Amplified fragment length polymorphism in Elymus …The amplified fragment length polymorphism (AFLP) tech-nique (Vos et al. 1995) is a robust and highly effective method of DNA fingerprinting
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Amplified fragment length polymorphism inElymus elymoides, Elymus multisetus, and otherElymus taxa
Steven R. Larson, Thomas A. Jones, Carrie L. McCracken, and Kevin B. Jensen
Abstract: The geographic and phylogenetic significance of amplified fragment length polymorphism within and among22 Elymus elymoides (Raf.) Swezey subsp. elymoides, 24 E. elymoides subsp. brevifolius (J.G. Sm.) Barkworth, and 13Elymus multisetus (J.G. Sm.) Burtt-Davy squirreltail accessions was assessed relative to six other North American andthree Eurasian Elymus taxa. Elymus elymoides and E. multisetus, comprising Elymus sect. Sitanion (Raf.) Á. Löve,were both monophyletic and closely related compared with other congeners. The monophyly of subsp. elymoides wasalso supported; subsp. brevifolius, however, was paraphyletic and separated into four genetically distinct groups. Esti-mates of nucleotide divergence among the five E. elymoides groups range from 0.0194 to 0.0288, with approximately0.0329 differences per site between E. elymoides and E. multisetus. Corresponding estimates of nucleotide divergencerange from 0.0243 to 0.0387 among North American taxa and from 0.0337 to 0.0455 between North American andEurasian taxa. DNA polymorphism among E. elymoides accessions was correlated with geographic provenance andpreviously reported quantitative traits. Distinct genetic groups of E. elymoides generally correspond to different geo-graphic regions, whereas divergent E. multisetus and E. elymoides accessions are sympatric. Thus, taxonomic ranks ofE. multisetus and E. elymoides were supported and geographic groups within E. elymoides were distinguished.
Résumé : Les auteurs ont évalué la signification géographique et phylogénétique du polymorphisme de la longueur desfragments de restriction, à l’intérieur et entre 22 accessions d’Elymus elymoides (Raf.) Swzey subsp. elymoides, 24 d’E.elymoides subsp. brevifolia (J. G. Sm.) Barkworth, et 13 d’Elymus multisetus (J. G. Sm.) Burtt-Davy, comparativementà six autres taxons nord-américains et trois taxons eurasiens d’Elymus. L’Elymus elymoides et l’Elymus multisetus,comprenant l’Elymus sect. Sitanion (Raf.) A. Löve, sont tous deux monophylétiques et étroitement reliés comparative-ment à d’autres congénères. Les données supportent également la monophylie de la subsp. elymoides; la subsp. brevifo-lius est cependant paraphylétique et se sépare en quatre groupes génétiquement distincts. L’estimation de la divergencedes nucléotides parmi les cinq groupes de l’E. elymoides se situe entre 0,0194 et 0,0288, avec une différence d’ envi-ron 0,0329 par site entre les E. elymoides et E. multisetus. Les estimés correspondants des divergences des nucléotidesvont de 0,0243 à 0,0387 entre les taxons nord-américains et 0,0337 à 0,0455 entre les taxons nord-américains et eura-siens. Le polymorphisme de l’ADN au sein des accessions de l’E. elymoides est corrélé avec la provenance géogra-phique et les caractères quantitatifs déjà rapportés. Les groupes distincts de l’E. elymoides correspondent généralementà différentes régions géographiques, alors que les accessions divergentes de l’E. multisetus et de l’E. elymoides sontsympatriques. Ainsi, les rangs taxonomiques de l’E. multisetus et de l’E. elymoides trouvent un support et on distinguedes groupes géographiques au sein de l’E. elymoides.
Mots clés : AFPL, Elymus, diversité des nucléotides, élyme.
[Traduit par la Rédaction] Larson et al. 804
Introduction
Large-scale seedings of perennial grasses, shrubs, andforbs on North American rangelands have had a major im-pact on fire cycles, weed suppression, soil stabilization, for-
age production, and habitat qualities. These plantings havealso impacted rangeland species composition, as evidencedby the widespread abundance of introduced crested wheat-grasses (Agropyron cristatum and Agropyron desertorum).Likewise, large-scale seedings of native plants may alsochange the natural abundance, distribution, and variability ofthe respective natural flora. However, the latter effects arenot easily discerned and the geographic significance of natu-ral genetic variability has not been well documented innative range grasses of western North America. A better ap-preciation of genetic diversity in these rangeland grasseswill help researchers and land managers develop and selectseed sources needed for large-scale revegetation.
Received 28 October 2002. Published on the NRC ResearchPress Web site at http://canjbot.nrc.ca on 4 September 2003.
S.R. Larson,1 T.A. Jones, and K.B. Jensen. USDA-ARS,Forage and Range Research Laboratory, Utah StateUniversity, Logan, UT 84341, U.S.A.C.L. McCracken. Laboratories of Analytical Biology,National Museum of Natural History, Smithsonian Institution,4210 Silver Hill Road, Suitland, MD 20746, U.S.A.
J:\cjb\cjb8108\B03-077.vpAugust 29, 2003 8:24:52 AM
Color profile: Generic CMYK printer profileComposite Default screen
bottlebrush squirreltail and Elymus multisetus (J.G. Sm.)Burtt Davy (Sitanion jubatum J.G. Sm.) big squirreltailgrasses of western North America (Barkworth 1997;Holmgren and Holmgren 1977). Taxonomists (Barkworth1997; Holmgren and Holmgren 1977) also recognized foursubspecific taxa of bottlebrush squirreltail: Elymuselymoides subsp. elymoides (Sitanion hystrix var. hystrix),Elymus elymoides subsp. brevifolius (J.G. Sm.) Barkworth(Sitanion hystrix var. brevifolium (J.G. Smith) C.L. Hitchc.),Elymus elymoides subsp. californicus (J.G. Sm.) Barkworth(Sitanion hystrix var. californicum (J.G. Sm.) F.D. Wilson),and Elymus elymoides subsp. hordeoides (Suksd.)Barkworth (Sitanion hystrix var. hordeoides (Suksd.) C.L.Hitchc.). These squirreltail taxa were transferred from genusSitanion Raf. to genus Elymus L. (Barkworth et al. 1983;Barkworth and Dewey 1985) on the basis of hybridizationand cytogenetic studies (Brown and Pratt 1960; Church1967a, 1967b; Dewey 1967, 1969; Stebbins et al. 1946;Stebbins and Snyder 1956; Stebbins and Vaarama 1954).Chromosome-pairing data suggest that virtually all of thenative North American species of Elymus are StStHHallotetraploids (2n = 4x = 28), derived from Pseudo-roegneria (St genome) and Hordeum (H genome) (Dewey1982, 1983a, 1983b, 1984). Little or no significant variationin nuclear DNA content has been detected among divergentallotetraploid Elymus species (Vogel et al. 1999). Likewise,the chloroplast ndhF DNA sequences of E. elymoides andother allotetraploid Elymus species are virtually identical(Redinbaugh et al. 2000; Mason-Gamer et al. 2002). Theplacement of E. elymoides in Elymus was also supported bythe phylogeny of granule-bound starch synthase nuclearDNA sequences (Mason-Gamer 2001). Yet, bottlebrushsquirreltail (E. elymoides) and big squirreltail (E. multisetus)taxa are distinguished from other Elymus species by havinga brittle rachis and subulate glumes extending into longawns (Wilson 1963), suggesting a sister relationship of thesetwo taxa. Consequently, E. elymoides and E. multisetus wereretained together as sect. Sitanion of genus Elymus (Löve1984). However, these explicitly phylogenetic studies(Redinbaugh et al. 2000; Mason-Gamer 2001; Mason-Gameret al. 2002) did not detect sufficient DNA polymorphism toevaluate genetic diversity or relationships within and be-tween E. elymoides and E. multisetus.
Squirreltail grasses are widely distributed across diverseelevation and precipitation zones of western North America(Wilson 1963). The disarticulating rachis and long, arcuatelydiverging glume and lemma awns promote wind dispersalacross open ground (Barkworth 1997). Like most Elymusspecies, squirreltail grasses are relatively short-lived, prolificseed producers. With the exception of several species(including E. lanceolatus and E. wawawaiensis examinedhere), most allotetraploid Elymus species are self-fertilizing(Keller 1948; Jensen et al. 1990; Smith 1944). Althoughsquirreltail grasses have not been regarded as important for-ages, these perennial grasses are gaining increasing attentionin rangeland revegetation partly because they are naturallyadapted to colonize disturbed areas and may help suppressthe invasion of weeds (Jones 1998). Consequently, Jones etal. (2002) evaluated 12 quantitative traits among fiveE. elymoides subsp. brevifolius, 17 E. elymoides subsp.elymoides, and four E. multisetus accessions and nine quan-
titative traits among 21 E. elymoides subsp. brevifolius, 10E. elymoides subsp. elymoides, and 16 E. multisetus acces-sions grown under uniform field conditions. Quantitativetraits measured in both evaluations (Jones et al. 2002) in-cluded days to emergence, heading date, leaf length, plantheight, root length, root to shoot ratio, seed mass, specificroot length, and total plant dry matter. Geographically di-verse collections evaluated by Jones et al. (2002) representthe three most abundant and widely distributed squirreltailtaxa, particularly in the Rocky Mountain and IntermountainFloristic provinces. In both germplasm comparisons, thesethree squirreltail taxa were effectively separated into differ-ent groups by multivariate principle components analysis.Moreover, the subsp. brevifolius accessions also separatedinto three well-defined subgroups designated A, B, and C.These squirreltail groups evidently display different adapta-tions and recognition of these groups may be important.However, the phylogenetic significance of these squirreltailgroups was uncertain (Jones et al. 2002).
The amplified fragment length polymorphism (AFLP) tech-nique (Vos et al. 1995) is a robust and highly effective methodof DNA fingerprinting that can be used to measure and esti-mate nucleotide diversity (Innan et al. 1999), as demonstratedin cross-pollinating (Larson et al. 2000) and self-pollinating(Larson et al. 2001) plant species. Nucleotide diversity and di-vergence are standard measures of DNA variation used to in-vestigate population dynamics, evolutionary relationships, andother biological phenomena (Nei 1987). In the context of nat-ural populations within plant species, amplified fragmentlength polymorphism may be correlated with geographic ori-gin (Larson et al. 2001; Massa et al. 2001). Moreover, theAFLP technique can also resolve phylogenetic relationshipsamong closely related (i.e., congeneric) grass species(Aggarwal et al. 1999; Massa et al. 2001).
This study investigates the geographic and phylogeneticsignificance of amplified fragment length polymorphismwithin and among squirreltail taxa and subtaxonomic groupspreviously distinguished by quantitative trait variation (Joneset al. 2002). Specific objectives of this investigation were to(i) test genetic relationships and compare rates of nucleotidevariation within and among E. elymoides subsp. elymoidesand E. elymoides subsp. brevifolius groups A, B, C, and Dthat were distinguished by quantitative trait evaluations(Jones et al. 2002), (ii) test genetic relationships and com-pare rates of nucleotide variation between E. elymoides andE. multisetus (sect. Sitanion) relative to nine other allote-traploid Elymus species, including six North American taxa(Elymus canadensis L., Elymus glaucus Buckl., Elymushystrix (Moench) Á. Löve, Elymus lanceolatus (Scribner &Smith) Gould, Elymus trachycaulus (Link) Gould exShinners, and Elymus wawawaiensis J.R. Carlson J.R.Carlson and Barkworth) and three Eurasian taxa (Elymuscaninus L., Elymus mutabilis (Drobov) Tzvelev, and Elymussibiricus L.) representing five other sections of Elymus(Löve 1984), and (iii) test overall correlations among quanti-tative traits (Jones et al. 2002), geographic provenance, andamplified fragment length polymorphism. This research willhelp identify natural germplasm sources that represent ge-netic diversity in squirreltail, particularly among the Inter-mountain and Rocky Mountain floristic provinces, andelucidate phylogenetic relationships among Elymus species.
J:\cjb\cjb8108\B03-077.vpAugust 29, 2003 8:24:53 AM
Color profile: Generic CMYK printer profileComposite Default screen
Methods
Plant materialsSeeds of each accession were originally collected and de-
fined by their native site origin (Table 1) and reproduced atleast one generation. The 46 E. elymoides accessions and 13E. multisetus accessions were grown near Logan, Utah, andclassified to species or subspecies (Table 1) using a dichoto-mous key (Wilson 1963) as described by Jones et al. (2002).Accessions with D-, DJ-, or T- collection numbers and theSand Hollow germplasm release were maintained in theUSDA-ARS Forage and Range Research Laboratory (Logan,Utah) germplasm collection along with detailed descriptionsfor each collection site (Jensen et al. 1997). Most of theseaccessions were submitted for preservation and distributionby the USDA-ARS National Plant Germplasm System(NPGS) Western Regional Plant Introduction Station (Pull-man, Wash.). Accessions with NRCS collection numberswere obtained directly from the USDA-NRCS or via NPGS.The MULT-13 accession (Table 1) was provided by theUnited States Department of Interior (USDOI) Bureau ofLand Management. Similarly, the ELYMb-01 accession (Ta-ble 1) Oregon State University (Corvallis, Oreg.). Seed forthe ELYMe-41, ELYMe-42, ELYMe-43, and ELYMe-44 ac-cessions (Table 1) was provided by the Maughan & Barton,Granite, Rainier, and Wind River seed companies, respec-tively. The remaining accessions (Table 1) were obtained di-rectly from NPGS. The latitude and longitude coordinates ofcollection sites for commercial seed sources, NRCS acces-sions, and other NPGS accessions were approximated for pur-poses of this study.
The E. canadensis (CANA), E. caninus (CANI), E. glaucus(GLAU), E. hystrix (HYST), E. lanceolatus (LANC),E. mutabilis (MUTA), E. sibiricus (SIBI), E. trachycaulus(TRAC), and E. wawawaiensis (WAWA) accessions (Table 1)were obtained directly from the USDA Forage and Range Re-search Laboratory (Jensen et al. 1997). All taxa and acces-sions are allotetraploid (2n = 4x = 28).
Seeds were germinated on moist blotter paper. Seven seed-lings of each accession were grown in single-plant containersin a greenhouse. Although morphological variation is readilyapparent among most of the squirreltail accessions, individualplants are generally very uniform within accessions (Jones etal. 2002). For the purposes of this study, two seedlings wererandomly sampled from each accession. Voucher specimens(listed in Table 1) were submitted to the Intermountain Her-barium at Utah State University (Logan, Utah).
DNA analysesSamples of 100 mg of leaf tissue were collected from
each seedling and placed in 2-mL microcentrifuge tubescontaining two steel bearings (5 mm in diameter). Thesesamples were subsequently frozen under liquid nitrogen andvortexed into a fine powder. One millilitre of extractionbuffer (2% hexadecyltrimethyl-ammonium bromide (CTAB),1.4 M NaCl, 20 mM ethylenediaminetetraacetic acid(EDTA), 100 mM Tris–HCl (pH 8.0), 0.2% β-mercapto-ethanol, and 0.1 RNAase mg/mL, 65 °C) was added to thefrozen leaf powder and incubated at 65 °C for at least 1 h. A24:1 (v/v) solution of chloroform – isoamyl alcohol wasadded and mixed vigorously prior to phase separation by
centrifugation (14 000g for 5 min). The upper aqueous phase(containing nucleic acids) was transferred to a 1.5-mLmicrocentrifuge tube and mixed with 0.7 mL of coldisopropanol. Nucleic acids were hooked out with a glass pi-pette and washed in a solution of 70% ethanol and 10 mMammonium acetate, air dried, and dissolved in TE buffer(10 mM Tris–HCl (pH 8.0) and 1 mM EDTA (pH 8.0)).Genomic DNA quantity and quality were evaluated byagarose gel electrophoresis.
DNA fingerprinting was conducted using the AFLP tech-nique according to the methods of Vos et al. (1995), exceptthat EcoRI selective amplification primers included a fluores-cent 6-carboxyfluorescein label on the 5′ nucleotide. Selectiveamplifications were performed using six EcoRI +3 – MseI +3primer pairs (e.g., E.AGC//M.CAG, E.AGC//M.CAT,E.AGC//M.CTG, E.AGG//M.CAA, E.AGG/M.CAC, E.AGG/M.CAG), where E and M designate the EcoRI and MseI adapt-ers with three selective nucleotides as described by Vos et al.(1995). The amplified DNA fragments were size fractionatedusing an ABI3100 instrument with 50-cm capillaries, POP-6polymer, GeneScan 400HD (rhodamine X) internal size stan-dards, and Genescan software (PE Applied Biosystems, Fos-ter City, Calif). The GeneScan sample files were subsequentlyanalyzed for the presence and absence of DNA fragments, be-tween 50 and 400 bp, using Genographer version 1.5(Benham et al. 1999). Although subjective, the first author at-tempted to score virtually all fragments into allelic categoriesbased on comparisons of fragments with similar relative mi-gration coefficients (determined primarily by the number ofnucleotide base pairs). Categories were separated by obviousor seemingly discrete differences in relative migration. Thus,some categories were more or less variable than others interms of relative migration units (estimated in nucleotide basepairs). However, most categories were at least 0.5 relative mi-gration unit apart. Virtually all fragments that showed asmooth fluorescent trace signal (i.e., clearly above stochasticbackground signals) were considered. However, possible frag-ment categories that did not show discrete differences fromstochastic background signals were ignored.
Data analysesThe total number of fragments per plant (M) and total
number of differences between plants (P) were computed di-rectly from binary data sets of fragment present (1) and frag-ment absent (0). The proportion of shared fragmentsbetween plants (F) was computed using the formula F =(MX + MY – P)/(MX + MY), where MX and MY denote the to-tal number of fragments for each of the two plants beingcompared. This formula for F is equivalent to that reportedby Nei and Li (1979). Estimates of total nucleotide diversitywithin taxa (πt), nucleotide diversity within accessions (π),and nucleotide divergence among subspecific groups or taxa(D) were estimated based on corresponding F and M valuesusing methods and software described by Innan et al. (1999).The corrected nucleotide divergence (DA) was calculated us-ing the formula DA = D – (πX – πY)/2, where πX and πY de-note the total nucleotide diversity (πt) within the two taxa orgroups being compared. Differentiation (gst) among acces-sions within taxa or subspecific groups was calculated usingthe formula gst = (πt – π)/πt. Similarly, differentiation amongtaxa or subspecific groups (Gst) was calculated using the for-
J:\cjb\cjb8108\B03-077.vpAugust 29, 2003 8:24:54 AM
Color profile: Generic CMYK printer profileComposite Default screen
mula Gst = DA/D. Neighbor-joining analyses (Saitou and Nei1987) of genetic relationships were based on a user-defineddistance matrix (1 – F) in PAUP* version 4.0b8 (Swofford2000). Bootstrap confidence levels (Efron and Tibshirani1991; Felsenstein 1985) were recovered from the 70% ma-jority-rule consensus of 1000 neighbor-joining searches ofrestriction-site distance (Nei and Li 1979) computed fromthe binary allele data set using PAUP*. Although the topolo-gies of the restriction-site distance tree and 1 – F distancetree are identical, the restriction-site distance scale is mean-ingless for the AFLP data. Thus, neighbor-joining trees wereconstructed using the user-defined distance matrix of 1 – Fwith bootstrap confidence levels obtained from the analysesof restriction-site distances. An unrooted neighbor-joiningtree was also constructed based on the total nucleotide diver-gence (D) within and among taxa or groups. Entries for eachtaxon were essentially duplicated except that each pair ofduplicated entries was distinguished by estimates of total nu-cleotide variation (πt) within taxa or groups. Graphic dis-plays of these neighbor-joining trees were developed usingTREEVIEW (Page 1996).
Correlations between matrices of geographical distance,quantitative trait variation (Jones et al. 2002), average P be-tween accessions, and corrected number of DNA polymor-phism between accessions (PA) were evaluated by theMantel (1967) test statistic (Z) using the MxComp procedureof NTSYS-pc (Rohlf 1998). Significance tests for these cor-relations were determined by comparing observed valueswith values obtained by 1000 random permutations (Smouseet al. 1986). Therefore, the upper-tail probability (p) that1000 random Z values are (by chance) less than observedvalues of Z is 0.002 or greater. Geographical distance (kilo-metres) matrices were computed from geographical coordi-nates using the formula described in Math Forum (1997):kilometres = arccos[cos(LATX)cos(LONGX)cos(LATY)cos(LONGY) + cos(LATX)sin(LONGX)cos(LATY) sin(LONGY) +sin(LATX)sin(LATY)]r, where LATX, LONGX, LATY, andLONGY are the latitude and longitude (expressed in radians)for the two accessions (X and Y) and r is 6378 km, the ra-dius of Earth. Quantitative trait variation within two assem-blages of accessions evaluated by Jones et al. (2002) wasanalyzed using standardized normal deviates for each trait.Taxonomic distances based on these standardized normaldeviates were computed using the SimInt procedure ofNTSYS-pc (Rohlf 1998). The corrected number of DNApolymorphisms among accessions (PA) was computed usingthe formula [PXY – (PX + PY)/2], where PX + PY are the av-erage numbers of differences within accessions (X and Y)and PXY is the average number of differences between acces-sions. Metric parameters of genetic distance (P and PA) wereused primarily because corresponding estimates of the num-ber of nucleotide differences and corrected number of nucle-otide differences among all pairwise comparisons ofaccessions would be tedious.
A geographical map of collection sites for the 46E. elymoides and 13 E. multisetus squirreltail accessions wasdeveloped using ArcMapTM 8.2 (ESRI®, Redlands, Calif.).
J:\cjb\cjb8108\B03-077.vpAugust 29, 2003 8:24:55 AM
Color profile: Generic CMYK printer profileComposite Default screen
MseI +3 selective primer pairs were obtained from 161plants representing 83 accessions (Table 2). In some in-stances, weak or failed polymerase chain reaction amplifica-tions were successfully repeated. However, five individualDNA samples repeatedly displayed weak or failed polymer-ase chain reaction amplifications, which can probably beattributed to one or more steps of the template DNA prepa-rations. Therefore, only one useable plant genotype wasobtained from five accessions: ELYMb-24, ELYMb-28,ELYMb-29, MULT-13, and SIBI-01. The total number ofDNA fragments amplified from each plant was very consis-tent within these 11 allotetraploid species (Table 2). Amongthe six primer pairs analyzed, the average size of the small-est fragment scored was 55 bp and the average size of thelargest fragment scored was 396 bp. From Table 2, it can bededuced that the average number of fragments per plant (M)per primer pair (i.e., six) ranged from approximately 52 forE. mutabilis to 61.3 for E. multisetus (Table 2). A total of1265 different DNA fragments were resolved among the 161Elymus plants, with only 27 monomorphic bands. However,the proportion of shared fragments between individual plantswas much higher than would be expected by chance alone,even among the most divergent comparisons such asE. mutabilis versus E. multisetus (Table 3). Methods used inthis study can resolve at least 341 different fragment catego-ries per primer pair in the size range of 55–396 bp. Wedetected approximately 0.18 fragment per category (61fragments/341 categories) for E. mutabilis and 0.15 frag-ment per category (52 fragments/341 categories) forE. mutabilis. Thus, the deduced probability that heterologousfragments cofractionate into the same category is approxi-mately 0.165. The proportion of shared fragments amongElymus species, ranging from 0.535 to 0.794 (Table 3), ismuch higher than expected by chance alone. These observa-tions indicate that the aforementioned Elymus taxa sharemany homologous DNA fragments detected using the AFLPtechnique.
A high degree of genetic identity was apparent within ac-cessions. With the exceptions of ELYMe-31, MULT-10, andMULT-11, individual plants group strictly by accession(Fig. 1). Genetic differentiation (gst) among accessionsranged from 0.506 to 0.923 within the self-fertileE. elymoides, E. multisetus, E. canadensis, E. hystrix,E. mutabalis, E. trachycaulus, E. glaucus, E. caninus, andE. sibiricus species (Table 2). Compared with most of theself-fertile taxa, E. multisetus displayed relatively less DNAvariation among accessions and (or) greater DNA variationwithin accessions (Table 2; Fig. 1). Therefore, gst amongE. multisetus accessions was substantially lower than amongthe other eight self-fertile taxa (Table 2). The self-incompatible E. lanceolatus and E. wawawaiensis speciesdisplayed considerably more DNA variation within acces-sions and lower gst values compared with the self-fertile taxa(Table 2).
Estimates of total nucleotide divergence (D) among theeight North American taxa range from 24.3 and 38.7 differ-ences per 1000 nucleotides (Table 4). Similarly, pairwisecomparisons of D among the Eurasian taxa E. caninus,E. mutabilis, and E. sibiricus range from 15.2 to 36.0 differ-ences per 1000. Elymus caninus and E. mutabilis were thetwo most similar taxa (i.e., D = 15.2 differences per 1000).
Slightly greater nucleotide divergence, 33.7–45.5 per 1000,was detected between North American and Eurasian taxa.Thus, Eurasian species seemingly form a natural outgroup(Fig. 2). Based on these empirical observations, the phylo-genetic tree (Fig. 1) was rooted using Eurasian E. caninus,E. mutabilis, and E. sibiricus as an outgroup. The apportion-ment of amplified fragment length polymorphism amongtaxa or subspecific groups (Fig. 1) is somewhat less pro-nounced than the apportionment of nucleotide variationamong taxa or subtaxonomic groups (Fig. 2). In particular,the proportion of polymorphic DNA fragments (1 – F)among accessions within taxa ranges from 0.081 to 0.253(Table 2) with corresponding estimates of 5.4–20.2 nucleo-tide differences per 1000 (Table 2), whereas the proportionof polymorphic DNA fragments among taxa ranges from0.206 to 0.465 (Table 3) with corresponding estimates of15.2–44.4 nucleotide differences per 1000 (Table 4). Like-wise, the proportion of polymorphic DNA fragments amongaccessions within subspecific groups ranges from 0.082 to0.183 (Table 5) with corresponding estimates of 5.8–11 (Ta-ble 5), whereas the proportion of polymorphic DNA frag-ments among subspecific groups ranges from 0.244 to 0.331(Table 6) with corresponding estimates of 19.4–28.8 nucleo-tide differences per 1000 (Table 7). Thus, estimates of nu-cleotide variation slightly accentuate divergence amonggroups relative to diversity within groups. Estimates of nu-cleotide differences are corrected for the probability thatheterologous DNA fragments cofractionate, by chancealone, as described by Innan et al. (1999). Moreover, thenumber of nucleotide differences corresponding to each am-plified fragment length polymorphism increases as a func-tion of overall genetic divergence among the genotypesbeing compared. In any case, the topographies of phylogen-etic relationships inferred from amplified fragment lengthpolymorphism per se (Fig. 1) and corresponding estimates ofnucleotide variation (Fig. 2) are similar.
The 10 Elymus species examined with more than one ac-cession were strictly monophyletic and strongly supportedby high bootstrap values for each respective group (Fig. 1).Moreover, four interspecific groups were also well supportedby phylogenetic analyses based on the proportion of ampli-fied fragment length polymorphism among individual plants(Fig. 1) and the average nucleotide divergence among taxa(Fig. 2): (i) E. elymoides and E. multisetus, (ii) E. cana-densis and E. hystrix, (iii) E. lanceolatus and E. wawa-waiensis, and (iv) the Eurasian E. caninus, E. mutabilis, andE. sibiricus accessions. The E. caninus and E. sibiricus ac-cessions were most similar (Fig. 1). Phylogenies based onthe proportion of amplified fragment length polymorphismamong individual plants (Fig. 1) and the average nucleotidedivergence among taxa (Fig. 2) are consistent with a mono-phyletic origin of Elymus sect. Sitanion (i.e., the E. ely-moides and E. multisetus group). However, estimates ofnucleotide divergence between E. elymoides and E. multi-setus are similar to or greater than D among the other NorthAmerican Elymus taxa examined (Table 4). The monophylyof E. elymoides subsp. elymoides was also supported.Elymus elymoides subsp. brevifolius, on the other hand, wasparaphyletic and separated into four genetically distinctgroups supported by high bootstrap confidence levels(Fig. 1). Estimates of nucleotide divergence among these
J:\cjb\cjb8108\B03-077.vpAugust 29, 2003 8:24:55 AM
Color profile: Generic CMYK printer profileComposite Default screen
five subspecific groups range from 19.4 to 28.8 differencesper 1000 bases (Table 7), values that are less than mostinterspecific comparisons (Table 4).
Associations among geographic provenance, quantitativetrait variation, and DNA polymorphism among accessionswere detectable in two assemblages of squirreltail accessionsthat were evaluated by Jones et al. (2002) and this study.The 12 quantitative traits measured by Jones et al. (2002) forgermplasm assemblage 1 (Table 8) included days to emer-gence, leaf length, total plant dry matter, root to shoot ratio,leaf area, specific leaf area, root length, heading date, seedmass, emergence index from 20 mm, emergence index from60 mm, and nitrate reductase activity. The nine quantitativetraits measured by Jones et al. (2002) for germplasm assem-blage 2 (Table 9) included days to emergence, leaf length,total plant dry matter, root to shoot ratio, root length, spe-cific root length, heading date, plant height, and seed mass.Correlation between DNA polymorphism (P and PA) andquantitative trait variation was greater when E. elymoidesand E. multisetus accessions were compared collectively(Tables 8 and 9). However, associations between DNA poly-morphism (P and PA) and geographic origin were dimin-ished when E. elymoides and E. multisetus accessions wereincluded together (Tables 8 and 9). Thus, genetically distinctE. multisetus and E. elymoides accessions were collectedfrom the same general region, whereas genetically distin-guishable groups within E. elymoides generally originatefrom different geographic regions (Fig. 3). WithinE. elymoides, correlations of amplified fragment length poly-morphism (P and PA) and quantitative trait variation (Ta-bles 8 and 9) were slightly better than correlations of DNApolymorphism and geographic distance (Tables 8 and 9).
Within E. elymoides germplasm assemblage 1 (Table 8), thecorrelation of geographic provenance and quantitative traitvariation was better than correlations of DNA polymorphism(P and PA) and quantitative trait variation. Conversely, inE. elymoides germplasm assemblage 2 (Table 9), the correla-tions of DNA polymorphism (P and PA) and quantitativetrait variation were better than the correlation of geographicprovenance and quantitative trait variation. Correlation ofquantitative trait variation and DNA polymorphism is evi-dent by the fact that the morphological groups described byJones et al. (2002) precisely correspond to four geneticallydistinct E. elymoides groups (subsp. brevifolius groups A, B,and C and subsp. elymoides) (Figs. 1 and 2). Only oneaccession from E. elymoides subsp. brevifolius group D(Fig. 1) was examined by Jones et al. (2002). Therefore, thislatter E. elymoides subsp. brevifolius group was not recog-nized by Jones et al. (2002).
The correlation between geographic provenance and aver-age DNA polymorphism among all 59 squirreltail accessionswas 0.42, or 0.45 corrected for DNA polymorphism withinaccessions. The correlation between geographic provenanceand average DNA polymorphism strictly among the 46E. elymoides accessions was 0.55, or 0.54 corrected forDNA polymorphism within accessions. The overall correla-tion between geographic provenance and DNA polymor-phism among pairwise comparisons of individual plants,within and among the 46 E. elymoides accessions, was 0.57.Thus, a high degree of genetic identity within accessions(Fig. 1) contributed slightly to the overall correlation ofDNA polymorphism and geographic provenance among the46 E. elymoides accessions (Fig. 3). All of these correlationsare significant (p ≤ 0.002).
Table 2. Summary of amplified fragment length polymorphism within Elymus taxa detected using six EcoRI +3 – MseI +3 selectiveprimer pairs including average values for the number of fragments per plant (M), number of differences between plants (P), proportionof shared fragments between plants (F), total nucleotide diversity within taxa (πt), nucleotide diversity within accessions (π), and ge-netic differentiation among accessions within taxa (gst).
Fig. 1. Neighbor-joining phylogeny based on proportions of amplified fragment length polymorphism (1 – F) among individual plantsof Elymus elymoides subsp. elymoides (ELYMe), E. elymoides subsp. brevifolius (ELYMb), E. multisetus (MULT), E. canadensis(CANA), E. hystrix (HYST), E. glaucus (GLAU), E. lanceolatus (LANC), E. wawawaiensis (WAWA), E. trachycaulus (TRAC),E. caninus (CANI), E. mutabilis (MUTA), and E. sibiricus (SIBI). Bootstrap confidence levels are indicated for clades present in the50% majority-rule consensus tree.
J:\cjb\cjb8108\B03-077.vpAugust 29, 2003 8:24:56 AM
Color profile: Generic CMYK printer profileComposite Default screen
J:\cjb\cjb8108\B03-077.vpAugust 29, 2003 8:24:58 AM
Color profile: Generic CMYK printer profileComposite Default screen
Discussion
AFLP provided new and useful measures of genetic varia-tion within and among E. elymoides, E. multisetus, and otherallotetraploid Elymus species. Two key observations supportthe taxonomic ranks of E. elymoides and E. multisetus:(i) amplified fragment length polymorphism and nucleotidedivergence among these taxa are similar to or greater thancorresponding genetic differences among well-known andmorphologically distinct Elymus species and (ii) E. ely-moides and E. multisetus accessions can be reliably classi-fied into genetically distinct monophyletic groups on thebasis of glume and floret structures (Wilson 1963). Elymuselymoides and E. multisetus are self-compatible in nature.Thus, we are not surprised to find genetically distinct E. ely-moides and E. multisetusplants growing at the same site orin the same regions (Fig. 3). Although most allotetraploidElymus species can hybridize and form partially or fully fer-tile hybrids, gene flow within and among these species isprobably controlled by self-fertilization (Jensen et al. 1990).
At least five natural groups within E. elymoides were dis-cerned by morphology (Jones et al. 2002) and DNA finger-printing. The monophyly of subsp. elymoides was supportedby DNA fingerprinting; subsp. brevifolius, however, wasparaphyletic and separated into four genetically distinctgroups. In particular, subsp. brevifolius group C was moreclosely related to subsp. elymoides than it was to othersubsp. brevifolius genotypes in groups A, B, and D. Interest-ingly, the brevifolius group C accessions originated from re-gions that are dominated by subsp. elymoides accessions(Fig. (3), at least in our germplasm assemblage. Thus, hy-
bridization or introgression may account for the paraphylyof subsp. brevifolius. Alternatively, subsp. elymoides may bea recently derived lineage of a more diverse ancestral group,subsp. brevifolius. Two other taxonomic groups, E. ely-moides subsp. californicus and E. elymoides subsp.hordeoides, were not examined. Thus, the monophyly ofsubsp. elymoides, or perhaps E. elymoides in general, maynot hold up with the inclusion of additional variants. Like-wise, the apparent paraphyly of subsp. brevifolius may beaffected in some way by the inclusion of subsp. californicusand subsp. hordeoides. However, subsp. hordeoides is soelusive that authors of this study have come to doubt its ex-istence. In any case, the general aspect of subsp. hordeoidesis similar to that of subsp. elymoides, and subsp. cali-fornicum intergrades with subsp. elymoides where they arecontiguous (Wilson 1963).
A significant correlation between amplified fragmentlength polymorphism and geographic distance was detected,especially within E. elymoides. Thus, geographic provenancewas discerned by quantitative trait variation (Jones et al.2002) and DNA fingerprinting (Fig. 3). However, geographicorigin per se may not be a very reliable indicator of geneticidentity or quantitative trait adaptations. For example, differ-ent accessions of E. elymoides subsp. brevifolius groupsA and B were collected from the same site in Colorado(Fig. 3). Likewise, E. elymoidies subsp. elymoides, E. ely-moides subsp. brevifolius, and E. multisetus accessions werecollected from common areas on the Idaho Snake RiverPlain (Fig. 3). Conversely, genetically similar accessions ofE. elymoides subsp. brevifolius group D (Fig. 2) are evi-dently widely dispersed across wide latitudes of the northern
Note: Above diagonal: nucleotide divergence (DA × 1000) corrected for nucleotide diversity (πt) within groups (Table 2) and genetic differentiation(Gst) (bold); below diagonal: estimates of the total nucleotide divergence (D × 1000).
aNot corrected for nucleotide diversity within E. mutabilis.
Table 4. Genetic differentiation among Elymus taxa.
J:\cjb\cjb8108\B03-077.vpAugust 29, 2003 8:24:59 AM
Color profile: Generic CMYK printer profileComposite Default screen
Great Plains region of western North America, sharply sec-tioned from other groups of bottlebrush squirreltail west ofthe Rocky Mountains (Fig. 3). Unfortunately, quantitativetraits of subsp. brevifolius group D were not evaluated byJones et al. (2002). Although subgroups of subsp. brevifoliuswere clearly distinguished by quantitative trait variation(Jones et al. 2002) and (or) amplified fragment length poly-morphism, E. elymoides subsp. brevifolius is paraphyletic,and it may be difficult to develop dichotomous keys thatwill reliably classify plants into these five subspecificgroups. Moreover, measures of amplified fragment lengthpolymorphism and nucleotide divergence among these sub-specific groups of E. elymoides are substantially less thancorresponding measures of genetic divergence amongE. elymoides, E. multisetus, and other Elymus taxa. A similarphenomenon was observed for South American accessionsof Bromus sect. Ceratochloa (Massa et al. 2001). With theexception of big squirreltail (E. multisetus), we believe thatthese squirreltail taxa should be retained within E. ely-moides. In addition to verifying two genetically distinctsquirreltail species, these results help identify naturally im-portant groups of bottlebrush squirreltail (E. elymoides).
Fig. 2. Unrooted neighbor-joining trees based on nucleotide di-vergence (D), the sum of average nucleotide diversity (πt) withinand corrected divergence (DA) among (A) Elymus taxa or(B) subspecific groups of E. elymoides.
Am
ong
acce
ssio
nsW
ithi
nac
cess
ions
Acc
essi
ons
(pla
nts)
M(S
D)
P(S
D)
F(S
D)
π t×
1000
P(S
D)
F(S
D)
π×
1000
g st
subs
p.br
evif
oliu
sgr
oup
A14
(27)
376.
3(9
.6)
120
(28.
3)0.
841
(0.0
38)
1120
.3(8
.2)
0.97
3(0
.011
)1.
80.
847
subs
p.br
evif
oliu
sgr
oup
B3
(6)
358.
5(4
.6)
132.
5(1
9.7)
0.81
7(0
.024
)13
.716
.3(1
0.1)
0.97
7(0
.014
)1.
50.
891
subs
p.br
evif
oliu
sgr
oup
C3
(6)
361.
8(1
3.0)
124.
5(1
2.5
0.82
8(0
.018
)12
.730
.3(1
1.2)
0.95
8(0
.015
)2.
80.
780
subs
p.br
evif
oliu
sgr
oup
D4
(8)
391.
6(4
.0)
64.1
(14.
8)0.
918
(0.0
18)
5.8
13.4
(11.
8)0.
978
(0.0
25)
1.5
0.74
1su
bsp.
elym
oide
s22
(42)
356.
0(6
.9)
117.
1(2
0.5)
0.83
6(0
.031
)12
.044
.5(2
4.1)
0.93
1(0
.045
)4.
70.
608
Tab
le5.
Sum
mar
yof
ampl
ifie
dfr
agm
ent
leng
thpo
lym
orph
ism
wit
hin
five
subs
peci
fic
grou
psof
Ely
mus
elym
oide
sde
tect
edus
ing
six
Eco
RI
+3
–M
seI
+3
sele
ctiv
epr
imer
pair
sin
clud
ing
aver
age
valu
esfo
rth
enu
mbe
rof
frag
men
tspe
rpl
ant
(M),
num
ber
ofdi
ffer
ence
sbe
twee
npl
ants
(P),
prop
orti
onof
shar
edfr
agm
ents
betw
een
plan
ts(F
),to
tal
nucl
eoti
dedi
vers
ity
wit
hin
grou
ps( π
t),nu
cleo
tide
dive
rsit
yw
ithi
nac
cess
ions
( π),
and
gene
tic
diff
eren
tiat
ion
amon
gac
cess
ions
wit
hin
grou
ps(g
st).
J:\cjb\cjb8108\B03-077.vpAugust 29, 2003 8:24:59 AM
Color profile: Generic CMYK printer profileComposite Default screen
AFLP also provided new and informative measures ofphylogenetic relationships and evolutionary divergenceamong three Eurasian and eight North American Elymustaxa. In the “Conspectus of the Triticeae”, Löve (1984) rec-
ognized several sections in Elymus. Section DasystachyaeLöve includes E. lanceolatus as the type species along withother long-anther species (Löve 1984). Recently describedE. wawawaiensis (Carlson and Barkworth 1997) was not
1. subsp. brevifolius group A — 179.3 (10.3) 226.3 (18.0) 206.4 (9.8) 219.1 (6.1)2. subsp. brevifolius group B 0.756 (0.014) — 232.3 (14.3) 202.4 (10.4) 217.5 (7.8)3. subsp. brevifolius group C 0.693 (0.022) 0.678 (0.016) — 249.8 (11.7) 205.8 (14.2)4. subsp. brevifolius group D 0.731 (0.013) 0.730 (0.013) 0.669 (0.012) — 237.1 (7.8)5. subsp. elymoides 0.701 (0.010) 0.696 (0.010) 0.714 (0.017) 0.683 (0.010) —
Note: Above diagonal: average number of amplified fragment length polymorphisms between plants (P) averaged among groups(SD in parentheses); below diagonal: average proportion of shared fragments between plants (F) averaged among groups (SD inparentheses).
Table 6. Amplified fragment length polymorphism among five subspecific groups of Elymus elymoides detected us-ing six EcoRI +3 – MseI +3 selective amplification primer pairs.
1 2 3 4 5
1. subsp. brevifolius group A — 6.7
0.345
13.7
0.529
13.3
0.602
13.0
0.5222. subsp. brevifolius group B 19.4 — 14.1
0.516
12.9
0.584
12.5
0.4943. subsp. brevifolius group C 25.9 27.3 — 19.6
0.681
11.1
0.4744. subsp. brevifolius group D 22.1 22.1 28.8 — 19.6
0.7265. subsp. elymoides 24.9 25.3 23.4 27.0 —
Note: Above diagonal: nucleotide divergence (DA × 1000) corrected for nucleotide diversity (πt)within groups (Table 5) and genetic differentiation (GST) (bold); below diagonal: estimates of the totalnucleotide divergence (D × 1000).
Table 7. Genetic differentiation among subspecific groups of Elymus elymoides.
Note: Above diagonal: matrix correlations (r) and corresponding significance values (p) (in parentheses) among 10Elymus elymoides and three E. multisetus accessions considered together; below diagonal: r and p (in parentheses)strictly among the 10 E. elymoides accessions.
Table 8. Associations of geographic provenance, quantitative trait variation, and amplified fragmentlength polymorhism for the first germplasm assemblage described by Jones et al. (2002).
Geographicdistance
Quantitative traitvariation
DNApolymorphism P
Corrected DNApolymorphism PA
Geographic distance — 0.52 (0.002) Not significant Not significantQuantitative trait variation 0.58 (0.002) — 0.73 (0.002) 0.75 (0.002)DNA polymorphim P 0.62 (0.002) 0.65 (0.002) — 0.98 (0.002)Corrected DNA polymorphism PA 0.57 (0.002) 0.66 (0.002) 0.97 (0.002) —
Note: Above diagonal: matrix correlations (r) and corresponding significance values (p) (in parentheses) among 21 Elymuselymoides and six E. multisetus accessions considered together; below diagonal: r and p (in parentheses) strictly among the21 E. elymoides accessions.
Table 9. Associations of geographic provenance, quantitative trait variation, and amplified fragment lengthpolymorhism for the second germplasm assemblage described by Jones et al. (2002).
J:\cjb\cjb8108\B03-077.vpAugust 29, 2003 8:24:59 AM
Color profile: Generic CMYK printer profileComposite Default screen
Fig. 3. Collecting sites of Elymus elymoides and E. multisetus squirreltail accessions distinguished by quantitative traits (Jones et al.2002) and DNA fingerprinting.
J:\cjb\cjb8108\B03-077.vpAugust 29, 2003 8:25:03 AM
Color profile: Generic CMYK printer profileComposite Default screen
classified into the sections recognized by Löve (1984); how-ever, a close relationship between E. wawawaiensis andE. lanceolatus (Figs. 1 and 2) is not unexpected (Carlsonand Barkworth 1997). Elymus lanceolatus and E. wawa-waiensis are distinguished from other species examined hereby having long anthers and self-incompatibility mechanisms.Section Elymus includes E. sibiricus as the type species,E. glaucus, and several other species. However, these lattertwo species did not form a group and E. glaucus did nothave a sister species in this study (Figs. 1 and 2). On theother hand, E. caninus and E. mutabilis of sect. Goulardia(Husnot) Tzvelev were both closely related to the Elymustype species E. sibiricus. Section Goulardia includesE. caninus as the type species, E. mutabilis, E. trachycaulus,and several other species with one spiklet per node anda tough rachis (Löve 1984). However, E. caninus andE. mutabilis were more like E. sibiricus, and E. trachycaulusdid not have an obvious sister among the species examinedhere (Figs. 1 and 2). Interestingly, an E. sibiricus accessionfrom Sichuan, P.R.C., displayed more random amplifiedpolymorphic DNA and microsatellite DNA similarity to Eur-asian E. caninus accessions than did an E. mutabilis acces-sion from Finland (Sun et al. 1997). Conversely, ourE. mutabilis accession from Kazakhstan displayed more am-plified fragment length polymorphism similarity to EurasianE. caninus accessions than did our E. sibiricus accessionsfrom P.R.C. and Russia (Figs. 1 and 2). In any case, DNAevidence does not seem to support distinction of E. caninusand E. mutabilis (sect. Goulardia) from the Elymus typespecies E. sibiricus (sect. Elymus). Elymus hystrix andE. canadensis are type species of sect. Hystrix (Moench) Á.Löve and sect. Macrolepis (Nevski) Jaaska, respectively,(Löve 1984). However, multiple accessions for bothE. hystrix and E. canadensis group together in onewell-defined lineage (Figs. 1 and 2). Thus, distinction ofsect. Hystrix and sect. Macrolepis may not be useful. Sec-tion Sitanion essentially includes E. elymoides as the typespecies and E. multisetus exclusively (Löve 1984). Thegrouping of E. elymoides and E. multisetus in sect. Sitanion(Löve 1984) was supported by DNA fingerprinting (Figs. 1and 2).
Estimates of nucleotide variation based on amplified frag-ment length polymorphism within and among E. elymoides,E. multisetus, and other Elymus taxa can be compared withthose of other species. Corresponding estimates of nucleo-tide variation among purple needlegrass (Nassella pulchra)populations range from 1.1 to 3.8 differences per 1000 nu-cleotides (Larson et al. 2001). These values are considerablylower than πt in E. elymoides, E. multisetus, and otherElymus taxa. Yet morphological variation among these pur-ple needlegrass populations is substantial (Knapp and Rice1998). The content and structure of coding and noncodingDNA display substantial variation among different grass spe-cies. Thus, rates of nucleotide variation may not correspondto phenotypic diversity in divergent species. Estimates of nu-cleotide variation in bluebunch wheatgrass (Pseudoroegneriaspicata) of the U.S. Palouse region approach 38 differencesper 1000 nucleotides (Larson et al. 2000). These estimatesare substantially higher than those of any of the Elymus spe-cies examined in this study, including self-incompatibleE. lanceolatus and E. wawawaiensis. Elymus and Pseudo-
roegneria genera share the St genome, which is evidentlyvery similar in size to the H genome of Hordeum andElymus (Vogel et al. 1999). Genetic similarity betweenElymus and Pseudoroegneria is also evident by the frequentdifficulty in distinguishing E. wawawaiensis and P. spicata.Thus, differences in nucleotide variation between Pseudo-roegneria and Elymus species cannot be easily attributed tomode of reproduction or genome structure. Interestingly, to-tal nucleotide variation among predominantly Elymus taxa iscomparable with nucleotide diversity within P. spicata(Larson et al. 2000). Pseudoroegneria spicata is a widelydistributed, cross-pollinating grass with no other congenersrecognized in North America. Thus, levels of DNA variationmaintained within one widely distributed cross-pollinatingspecies, P. spicata, may be comparable with DNA variationpartitioned among numerous self-pollinating Elymus taxa.Although estimates of nucleotide variation based on ampli-fied fragment length polymorphism may come into question,these standard parameters of genetic diversity and phylogen-etic relationships provide a useful reference. Corrected nu-cleotide divergence (DA) is very much dependent on thesampling of genotypes representing diversity (πt) within thetaxa or groups being compared. Therefore, the apportion-ment of DA and πt within Elymus species (Fig. 2) should beviewed skeptically. However, estimates of total nucleotidedivergence (D) among Elymus taxa or groups should belargely independent of sampling within species. Comparedwith simple measures of amplified fragment length polymor-phism (Fig. 1), we believe that estimates of nucleotide diver-gence should provide a more accurate assessment ofphylogenetic relationships among divergent taxa or sub-specific groups (Fig. 2).
Acknowledgements
This study was funded by joint contributions of the U.S.Department of Agriculture, Agriculture Research Service,and the U.S. Department of Interior, Bureau of Land Man-agement, Great Basin Restoration Initiative. We thankDr. Lynn Clark, Dr. Elizabeth Zimmer, and other reviewersfor their suggestions.
References
Aggarwal, R.K., Brar, D.S., Nandi, S., Huang, N., and Khush, G.S.1999. Phylogenetic relationships among Oryza species revealedby AFLP markers. Theor. Appl. Genet. 98: 1320–1328.
Barkworth, M.E. 1997. Taxonomic and nomenclatural commentson the Triticeae in North America. Phytologia, 83: 302–311.
Barkworth, M.E., and Dewey, D.R. 1985. Genomically based gen-era in the perennial Triticeae of North America: identificationand membership. Am. J. Bot. 72: 767–776.
Barkworth, M.E., Dewey, D.R., and Atkins, R.J. 1983. New con-cepts in the Triticeae of the Intermountain Region: key and com-ments. Great Basin Nat. 43: 561–572.
Benham, J., Jeung, J.-U., Jasieniuk, M., Kanazin, V., and Blake, T.1999. Genographer: a graphical tool for automated AFLP andmicrosatellite analysis. J. Agric. Genomics, 4: article 4. Avail-able from http://www.ncgr.org/jag/ [cited 15 March 2002].
Brown, W.V., and Pratt, G.A. 1960. Hybridization and intro-gression in the grass genus Elymus. Am. J. Bot. 47: 669–676.
Dewey, D.R. 1969. Synthetic hybrids of Agropyron albicans ×A. dasystachyum, Sitanion hystrix, and Elymus canadensis. Am.J. Bot. 56: 664–670.
Dewey, D.R. 1982. Genomic and phylogenetic relationships amongNorth American perennial Triticeae. In Grasses and grasslands.Edited by J.R. Estes. Oklahoma University Press, Norman, Okla.pp. 51–88.
Dewey, D.R. 1983a. New nomenclatural combinations in the NorthAmerican perennial Triticeae. Brittonia, 35: 30–33.
Dewey, D.R. 1983b. Historical and current taxonomic perspectivesof Agropyron, Elymus, and related genera. Crop Sci. 23: 637–642.
Dewey, D.R. 1984. The genomic system of classification as a guideto intergeneric hybridization within Triticeae. In Gene manipula-tion in plant improvement. Edited by J.P. Gustafsen. PlenumPress, New York. pp. 209–279.
Efron, B., and Tibshirani, R. 1991. Statistical data analysis in thecomputer age. Science (Washington, D.C.), 252: 390–395.
Felsenstein, J. 1985. Confidence limits on phylogenies: an ap-proach using the bootstrap. Evolution, 39: 783–791.
Holmgren, A.H., and Holmgren, N.H. 1977. Poaceae. In Inter-mountain flora. Vol. 6. Edited by A. Cronquist, A.H. Holmgren,N.H. Holmgren, J.L. Reveal, and P.K. Holmgren. Columbia Uni-versity Press, New York. pp. 175–464.
Innan, H., Terauchi, R., Kahl, G., and Tajima, F. 1999. A methodfor estimating nucleotide diversity from AFLP data. Genetics,151: 1157–1164.
Jensen, K.B., Zhang, Y.F., and Dewey, D.R. 1990. Mode of polli-nation of perennial species of the Triticeae in relation togenomically defined genera. Can. J. Plant Sci. 70: 215–225.
Jensen, K.B., Asay, K.H., and Jones, T.A. 1997. ARS germplasmholdings of grasses and legumes including the Triticeae collec-tion, native plant collection, and native and introduced legumesat Logan, UT. In Systematic collections of the Agricultural Re-search Service. Edited by J.R. Lichtenfels. U.S. Dep. Agric.Misc. Publ. No. 1343.
Jones, T.A. 1998. Viewpoint: The present status and future pros-pects of squirreltail research. J. Range Manage. 51: 326–331.
Jones, T.A., Nielson, D.C., Arredondo, J.T., and Redinbaugh, M.G.2002. Characterization of diversity among three squirreltail taxa.J. Range Manage. 56: 474–482.
Keller, W. 1948. Interpretation of self-fertility in grasses by fre-quency distributions. J. Am. Soc. Agron. 40: 894–900.
Knapp, E.E., and Rice, K.J. 1998. Comparison of isozymes andquatitative traits for evaluating patterns of genetic variation inpurple needlegrass (Nassella pulchra). Conserv. Biol. 12:1031–1041.
Larson, S.R., Jones, T.A., Hu, Z.-M., McCracken, C.L., and Palazzo,A. 2000. Genetic diversity of bluebunch wheatgrass cultivars anda multiple-origin polycross. Crop Sci. 40: 1142–1147.
Larson, S.R., Cartier, E., McCracken, C.L., and Dyer, D. 2001.Mode of reproduction and amplified fragment length polymor-phism variation in purple needlegrass (Nassella pulchra): utili-zation of natural germplasm sources. Mol. Ecol. 10: 1165–1177.
Löve, A. 1984. Conspectus of the Triticeae. Feddes Repert. 95:425–521.
Mantel, N.A. 1967. The detection of disease clustering and a gen-eralized regression approach. Cancer Res. 27: 209–220.
Mason-Gamer, R.J. 2001. Origin of North American Elymus(Poaceae: Triticeae) allotetraploids based on granule-boundstarch synthase gene sequences. Syst. Bot. 26: 757–768.
Mason-Gamer, R.J., Orme, N.L., and Anderson, C.M. 2002. Phylo-genetic analysis of North American Elymus and the mono-genomic Triticeae (Poaceae) using three chloroplast DNA datasets. Genome, 45: 991–1002.
Math Forum. 1997. Ask Dr. MathTM. Available from http://www.mathforum.com/dr.math/problems/longandlat.html [cited 27 March2002].
Nei, M. 1987. Molecular evolutionary genetics. Columbia Univer-sity Press, New York.
Nei, M., and Li, W.H. 1979. Mathematical model for studying ge-netic variation in terms of restriction endonucleases. Proc. Natl.Acad. Sci. U.S.A. 76: 5269–5273.
Page, R.D. 1996. TREEVIEW: an application to display phylogen-etic trees on personal computers. Comput. Appl. Biosci. 12:357–358.
Redinbaugh, M.G., Jones, T.A., and Zhang, Y. 2000. Ubiquity ofthe St chloroplast genome in St-containing Triticeae polyploids.Genome, 43: 846–852.
Rohlf, F.J. 1998. NTSYSpc numerical taxonomy and multivariateanalysis system version 2.0. Exeter Software, Setauket, N.Y.
Saitou, N., and Nei, M. 1987. The neighbor-joining method: a newmethod for reconstructing phylogenetic trees. Mol. Biol. Evol.4: 406–425.
Smith, D.C. 1944. Pollination and seed formation in grasses. J.Agric. Res. 68: 79–95.
Smouse, P.E., Long, J.C., and Sokal, R.R. 1986. Multiple regres-sion and correlation extensions of the Mantel test of matrixcorrespondance. Syst. Zool. 35: 627–632.
Stebbins, G.L., and Snyder, L.A. 1956. Artificial and natural hy-brids in the Gramineae, tribe Hordeae. V. Am. J. Bot. 43:305–312.
Stebbins, G.L., and Vaarama, A. 1954. Artificial and natural hy-brids in the Gramineae, tribe Hordeae. VII. Hybirds and allo-polyploids between Elymus glaucus and Sitanion spp. Genetics,39: 378–395.
Stebbins, G.L., Valencia, J.I., and Valencia, R.M. 1946. Artificialand natural hybrids in the Gramineae, tribe Hordeae. I. Elymus,Sitanion, and Agropyron. Am. J. Bot. 33: 338–351.
Sun, G.-L., Salomon, B., and von Bothmer, R. 1997. Analysis oftetraploid Elymus species using wheat microsatellite markersand RAPD markers. Genome, 40: 806–814.
Swofford, D.L. 2000. PAUP*. Phylogenetic analysis using parsi-mony (*and other methods). Version 4b10. Sinaur Associates,Sunderland, Mass.
Vogel, K.P., Arumuganathan, K., and Jensen, K.B. 1999. NuclearDNA content of perennial grasses of the Triticeae. Crop Sci. 39:661–667.
Vos, P.R., Hogers, M., Bleeker, M., Reijans, T., Van de Lee, T.,Hornes, M., Frijters, L., Pot, J., Peleman, M., Kuiper, M., andZabeau, M. 1995. AFLP: a new technique for DNA fingerprint-ing. Nucleic Acids Res. 23: 4407–4414.
Wilson, F.D. 1963. Revision of Sitanion (Triticeae, Gramineae).Brittonia, 15: 303–323.