene sequences Evo-lution 56 1388ndash1402

EMSHWILLER E and J J DOYLE 1998 Origins of domesticationand polyploidy in oca (Oxalis tuberosa Oxalidaceae) nrDNAITS data American Journal of Botany 85 975ndash985

mdashmdashmdash and mdashmdashmdash 2002 Origins of domestication and polyploidyin oca (Oxalis tuberosa Oxalidaceae) 2 Chloroplast-expressedglutamine synthetase data American Journal of Botany 891042ndash1056

ERNST K and P WESTHOFF 1997 The phosphoenolpyruvate car-boxylase (ppc) gene family of Flaveria trinervia (C4) and Fpringlei (C3) molecular characterization and expression anal-ysis of the ppcB and ppcC genes Plant Molecular Biology 34427ndash443

FELDMAN M I STRAUSS and A VARDI 1979 Chromosome pair-ing and fertility of F1 hybrids of Ae longissima and Ae searsiiCanadian Journal of Genetics and Cytology 21 261ndash272

FELSENSTEIN J 1981 Evolutionary trees from DNA sequences amaximum likelihood approach Journal of Molecular Evolution17 368ndash376

mdashmdashmdash 1985 Confidence limits on phylogenies an approach usingthe bootstrap Evolution 39 783ndash791

FRATI F C SIMON J SULLIVAN and D L SWOFFORD 1997 Evo-lution of the mitochondrial cytochrome oxidase II gene inCollembola Journal of Molecular Evolution 44 145ndash158

GALE M D and K M DEVOS 1998 Plant comparative geneticsafter ten years Genome 282 656ndash659

GE S T SANG B-R LU and D-Y HONG 1999 Phylogeny of ricegenomes with emphasis on origins of allotetraploid speciesProceedings of the National Academy of Sciences USA 96 14400ndash14405

GEHRIG H H V HEUTE and M KLUGE 2001 New partial se-quences of phosphoenolpyruvate carboxylase as molecularphylogenetic markers Molecular Phylogenetics and Evolution 20262ndash274

GOLDMAN N 1993 Statistical tests of models of DNA substitutionJournal of Molecular Evolution 36 182ndash198

GONZALEZ M C L OSUNA C ECHEVARRIA J VIDAL and F JCEJUDO 1998 Expression and localization of phosphoenol-pyruvate carboxylase in developing and germinating wheatgrains Plant Physiology 116 1249ndash1258

GU X Y-X Fu and W-H Li 1995 Maximum likelihood esti-mation of the heterogeneity of substitution rate among nu-cleotide sites Molecular Biology and Evolution 12 546ndash557

HASEGAWA M H KISHINO and T YANO 1985 Dating of thehuman-ape split by a molecular clock of mitochondrial DNAJournal of Molecular Evolution 22 160ndash174

2004] 857HELFGOTT AND MASON-GAMER ELYMUS PHYLOGENY

primary criterion for both grouping and ranking oftaxa in the Triticeae (eg Baum et al 1987 Kellogg1989 Seberg and Petersen 1998) However the PepCphylogeny corroborates the cytogenetic findings andis in agreement with previous studies based on bothisozyme (Jaaska 1995 1998) and waxy gene sequencedata (Mason-Gamer 2001) A classification systembased on chromosome pairing provides a reasonablestarting point in that it groups taxa based on a char-acter that appears to reflect evolutionary relatednessHowever its utility is limited because it cannot provideinsights into the hierarchical relationships among thetaxa examined (eg Kellogg 1989)

Elymus californicusmdashis it a Member of ElymusThe octoploid Elymus californicus is the only Elymusspecies in the present study in which the St and Hgenomes have not been detected This is not surpris-ing because the classification of E californicus has beenunstable since it was first described in 1880 and itsgenomic makeup remains uncertain (Love 1984 MaryBarkworth pers comm) In an attempt to gain in-sights into the genomic complement of this enigmaticspecies Jensen and Wang (1997) screened for the Stand Ns genomes using genome-specific RAPD mark-ers Their study found no evidence of the St genomebut did detect a marker specific to the Ns genomewhich is found in the diploid genus PsathyrostachysNevski and in the allopolyploid genus Leymus HochstFrom this information Jensen and Wang (1997) sug-gested that E californicus should be transferred to Ley-mus The PepC data neither refute nor confirm thesefindings directly placing Elymus californicus at the baseof the tree Nevertheless the PepC phylogeny clearlydemonstrates that E californicus does not belong withinElymus The waxy gene phylogeny by Mason-Gamer etal (2001) was also consistent with Jensen and Wangrsquos(1997) findings revealing a weak association betweenPsathyrostachys and E californicus in most of the most-parsimonious trees Within the chloroplast DNA phy-logeny by Mason-Gamer et al (2002) the placement ofE californicus is equivocal E californicus is weaklygrouped with Hordeum in some of the shortest treesThe PepC data add to the mounting evidence suggest-ing that E californicus does not belong within Elymusbut additional sampling and phylogenetic analyses areneeded to develop a firm understanding of how thisspecies is related to other members of the Triticeae

Elymus H and St Genomes The PepC data indicatethat all North American tetraploid species of Elymushave St 1 H genome content but there is little reso-lution among Elymus individuals in either the St or theH clade This pattern is consistent with the hypothesisthat the Elymus tetraploids are formed from a singleset of genome donors or multiple closely related setsof donors but the specific progenitor species involvedremain a mystery The Elymus H genome sequences

form a strongly supported clade with the two NorthAmerican Hordeum taxa included in the study (Hbrachyantherum and H jubatum) while the Elymus Stsequences form an unresolved polytomy with P spicatathe only North American Pseudoroegneria species (Fig2) The lack of resolution among the Elymus St copiesis similar to that previously observed among chloro-plast DNA sequences (Mason-Gamer et al 2002) In astudy based on the waxy gene (Mason-Gamer et al2001) St sequences formed two distinct clades whichdid not correspond to geographical or morphologicaldifferentiation It was unclear whether the pattern wasthe result of allelic variation or recent gene duplicationprior to the divergence of the St group The PepC andwaxy studies have provided hints on which Hordeumspecies may be most closely related to Elymus but ad-ditional sampling of both Hordeum and Pseudoroegneriais needed before we can develop a better picture of theorigin of the H and St genomes

Distinction Among Hordeum Genomes A detaileddiscussion of Hordeum evolution in light of the PepCphylogeny is impossible given the limited number oftaxa sampled However the data do resolve the fourHordeum genomes (H Xa I and Xu) and providesome insights into how these groups are related (Fig2) Taxa with genomes I (Hordeum vulgare) or Xu (Hor-deum murinum) form a strongly supported monophy-letic group which in turn is sister to a second well-supported clade that includes species with the H or Xa(Hordeum marinum) genomes This corroborates thestudy by Komatsuda et al (1999) of Hordeum diploidtaxa using Vrs1 nuclear intron sequence data The PepCresults are also in accordance with Petersen and Se-bergrsquos (2003) analysis of diploid Hordeum taxa inwhich they combined the chloroplast data of Doebleyet al (1992) and Komatsuda et al (1999) with datafrom the nuclear marker DMC1 (Petersen and Seberg2002 2003) and the chloroplast rbcL gene (Petersen andSeberg 1998) Nuclear and chloroplast DNA data werepartly contradictory but in the combined data set theXu and I genomes formed a sister group to the H andXa taxa although one H genome taxon (H brevisubu-latum Link) was sister to H marinum (Xa) Data fromthe nuclear ribosomal internal transcribed spacers(Frank Blattner pers comm) also recovered four ma-jor clades corresponding to the four Hordeum genomesbut the arrangement of the clades was somewhat dif-ferent Genomes H 1 Xa formed a clade sister to Xuand this clade is sister to I though the placement ofthe Xu clade is weakly supported It is clear from theseHordeum studies that the genome groups are in factphylogenetically informative

Conflict Between PepC and waxy Results from theSH test suggest that the PepC and waxy trees are sig-nificantly incongruent Constrained analyses confirmthese findings suggesting that there is disagreement

858 [Volume 29SYSTEMATIC BOTANY

involving some of the clades in the PepC and waxy phy-logenies (Figs 3 4) There are two cases within Hor-deum where taxa have different positions in the twotrees but are members of clades with moderate to highbootstrap support The first is Hordeum marinum in thePepC tree it is sister to the H clade while in the waxytree it is sister to H murinum (Xu) In the second casethe same accession of H jubatum has a different posi-tion in the PepC and waxy trees In the PepC phylogenyit is nested within the well-supported H clade whilein the waxy phylogeny it is found at the base of a gradeincluding H marinum (Xa) H murinum (Xu) and Hvulgare (I) While the reason for the conflicting place-ment of H jubatum is not known it is an allotetraploid(eg Joslashrgensen 1986 Bothmer et al 1987) and the twodata sets may have sampled gene copies representingthe two different homoeologous genomes

The placement of Taeniatherum caput-medusae is alsoat odds between the PepC and waxy phylogenies (Figs3 4) In the PepC tree T caput-medusae forms a well-supported clade (BS 5 95) with Aegilops comosa whilein the waxy tree it is sister to Eremopyrum orientale (BS5 50) which in turn is sister to Agropyron cristatum (BS5 70) The association seen in the waxy topology hasnot been seen in other gene trees but a close relation-ship between T caput-medusae and Aegilops was seen inthe chloroplast DNA phylogeny (Mason-Gamer et al2002) The PepC data strongly conflict with the waxydata on the placement of Taeniatherum with 36 addi-tional steps required in the constrained analysis Thewaxy data show far less conflict with Taeniatherumrsquosplacement in the PepC tree requiring only two addi-tional steps in the constrained analysis There may besome underlying signal within the waxy data set thatsupports a more congruent placement of TaeniatherumIn the PepC phylogeny E orientale forms a stronglysupported clade with A cristatum (BS 5 100) ratherthan Taeniatherum in agreement with the chloroplastDNA phylogeny (Mason-Gamer et al 2002) and theDMC1 phylogeny (Petersen and Seberg 2002) in whichAgropyron cristatum was grouped with two other Ere-mopyrum species E distans (KKoch) Nevski and E tri-ticeum (Gaertn) Nevski

Secale is well-supported on both trees (BS 5 100)which is not surprising A correlation between mono-genomic groups and clades on molecular phylogenetictrees has been previously reported (Hsiao et al 1995Kellogg and Appels 1995 Kellogg et al 1996 Mason-Gamer and Kellogg 1996a 1996b Petersen and Seberg1997) In the waxy tree the Secale clade is sister to Het-eranthelium piliferum forming a weakly supportedclade (BS 5 68) this association was previously sup-ported in Mason-Gamerrsquos chloroplast DNA phylogeny

Phylogenetic conflict was previously demonstratedamong Triticeae gene trees with the most extensivedifferences documented in comparisons between chlo-

roplast and nuclear data (Kellogg et al 1996 Mason-Gamer and Kellogg 1996b) There were a few instancesof statistically significant incongruence among the nu-clear data sets and the present comparison betweenPepC and waxy shows conflict involving several taxa Ifthe conflict is genuine the nuclear gene trees may re-flect differences among phylogenies of the chromo-somes or sections of chromosomes on which thegenes reside The discordance between PepC and waxygene tree topologies (as well as many others) high-lights the need for continued research on chromosomeevolution in the Triticeae in the hopes of better under-standing mechanisms behind the complex historicalpatterns that are emerging from phylogenetic studiesof this group

Two different historical scenarios provide possibleexplanations for the reticulate history of the tribe (Kel-logg et al 1996) First the pattern may result fromperiods of cladogenesis and dispersal interspersedwith hybridization and introgression between speciesThe level of hybridization slowed or stopped at somepoint in the past most genera in the tribe are inter-sterile and hybrids among them generally exhibit alack of chromosome pairing (eg Dewey 1984 and ref-erences therein) However some degree gene exchangeamong genera is possible if a small percentage of vi-able hybrid seeds are produced through chance aloneIt is also possible that present-day gene flow betweendiploid taxa is taking place through a polyploid inter-mediate Gene flow between diploids via a tetraploidbridge requires the formation of partially fertile trip-loids as has been reported for species of Aegilops byVardi (1973) Feldman et al (1979) and Kimber andFeldman (1987) There is also phylogenetic evidence ofhybridization or introgression Mason-Gamer and Kel-logg (2000) reported that an accession of Pseudoroeg-neria strigosa (M Bieb) ALove which was expected tobe a monogenomic St species has both St and H cop-ies of the waxy gene and an St chloroplast The resultwas consistent with both the allotetraploid-bridge anddirect hybridization hypotheses

A second possible explanation for the reticulate his-tory of the Triticeae is that it reflects a burst of diver-sification with rapid lineage sorting and fixation ofdifferent alleles into lineages which quickly becameintersterile (Kellogg et al 1996) Given the need forextensive ancestral polymorphism in this hypothesisit seems unlikely that rapid diversification and lineagesorting was the major driving force behind the retic-ulate history of the Triticeae However it is not clearhow we could determine which scenario (lineage sort-ing and hybridization-introgression) has had the big-ger impact on the history of the Triticeae In fact thesehypotheses are not mutually exclusive and it may bethat both processes have played some role in shapingthe history of the Triticeae

2004] 859HELFGOTT AND MASON-GAMER ELYMUS PHYLOGENY

In conclusion this study verifies Deweyrsquos (19821983) and Loversquos (1982 1984) insightful chromosome-based revisions of the Triticeae The PepC results clear-ly show that North American tetraploid members ofElymus have an St 1 H genomic content and are there-fore derived from Hordeum and Pseudoroegneria How-ever the current data reveal little genetic differentia-tion within North American Elymus Consequently therelationships among Elymus species and their relation-ships to specific progenitor species await a better-re-solved phylogeny Additional studies which integratemorphological cytological and phylogenetic informa-tion are necessary to understand better the historicalrelationships within Elymus The PepC-based analysisof North American Elymus is a small step forward inthis process

ACKNOWLEDGEMENTS The authors wish to thank Mary Bark-worth for sharing plant material the USDArsquos Germplasm Resourc-es Information Network for seeds Gregory Plunkett and twoanonymous reviewers for helpful comments on the manuscriptThe National Science Foundation (DEB-9974181) funded this work

LITERATURE CITED

BARKWORTH M E and D R DEWEY 1985 Genomically basedgenera in the perennial Triticeae of North America identifi-cation and membership American Journal of Botany 72 767ndash776

BAUM B R J R ESTES and P K GUPTA 1987 Assessment of thegenomic system of classification in the Triticeae AmericanJournal of Botany 74 1388ndash1395

BESNARD G B OFFMANN C ROBERT and C ROUCH 2002 As-sessment of the C4 phosphoenolpyruvate carboxylase genediversity in grasses (Poaceae) Theoretical and Applied Genetics105 404ndash412

BOTHMER R VON J FLINK and T LANDSTROM 1987 Meiosis inHordeum interspecific hybrids II Triploid hybrids Evolution-ary Trends in Plants 1 41ndash50

CRETIN C S SANTI E KERYER L LEPINIEC D TAGU J VIDALand P GADAL 1991 The phosphoenolpyruvate carboxylasegene family in Sorghum promoter structure amino acid se-quences and expression of genes Gene 99 87ndash94

CRONN R C R L SMALL T HASELKORN and J F WENDEL 2003Cryptic repeated genomic recombination during speciation inGossypium gossypioides Evolution 57 2475ndash2489

mdashmdashmdash X ZHAO A H PATERSON and J F WENDEL 1996 Poly-morphism and concerted evolution in a tandemly repeatedgene family 5S ribosomal DNA in diploid and allopolyploidcottons Journal of Molecular Evolution 42 685ndash705

CRONQUIST A 1987 A botanical critique of cladism Botanical Re-view 53 1ndash52

CUSHMAN J C and H J BOHNERT 1989a Nucleotide sequence ofthe Ppc2 gene encoding a housekeeping isoform of PEPCfrom Mesembryanthemum crystallinum Nucleic Acids Research17 6743ndash6744

mdashmdashmdash and mdashmdashmdash 1989b Nucleotide sequence of the gene en-coding a CAM specific isoform of PEPC from Mesembryan-themum crystallinum Nucleic Acids Research 17 6745ndash6746

DAVIS J I and R J SORENG 1993 Phylogenetic structure in thegrass family (Poaceae) as determined from chloroplast DNArestriction site variation American Journal of Botany 80 1444ndash1454

DEVOS K M J DUBCOVSKY J DVORAK C N CHINOY and M DGALE 1995 Structural evolution of wheat chromosomes 4A

5A and 7B and its impact on recombination Theoretical andApplied Genetics 91 282ndash288

DEWEY D R 1982 Genomic and phylogenetic relationshipsamong North American perennial Triticeae Pp 51ndash88 inGrasses and grasslands eds J R Estes R J Tyrl and J NBrunken Norman Oklahoma University Press

mdashmdashmdash 1983 Historical and current taxonomic perspectives of Ag-ropyron Elymus and related genera Crop Science 23 637ndash642

mdashmdashmdash 1984 The genomic system of classification as a guide tointergeneric hybridization within the perennial Triticeae Pp209ndash279 in Gene manipulation in plant improvement Proceedingsof the 6th Stadler genetics symposium ed J P Gustafson NewYork Columbia University Press

DOEBLEY J R VON BOTHMER and S LARSON 1992 ChloroplastDNA variation and the phylogeny of Hordeum (Poaceae)American Journal of Botany 79 576ndash584

DOYLE J J and J L DOYLE 1987 A rapid DNA isolation proce-dure for small quantities of fresh leaf tissue PhytochemicalBulletin 19 11ndash15

mdashmdashmdash mdashmdashmdash A H D BROWN and R G PALMER 2002 Ge-nomes multiple origins and lineage recombination in theGlycine tomentella complex histone H3-D gene sequences Evo-lution 56 1388ndash1402

EMSHWILLER E and J J DOYLE 1998 Origins of domesticationand polyploidy in oca (Oxalis tuberosa Oxalidaceae) nrDNAITS data American Journal of Botany 85 975ndash985

mdashmdashmdash and mdashmdashmdash 2002 Origins of domestication and polyploidyin oca (Oxalis tuberosa Oxalidaceae) 2 Chloroplast-expressedglutamine synthetase data American Journal of Botany 891042ndash1056

ERNST K and P WESTHOFF 1997 The phosphoenolpyruvate car-boxylase (ppc) gene family of Flaveria trinervia (C4) and Fpringlei (C3) molecular characterization and expression anal-ysis of the ppcB and ppcC genes Plant Molecular Biology 34427ndash443

FELDMAN M I STRAUSS and A VARDI 1979 Chromosome pair-ing and fertility of F1 hybrids of Ae longissima and Ae searsiiCanadian Journal of Genetics and Cytology 21 261ndash272

FELSENSTEIN J 1981 Evolutionary trees from DNA sequences amaximum likelihood approach Journal of Molecular Evolution17 368ndash376

mdashmdashmdash 1985 Confidence limits on phylogenies an approach usingthe bootstrap Evolution 39 783ndash791

FRATI F C SIMON J SULLIVAN and D L SWOFFORD 1997 Evo-lution of the mitochondrial cytochrome oxidase II gene inCollembola Journal of Molecular Evolution 44 145ndash158

GALE M D and K M DEVOS 1998 Plant comparative geneticsafter ten years Genome 282 656ndash659

GE S T SANG B-R LU and D-Y HONG 1999 Phylogeny of ricegenomes with emphasis on origins of allotetraploid speciesProceedings of the National Academy of Sciences USA 96 14400ndash14405

GEHRIG H H V HEUTE and M KLUGE 2001 New partial se-quences of phosphoenolpyruvate carboxylase as molecularphylogenetic markers Molecular Phylogenetics and Evolution 20262ndash274

GOLDMAN N 1993 Statistical tests of models of DNA substitutionJournal of Molecular Evolution 36 182ndash198

GONZALEZ M C L OSUNA C ECHEVARRIA J VIDAL and F JCEJUDO 1998 Expression and localization of phosphoenol-pyruvate carboxylase in developing and germinating wheatgrains Plant Physiology 116 1249ndash1258

GU X Y-X Fu and W-H Li 1995 Maximum likelihood esti-mation of the heterogeneity of substitution rate among nu-cleotide sites Molecular Biology and Evolution 12 546ndash557

HASEGAWA M H KISHINO and T YANO 1985 Dating of thehuman-ape split by a molecular clock of mitochondrial DNAJournal of Molecular Evolution 22 160ndash174

858 [Volume 29SYSTEMATIC BOTANY

involving some of the clades in the PepC and waxy phy-logenies (Figs 3 4) There are two cases within Hor-deum where taxa have different positions in the twotrees but are members of clades with moderate to highbootstrap support The first is Hordeum marinum in thePepC tree it is sister to the H clade while in the waxytree it is sister to H murinum (Xu) In the second casethe same accession of H jubatum has a different posi-tion in the PepC and waxy trees In the PepC phylogenyit is nested within the well-supported H clade whilein the waxy phylogeny it is found at the base of a gradeincluding H marinum (Xa) H murinum (Xu) and Hvulgare (I) While the reason for the conflicting place-ment of H jubatum is not known it is an allotetraploid(eg Joslashrgensen 1986 Bothmer et al 1987) and the twodata sets may have sampled gene copies representingthe two different homoeologous genomes

The placement of Taeniatherum caput-medusae is alsoat odds between the PepC and waxy phylogenies (Figs3 4) In the PepC tree T caput-medusae forms a well-supported clade (BS 5 95) with Aegilops comosa whilein the waxy tree it is sister to Eremopyrum orientale (BS5 50) which in turn is sister to Agropyron cristatum (BS5 70) The association seen in the waxy topology hasnot been seen in other gene trees but a close relation-ship between T caput-medusae and Aegilops was seen inthe chloroplast DNA phylogeny (Mason-Gamer et al2002) The PepC data strongly conflict with the waxydata on the placement of Taeniatherum with 36 addi-tional steps required in the constrained analysis Thewaxy data show far less conflict with Taeniatherumrsquosplacement in the PepC tree requiring only two addi-tional steps in the constrained analysis There may besome underlying signal within the waxy data set thatsupports a more congruent placement of TaeniatherumIn the PepC phylogeny E orientale forms a stronglysupported clade with A cristatum (BS 5 100) ratherthan Taeniatherum in agreement with the chloroplastDNA phylogeny (Mason-Gamer et al 2002) and theDMC1 phylogeny (Petersen and Seberg 2002) in whichAgropyron cristatum was grouped with two other Ere-mopyrum species E distans (KKoch) Nevski and E tri-ticeum (Gaertn) Nevski

Secale is well-supported on both trees (BS 5 100)which is not surprising A correlation between mono-genomic groups and clades on molecular phylogenetictrees has been previously reported (Hsiao et al 1995Kellogg and Appels 1995 Kellogg et al 1996 Mason-Gamer and Kellogg 1996a 1996b Petersen and Seberg1997) In the waxy tree the Secale clade is sister to Het-eranthelium piliferum forming a weakly supportedclade (BS 5 68) this association was previously sup-ported in Mason-Gamerrsquos chloroplast DNA phylogeny

Phylogenetic conflict was previously demonstratedamong Triticeae gene trees with the most extensivedifferences documented in comparisons between chlo-

roplast and nuclear data (Kellogg et al 1996 Mason-Gamer and Kellogg 1996b) There were a few instancesof statistically significant incongruence among the nu-clear data sets and the present comparison betweenPepC and waxy shows conflict involving several taxa Ifthe conflict is genuine the nuclear gene trees may re-flect differences among phylogenies of the chromo-somes or sections of chromosomes on which thegenes reside The discordance between PepC and waxygene tree topologies (as well as many others) high-lights the need for continued research on chromosomeevolution in the Triticeae in the hopes of better under-standing mechanisms behind the complex historicalpatterns that are emerging from phylogenetic studiesof this group

Two different historical scenarios provide possibleexplanations for the reticulate history of the tribe (Kel-logg et al 1996) First the pattern may result fromperiods of cladogenesis and dispersal interspersedwith hybridization and introgression between speciesThe level of hybridization slowed or stopped at somepoint in the past most genera in the tribe are inter-sterile and hybrids among them generally exhibit alack of chromosome pairing (eg Dewey 1984 and ref-erences therein) However some degree gene exchangeamong genera is possible if a small percentage of vi-able hybrid seeds are produced through chance aloneIt is also possible that present-day gene flow betweendiploid taxa is taking place through a polyploid inter-mediate Gene flow between diploids via a tetraploidbridge requires the formation of partially fertile trip-loids as has been reported for species of Aegilops byVardi (1973) Feldman et al (1979) and Kimber andFeldman (1987) There is also phylogenetic evidence ofhybridization or introgression Mason-Gamer and Kel-logg (2000) reported that an accession of Pseudoroeg-neria strigosa (M Bieb) ALove which was expected tobe a monogenomic St species has both St and H cop-ies of the waxy gene and an St chloroplast The resultwas consistent with both the allotetraploid-bridge anddirect hybridization hypotheses

A second possible explanation for the reticulate his-tory of the Triticeae is that it reflects a burst of diver-sification with rapid lineage sorting and fixation ofdifferent alleles into lineages which quickly becameintersterile (Kellogg et al 1996) Given the need forextensive ancestral polymorphism in this hypothesisit seems unlikely that rapid diversification and lineagesorting was the major driving force behind the retic-ulate history of the Triticeae However it is not clearhow we could determine which scenario (lineage sort-ing and hybridization-introgression) has had the big-ger impact on the history of the Triticeae In fact thesehypotheses are not mutually exclusive and it may bethat both processes have played some role in shapingthe history of the Triticeae

2004] 859HELFGOTT AND MASON-GAMER ELYMUS PHYLOGENY

In conclusion this study verifies Deweyrsquos (19821983) and Loversquos (1982 1984) insightful chromosome-based revisions of the Triticeae The PepC results clear-ly show that North American tetraploid members ofElymus have an St 1 H genomic content and are there-fore derived from Hordeum and Pseudoroegneria How-ever the current data reveal little genetic differentia-tion within North American Elymus Consequently therelationships among Elymus species and their relation-ships to specific progenitor species await a better-re-solved phylogeny Additional studies which integratemorphological cytological and phylogenetic informa-tion are necessary to understand better the historicalrelationships within Elymus The PepC-based analysisof North American Elymus is a small step forward inthis process

ACKNOWLEDGEMENTS The authors wish to thank Mary Bark-worth for sharing plant material the USDArsquos Germplasm Resourc-es Information Network for seeds Gregory Plunkett and twoanonymous reviewers for helpful comments on the manuscriptThe National Science Foundation (DEB-9974181) funded this work

LITERATURE CITED

BARKWORTH M E and D R DEWEY 1985 Genomically basedgenera in the perennial Triticeae of North America identifi-cation and membership American Journal of Botany 72 767ndash776

BAUM B R J R ESTES and P K GUPTA 1987 Assessment of thegenomic system of classification in the Triticeae AmericanJournal of Botany 74 1388ndash1395

BESNARD G B OFFMANN C ROBERT and C ROUCH 2002 As-sessment of the C4 phosphoenolpyruvate carboxylase genediversity in grasses (Poaceae) Theoretical and Applied Genetics105 404ndash412

BOTHMER R VON J FLINK and T LANDSTROM 1987 Meiosis inHordeum interspecific hybrids II Triploid hybrids Evolution-ary Trends in Plants 1 41ndash50

CRETIN C S SANTI E KERYER L LEPINIEC D TAGU J VIDALand P GADAL 1991 The phosphoenolpyruvate carboxylasegene family in Sorghum promoter structure amino acid se-quences and expression of genes Gene 99 87ndash94

CRONN R C R L SMALL T HASELKORN and J F WENDEL 2003Cryptic repeated genomic recombination during speciation inGossypium gossypioides Evolution 57 2475ndash2489

mdashmdashmdash X ZHAO A H PATERSON and J F WENDEL 1996 Poly-morphism and concerted evolution in a tandemly repeatedgene family 5S ribosomal DNA in diploid and allopolyploidcottons Journal of Molecular Evolution 42 685ndash705

CRONQUIST A 1987 A botanical critique of cladism Botanical Re-view 53 1ndash52

CUSHMAN J C and H J BOHNERT 1989a Nucleotide sequence ofthe Ppc2 gene encoding a housekeeping isoform of PEPCfrom Mesembryanthemum crystallinum Nucleic Acids Research17 6743ndash6744

mdashmdashmdash and mdashmdashmdash 1989b Nucleotide sequence of the gene en-coding a CAM specific isoform of PEPC from Mesembryan-themum crystallinum Nucleic Acids Research 17 6745ndash6746

DAVIS J I and R J SORENG 1993 Phylogenetic structure in thegrass family (Poaceae) as determined from chloroplast DNArestriction site variation American Journal of Botany 80 1444ndash1454

DEVOS K M J DUBCOVSKY J DVORAK C N CHINOY and M DGALE 1995 Structural evolution of wheat chromosomes 4A

5A and 7B and its impact on recombination Theoretical andApplied Genetics 91 282ndash288

DEWEY D R 1982 Genomic and phylogenetic relationshipsamong North American perennial Triticeae Pp 51ndash88 inGrasses and grasslands eds J R Estes R J Tyrl and J NBrunken Norman Oklahoma University Press

mdashmdashmdash 1983 Historical and current taxonomic perspectives of Ag-ropyron Elymus and related genera Crop Science 23 637ndash642

mdashmdashmdash 1984 The genomic system of classification as a guide tointergeneric hybridization within the perennial Triticeae Pp209ndash279 in Gene manipulation in plant improvement Proceedingsof the 6th Stadler genetics symposium ed J P Gustafson NewYork Columbia University Press

DOEBLEY J R VON BOTHMER and S LARSON 1992 ChloroplastDNA variation and the phylogeny of Hordeum (Poaceae)American Journal of Botany 79 576ndash584

DOYLE J J and J L DOYLE 1987 A rapid DNA isolation proce-dure for small quantities of fresh leaf tissue PhytochemicalBulletin 19 11ndash15

mdashmdashmdash mdashmdashmdash A H D BROWN and R G PALMER 2002 Ge-nomes multiple origins and lineage recombination in theGlycine tomentella complex histone H3-D gene sequences Evo-lution 56 1388ndash1402

EMSHWILLER E and J J DOYLE 1998 Origins of domesticationand polyploidy in oca (Oxalis tuberosa Oxalidaceae) nrDNAITS data American Journal of Botany 85 975ndash985

mdashmdashmdash and mdashmdashmdash 2002 Origins of domestication and polyploidyin oca (Oxalis tuberosa Oxalidaceae) 2 Chloroplast-expressedglutamine synthetase data American Journal of Botany 891042ndash1056

ERNST K and P WESTHOFF 1997 The phosphoenolpyruvate car-boxylase (ppc) gene family of Flaveria trinervia (C4) and Fpringlei (C3) molecular characterization and expression anal-ysis of the ppcB and ppcC genes Plant Molecular Biology 34427ndash443

FELDMAN M I STRAUSS and A VARDI 1979 Chromosome pair-ing and fertility of F1 hybrids of Ae longissima and Ae searsiiCanadian Journal of Genetics and Cytology 21 261ndash272

FELSENSTEIN J 1981 Evolutionary trees from DNA sequences amaximum likelihood approach Journal of Molecular Evolution17 368ndash376

mdashmdashmdash 1985 Confidence limits on phylogenies an approach usingthe bootstrap Evolution 39 783ndash791

FRATI F C SIMON J SULLIVAN and D L SWOFFORD 1997 Evo-lution of the mitochondrial cytochrome oxidase II gene inCollembola Journal of Molecular Evolution 44 145ndash158

GALE M D and K M DEVOS 1998 Plant comparative geneticsafter ten years Genome 282 656ndash659

GE S T SANG B-R LU and D-Y HONG 1999 Phylogeny of ricegenomes with emphasis on origins of allotetraploid speciesProceedings of the National Academy of Sciences USA 96 14400ndash14405

GEHRIG H H V HEUTE and M KLUGE 2001 New partial se-quences of phosphoenolpyruvate carboxylase as molecularphylogenetic markers Molecular Phylogenetics and Evolution 20262ndash274

GOLDMAN N 1993 Statistical tests of models of DNA substitutionJournal of Molecular Evolution 36 182ndash198

GONZALEZ M C L OSUNA C ECHEVARRIA J VIDAL and F JCEJUDO 1998 Expression and localization of phosphoenol-pyruvate carboxylase in developing and germinating wheatgrains Plant Physiology 116 1249ndash1258

GU X Y-X Fu and W-H Li 1995 Maximum likelihood esti-mation of the heterogeneity of substitution rate among nu-cleotide sites Molecular Biology and Evolution 12 546ndash557

HASEGAWA M H KISHINO and T YANO 1985 Dating of thehuman-ape split by a molecular clock of mitochondrial DNAJournal of Molecular Evolution 22 160ndash174

2004] 859HELFGOTT AND MASON-GAMER ELYMUS PHYLOGENY

In conclusion this study verifies Deweyrsquos (19821983) and Loversquos (1982 1984) insightful chromosome-based revisions of the Triticeae The PepC results clear-ly show that North American tetraploid members ofElymus have an St 1 H genomic content and are there-fore derived from Hordeum and Pseudoroegneria How-ever the current data reveal little genetic differentia-tion within North American Elymus Consequently therelationships among Elymus species and their relation-ships to specific progenitor species await a better-re-solved phylogeny Additional studies which integratemorphological cytological and phylogenetic informa-tion are necessary to understand better the historicalrelationships within Elymus The PepC-based analysisof North American Elymus is a small step forward inthis process

ACKNOWLEDGEMENTS The authors wish to thank Mary Bark-worth for sharing plant material the USDArsquos Germplasm Resourc-es Information Network for seeds Gregory Plunkett and twoanonymous reviewers for helpful comments on the manuscriptThe National Science Foundation (DEB-9974181) funded this work

LITERATURE CITED

BARKWORTH M E and D R DEWEY 1985 Genomically basedgenera in the perennial Triticeae of North America identifi-cation and membership American Journal of Botany 72 767ndash776

BAUM B R J R ESTES and P K GUPTA 1987 Assessment of thegenomic system of classification in the Triticeae AmericanJournal of Botany 74 1388ndash1395

BESNARD G B OFFMANN C ROBERT and C ROUCH 2002 As-sessment of the C4 phosphoenolpyruvate carboxylase genediversity in grasses (Poaceae) Theoretical and Applied Genetics105 404ndash412

BOTHMER R VON J FLINK and T LANDSTROM 1987 Meiosis inHordeum interspecific hybrids II Triploid hybrids Evolution-ary Trends in Plants 1 41ndash50

CRETIN C S SANTI E KERYER L LEPINIEC D TAGU J VIDALand P GADAL 1991 The phosphoenolpyruvate carboxylasegene family in Sorghum promoter structure amino acid se-quences and expression of genes Gene 99 87ndash94

CRONN R C R L SMALL T HASELKORN and J F WENDEL 2003Cryptic repeated genomic recombination during speciation inGossypium gossypioides Evolution 57 2475ndash2489

mdashmdashmdash X ZHAO A H PATERSON and J F WENDEL 1996 Poly-morphism and concerted evolution in a tandemly repeatedgene family 5S ribosomal DNA in diploid and allopolyploidcottons Journal of Molecular Evolution 42 685ndash705

CRONQUIST A 1987 A botanical critique of cladism Botanical Re-view 53 1ndash52

CUSHMAN J C and H J BOHNERT 1989a Nucleotide sequence ofthe Ppc2 gene encoding a housekeeping isoform of PEPCfrom Mesembryanthemum crystallinum Nucleic Acids Research17 6743ndash6744

mdashmdashmdash and mdashmdashmdash 1989b Nucleotide sequence of the gene en-coding a CAM specific isoform of PEPC from Mesembryan-themum crystallinum Nucleic Acids Research 17 6745ndash6746

DAVIS J I and R J SORENG 1993 Phylogenetic structure in thegrass family (Poaceae) as determined from chloroplast DNArestriction site variation American Journal of Botany 80 1444ndash1454

DEVOS K M J DUBCOVSKY J DVORAK C N CHINOY and M DGALE 1995 Structural evolution of wheat chromosomes 4A

5A and 7B and its impact on recombination Theoretical andApplied Genetics 91 282ndash288

DEWEY D R 1982 Genomic and phylogenetic relationshipsamong North American perennial Triticeae Pp 51ndash88 inGrasses and grasslands eds J R Estes R J Tyrl and J NBrunken Norman Oklahoma University Press

mdashmdashmdash 1983 Historical and current taxonomic perspectives of Ag-ropyron Elymus and related genera Crop Science 23 637ndash642

mdashmdashmdash 1984 The genomic system of classification as a guide tointergeneric hybridization within the perennial Triticeae Pp209ndash279 in Gene manipulation in plant improvement Proceedingsof the 6th Stadler genetics symposium ed J P Gustafson NewYork Columbia University Press

DOEBLEY J R VON BOTHMER and S LARSON 1992 ChloroplastDNA variation and the phylogeny of Hordeum (Poaceae)American Journal of Botany 79 576ndash584

DOYLE J J and J L DOYLE 1987 A rapid DNA isolation proce-dure for small quantities of fresh leaf tissue PhytochemicalBulletin 19 11ndash15

mdashmdashmdash mdashmdashmdash A H D BROWN and R G PALMER 2002 Ge-nomes multiple origins and lineage recombination in theGlycine tomentella complex histone H3-D gene sequences Evo-lution 56 1388ndash1402

EMSHWILLER E and J J DOYLE 1998 Origins of domesticationand polyploidy in oca (Oxalis tuberosa Oxalidaceae) nrDNAITS data American Journal of Botany 85 975ndash985

mdashmdashmdash and mdashmdashmdash 2002 Origins of domestication and polyploidyin oca (Oxalis tuberosa Oxalidaceae) 2 Chloroplast-expressedglutamine synthetase data American Journal of Botany 891042ndash1056

ERNST K and P WESTHOFF 1997 The phosphoenolpyruvate car-boxylase (ppc) gene family of Flaveria trinervia (C4) and Fpringlei (C3) molecular characterization and expression anal-ysis of the ppcB and ppcC genes Plant Molecular Biology 34427ndash443

FELDMAN M I STRAUSS and A VARDI 1979 Chromosome pair-ing and fertility of F1 hybrids of Ae longissima and Ae searsiiCanadian Journal of Genetics and Cytology 21 261ndash272

FELSENSTEIN J 1981 Evolutionary trees from DNA sequences amaximum likelihood approach Journal of Molecular Evolution17 368ndash376

mdashmdashmdash 1985 Confidence limits on phylogenies an approach usingthe bootstrap Evolution 39 783ndash791

FRATI F C SIMON J SULLIVAN and D L SWOFFORD 1997 Evo-lution of the mitochondrial cytochrome oxidase II gene inCollembola Journal of Molecular Evolution 44 145ndash158

GALE M D and K M DEVOS 1998 Plant comparative geneticsafter ten years Genome 282 656ndash659

GE S T SANG B-R LU and D-Y HONG 1999 Phylogeny of ricegenomes with emphasis on origins of allotetraploid speciesProceedings of the National Academy of Sciences USA 96 14400ndash14405

GEHRIG H H V HEUTE and M KLUGE 2001 New partial se-quences of phosphoenolpyruvate carboxylase as molecularphylogenetic markers Molecular Phylogenetics and Evolution 20262ndash274

GOLDMAN N 1993 Statistical tests of models of DNA substitutionJournal of Molecular Evolution 36 182ndash198

GONZALEZ M C L OSUNA C ECHEVARRIA J VIDAL and F JCEJUDO 1998 Expression and localization of phosphoenol-pyruvate carboxylase in developing and germinating wheatgrains Plant Physiology 116 1249ndash1258

GU X Y-X Fu and W-H Li 1995 Maximum likelihood esti-mation of the heterogeneity of substitution rate among nu-cleotide sites Molecular Biology and Evolution 12 546ndash557

HASEGAWA M H KISHINO and T YANO 1985 Dating of thehuman-ape split by a molecular clock of mitochondrial DNAJournal of Molecular Evolution 22 160ndash174

860 [Volume 29SYSTEMATIC BOTANY

HENDY M D and D PENNY 1982 Branch and bound algorithmsto determine minimal evolutionary trees Mathematical Biosci-ences 59 277ndash290

HITCHCOCK A S 1951 Manual of the grasses of the United Statesed 2 revised by A Chase USDA Miscellaneous Publications200 Washington D C U S Government Printing Office

HSIAO C N J CHATTERTON K H ASAY and K B JENSEN 1995Phylogenetic relationships of the monogenomic species of thewheat tribe Triticeae (Poaceae) inferred from nuclear rDNA(internal transcribed spacer) sequences Genome 38 211ndash223

HUDSPETH R L and J W GRULA 1989 Structure and expressionof the maize gene encoding the phosphoenolpyruvate car-boxylase isozyme involved in C4 photosynthesis Plant Molec-ular Biology 12 579ndash589

HUELSENBECK P H and K A CRANDALL 1997 Phylogeny esti-mation and hypothesis testing using maximum likelihoodAnnual Review of Ecology and Systematics 28 437ndash466

mdashmdashmdash and B RANNALA 1997 Phylogenetic methods come of agetesting hypotheses in an evolutionary context Science 276227ndash232

mdashmdashmdash and F RONQUIST 2001 MRBAYES Bayesian inference ofphylogenetic trees Bioinformatics 17 754ndash755

HULL D L 1979 The limits of cladism Systematic Zoology 28 416ndash440

IZUI K S ISHIJIMA Y YAMAGUCHI F KATAGIRI T MURATA KSHIGESADA T SUGIYAMA and H KATUKI 1986 Cloning andsequence analysis of cDNA encoding active PEPC of the C4

photosynthesis in maize Nucleic Acids Research 14 1615ndash1628JAASKA V 1995 Isozyme data on the diploid progenitors of allo-

tetraploid Elymus species Pp 165ndash168 in Proceedings of the 2ndInternational Triticeae Symposium eds R R-C Wang K BJensen and C Jaussi Logan Utah USDA Forage and RangeLaboratory

mdashmdashmdash 1998 Isoenzyme data on the origin of North Americanallotetraploid Elymus species Pp 209ndash216 in Triticeae III edA A Jaradat Enfield New Hampshire Science PublishersInc

JENSEN R and R-C WANG 1997 Cytological and molecular ev-idence for transferring Elymus coreanus from the genus Elymusto Leymus and molecular evidence for Elymus californicus (Po-aceae Triticeae) International Journal of Plant Sciences 158872ndash877

JoslashRGENSEN R B 1986 Relationships in the barley genus (Hor-deum) an electrophoretic examination of proteins Hereditas104 273ndash291

JUKES T H and C R CANTOR 1969 Evolution of protein mole-cules Pp 21ndash132 in Mammalian protein metabolism ed H NMunro New York Academic Press

KAWAMURA T K SHIGESADA H TOH S OKUMURA S YANAGI-SAWA and K IZUI 1992 Molecular evolution of phospho-enolpyruvate carboxylase for C4 photosynthesis in maizecomparison of its cDNA sequence with a newly isolatedcDNA encoding an isozyme involved in the anaplerotic func-tion Journal of Biochemistry 112 147ndash154

KELLOGG E A 1989 Comments on genomic genera in the Triti-ceae (Poaceae) American Journal of Botany 76 796ndash805

mdashmdashmdash 1992 Tools for studying the chloroplast genome in theTriticeae (Gramineae) an EcoRI map a diagnostic deletionand support for Bromus as an outgroup American Journal ofBotany 79 186ndash197

mdashmdashmdash and R APPELS 1995 Intraspecific and interspecific varia-tion in 5S RNA genes are decoupled in diploid wheat rela-tives Genetics 140 325ndash343

mdashmdashmdash mdashmdashmdash and R J MASON-GAMER 1996 When gene treestell different stories the diploid genera of Triticeae (Grami-neae) Systematic Botany 21 321ndash347

KIMBER G and M FELDMAN 1987 Wild wheat an introduction Co-lumbia College of Agriculture University of Missouri

KIMURA M 1980 A simple method for estimating evolutionaryrate of base substitution through comparative studies of nu-cleotide sequences Journal of Molecular Evolution 16 111ndash120

KLEINHOFS A 1997 Integrating barley RFLP and classical markermaps Barley Genetics Newsletter 27 105ndash112

KOMATSUDA T K-I TANNO B SALOMON T BRYNGELSSON and RVON BOTHMER 1999 Phylogeny in the genus Hordeum basedon nucleotide sequences closely linked to the vrs1 locus (rownumber of spikelets) Genome 42 973ndash981

KORZUN V S MALYSHEV A VOYLOKOV and A BORNER 1997RFLP-based mapping of three mutant loci in rye (Secale cerealeL) and their relation to homoeologous loci within the Gra-mineae Theoretical and Applied Genetics 95 468ndash473

LEPINIEC L E KERYER H PHILIPPE P GADAL and C CRETIN1993 Sorghum phosphoenolpyruvate carboxylase gene fami-ly structure function and molecular evolution Plant Molec-ular Biology 21 487ndash502

mdashmdashmdash mdashmdashmdash D TAGU P GADAL and C CRETIN 1992 Com-plete nucleotide sequence of a Sorghum gene coding for thephosphoenolpyruvate carboxylase involved in C4 photosyn-thesis Plant Molecular Biology 19 339ndash342

mdashmdashmdash S SANTI E KERYER V AMIET J VIDAL P GADAL and CCRETIN 1991 Complete nucleotide sequence of one memberof the Sorghum phosphoenolpyruvate carboxylase gene fam-ily Plant Molecular Biology 17 1077ndash1079

mdashmdashmdash J VIDAL R CHOLLET P GADAL and C CRETIN 1994Phosphoenolpyruvate carboxylase structure regulation andevolution Plant Science 99 111ndash124

LOVE A 1982 Generic evolution of the wheatgrasses BiologischesZentralblatt 101 199ndash212

mdashmdashmdash 1984 Conspectus of the Triticeae Feddes Repertorium 95425ndash521

MADDISON W P and D R MADDISON 2002 MacClade ver 40Analysis of phylogeny and character evolution SunderlandSinauer Associates

MALCOMBER S T 2002 Phylogeny of Gaertnera Lam (Rubiaceae)based on multiple DNA markers evidence of a rapid radia-tion in a widespread morphologically diverse genus Evolu-tion 56 42ndash57

MASON-GAMER R J 2001 Origin of North American Elymus (Po-aceae Triticeae) allotetraploids based on granule-boundstarch synthase gene sequences Systematic Botany 26 757ndash768

mdashmdashmdash 2004 Reticulate evolution introgression and intertribalgene capture in an allohexaploid grass Systematic Biology 5325ndash37

mdashmdashmdash and E A KELLOGG 1996a Chloroplast DNA analysis ofthe monogenomic Triticeae phylogenetic implications andgenome-specific markers Pp 301ndash325 in Methods of genomeanalysis in plants ed P P Jauhar Boca Raton CRC Press

mdashmdashmdash and mdashmdashmdash 1996b Testing for phylogenetic conflict amongmolecular data sets in the tribe Triticeae (Gramineae) Sys-tematic Biology 45 524ndash545

mdashmdashmdash and mdashmdashmdash 2000 Phylogenetic analysis of the Triticeaeusing the starch synthase gene and a preliminary analysisof some North American Elymus species Pp 102ndash109 in Grasssystematics and evolution eds S W L Jacobs and J EverettMelbourne CSIRO Publishing

mdashmdashmdash N L ORME and C M ANDERSON 2002 Phylogeneticanalysis of North American Elymus and the monogenomicTriticeae (Poaceae) using three chloroplast DNA data setsGenome 45 991ndash1002

mdashmdashmdash C F WEIL and E A KELLOGG 1998 Granule-boundstarch synthase structure function and phylogenetic utilityMolecular Biology and Evolution 15 1658ndash1673

MCDADE L A 1992 Hybrids and phylogenetic systematics II Theimpact of hybrids on cladistic analysis Evolution 46 1329ndash1346

2004] 861HELFGOTT AND MASON-GAMER ELYMUS PHYLOGENY

mdashmdashmdash 1997 Hybrids and phylogenetic systematics III Compar-ison with distance methods Systematic Botany 22 669ndash683

PETERSEN G and O SEBERG 1997 Phylogenetic analysis of theTriticeae (Poaceae) based on rpoA sequence data MolecularPhylogenetics and Evolution 7 217ndash230

mdashmdashmdash and mdashmdashmdash 1998 Molecular studies on the phylogeny ofthe genus barley (Hordeum Poaceae) Pp 437ndash440 in Moleculartools for screening biodiversity plants and animals eds A KarpP G Isaac and D S Ingram New York Chapman and Hall

mdashmdashmdash and mdashmdashmdash 2002 Molecular evolution and phylogeneticapplication of DMC1 Molecular Phylogenetics and Evolution 2243ndash50

mdashmdashmdash and mdashmdashmdash 2003 Phylogenetic analyses of the diploid spe-cies of Hordeum (Poaceae) and a revised classification of thegenus Systematic Botany 28 293ndash306

RAUSCHER J T J J DOYLE and A H D BROWN 2002 Internaltranscribed spacer repeat-specific primers and the analysis ofhybridization in the Glycine tomentella (Leguminosae) poly-ploid complex Molecular Ecology 11 2691ndash2702

ROHDE W D BECKER and F SALAMINI 1988 Structural analysisof the waxy locus from Hordeum vulgare Nucleic Acids Research16 7185ndash7186

SANDERSON M J 1998 Estimating rate and time in molecular phy-logenies beyond the molecular clock Pp 242ndash264 in Molec-ular systematics of plants II DNA sequencing eds D E SoltisP S Soltis and J J Doyle Boston Kluwer Academic Pub-lishers

mdashmdashmdash and M J DONOGHUE 1989 Patterns of variation in levelsof homoplasy Evolution 43 1781ndash1795

SANG T and D ZHANG 1999 Reconstructing hybrid speciationusing sequences of low copy nuclear genes hybrid origins offive Paeonia species based on Adh gene phylogenies System-atic Botany 24 148ndash163

SEBERG O and G PETERSEN 1998 A critical review of conceptsand methods used in classical genome analysis The BotanicalReview 64 372ndash417

SEELANAN T A SCHNABEL and J F WENDEL 1997 Congruenceand consensus in the cotton tribe (Malvaceae) Systematic Bot-any 22 259ndash290

SHIMODAIRA H and M HASEGAWA 1999 Multiple comparisonsof log-likelihoods with applications to phylogenetic inferenceMolecular Biology and Evolution 16 1114ndash1116

SMALL R L and J F WENDEL 2000 Phylogeny duplication and

intraspecific variation of Adh sequences in New World diploidcottons Molecular Phylogenetics and Evolution 16 73ndash84

SMEDMARK J E E T ERIKSSON R C EVANS and C S CAMPBELL2003 Ancient allopolyploid speciation in Geinae (Rosaceae)evidence from nuclear granule-bound synthase (GBSSI) genesequences Systematic Biology 52 374ndash385

SULLIVAN J A MARKERT and C W KILPATRICK 1997 Phylo-geography and molecular systematics of the Peromyscus az-tecus species group (Rodentia Muridae) inferred using parsi-mony and likelihood Systematic Biology 46 426ndash440

SVENSSON P O E BLASING and P WESTHOFF 2003 Evolution ofC4 phosphoenolpyruvate carboxylase Archives of Biochemistryand Biophysics 414 180ndash188

SWOFFORD D L 2002 PAUP Phylogenetic analysis using parsi-mony (and other methods) version 4b10 Sunderland Sin-auer Associates

mdashmdashmdash G J OLSEN P J WADDELL and D M HILLIS 1996 Phy-logenetic inference Pp 407ndash514 in Molecular systematics 2nded eds D M Hillis C Moritz and B K Mable SunderlandSinauer Associates

THOMPSON J D D G HIGGINS and T J GIBSON 1994 CLUSTALW improving the sensitivity of progressive multiple se-quence alignment through sequence weighting position spe-cific gap penalties and weight matrix choice Nucleic AcidsResearch 22 4673ndash4680

VARDI A 1973 Introgression between different ploidy levels inthe wheat group Pp 131ndash141 in Proceedings of the Fourth In-ternational Wheat Genetics Symposium eds E R Sears and LM S Sears Columbia University of Missouri Press

WADDELL P J and D PENNY 1996 Evolutionary trees of apes andhumans from DNA sequences Pp 53ndash73 in Handbook of sym-bolic evolution eds A J Locke and C R Peters Oxford Clar-endon Press

WAGNER A N BLACKSTONE P CARTWRIGHT M DICK B MISOFP SNOW G P WAGNER J BARTELS M MURTHA and J PEN-DLETON 1994 Surveys of gene families using polymerasechain reaction PCR selection and PCR drift Systematic Biol-ogy 43 250ndash261

YANG Z 1993 Maximum likelihood estimation of phylogeny fromDNA sequences when substitution rates differ over sites ap-proximate methods Molecular Biology and Evolution 10 1396ndash1401

mdashmdashmdash 1994 Estimating the pattern of nucleotide substitutionJournal of Molecular Evolution 39 105ndash111