A Phylogenetic Analysis of Epilobium (Onagraceae) Based on Nuclear Ribosomal DNA Sequences

Systematic Botany(1994), 19(3): pp. 363-388 ? Copyright1994 by the American Society of Plant Taxonomists

A Phylogenetic Analysis of Epilobium (Onagraceae) Based on Nuclear Ribosomal DNA Sequences DAVID A. BAUM and KENNETH J. SYTSMA

ofWisconsin,Madison,Wisconsin53706-1831 Department of Botany,University PETER C. HOCH MissouriBotanicalGarden,P.O. Box 299,St. Louis,Missouri63166-0299 spacersand 5.8S cistronof nuclearribosomalDNA were ABSTRACT.The internaltranscribed and two outgroups.Phylogenieswere inferredfromthe sequencedfrom22 speciesof Epilobium analysis. maximum-likelihood, and compatibility neighbor-joining, sequencesusing parsimony, (indels)vs. weightsforinsertions/deletions Withparsimony we exploredthe effectof different character vs. transversions, and a posteriori codingvs. non-coding regions,transitions substitutions, was foundto be sisterto therestof thegenus.The remainder reweighting. SectionChamaenerion clade,which ofEpilobium fellintotwomainclades: 1) sect.Epilobium,and 2) the"xerophytic" hadan uncertain (sect.Epilobium) theremaining sixsections.TheunusualspeciesE.rigidum comprises clade,or as the base of the "xerophytic" position,appearingeitherat the base of sect.Epilobium, clade,relationships Withinthe "xerophytic" sisterto the entiregenusexceptsect.Chamaenerion. groupconsistingof the annual sects.Boisduvalia resolvedbut a monophyletic were incompletely and theperennial,hummingbirdand Currania thegenusBoisduvalia) constituting (bothformerly was supported.The molecularphylogenyis compatiblewith some pollinatedsect.Zauschneria and cytological and is usefulforinterpreting morphological important independentcharacters evolutionin Epilobium. Epilobium is the largest genus of Onagraceae, comprisingapproximately170 species of mainly temperateherbs(Raven 1976,1988; Hoch and Raven 1992). The genus is remarkable for its morphological, ecological and cytological diversity.This variation is manifestedprimarily at the sectional level with eight sections being recognized. The sections are highly distinctive showing variationin vegetativeand floralmorphology, anatomy,palynology, cytology,phybiogeography,ecology,and breedtochemistry, ing systems (reviewed in Raven 1976, 1988). However, in the absence of a phylogeny for Epilobium, the patternofevolution of these characterscannotbe resolved. In thispaper we present a molecular phylogeneticstudyof Epilobium based on sequences of the internaltranscribed spacersand 5.8S gene of nuclear ribosomalDNA (rDNA) and then utilize the resultsto elucidate patternsof characterevolution in the genus. As well as clarifyingspecific issues in the systematicsof Epilobium,this study addresses some general issues in plant molecular systematics. We consider problems of aligning and scoring non-coding sequences subject to insertion/deletions (indels). Also, we explore the effectof utilizing differenttree-buildingalgo-

rithms (parsimony, neighbor-joining, maximum-likelihoodand compatibility)and different weighting schemes on the phylogenetic is one of the relconclusions. Finally,Epilobium atively few plant groups for which phylogenetic analysis has been completed from morphological data (Hoch and Crisci, in mss.) and both nuclear (this study) and organellar (Sytsma et al. 1991; unpubl. data) molecular data. The resultsof this study are thereforerelevant to ongoing debates over the relative utilityof molecularand morphologicaldata in plant phylogenetics. SYSTEMATIC BACKGROUND

is distinguishedfromthe The genus Epilobium rest of the familyby the combination of dotlike, heteropycnoticchromosomes (found also in Ludwigia;Kurabayashi et al. 1962), pollen usually shed in tetrads(found in some sections of Ludwigiaand a few species of Camissonia; phloem (CarlquSkvarla et al. 1975),interxylary ist 1975), multicellular papillae on stigmata (Heslop-Harrison 1990), and seeds usually adorned with a tuftof trichomes(a "coma") at the chalazal end (Raven 1976, 1988; Hoch et al.

363

This content downloaded from 128.104.46.196 on Mon, 02 Mar 2015 14:42:38 UTC All use subject to JSTOR Terms and Conditions

364

SYSTEMATIC BOTANY

[Volume 19

1993). In view of its distinctiveness,Epilobium sphere but with a majorradiationin Australasia and segregategenera have been placed in their (Raven and Raven 1976) and smaller ones in own tribeEpilobieae. temperateSouth America and southernAfrica. The relationshipbetween Epilobieae and the All species of sect. Epilobiumhave n = 18, but rest of Onagraceae was considered enigmatic several chromosomalrearrangementsappear to (Raven 1988), but recent cladistic analyses of delimit groups of species within the section both molecularand morphologicaldata support (Seavey and Raven 1977a, b, 1978). a close relationshipbetween Epilobieae and the Section Chamaenerion, with eight species, has other estipulate tribe, Onagreae (Crisci et al. been recognized at the generic level particu1990; Sytsmaet al. 1991; Bult and Zimmer 1993; larly by European botanists (e.g., Holub 1972). Conti et al. 1993; Hoch et al. 1993). Indeed, apart This section includes the widespread circumfromthe basal position of Ludwigia(supported boreal species Epilobiumangustifolium and E. laby all cladisticanalyses), an Onagreae plus Epi- tifoliumand six other Eurasian species. Two lobieae clade is the best supportedphylogenetic distinctivesubsectionsare recognized (Rosmarrelationshipin the family(Conti et al. 1993). inifolium and Leiostylae)with fourspecies in each Traditionally(e.g., Munz 1965), tribe Epilo- (Raven 1976). Membersofsect.Chamaenerion are bieae has been considered to include up to three protandrous,bumble-bee pollinated perennials genera apart from Epilobium(Boisduvalia,Cha- and have several characters that distinguish maenerionand Zauschneria).Raven (1976) rec- them fromthe restof Epilobium (e.g., floraltube ognized two genera, Boisduvaliawith two sec- absent, spiral phyllotaxis,pollen not shed in tions(Raven and Moore 1965) and Epilobium with tetrads).They have a base chromosomenumber six sections (including sects. Chamaenerion and ofx = 18,with polyploids in at least two species. Zauschneria).Raven proposed thatthe coma-less Sections Boisduvaliaand CurraniacompriseanBoisduvalia,which includes the only species in nual, mostlyautogamous herbsthatlack a coma the tribethathe interpretsas having the diploid and together have been recognized traditionchromosomenumber (n = 9, 10), probably sep- ally at the generic level as Boisduvalia(Munz aratedfromthe Epilobium line priorto evolution 1965; Raven 1976). Hoch and Raven (1992) in of the characteristicseed coma. Preliminary transferringspecies of Boisduvaliato Epilobium phylogenetic analyses of morphological and maintained the two sections (Boisduvaliaand moleculardata (including those reportedin this Currania;Raven and Moore 1965; Raven 1976), paper) suggested flawsin this model, such that consistentwith Raven's (1988) suggestion that Hoch and Raven (1992) placed the sections of the coma could have been lost independently Boisduvaliawithin Epilobium, resultingin a sin- in these two groups. Epilobiumconcinnum (sect. gle broadly defined genus with eight sections. Boisduvalia)and populationsofE. pygmaeum (sect. As currently circumscribed, the genus is Currania)occur in South America,whereas the marked by only a single synapomorphy: dry other species of both sections are restrictedto multicellular papillae on the stigma (Heslop- western North America. Species of sect. BoisHarrison 1990; Hoch et al. 1993). Nonetheless, duvalia have gametic chromosome numbers of all the phylogeneticanalyses of both molecular n = 9 (one species), n = 10 (two species) and n and morphological data thatincluded multiple = 19 (one species). The two species of sect. CurrepresentativesofEpilobieae have supportedits rania have n = 15. monophyly (Martin and Dowd 1986; Crisci et The two species in sect. Zauschneriaare peal. 1990; Sytsma et al. 1991; Hoch et al. 1993; rennials with slightly zygomorphic, tubular, Sytsma,unpubl. data). red-orange, hummingbird-pollinatedflowers, Raven (1976) delimited sections in Epilobium restrictedto western North America. In view based on morphological (including palynolog- oftheirdistinctiveflowers,species in sect.Zausical and anatomical) and cytologicalcharacters. chneriahave been recognized traditionallyas a The largestis sect. Epilobium, with approximate- segregate genus (e.g., Munz 1965). However, ly 150 species (Raven 1988; Hoch, unpubl. data). apart fromthe suite of floralcharactersassociThese are perennial, primarily autogamous ated with hummingbird-pollination,the two herbs, often of mesic habitats,mostly in mon- species clearlyfallwithinthe range ofvariation tane and boreal areas of the northern hemi- in the otherwisebee-pollinated genus Epilobium

This content downloaded from 128.104.46.196 on Mon, 02 Mar 2015 14:42:38 UTC All use subject to JSTOR Terms and Conditions

1994]

365

BAUM ET AL.: EPILOBIUM

(Raven 1976), and specificallyshare with Epilobiumsome specializations of the seeds (Seavey et al. 1977a), pollen (Skvarla et al. 1978) and morphology(Raven 1976, 1988). The base chromosome number is x = 15 with n = 30 tetraploids occurringin E. canum. The last threesectionsare all xerophyticwith restricteddistributionsin southwesternNorth America. Section Cordylophorum includes three perennial species in two subsections,all with n = 15. The one species in subsect. Nuttaliahas cream-colored,slightly zygomorphic flowers, whereas those in subsect.Petrolobium have rosepurple, actinomorphic flowers. Nonetheless, hybridsbetween the subsectionscan be formed (Seavey and Raven 1977c). The sole species in sect. Xerolobium (Epilobium is a highly polymorphic, sumbrachycarpum) mer-floweringannual with n = 12. It exhibits similaritiesto sect. Cordylophorum, but members of the two sections do not successfullyhybridize (Raven 1976). Finally, sect. Crossostigma comprises two diminutive, spring-floweringannuals with reduced, often cleistogamous flowers.Although placed togetherbecause of general similarityin manyvegetativeand reproductivefeatures,the two species have differentchromosome numbers (n = 13 or n = 16) and are not known to hybridizedespite frequentsympatry(Seavey et al. 1977b). Their gross similaritymay be attributed less to close ancestrythan to extremereduction common in many annual plants. Despite the wealth of informationon variation in morphology, anatomy, cytology, and phytochemistry,phylogenetic relationships withinEpilobium have remained obscure. Raven oc(1976) suggested that sect. Cordylophorum cupied a ratherbasal position,linkingwithsect. Zauschneria(both sections n = 15) on the one hand, and with sect. Xerolobium (similar pollen and seeds) on the other.He clearlyviewed sect. Chamaenerion as being specialized and considered it close to sect. Epilobium(both with x = 18), linked via E. rigidum (sect. Epilobium).In the case of sects. Boisduvaliaand Curraniaa basal positionwas inferredbecause of the absence of a coma and the occurrence of the only "diploids" in the tribe(n = 9, 10 in sect. Boisduvalia). Also, the chromosomenumberof n = 15 in sect. Curraniasuggesteda possible relationshipto the putatively "basal" Epilobiumspecies with the

and Zausame number (sects. Cordyolophorum schneria).However, Raven (1988) laterraised the possibilityof coma loss and, hence, a non-basal position forsects. Boisduvaliaand Currania. Raven (1976) and Seavey et al. (1977b) viewed as havthe reduced annuals in sect. Crossostigma ing obscurerelationships.They consideredthem distant fromsects. Epilobiumand Chamaenerion and ruled out an affinitywith the monotypic They implied either that Crossect. Xerolobium. sostigmais basal to all sections except Boisduvalia I and Curraniaor thatit is close to Cordylophorum Zauschneria. In this study,our aims were to: 1) test the monophylyof the sections of Epilobium(except the monotypicXerolobium); 2) infer the phylogenetic relationships between the sections, particularlyto test the hypotheses of Raven (1976, 1988), and 3) reexamine morphological and chromosomal evolution in the context of these molecular results.To this end we unusdertooka phylogeneticanalysis of Epilobium ing sequences of the two internal-transcribed spacers (ITS1 and ITS2) and the 5.8S cistronof rDNA. This region has been found to provide good phylogeneticresolutionat the subgeneric level (e.g., Baldwin 1992, 1993; Wojciechowski et al. 1993; Suh et al. 1993; Kim and Jansen1994) and seemed well suited to this study. ANDMETHODS MATERIAL Terminal Taxa. Table 1 liststhe ingrouptaxa used in this analysis and voucher information. Ribosomal DNA sequences were collected for all representativespecies of the smaller secniveumBrand. in sect. tions,except forEpilobium subsect. Petrolobium.We used Cordylophorum carefully chosen representatives of the two largest sections, Chamaenerionand Epilobium, sampling both subsections of the formerand representatives of each major chromosome group (see Seavey and Raven 1977a, b, 1978) in the latter,plus the putatively basal species E. obcordatum and E. rigidum. Two representativesof Onagreae were used nutas outgroups,Clarkiabottaeand Gayophytum tallii.Clarkiawas selected because it shares a dry, commisuralstigma with Epilobium.Gayophytum was selected because it shares a number of embryological features with Epilobiumincluding thin ovule parietal tissue (thick in other Ona-

This content downloaded from 128.104.46.196 on Mon, 02 Mar 2015 14:42:38 UTC All use subject to JSTOR Terms and Conditions

[Volume 19

SYSTEMATIC BOTANY

366 TABLE

1. The taxonomyof Epilobium and materialused in the study. Number of species

Taxon

EpilobiumL. 1. Sect. Boisduvalia

4

Distribution

N. and S. America

Chromosome number

n= 9

n = 19

n = 10

n = 10

2. Sect. Currania

2

N. and S. America

n = 15

n = 15

3. Sect. Cordylophorum 3a. Subsect. Nuttalia

Species studied and voucher

E. torreyi (S. Watson) Hoch & Raven Cult. MO (M3599); U.S.A., California, Fresno Co., Seaveyin 1974 (MO). E. concinnum (D. Don) Hoch & Raven Cult. MO (M3646); Chile, Region VIII, Munioz2383 (MO). E. pallidum(Eastwood) Hoch & Raven Cult. MO (M3598); U.S.A., California, Butte Co., Oswald 554 (CHSC). E. densiflorum (Lindl.) Hoch & Raven Cult. MO (M3596); U.S.A., California, Butte Co., Oswald 794 (CHSC). E. clesitogamum (Curran) Hoch & Raven Cult. MO (M3603); U.S.A., California, Butte Co., Broyles 1089 (CHSC, MO). E. pygmaeum (Speg.) Hoch & Raven Cult. MO (M3601); U.S.A., California, Butte Co., Broyles 1090 (CHSC, MO).

1

W. United States

n = 15

2

W. United States

n = 15

4. Sect. Xerolobium

1

W. North America & S. South America

n = 12

E. brachycarpum Presl U.S.A, California,Davis, Sytsmas.n. (WIS).

5. Sect. Zauschneria

2

W. North America

n = 15, 30

E. canum(E. Greene) Raven subsp. canum. Cult. Universityof CaliforniaBotanical Garden, Berkeley

3b. Subsect. Petrolobium

E. nevadenseMunz U.S.A., Nevada, Clark Co., Hoch 3440 (MO). E. suffruticosum Nutt. U.S.A., Wyoming,Teton Co., Ravens.n. (MO).

This content downloaded from 128.104.46.196 on Mon, 02 Mar 2015 14:42:38 UTC All use subject to JSTOR Terms and Conditions

1994]

367

BAUM ET AL.: EPILOBIUM TABLE 1. Number of species

Taxon

Continued.

Distribution

Chromosome number

n = 15, 30

n = 15

6. Sect. Crossostigma

2

W. North America

n = 13

n = 16

7. Sect. Epilobium

ca. 150

Cosmopolitan

n = 18

n = 18

n = 18

n = 18

Species studied and voucher

(59.1378); U.S.A., California,Los Angeles Co., Topanga Canyon, Beardand Beard in 1959 (UC). E. canumsubsp. latifolium(Hook.) Raven Cult. Universityof California Botanical Garden, Berkeley(No. 54.0809); U.S.A., California,ButteCo., Cresta Reservoir,See in 1954. (UC). E. septentrionale (Keck) Raven Cult. Universityof CaliforniaBotanical Garden, Berkeley(No. 83.1196); U.S.A., California,Mendocino Co., near Piercy, Raiche30551 (UC). Lindl. ex E. minutum Lehm. Cult. MO (M3607); U.S.A., Oregon, Curry Co., Chambers4847 (MO, OSC). E. foliosum(Nutt. ex Torr. & A. Gray) Suksd. Cult. MO (M3607); U.S.A., California, Butte Co., Schlisingin 1987 (MO). E. rigidumHausskn. U.S.A., California,Del Norte Co., Wiens6797 (MO). A. Gray E. obcordatum U.S.A., Oregon, Harney Co., Seavey 1151 (MO). E. ciliatumRaf. U.S.A., Oregon, Multnomah Co., Seavey 1149 (MO). E. luteumPursh U.S.A., Oregon, Marion Co., Seavey 1452 (MO).

This content downloaded from 128.104.46.196 on Mon, 02 Mar 2015 14:42:38 UTC All use subject to JSTOR Terms and Conditions

368

[Volume 19

SYSTEMATIC BOTANY TABLE 1. Number of species

Taxon

8. Sect. Chamaenerion 8a. Subsect. Rosmarinifolium

8b. Subsect. Leiostylae

Continued.

Distribution

Chromosome number

n = 18

E. siskiyouense (Munz) Hoch and Raven U.S.A. Oregon, Jackson Co., Seavey 1150 (MO). E. dodonaeiVill. Cult. (MO3609); Switzerland, Geneva Botanical Garden (MO). E. angustifolium L. U.S.A., Wisconsin, Sytsma 5500 (WIS). E. latifolium L. Canada, BritishColumbia, Mt. Robson Park, Conti369 (MO).

4

W. Eurasia

n = 18

4

Circumboreal and Himalayas

n = 18, 36, 54 n = 18, 36

Outgroups ClarkiaPursh.

A. Juss. Gayophytum

ca. 30

N. & S. America

n= 9

11

N. & S. America

n= 7

greae) and retarded development of the inner integument (Hoch et al. 1993). A recent morphological cladistic analysis, including these as embryologicalcharacters,placed Gayophytum sistergroup to Epilobieae (Hoch et al. 1993). DNA Extraction. Most of the DNA's were extractedusing the modified Zimmer protocol (Smith et al. 1991). DNA's were extractedfrom E. minutum, E. a few species (Epilobium latifolium, E. cleistogamum) foliosum, using a CTAB method (Saghai-Maroof et al. 1984; Doyle and Doyle 1987; Smith et al. 1991). We modified the extractionformicrocentrifugetubes and used 6% broCTAB (6% Hexadecyltrimethylammonium mide, 50mM Tris-HCl, 2.1M NaCl, lOmM EDTA). DNA was precipitatedwith ethanol and either sodium chloride or ammonium acetate,

Species studied and voucher

C. bottae(Spach) Lewis & Lewis U.S.A., California,Los Angeles Co., Mulholland Drive, Weeden 35-4 (DAV). G. nuttalliiTorr. & A. Gray U.S.A., California,Alpine Co., State Highway 88 near Caples Lake (west of Carson Pass), B. Baldwin(no voucher).

because these cause less polysaccharide to coprecipitatewith the DNA than is the case with isopropanol or with ethanol in the presence of sodium acetate (Bult et al. 1992). Amplification and Sequencing of rDNA. The internaltranscribedspacers,the 5.8S gene, and flankingregions of the 18S and 26S genes were amplifiedfromtotal genomic DNA using the polymerase chain reaction (PCR; Mullis et al. 1986). The reactions were 1OO,uland contained 0.5,1 (2.5 units) AmpliTaqE (Perkin Elmer Cetus), 1OAlof 10 x reactionbuffer(Perkin Elmer Cetus), 2,1I of each dNTP (10mM stock solution), l,uleach ofprimers"ITS5" and "ITS4" (10mM stock solution) and 1OAlof diluted template DNA. Generally,optimalamplificationwas achieved when the template DNA was diluted

This content downloaded from 128.104.46.196 on Mon, 02 Mar 2015 14:42:38 UTC All use subject to JSTOR Terms and Conditions

1994]

BAUM ET AL.: EPILOBIUM

1:10 or 1:100. The primer sequences were derived fromWhite et al. (1990) and synthesized commercially(Oligos Etc. Inc.). The reactions were run through30 cycles of 1.5 min at 94?C, 2 min at 48?C and 3 min at 72?C. SuccessfulPCR reactionsresulted in a single band in a minigel correspondingto approximately700bp. The successfulreactions were cleaned using the GeneClean II9 kit (Bio 101) and then sequenced directlyusing the Sequenase 2.09 kit (United StatesBiochemical). The sequencing reactions utilized a protocol modifiedfordoublestrandedDNA that involved flash-freezingdenatured annealing reactionsin liquid nitrogen and thawing them at the beginning of the extension reaction (Gyllensten 1989; Conti et al. 1993; Rodman et al. 1993). The main differences fromthese published procedures were thatthe terminationreactions were carried out in the presence of 1% NP-40 (Sigma Chemical Co.) and thatthe stop bufferincluded 20mM sodium hydroxide.Boththese measureshelped reduce the occurrenceof compressionsand othersequencing artifacts.The sequencing reactionswere run through 0.4mm thick, denaturing, 6% polyacrylamidegels at 60 watts for2-4 hours. Gels were fixed(10% Acetic Acid, 12% Ethanol) and dried on a vacuum dryer.The dried gels were exposed to X-rayfilm(Kodak XAR) for2-7 days at room temperature. All species reported in the study were sequenced at least once using each of the four primers "ITS1," "ITS2," "ITS3B," and "ITS4" [primersequences as reported in White et al., (1990) except that "ITS3B" differsfrom"ITS3" by one base, having the sequence: 5' GCATCGATGAAGAACGTAGC 3']. Using all fourprimersmeantthatall partsof the reported sequences were read in both directions except fora shortstretchat each end and sometimesa region in the middle of the 5.8S gene. It was found thatregions thatwere difficultto read in one directionbecause of compressions or hard stops were, without exception, clearly interpretablewhen the otherstrandwas sequenced. Although it is possible that sequential nucleotides of the same base were compressed in both directions,there is no reason to think that this could have introduced any phylogenetic bias. Sequence Analysis. Sequences were manually aligned using SeqApp 1.8a (Gilbert 1992) and MacClade ver. 3.0 (Maddison and Maddison 1992). In most cases there was little ambi-

369

guityin alignment.However, when therewere alternatives that seemed equally likely, the alignment chosen was the one that generated the fewestpotentiallyinformativecharacters.A limited explorationof these alternativearrangments found that they gave longer trees when analyzed using PAUP. When more than one alignment generated the same number of informativesites we chose the one giving the shortesttrees (indicating a higher congruence with the remainder of the data set). In some regions alternative alignments gave identical phylogenetic conclusions, so one arrangement was arbitrarilychosen for the remaining analyses. Some previous studies of sequence data have excluded indels from the initial phylogenetic analysis (e.g., Baldwin 1993; Hibbett and Vilgalys 1993; Suh et al. 1993). This approach is conservative when alignments are questionable, but otherwise indels should be included because they may contain phylogenetic information and, indeed, may provide particularly clear indications of relationships (Lloyd and Calder 1991; Revera and Lake 1992; Baldwin 1993). There are currentlytwo main ways that indels may be incorporated into phylogenetic analysis(Wojciechowskiet al. 1993;Johnsonand Soltis 1994). Each gap position can be treated as a separate characterstate as if it were a fifth base (using GAPMODE=NEWSTATE in PAUP). Alternatively,gap positions can be treated as missing in the body of the data matrix(GAPMODE=MISSING) and then each indel scored and entered as a separate character(Swofford 1991, 1993; Appendix 1). The second approach was employed in this study because it permits indels to be weighted without also weighting nucleotide changes in the insertionsand avoids long gaps being weighted excessively. However, scoring indels as separate charactersruns the riskofoverweightingindels ifadjacent gaps are non-independent (e.g., because of erroneous decisions made during alignment). To counteract this possibility, we conducted the phylogenetic analysis with a range of relative weights both above and below 1. An additional concern is thatthe missingvalues scored in the body of the matrixcan lead to artifactualresolution of clades (Maddison 1994). However, this appears not to be a problem for our data, because excluding all sites with gaps resulted

This content downloaded from 128.104.46.196 on Mon, 02 Mar 2015 14:42:38 UTC All use subject to JSTOR Terms and Conditions

370 TABLE

study.

[Volume 19

SYSTEMATIC BOTANY

2. Sequence characteristicsof the internal transcribedspacers and 5.8S cistron sequenced in this

Mean length (bp) Length range (bp) Aligned length (bp) Number of indels Proportionambiguous GC content Transitions(minimum) Transversions(minimum) Transversion/transition Variable sites Informativesites

ITS 1

5.8S

ITS2

Overall

242.42 240-244 259 22 1.48% 58.90% 71 60 0.85 104 56

164.04 164-165 165 1 1.24% 52.93% 8 10 1.25 17 3

212.46 211-216 219 5 0.71% 57.96% 49 33 0.67 69 44

618.92 616-622 643 28 1.15% 56.65% 128 103 0.81 190 103

in the same base-line consensus tree (see below). The GC content was calculated for each region (ITS1, 5.8S, ITS2) in each species (Table 3). These ratioswere based only upon unambiguous sequence, although W (A or T) and S (G or C) were added to their respective categories. Minimal transition:transversion ratioswere calculated by counting up the fewest number of each required to account forthe characterstates at each position.When all fournucleotideswere presentat a site we scored one transversionand two transitions(Table 2). PhylogeneticAnalysis. Parsimonyanalysis of the sequence data was conducted using the computerprogramPAUP, version3.1 (Swofford 1993). The strategyinvolved a detailed "baseline" study under the Fitch criterion (equal weighting of all regions of the sequence, including gaps) followed by exploration of various alternativeweighting regimens. The baseline studystartedby findingmostparsimonious trees using heuristic searches with tree bisection reconnection(TBR) branch swapping. The possibilityof therebeing undiscovered islands of mostparsimonioustrees(Maddison 1991) was

minimized by the use of random addition sequences and steepestdescent. Many of the trees found by PAUP are identical once branches supported by missing or ambiguous data are collapsed (this is true even when the COLLAPSE option is activated). These branches can be identified because their minimum length (displayed in the "Table of Linkages") is zero. In the textwe reportthe numberof treesfound during searches with uninformativecharacters excluded and the COLLAPSE option activated and also the numberof distincttreesremaining afterbranches with a minimal length of zero are manually collapsed. The amount of phylogenetic informationin the base-line analysis was estimated using the consistencyindex (CI; Kluge and Farris 1969), retentionindex (RI; Archie 1989; Farris 1989) and gl statistic(Hillis 1991; Huelsenbeck 1991; Hillis and Huelsenbeck 1992). This last measure was estimatedfrom10,000 random trees using the RANDOM TREES option in PAUP. Branch lengths (available in PAUP) and two other statisticswere examined in order to estimatethe relativeinternalsupportfordifferent elements in the most parsimonious trees. The

TABLE 3. GC content by taxonomic group and region of the sequence. The values given are percent of sequenced bases that are G, C or S (G or C). For each taxonomicgroup, mean is given followed by the range of observed values in brackets:N = number of species in the study. Taxonomic group

Outgroups Sect. Chamaenerion Sect. Epilobium Others

N

ITS1

5.8S

ITS2

OVERALL

2 3 5 14

58.3 (57.9-58.7) 53.8 (53.1-54.9) 57.9 (57.4-58.7) 60.4 (58.8-62.0)

52.8 (52.5-53.0) 54.0 (53.7-54.3) 52.9 (52.0-53.4) 52.8 (51.3-53.7)

53.8 (53.6-54.0) 54.4 (53.6-55.0) 56.6 (56.3-56.9) 58.1 (56.9-59.7)

55.3 (55.3-55.3) 56.0 (53.7-54.4) 56.1 (55.9-56.4) 57.6 (56.9-58.4)

This content downloaded from 128.104.46.196 on Mon, 02 Mar 2015 14:42:38 UTC All use subject to JSTOR Terms and Conditions

1994]

BAUM ET AL.: EPILOBIUM

decay function(Bremer 1988; Donoghue et al. 1992) was calculated by carryingout heuristic TBR searches constrained to search only trees lacking the clade being studied (using the CONVERSE ENFORCE command). These searches had to be conducted using multiple, random addition sequences (ADDSEQ= RANDOM) because other addition sequences tend to findislands of treesclose to the unconstrained searches, thereby overestimatingthe decay index. The bootstrapsupport(Felsenstein 1985) forclades found in the combinable componentconsensusofthe mostparsimonioustrees was calculated from 100 replicates using the BOOTSTRAP command in PAUP. Despite the limitationsof the bootstrap (Sanderson 1989) and decay index, they do give some indication as to the relative internalsupport fordifferent clades in the base-line trees. Having examined the phylogenyunder baseline conditions, we explored the effectof six alternative characterweighting regimens: 1) Coding (i.e., 5.8S) characters= 2, non-coding (i.e., ITS) characters = 1; 2) Indels = 2, substitutions = 1; 3) Indels = 5, substitutions = 1; 4) Indels = 1, substitutions = 2; 5) Indels = 0, substitutions= 1, and 6) a posteriorire-

weighting of charactersbased on the base-line trees.Weightingregimens 1-5 were applied using the WTSETS command in PAUP, whereas a posteriori reweighting(regimen 6) was accomplished using the REWEIGHT CHARACTERS option in PAUP (using the maximal value of the rescaled consistencyindex). In addition,we evaluated the effectof a 1.1:1 or 2:1 character state weighting of transversions over transitions (using the USERTYPE STEPMATRIX command in PAUP). In addition to parsimony analysis, the data were analyzed using a number of other methods of phylogenetic inference. Distance trees were calculated using the neighbor-joiningalgorithm (Saitou and Nei 1987), implemented using the NEIGHBOR programin the PHYLIP 3.4 computer package (Felsenstein 1992). Distance matrices were calculated using the DNADIST program in PHYLIP using: 1) no transition: transversionbias; 2) 1.1:1.0 bias, and 3) a 2:1 bias. These programsignore indels. Maximum likelihood phylogeny estimation was explored utilizing the fastDNAML program (Olsen et al. 1992). We inferredthe maximum likelihood tree assuming transitionsare

371

twiceas likelyas transversionswith eitherequal rates of change at all positions, or with twice the probability of change at positions in the spacers relative to the 5.8S cistron. The maximum-likelihood algorithm employed ignores indels. Compatibilityanalysis was carried out using the DNACOMP program (PHYLIP 3.4; Felsenstein 1992),withgaps and ambiguous sitestreated as missing. The implicationsof the molecular data forthe interpretationof morphological and chromosomal evolution in Epilobiumwas evaluated by reconstructingthe most parsimonious pattern of characterevolution on the molecular trees. For morphologicalcharacterswe assumed unordered character states and determined parsimonious reconstructionsmanually. For chromosome number we used MacClade 3.0 (Maddison and Maddison, 1992) under two models of aneuploid evolution (allopolyploidy could not be modeled). In the firstmodel, the number of aneuploid events was minimized, regardlessof the number of chromosomes lost or gained. In the second model, the total number of chromosomesgained or lost on the tree was minimized (using a step-matrixin which the cost of a change between two states was equal to the differencein chromosome number). In eithercase, the clearly allopolyploid E. concinnum was scored as polymorphicforn = 9 and n = 10. RESULTS

Sequence Analysis. The aligned sequences are provided in Appendix 1. Characteristicsof these sequences are summarized in Table 2. In all species, ITS1 (240-244 bp) is longer than ITS2 (211-216 bp). This is similarto the findings for ViciafabaL. (Yokata et al. 1989), Sinapisalba L. (Rathgeber and Capesius 1989) and for representativesof Asteraceae (Baldwin 1992, 1993; Kim and Jansen1994) and Winteraceae (Suh et al. 1993), but contraststhe situation in other angiosperms(Baldwin 1992, 1993). The 5.8S cistron was 164bp long in all species (except for which has one extrabase), Epilobium angustifolium which is similarto otherangiosperms(Takaiwa et al. 1985; Yokata et al. 1989; Venkateswarlu and Nazar 1991; Baldwin 1992; Suh et al. 1993; Kim and Jansen 1994). The GC content of both ITS regions and the

This content downloaded from 128.104.46.196 on Mon, 02 Mar 2015 14:42:38 UTC All use subject to JSTOR Terms and Conditions

[Volume 19

SYSTEMATIC BOTANY

372

Gayophytum Clarkia

100

+17

100

+14

100 +8

100

|

|92.4

1E

dodonaei Epilobium E.angustifolium Chamaenerion E. latifolium E luteum E. siskiyouense Epilobium E. ciliatum E. obcordatum E rigidum E brachycarpum Xerolobium E foliosum |Crossosti Cossim E. minutum

60.6 +1

50.9 +1

+20

J

+3

E nevadense

99.0 +6

E suffruticosum E pygmaeum

80.5

50.9 +1

Currania

~~E.cleistogamumIClade

+1

91.2 +3

|

Cordylophorum

E. densiflorum E concinnum

8_93

Boisduvalia

E. pallidum

+3

IE.

73j4

"Xerophytic"

=

torreyi E canumcanum E canumlatifolium Zauschneria E septentrionale

FIG. 1. Strictconsensus of the most parsimonious unrooted trees fromthe base-line study. The bootstrap support (%) is shown above branches and decay index is given below.

5.8S cistron fall within the range previously reportedforangiosperms,being somewhat GC rich (Baldwin 1992). However, there are noticeable differencesin GC contentbetween taxonomic groups for the ITS regions, as summarized in Table 3. In particular,Epilobiumsects. and Epilobiumand the outgroups Chamaenerion are less GC rich than the remaining sections of Epilobium. Using the aligned sequences, and treating gaps as missing, the sequence divergence between species pairs was calculated. They ranged from0.0-12.9%,the highestbeing between sect. Chamaenerionand sect. Boisduvalia,the lowest being between species within sect. Epilobium. The outgroups differedfrom the ingroups by 8.5-10.4%. As reported previously (Baldwin 1992, 1993; Kim and Jansen1994), sequence divergencewas approximately50-100% higherin ITS1 than ITS2, and both these regions showed many times greater divergence than the 5.8S cistron. There is a total of 190 variable sites,of which 104 (54.7%) are in ITS1, 17 (8.9%) in the 5.8S

cistron,and 69 (36.3%) in ITS2. The variable sites include 103 with potentially informative distributions(54.2% of the variable sites). Of the informativesites 56 (54.3%) are in ITS 1, 3 (2.9%) in the 5.8S cistronand 44 (42.7%) in ITS2. The alignment required the insertion of 28 gaps ranging from one to four base pairs in length. All but six indels are located in ITS1. Thirteen indels are autapomorphic,having an insertion or deletion unique to one taxon, whereas the other 15 are potentially informative. PhylogeneticAnalysis. The base-line study found 21 most parsimonious trees,whose strict consensus tree is shown in Figures 1, 2 [the combinablecomponentconsensus(Bremer1990) is identical]. However, when branches supported only by missing data were collapsed, only five distincttreeswere recognized. These five trees differonly in the resolution of the + Xerolobium+ Crossostigma+ Cordylophorum (Zauschneria+ Boisduvalia+ Currania)polytomy (Fig. 3). These trees have a length of 193 steps when only informativecharacterswere includ-

This content downloaded from 128.104.46.196 on Mon, 02 Mar 2015 14:42:38 UTC All use subject to JSTOR Terms and Conditions

1994]

373

BAUM ET AL.: EPILOBIUM

t

S

E

C'n <

N

S

d

w

0 .e~~~~~~~ co

2

U

I_~~~~~~

<

t S

S

aci

U

S

W=>~~~~

1

8U?zz1

4~~~~~~~~~~~~~~~ CCb4

C) w

! F

g

t >

+

e

i

!

-15

12 E. brachycarpum

Epilobium Xerolobium

|

Crossostigma

E fminutum

>- 1613

15 E. nevadense rz~

15 E. suffruticosum

I

15 E. cleistogamum

I

15 E. pygmaeum

15 E. canumcanum

a

IChamaenerion

Cordylophorum C Currania Zauschneria

30 E. canum latifolium 15 E. septentrionale 10 E. densiflorum

-

10+9 10E. pallidum

_

Boisduvalia

>~~~9 19 E. concinnum 9 E. torreyi

9+9

I~

9

e

)- 6

_ >81 7

~8

x 8~

b

6+7

6+10

8 X >10

6+6 6+6

x

|=

.15

18 Epilobiumdodonaei 18 E. angustifolium Chamaenerion 18 E. latifolium 18 E. luteum 18 E. siskiyouense 18 E. ciliatum Epilobium 18 E. obcordatum ~18 F. rigidum 12 E. brachycarpum IXerolobium 16 E. minutum

I

13 E.foliosum 15 E. nevadense *Cordylophorum E. suffruticosum 15 E. pygmaeum Currania 15 E. cleistogamum I 15 E. canumcanum 30 E. canumlatifolium Zauschneria 15 E. septentrionale 10 E. densiflorum

I

I

10+9 10 E. pallidum

19 E. concinnum

Boisduvalia

9 F. torreyi

This content downloaded from 128.104.46.196 on Mon, 02 Mar 2015 14:42:38 UTC All use subject to JSTOR Terms and Conditions

378

SYSTEMATIC BOTANY

[Volume 19

These two hypotheses of chromosomal evo- some morphological characters that occur in lution are superimposed on the ITS phylogeny species of the clade but not in sect.Chamaenerion in Figure 4. Assuming aneuploid evolution, n nor in mostof the likelyoutgroupsto the genus = 18 is inferredto be ancestral for Epilobium Epilobium.These putative synapomorphies are: whether one minimizes the number of aneu- pollen released in tetrads(although some poprelease pollen in moploid events or the total number of chromo- ulations ofE. brachycarpum somes lost or gained (see Methods; Fig. 4a). This nads; Raven 1976), petals notched (also occurs is concordantwith the second hypothesispro- in some Clarkiaspecies, see above) and perhaps posed by Raven (1988). However, the data can- flowerswith a distinctfloraltube. Furtherwork and is clearlyneeded to see whetheradditional mornot entirelyrejectn = 18 in sects. Epilobium being derived through indepen- phological charactersare synapomorphiesofthis Chamaenerion dent doubling fromn = 9 ancestors(Fig. 4b) as clade. Section Epilobium. Section Epilobiuminsuggested by Stebbins (1971) and Raven (1976, cludes about 150 species (Raven 1976; Hoch, 1988). Section Chamaenerion. Section Chamaener- unpubl. data), or nearly 90% of the species in ionis sisterto the restof the genus, contradict- the genus. Raven (1976), following Hausing the hypothesis that it is a specialized de- sknecht(1884) and others,delimitedthe section rivativeof sect. Epilobium(Raven 1976). Species by a combination of charactersincluding gain sect.Chamaenerion have several charactersthat metic chromosome number n = 18, leaves opdistinguish them from the rest of the genus posite below, flowersactinomorphic,floraltube (Chen et al. 1992): leaves spirally arranged (vs. present,petals deeply notched, pollen shed in opposite,at least below the inflorescence),flow- tetrads, distal pollen endexine solid, viscin ers slightly zygomorphic (vs. actinomorphic, threads tightlycompound, and seeds comose. except in sect. Zauschneriaand Epilobiumsuffru- None of these charactersare considered unique and thusitis not clearlymonoticosum),pollen in monads (vs. tetrads,except to sect.Epilobium, some plants of E. brachycarpum), floraltube ab- phyletic.Furthersampling would thus be usesent (vs. present),stamens in one subequal set ful to ensure that the five membersof the secof eight (vs. two unequal sets of four), style tion used in this studyare representativeof the initiallydeflexed (vs. straight),and petal notch remainder. One of the most divergent and problematic absent (vs. present). Of these, zygomorphic flowers, style deflexed, alternate leaves, and species in sect. Epilobiumis Epilobium rigidum.Its perhaps lack of a floral tube are apomorphies large flowers with very short floral tube and whereas pollen shed in "hard" (subcoriaceous) leaves (Raven 1976) are of sect. Chamaenerion, monads,stamenssubequal, and petals subentire unusual or unique in the section, much more Furtherbeing like charactersin sect. Chamaenerion. are probablyplesiomorphicforEpilobium, and the outgroups more, its seeds are of similar size and morpresentin sect. Chamaenerion Zausphology to those in sects. Cordylophorum, in Onagreae. into two chneria,and Xerolobium(Seavey et al. 1977a). The separation of sect. Chamaenerion well demarcated subsections based on seed Nonetheless, in other features(especially pol(Ramorphology and style indumentum (Raven len characters)it is typicalof sect.Epilobium 1976) gains support from the molecular phy- ven 1976). Although E. rigidumreadily forms (and logeny. It is noteworthythat at the molecular hybridswith otherspecies of sect.Epilobium are more not with species of any other sections),the hylevel, the subsectionsof Chamaenerion divergent than are sections within the "xero- brids are entirelysterileand the chromosomes phytic" clade (5.9-6.1% vs. 0.4-5.0% sequence fail to form bivalents at meiotic metaphase divergence),implyingeitheran ancient split or (Seavey and Raven 1977a,b). These resultsconaccelerated molecular evolution. trastwith the otherwis-enearly universal forThe Genus Excluding Sect. Chamaenermation of bivalents in hybridswithin sect. Epion. The monophylyof the sisterto sect. Cha- ilobiumand again suggest a somewhat isolated maenerion(i.e., the rest of the genus) is well position forE. rigidum. supported by the ITS sequence data. Although Parsimony analyses of ITS using indels and this particularclade, including the formergen- neighbor-joining analysis support the placeera Boisduvaliaand Zauschneria, has notbeen rec- ment of Epilobium rigidumwithin sect. Epilobium ognized previously as monophyletic,there are as sisterto the restof the section.However, this

This content downloaded from 128.104.46.196 on Mon, 02 Mar 2015 14:42:38 UTC All use subject to JSTOR Terms and Conditions

1994]

BAUM ET AL.: EPILOBIUM

379

topology has relativelylow internalsupport as mative.Furthersampling within the large sect. indicated by both bootstrap and decay index. Epilobium, possibly using more rapidlyevolving occur in other DNA sequences and!or a broadersample oftaxa, Two otherpositions of E. rigidum phylogenetic analyses: sister to the whole ge- may yet resolve them. The factthatso littlecan nus except sect. Chamaenerion(parsimony ex- be deduced about relationships among even cluding indels, compatibility)or as sisterto the widely divergentspecies within sect. Epilobium "xerophytic"clade (maximum-likelihood).Thus, could suggest that the group has undergone based on the ITS data the position of thiscritical rapid "star-burst"radiation,possibly driven or species remains uncertain. However, prelimi- accompanied by hybridization,as proposed by naryresultsof chloroplastDNA restrictionsite Raven and Raven (1976). The neighbor-joiningalgorithmdid suggest analysis favor the topology suggested by maxa possible resolutionof sect. Epilobium s. str.Epimum-likelihood(Sytsma et al., unpubl. data). Two species often considered to be closely ilobiumobcordatum was inferredto be the sister relatedto Epilobium are E. obcordatum and group to the remaining three species and, of rigidum (BB) and E. luteum(CC) are E. siskiyouense (Hoch and Raven 1980). Like E. these, E. siskiyouense both have large flowers,relativelylarge closer to each other than either is to E. ciliatum rigidum, seeds, and a low, clumped habit,and both occur (AA). This topology is compatible with the in ratherxeric habitats in northernCalifornia chromosome data in suggesting that the "BB" (E. obcordatum occurs more widely in western arrangementis ancestraland thatthe "AA" and North America). Although all threespecies are "CC" arrangements were derived indepenamply distinct,morphologyand distributional dentlyfrom"BB." However, the resolutionimdata suggest the possibility that E. siskiyouense plied by the neighbor-joininganalysis could be may have originated following hybridization an artifactcaused by unequal ratesof evolution between the othertwo (Hoch and Raven 1980). and mustbe considered tentativeat best in the The analysis of the ITS data supportsthe inclu- absence of confirmationfromother tree buildsion of both E. obcordatum and E. siskiyouense ing methods. within sect. Epilobiums. str.(i.e., sect. Epilobium The Base of the "Xerophytic" Clade. The minus E. rigidum)except in parsimonyanalyses diverse "xerophytic"clade, which is well supwith gaps excluded, in which case neithersect. ported by the molecular data, may also have Epilobiumnor sect. Epilobiums. str. were sup- some support from pollen morphology. Most ported. species of the "xerophytic"clade have incised Seavey and Raven (1977a, b; 1978; Seavey, compound or smooth viscin threads and distal pers. comm.) have reportedthe existenceof ma- pollen walls with large endexine channels, in jor chromosomal translocationsin sect. Epilobi- contrastto tightlycompound viscin threadsand um,and have delimited three major transloca- distal walls with solid endexine found in sects. and Chamaenerion (Raven 1976; Skvartion groups and several minor ones. The most Epilobium widespread arrangement,found in E. obcorda- la et al. 1976, 1978). most Eurasian, and all tumand E. siskiyouense, The "xerophytic" clade comprises four linthe Australasian species (Raven and Raven 1976), eages: sect.Crossostigma, sect.Cordylophorum, and probably the plesiomorphic characterstate monotypicsect. Xerolobium, and the clade com(Seavey and Raven 1977a, b; 1978), is the "BB" prising sects. Zauschneria,Boisduvalia,and Curarrangement.Separated fromthe "BB" arrange- rania(discussed in the next section). The monoment by single translocationsare the "AA" ar- phyly of sect. Crossostigma is supported, rangement,found mainly in the New World confirmingthe taxonomyproposed by Seavey (exemplifiedin this studyby E. ciliatum;Seavey et al. (1977b). In contrast,the molecular data and Raven 1977a,b) and the "CC" arrangement, neither support nor clearly contradict the Raven (1976) occurringin the circumboreal"Alpinae" group monophylyof sect.Cordylophorum. (as well as the NorthAmericanE. luteum;Seavey recognized this group of threewoody, clumped and Raven 1978). perennials with n = 15 as distinctfrom,but very Parsimonyanalysis of sequence data failed to close to sects. Zauschneria(also x = 15) and Xeresolve the relationships among the transloca- rolobium. These three sections share many seed, tion groups sampled in this study. Although pollen and other characters,but Raven (1976) these species differedby as many as eight mu- could delimitsect.Cordylophorum only by a comtations, none of them were potentially infor- bination of characterswith no obvious syna-

This content downloaded from 128.104.46.196 on Mon, 02 Mar 2015 14:42:38 UTC All use subject to JSTOR Terms and Conditions

380

SYSTEMATIC BOTANY

pomorphy. Thus, neither the molecular nor morphologicalcharacterscan rule out sect. Cordylophorumbeing a paraphyletic assemblage united by a suite of plesiomorphies. The relationshipsamong the four major lineages in the "xerophytic" clade cannot be inferredfromthe ITS sequence data. It mightbe hoped that morphological informationcould help resolve this polytomy, particularly distinctivecharacterssuch as fusion of the pedicel with the flower-subtendingleaf and the presence of an apiculus of oily cells at the leaf apex, both discussed at lengthby Raven (1976). However, since Raven's study it has become clear that the distributionof these charactersis far more complex than was previously thought (Hoch, unpubl. data), and until furthermorphological work is conducted they cannot be used to illuminate phylogenetic affinities. Chromosomenumbersare extremelyvariable in the species of the "xerophytic"clade (Table 1), and Raven (1976) relied substantiallyon this evidence in delimitingthe sections included in the clade. Stebbins (1971) and Raven (1976) suggested that the numbers in the "xerophytic" clade were derived followinghybridizationand polyploidizationof aneuploid derivativesof the ancestralx = 9 progenitor(Fig. 4b). They argued that all such diploid level aneuploids are extinct.The alternativehypothesis,raised by Raven (1988), posits an ancestral number of x = 18 forEpilobiumwith aneuploidy giving rise to chromosomenumbersfromn = 9 to 16 (Fig. 4a). From the base number of x = 15 for the "xerophytic"clade, fouraneuploid descentsand one ascent are needed to minimizeboth the number of chromosomesgained or lost and the number of aneuploid events; 15 to 13, 15 to 12, 15 to 10, 10 to 9, 15 to 16. When minimizingthe number of aneuploid events it is equally parsimonious forx = 16 to be basal in the "xerophytic"clade ifsect.Crossostigma is sisterto the othersections in the clade. In this case, five aneuploid descents are implicated; 16 to 13, 16 to 15, 15 to 12, 15 to 10,and 10 to 9. In eithercase, the South American endemic, E. concinnum,n = 19, is

[Volume 19

molecular phylogenyforthe genus. Looking at the whole genus (discountingpolyploid events in sect.Chamaenerion, Epilobium canumand E. concinnum) the Stebbins/Raven hypothesis requires a minimum of three euploid doublings (9+9, 9+9 and 6+6), three aneuploid losses (9 to 8, 8 to 7 and 7 to 6), one aneuploid gain (9 to 10), four amphidiploid events (7+8 , 7+8, 6 +7 and 6 + 10) and threeextinctionevents (ancestral taxa with n = 6, 7, and 8). In contrast, the x = 18 hypothesisrequires no euploid doubling, amphidiploidy, or extinction of hypotheticalancestors,but a minimum of six aneuploid changes. It is important to determine which of the hypothesesis correctnot only because it is necessaryfora fullerunderstanding of evolution in Epilobium,but also because it would be a step towards generalizing as to the relative probabilityof aneuploidy versus other modes of chromosome evolution in angiosperms. Three lines of evidence could potentiallybe used to discriminatebetween the two hypotheses of chromosomal evolution. The Stebbins/ Raven hypothesisinvokes extensivehybridization within the "xerophytic" clade, and this predictsthatdifferent, unlinked molecular data sets (for example ITS and chloroplast DNA) could give conflictingtopologies for the "xerophytic"clade. The Stebbins/Ravenhypothesis also suggests that the n = 9 and n = 10 species

in sect. Boisduvaliaare diploids, whereas the x = 18 hypothesis considers these same species to be tetraploids.Thus, these hypothesesmight be discriminatedby looking for isozyme duplications to determinewhether the n = 9 and n = 10 are diploid or tetraploid (see Gottlieb 1983). Finally,itwould be usefulto characterize the morphology and size of Epilobiumchromosomes to see whetherthereis evidence of major structuralrearrangementsthatcould explain reductions in chromosomenumber,as envisaged by the n = 18 hypothesis,withoutrequiringthe deletion of large quantities of genetic material. The Zauschneria + Boisduvalia + Currania Clade. The monophyletic group comprising viewed as being a tetraploid based on n = 10 sects.Currania, Boisduvaliaand Zauschneriais perand n = 9 (Raven and Moore 1965; Raven 1988; haps the least expected element of the molecSeavey 1992), and populations of E. canumwith ular phylogeny. Previous taxonomic work has n = 30 are seen as autotetraploids(Raven 1976). viewed sects. Boisduvaliaand Currania(as genus The two competing hypotheses of chromo- Boisduvalia)as being only distantlyrelated to somal evolution differin the type and number sect. Zauschneria(Munz 1965; Raven 1976). The of events of chromosomal evolution that must monophyly of Currania+ Boisduvalia+ Zausbe invoked in order to be compatible with the chneriahas nonethelessbeen confirmedby chlo-

This content downloaded from 128.104.46.196 on Mon, 02 Mar 2015 14:42:38 UTC All use subject to JSTOR Terms and Conditions

1994]

BAUM ET AL.: EPILOBIUM

roplast restrictionsite analysis (Sytsma et al., unpubl. data). The monophyly of all three sections in this clade is supported, albeit weakly, by the molecular data (Figs. 1, 2). Morphological characters strongly support sect. Zauschneria; the species share long, tubular,hummingbird-pollinated flowersfound nowhere else in the genus. The two species of sect. Curraniashare at least two derived characters:seeds in two rows per locule, and capsule walls tough and splitting only in the upper third.In contrast,there are no morphological synapomorphies supportingthe monophyly of sect. Boisduvalia. The ITS data imply that sect. Zauschneriais closer to sect. Boisduvaliathan either is to sect. Currania.This topologyconflictswith threenonmolecularcharactersshared by sects.Boisduvalia and Currania:the lack of the coma, the occurrence of angular-fusiformseeds (Seavey et al. 1977a),and the annual habit.The moleculardata require only one extrastep forsects. Boisduvalia and Curraniato be placed as sister taxa. As a result,a combined analysis of both molecular and morphological data (Barrett et al. 1991) would mostlikely favora Currania+ Boisduvalia clade. Nonetheless, a parallel loss of the coma in sects.Boisduvaliaand Curraniacannot be ruled out because the coma has been lost independently in two species in sect. Epilobium(E. curtisiaeRaven and some populations of E. ciliatum; Raven, 1976, 1988). Likewise, the annual habit and Xero(occurringalso in sects. Crossostigma lobium)is clearlyhomoplastic.Furthermore,the seed morphology of sects. Boisduvaliaand Currania,although distinctive,mightbe correlated to coma loss and, therefore,it too could be homoplastic.Thus, furtherwork is needed to clarify the status of the morphological characters supportinga Currania+ Boisduvaliaclade and to resolve the relationships between these two sections and sect. Zauschneria. Based on thismolecularphylogenyand available morphological, cytological, and biogeographical information,it is possible to summarize the probable evolutionary history of Epilobium.The genus originated in western NorthAmerica,correspondingto the main concentrationin the closely relatedtribeOnagreae, withinwhich Epilobieae is likely nested (Hoch et al. 1993; Baum and Sytsma,unpubl. data). The gave mostrecentcommon ancestorof Epilobium rise to two distinctclades, the largely Eurasian and a clade comprising the sect. Chamaenerion

381

otherseven sections. This latterclade diverged a species-rich into two groups: sect. Epilobium, group ofmoisthabitatsthathas dispersedworld wide, and the "xerophytic"clade concentrated in drier areas of western and southwestern North America. The "xerophytic" clade, alspecies, though comprisingonly 8% ofEpilobium has nonetheless undergone dramatic morphological and cytological evolution resulting in the 14 species being placed in six separate sections. These species range fromoutcrossingperennialswith zygomorphic,hummingbird-pollinated flowers(sect.Zauschneria)to autogamous annuals with actinomorphic, minute flowers (e.g., sect. Boisduvalia). Based on the molecular phylogeny, characterizationof the most recent common ancestor of extantEpilobiumis possible. It was probably a perennial herb with opposite leaves (at least basally), entire petals, pollen with solid distal walls and shed in monads, comose seeds, rosepurple petals, actinomorphic flowers,a chromosome number of n = 18 (or possibly n = 9) and a dry papillate and probably four-lobed stigma. The sequences of the ITS regions and 5.8S cistron of rDNA greatly enhance our knowledge of the phylogeny of Epilobiumand help clarifythe patternof morphological and cytological evolution. However, a number of importantquestions remain unresolved. First,the relationshipsbetween the basal lineages in the "xerophytic"clade remain unclear and should be a priorityfor futurework. These branches hold the key to a clearerunderstandingof morphological and cytologicalevolutionwithinthis diverse clade. Secondly, the position of Epilobiumrigidumneeds furtherstudy. Thirdly,the relationships among sects. Boisduvalia,Zauschneriaand Curranianeed to be confirmedin light of the hypothesized parallel loss of the coma and gain of angular-fusiformseeds implied by the molecular data. Finally, further work is needed to determine relationships in Ongoing cladisthe species-richsect. Epilobium. tic analyses based on morphology (Hoch and Crisci,in mss.) and on chloroplastDNA restriction analysis (Sytsma et al., unpubl. data) have the potential to clarifysome of these problems and will serve as importantindependent tests of these ITS sequence results. ACKNOWLEDGMENTS. This work was supported by the National Science Foundation (BSR-9020055 to K.

This content downloaded from 128.104.46.196 on Mon, 02 Mar 2015 14:42:38 UTC All use subject to JSTOR Terms and Conditions

382

SYSTEMATIC BOTANY

[Volume 19

Sytsma and BSR-8906848 to Peter Raven). We are Dipsacales based on rbcL sequences. Annals of gratefulto the colleagues who sent material(see Table the Missouri Botanical Garden 79: 333-345. 1), JimSmithfordoing many of the DNA extractions, DOYLE, J. J.and J.L. DOYLE. 1987. A rapid DNA Kenneth Karol and Elena Conti for advice on seisolation procedure forsmall quantities of fresh leaf tissue. PhytochemicalBulletin 19: 11-15. quencing, and Charles Delwiche forhelp with analysis. This paper was much improved by comments FARRIS, J. S. 1989. The retentionindex and homofromBruce Baldwin, Peter Raven, and an anonymous plasy excess. SystematicZoology 38: 406-407. reviewer. FELSENSTEIN, J. 1985. Confidence limits of phylogenies: An approach using the bootstrap.Evolution 39: 783-791. LITERATURE CITED . 1992. PHYLIP manualversion3.4. Berkeley, ARCHIE,J. W. 1989. Homoplasy excess ratios: New California: UniversityHerbarium of the Univerindices for measuring levels of homoplasy in sityof California. phylogenetic systematicsand a critique of the GILBERT, D. 1992. SeqApp version 1.8a, a multiple consistencyindex. SystematicZoology 38: 253sequence editor for Macintosh computers. Pub269. lished electronically on the Internet,available B. G. 1992. Phylogeneticutilityof the inBALDWIN, via anonymous ftpfromftp.bio.indiana.edu. ternal transcribedspacers of nuclear ribosomal GOTTLIEB, L. D. 1983. Isozyme number and phylogDNA in plants: An example fromthe Composieny. Pp.209-221 in Proteinsand Nucleic Acids in tae. Molecular Phylogeneticsand Evolution 1: 3Plant Systematics, eds. U. Jensen and D. E. Fair16. brothers.Berlin: Springer-Verlag. 1993. Molecular phylogeneticsof Calycaden- GYLLENSTEN, U. 1989. Direct sequencing of in vitro ia (Compositae) based on ITS sequences of nuamplified DNA. Pp. 45-60 in PCR Technology: clear ribosomal DNA: Chromosomal and morPrinciples and applications forDNA amplification, ed. phological evolution reexamined. American H. A. Erlich. New York: StocktonPress. Journalof Botany 80: 222-238. C. 1884. Monographieder GattungEpiHAUSSKNECHT, BARRETT,M., M. J. DONOGHUE, and E. SOBER. 1991. lobium.Jena:Gustav Fischer. Against consensus. SystematicZoology 40: 486- HESLoP-HARRISON, Y. 1990. Stigma formand surface 493. in relation to self-incompatibility in the OnagraK. 1988. The limitsof amino acid sequence BREMER, ceae. Nordic Journalof Botany 10: 1-19. data in angiosperm phylogeneticreconstruction. HIBBETT, D. S. and R. VILGALYS. 1993. Phylogenetic Evolution 42: 795-803. relationships of Lentinus(Basidiomycotina) in. 1990. Combinable component consensus. ferredfrommolecular and morphological charCladistics 6: 369-372. acters. SystematicBotany 18: 409-433. BULT, C., M. KALLERSJO,and Y. SUH. 1992. AmplifiHILLIS, D. M. 1991. Discriminating between phycation and sequencing of 16S/18S rDNA from logenetic signal and random noise in DNA segel-purifiedtotalplant DNA. Plant Molecular Biquences. Pp. 278-294 in Phylogenetic analysisof ology Reporter10: 272-284. DNA sequences,eds. M. M. Miyamoto and J.CraAND E. A. ZIMMER. 1993. Nuclear ribosomal craft.Oxford: Oxford Univ. Press. RNA sequences forinferringtribalrelationships and J. P. HUELSENBECK. 1992. Signal, noise, within Onagraceae. SystematicBotany 18: 48-63. and reliabilityin molecular phylogenetic analyses. Journalof Heredity 83: 189-195. S. 1975. Wood anatomy of Onagraceae, CARLQUIST, with notes on alternative modes of photosyn- HOCH, P. C., J. V. CRISCI,H. TOBE,and P. E. BERRY. 1993. A cladisticanalysis of the plant familyOnthate movement in dicotyledon woods. Annals of the Missouri Botanical Garden 62: 386-424. agraceae. SystematicBotany 18: 31-47. and P. H. RAVEN. 1980. A new combination CHEN,C.-J.,P. C. HOCH,and P. H. RAVEN. 1992. Sysin Epilobium(Onagraceae). Madrofio 27: 146. tematicsof Epilobium (Onagraceae) in China. Sys. 1992. Boisduvalia,a coma-lessEpand tematicBotany Monographs. 34: 1-209. ilobium(Onagraceae). Phytologia 73: 456-459. CONTI, E., A. FISCHBACH and K. J.SYTSMA. 1993. Tribal relationshipsin Onagraceae: Implicationsfrom HOLUB, J. 1972. Taxonomic and nomenclatural remarkson Chamaenerion. Folia Geobotanica et PhyrbcL sequence data. Annals of the Missouri Bototaxonomica7: 81-90. tanical Garden 80: 672-685. CRISCI,J.V., E. A. ZIMMER,P. C. HOCH,G. B. JOHNSON, HUELSENBECK, J. P. 1991. Tree-length distribution skewness: An indicator of phylogenetic inforC. MUDD, and N. S. PAN. 1990. Phylogenetic mation. SystematicZoology 40: 257-270. implications of ribosomal DNA restrictionsite variationin the plant familyOnagraceae. Annals JOHNSON, L. A. and D. E. SOLTIS. 1994. matK DNA of the Missouri Botanical Garden 77: 523-538. sequences and phylogenetic reconstructionin Saxifragaceae s. str. SystematicBotany 19: 143DONOGHUE,M. J.,R. G. OLMSTEAD, J.F. SMITH,and J. 156. D. PALMER.1992. Phylogenetic relationshipsof

This content downloaded from 128.104.46.196 on Mon, 02 Mar 2015 14:42:38 UTC All use subject to JSTOR Terms and Conditions

1994]

BAUM ET AL.: EPILOBIUM

KIM, K.-J.and R. K. JANSEN. 1994. Congruence of phylogenetic hypotheses among differentdata sets in dwarfdandelions (Krigia):Additional informationfrom internal transcribed spacer sequences of nuclear ribosomal DNA. Plant Systematicsand Evolution (;n press). KLUGE, A. G. and J.S. FARRIS. 1969. Quantitative phyleticsand the evolution of anurans. Systematic Zoology 18: 1-32. KURABAYASHI,M., H. L.wIs, and P. H. RAVEN. 1962. A comparative study of mitosis in Onagraceae. Annals ofthe MissouriBotanicalGarden49: 10031026. LLOYD, D. G. and V. L. CALDER. 1991. Multi-residue gaps, a class of molecular characterswith exceptional reliabilityforphylogeneticanalyses. Journal of EvolutionaryBiology 4: 9-21. MADDISON, D. R. 1991. The discovery and importance of multiple islands of most-parsimonious trees. SystematicZoology 40: 315-328. MADDISON, W. P. 1994. Missing data versus missing charactersin phylogenetic analysis. Systematic Biology 42: 576-581. and D. R. MADDISON. 1992. MacClade version 3. Sunderland, Massachusetts: Sinauer Associates, Inc. MARTIN, P. G. and J.M. DOWD. 1986. Phylogenetic studies using proteinsequences within the order Myrtales.Annals of the Missouri Botanical Garden 73: 441-448. MULLIS, K., F. FALOONA, S. SCHARF, R. SIAKI, G. HORN, and H. ERLICH. 1986. Specificenzymaticamplificationof DNA in vitro:The polymerase chain reaction.Cold Spring Harbor Symposia of QuantitativeBiology 51: 263-273. MUNZ, P. A. 1965. Onagraceae. Pp. 1-278 in North AmericanFlora,Ser. II, Part 5, ed. C. T. Rogerson. New York: New York Botanical Garden. OLSEN, G. J.,R. OVERBEEK,N. LARSEN, T. L. MARSH, M. J. MCCAUGHEY, M. A. MACIUKENAS, W.-M. KUAN, T. J.MACKE, Y. XING, and C. R. WOESE. 1992. The ribosomal database project. Nucleic Acids Research 20 (supplement): 2199-2200. RATHGEBER,J.and I. CAPESIUS. 1989. Nucleotide sequences of the 18S-25S spacer region frommustard DNA. Nucleic Acids Research 17: 7522. RAVEN, P. H. 1976. Generic and sectional delimitation in Onagraceae, tribeEpilobieae.Annals of the Missouri Botanical Garden 63: 326-340. . 1979. A survey of reproductivebiology in Onagraceae. New Zealand Journalof Botany 17: 575-593. 1988. Onagraceae as a model of plant evolution. Pp. 85-107 in Plant evolutionary biology, eds. L. D. Gottlieb and S. K. Jain. New York: Chapman and Hall. and D. M. MOORE. 1965. A revision of Boisduvalia(Onagraceae). Brittonia17: 238-254.

383

and T. E. RAVEN. 1976. The genus Epilobium in Australia:A systematicand evolutionarystudy. New Zealand Department of Scientificand Industrial Research Bulletin 216: 1-321. REVERA,M. C. and J.A. LAKE. 1992. Evidence that eukaryotesand eocyteprokaryotesare immediate relatives.Science 257: 74-76. RODMAN, J.,R. PRICE, K. KAROL, E. CONTI, K. SYTSMA, and J. PALMER. 1993. Nucleotide sequences of the rbcLgene indicate monophylyof mustardoil plants. Annals of the Missouri Botanical Garden 80: 686-699. SAGAI-MAROOF, M. A., K. M. SOLIMAN, R. A. JORGENSEN,and R. W. ALLARD. 1984. Ribosomal DNA spacer length polymorphismin barley: Mendelian inheritance,chromosomallocation and population dynamics. Proceedings of the National Academy of Sciences, U.S.A. 81: 8014-8018. SAITOU, N. and M. NEI. 1987. The neighbor-joining method: A new method for reconstructingevolutionarytrees.Molecular Biology and Evolution 4: 406-425. SANDERSON, M. J. 1989. Confidence limits on phylogenies:The bootstraprevisited.Cladistics5: 113129. SEAVEY, S. R. 1992. Experimentalhybridizationand chromosomehomologies in Boisduvaliasect. Boisduvalia (Onagraceae). SystematicBotany 17: 8490. , R. E. MAGILL, and P. H. RAVEN. 1977a. Evolution ofseed size, shape and surfacearchitecture in the tribeEpilobieae(Onagraceae). Annals of the Missouri Botanical Garden 64: 18-47. and P. H. RAVEN. 1977a. Chromosomal evolution in Epilobiumsect. Epilobium(Onagraceae). Plant Systematicsand Evolution 127: 107-119. . 1977b. Chromosomal evolution and in Epilobium sect. Epilobium (Onagraceae), II. Plant Systematicsand Evolution 128: 195-200. . 1977c. Experimentalhybridsin and Epilobium(including Zauschneria)species with n = 15 (Onagraceae). American Journalof Botany 64: 439-442. . 1978. Chromosomal evolution and (Onagraceae). III. Plant in Epilobium sect.Epilobium Systematicsand Evolution 130: 79-83. , P. WRIGHT, and P. H. RAVEN. 1977b. A comand E. foliosum(Onminutum parison of Epilobium agraceae). Madrofio 24: 6-12. SKVARLA,J.J.,P. H. RAVEN, and J.PRAGLOWSKI. 1975. The evolution of pollen tetradsin Onagraceae. American Journalof Botany 62: 6-35. . 1976. Ultrastructural , and 1 surveyof Onagraceae pollen. Pp. 447-479, in The significance of theexine,eds. I. K. Ferevolutionary guson and J.Muller. London: Linnean Society. , W. F. CHISSOE, and M. SHARP. 1978. I An ultrastructural studyof viscin threadsin Onagraceae pollen. Pollen et Spores 22: 5-143.

This content downloaded from 128.104.46.196 on Mon, 02 Mar 2015 14:42:38 UTC All use subject to JSTOR Terms and Conditions

384

[Volume 19

SYSTEMATIC BOTANY

DNA analysis of tribaland generic relationships within Onagraceae. American Journalof Botany cellular DNA extractionprotocols. Phytochemi78: 222 (Abstract). cal Bulletin 23: 2-9. TAKAIWA, F., K. OONO, and M. SUGIURA. 1985. Nuevolutionin higher cleotide sequence of the 17S-25S spacer region STEBBINS,G. L. 1971. Chromosomal fromrice rDNA. Plant Molecular Biology 4: 355plants.Reading, Massachusetts: Addison-Wesley Pub. Co. 364. SUH, Y., L. B. THIEN, H. E. REEVE, and E. A. ZIMMER. VENKATESWARLU,K. and R. NAZAR. 1991. A con1993. Molecular evolution and phylogeneticimserved core structurein the 18-25S rRNA interplications of internaltranscribedspacer sequencgenic region fromtobacco,Nicotianarustica.Plant es of ribosomal DNA in Winteraceae. American Molecular Biology 17: 189-194. Journalof Botany 80: 1042-1055. WHITE, T. J.,T. BIRNS, S. LEE, and J.TAYLOR. 1990. SWOFFORD,D. 1991. PAUP: Phylogenetic analysisusing Amplificationand direct sequencing of fungal parsimony, version3.0. Champaign: Illinois NaturibosomalRNA genes forphylogenetics.Pp. 315ral HistorySurvey. 322 in PCR protocols:A guide to methodsand ap. 1993. PAUP: Phylogenetic analysisusingparplications,eds. M. Innis, D. Gelfand, J. Sninsky, simony,version3.1. Champaign: Illinois Natural and T. White. San Diego: Academic Press. HistorySurvey. WOJCIECHOWSKI,M. F., M. J.SANDERSON, B. G. BALDSYTSMA, K. J.and L. D. GOTTLIEB. 1986. Chloroplast WIN, and M. J.DONOGHUE. 1993. Monophyly of DNA evolution and phylogenetic relationships aneuploid Astragalus(Fabaceae): Evidence from in Clarkiasect. Peripetasma(Onagraceae). Evolunuclear ribosomal DNA internal transcribed tion 40: 1248-1261. spacer sequences. AmericanJournalofBotany80: 711-722. -and J.F. SMITH. 1992. Molecular systematics of Onagraceae: Examples fromClarkiaand Fuch- YOKATA, Y., T, KAWATA,Y. lIDA, A. KATO, and S. TAsia. Pp. 295-323 in Molecularsystematics ofplants, NIFUJI. 1989. Nucleotide sequences of the 5.8S eds. P. S. Soltis, D. E. Soltis, and J.J.Doyle. New rRNA gene and internal transcribed spacer York: Chapman and Hall. regionsin carrotand broad bean ribosomalDNA. , and P. HOCH. 1991. A chloroplast Journalof Molecular Evolution 29: 294-301. f SMITH, J.F., K. S. SYTSMA, J. S. SHOEMAKER,and R. L. SMITH. 1991. A qualitative comparison of total

APPENDIX 1. Aligned sequences of the internal transcribedspacers and 5.8S cistron.Ambiguitycodes are used: M = {AC}; R = {AG}; W = {AT}; S = {CG}; Y = {CT}; K = {GT}; Q = {C-}; J= {G-}; E = {T-}. Indels are numbered above and scored as additional binaryor multistatecharacters(higher numberscorresponding to longer gaps) at the end of the matrix.All indels were treatedas unordered charactersin the analysis (see text).The sequence is numbered fromthe 5' end of ITS1 throughthe 3' end of ITS2. Positions 1-259 = ITS1; 260-424 = 5.8S; 425-643 = ITS2. Genome Sequence DataBase accession numbers are given at the end of the matrix. Indels: Gayophytum Clarkia E. dodonei E. angustifolium E. 1atifolium E. luteum E. ciliatum E. obcordatum E. rigidum E. siskiyouense E. canumcanum E. canumlatifolium E. septentrionale E. nevadense E. suffruticosum E. brachycarpum E. concinnum E. torrey E. pallidum E. densiflorum E. pygmaeum E. cleistogamum Efoliosum E. minutum

10

20

30

40

50

1

2

60

3

4

70

5

80

GTCGAATCCTGCACAGCAGA ACAACCCGAG AACCGGTTAACAACCACTTGGGAGACT-GGGG--CAT--G TCCCC-TGCG GTCATATCYYGCACAGCAGA ACAACCCGAG AACCGJTTAACAACCACTTTGGAGACT-GGGG--CAC--G TCCTC-TGCG ACAACCTGAGAACCGGTTAATAACCRATTGGGAGAAT-JJJJJGCAT--G CCCCC-TGTG GTCGAATCCTGCACAGCAGA CCCCC-TGTG ACAACCTGAG AACCGGTTAATAACCAACTGGGATAAT-GGGGGGCATATG GTCGAATCCTGCACAGCAGA ACAACCTNNGAACSGGTTAATAACCAACTGGG-TAAT-GGGGGGCAT--GCCCCC-TGTG GNNGAATCCTGCACAGCAGA AACCGGTTAACAACCAGTTGGRAGA---CJ JJGGCAT--C GCCCCTTGCG GTCGAATCCTGCATAGCAGA ACAACCCGAG JJGGCAT--C GCCCCTTGCG AACCGGTTAACAACCAGTTGGNAGACGACJ GTCGAATCCTGNATAGCAGA ACAACCNGAG ACAACCCGAGAACCGGTTAACAACCAGTTGGGAGA---CG GGGGCAT--C GCCCCTTGCG GTCGAATCCTGCATAGCAGA ACAACCCGAG AACCGGTTAACAACCAGTTGGGAGA---CG GGGGCCT--C GCCCCCTGCG GTCGAATCCTGCATAGCAGA ACAACCCGAG AACCGGTTAACAACCAGTTGGGAGA---CG GGGGCAT--QGCCCCTTGCG GTCGAATCCTGCATAGCAGA ACAACCSSAGAACCGGTTAACAACCAGTTTGGAGA---CG GGGGCAC--RGCCCC-TGCG NNNNNNNNNT GCANNGCAGN ACAACCQGAG AACCGGTTAACAACCAGTTTGGAGA---CG GGGGCAC--CGCCCC-TGCG GTCGAATCCTGCAKAGCAGA ACAACCCGAGAACCGGTTAACCACCAGTTTGGAGA---CG GGGGCAC--CGCCCC-TGCG NNNNNNNNNN NCANAGCAGA ACAACCCGAGAACCGGTTAACAACCAGTTTGGAGA---CG GGGGCAC--CGCCCC-TGCG GTCGAATCCTGCATAGCAGA GTCGAATCCTGCATAGCAGA ACAACCCGAGAACCGGTTAACAACCAGTTTGGAGA---CG GGG-CAC--C GCCCC-TGCG ACAACCCGAGAACCGGTTAACAACCAGTTTGGAGA---CG GGGGCAC--CGCCCC-TGCG GTCGAATCCTGCACAGCAGA GTCGAATCCTGCACAGCAGA ACAACCCGAG AACCGGTTAACAACCAGTTTGGAGA---CG GGGGCAC--CGCCCC-CGCG GTCGAATCCTGCACAGCAGA ACAACCCGAG AACCGGTTAACAACCAGTTTGGAGA---CG GGGGCAC--CGCCCC-TGCG AACCGGTTAACAACCAGTTTGGAGA---CG GGGGCAC--CGCCCC-TGCG GYCGAATCCTGCACAGCAGA ACAACCCGAG GTCGAATCCTGCACAGCAGA ACAACCCGAG AACCGGTTAACAACCAGTTTGGAGA---CG GGGGCAM--CGCCCC-TGCG GTCGAATCCTGCATAGCAGA ACAACCCGAG AACCGGTTAACAACCAGTTTGGAGA---CJ JJGGCAC--C GCCCC-TGCG GTCGAATCCTGCATAGCAGA ACAACCCGAG AACCGGTTAACAACCAGTTTGGAGA---CG GGGGCAC--CGCCCC-TGCG GTCGAATCCTGCATAGCAGA ACAACCCGAG AACCGGTTAACAACCAGTTTGGAGA---CG GGGGCAC--CGCCCC-TGCG GTCGAATCCTGCATAGCAGA ACAACCCGAGAACCGGTTAACAACCAGTTTGGAGA--ACGGGGGCAC--CGCCCC----G

This content downloaded from 128.104.46.196 on Mon, 02 Mar 2015 14:42:38 UTC All use subject to JSTOR Terms and Conditions

1994]

BAUM ET AL.: EPILOBIUM APPENDIX

Indels: Gayophytum Clarkia E. dodone: E. angustifolium E. latlfolum E. luteum E. citliatum E. obcordatum E. rgidum E. s:skiyouense E. canumcanum E. canumlatifolium E. septentrionale E. nevadense E. suffruticosum E. brachycarpum E. concinnum E. torreyi E. pallidum E. densiflorum E. pygmaeum E. cleistogamum E. folhosum E. minutum

90

6

78 9 100

Gayophytum Clarkia E. dodonez E. angustifolium E. latifolium E. luteum E. ciliatum E. obcordatum E. rigidum E. srskiyouense E. canumcanum E. canumlatifolium E. septentrionale E. nevadense E. suffruticosum E. brachycarpum E. concinnum E. torrey: E. pallidum E. densiflorum E. pygmaeum E. cleistogamum E foliosum E. minutum

110

Continued. 10

11 12 120

131415 130

16 150

140

160

CTCCCAAACCCCGATTCC-G T-GGGTAGCCCCC-GATAAGGGGGTCCACGGCTT-CGGGTAACAACGAAA -CCCGGCACG CTCCCAAATCCMGTCTCC-GT-GGGCAGCCCCCCCAT-GG GGGGTCCATGGATT-CGGGCAATAACGAAA -CCCGGCACG -CTCCCGAATTCCGCTTG-TC TTGGGTAGCC CATCGGG---TCCAAG TCTT-CGGAAAGTAACGAAA -CCCGGCACA C'TCCCGAGTTCCG-TTGCTC TTGGGTAGCC----CAACGG G---TCCAAG ACTT-CGKGAAGTAACGAAA -CCCGGCACG CTCCCGAGTTCCG-TTGCTC TTGGGTAGCC----CAACGG G---TCCAAG ACTT-CGTGAAGTAACGAAA -CCCGGCACG CTCACAAACCCCGCTTGCTGT-GGGTAGCCC---CACCGG G---TCCACA ACACGCGGGN ATCAACGAAA -CCCGGCACG CTCACAAACCCCGCTTGCTGT-GGGTAGCCC---CACCGG G---TCCACA GCACGCGGGCATCAACGAAA -CCCGGCACG CTCACAAACCCCGCTTGCTGT-GGGCAGCCC---CACCGG G---TCCACA ACACGCGGGCATCAACGAAA -CCCGGCACG CTCACAAACCTCGCTTGTTGT-GGGYAGCCC---CACCGG G---TCCACA AC-TGCGGGCATCAACGAAA-CCCGGCACG CTCACAAACCCCGCTTGYTGT-GGGTAGCCC---CACCGG G---TCCACA ACACGCGGGCATCAACGAAA -CCCGGCACG CTCTCAAACCCCGCGTGCTGT-GGGTAGCCCCC-CTTCGG G---TCCACK -MACGCGGGCATCAACGAAA -CCCGGCACG CTCTCAAACCCCGCNTGCTGT-GGGTAGCCCCC-CTTCGG G---TCCACG -CACGCGGGCATCAACGAAA -CCCGGCACG -CCCGGCACG CTCTCAAACCCCGCGTGCTGT-GGGTAGCCCCC-CTTCGG G---TCCACK -CACGCGGGCATCAACGAAA CYCTCAAACACCGCTTGCTGT-GGGTAGCCCCC-CATCGG G---TCCACA -CCCGCGGGCATCAACGAAA -CCCGGCACG CTCTCAAACCCCGCTTGCTGT-GGGTAGCCCCC-CATCGG G---TCCACA -CCCGCGGGCATCAACGAAA -CCCGGCACG CTCTCAAACCCCGCTTGCTGT-GGGTAGCCCCC-CATCGG G---TCCACA -CCCGCGGGCATCAACGAAA -CCCGGCACG CTCTCAAACACCGCGTGCTGT-GGGTAGCCCCC-CTTCGG GG--TCCACG -CACGCGGGCATCAACGAAA-CCCGGCACG CTCTCAAACCCCGCGCGCCGT-GGGTAGCCCCC-CTTCGG G---TCCACG -CACGCGGGCATCAACGAAA -CCCGGCACG CTCTCAAACACCGCGTGCTGT-GGGTAGCCCCC-CTTCGG G---TCCACG -CACGCGGGCATCAACGAAA-CCCGGCACG -CCCGGCACG CTCTCAAACCCCGCGTGCTGT-GGGTAGCCCCC-CTTCGG G---TCCACG -CACGCGGGCATCAACGAAA CTCTCAAACCCCACGTGGTGT-GGGTAGCCCCA-CATCGGGG--TCCACA -CACGCGGGCATCAACGAAA-CCCGGCACG -CCCGGCACG CTCTCAAACCCCGCGTGCTGT-GGGTAGCCCCC-CATCGG GG--TCCACA -CACGCGGGCATCAACGAAA ACCCGGCACG CTCTCAAACCCCGCTTGCTGT-GGGTAGCCCCC-CATCGG G---TCCACA -CCCGCGGGCATCAACGAAA -CCCGGCACG CTCTCAAACCCCGCTTKCTGT-GGGTAGCCCCC-CATCGG G---TCCACA -CCCGCGGGCATCAACGAAA

APPENDIX Indels:

1.

385

170

180

1.

190

Continued. 17 200

210

220

18

19 2021 230

240

CGCAGTCCG-CGACCCCCGTCCGCGGGGTGTCTC-GGGAT-CA-CGGNCC ATAAGAGAAG GAATGTGCCAAGGAAATCGA CGCTGTCCCCCAGCCCCCGTTCGCGGGGTGTCTC-GGGAT -CA-CGCGCC ATAAGAGAAG GAATGTGCCAAGGAAATCGA ATAAGAGAAG TGCAGTCCTGCTACCCCCGT TCGCGGGGTGTCTC-GGGAT -CAACGCGCC GAACGTGCCAAGGAACTCGA ACAAGAKAAG TGCAGTCCTGYTACCCCCGT TCGCGGGGTGTYTC-CGGAT -CAAYGCGCA GAATGTGCCAAGGAACCTGA ACAAGAGATG TGCAGTCCTGCTACCCCCGT TCGCGGGGTGTCTC-GGGAT -CAACGCGCA GAACGTGCCAAGGAACCTGA ATAAGAGAAG CGCAGTCTCGGCACCCCCGTTCGCGGGGTGTGTC-GTGAT -CAA-GCGCA GAACGTGCCAAGGAACTCGA ATAAGAGAAG CGCAGTCTCGGCACCCCCGTTCGCGGJGTGTGTY-GTGAT -CAA-GCGCA GAACGTGCCAAGGAACTCGA ATAAGAGAAG CGCAGTCTCGTCACCCCCGT TCGCGGGGCGTGTC-GTGAT -CAA-GCGCA GAACGTGCCAAGGAACTCGA ATAAGAGAAG CGCAGTCTCGGAACCCCCGTTCGCGGGGTGTGTC-GTGAT -CAA-GCGCA GAACGTGCCAAGGAACTCGA CGCAGTCTCGGCACCCCCGTTCGCGGGGTGTGTC-GTGAT -CAA-GCGCA ATAAGAGAAG GAACGTGCCAAGGAACTCGA CGCGGTCTCGGCACCCCCGTTCGCGGGGCGTGAC-GTGAT-CCA-GCGCA ATAAGAGAAG GAACGTGCCAAGGAACTCGA ATAAGAGAAG CGCGGTCTCGGCACCCCCGTTCGCGGGGCGTGAC-GTGAT-CCA-GCGCA GAACGTGCCAAGGAACTCGA ATAAGAGAAG CGCGGTCTCGGCACCCCCGTTCGCGGGGCGTGAC-GTGAT-CCA-GCGCA GAACGTGCCAAGGAACTCRA ATAAGAGARG CGCGGTCTCGGCACCCCCGTTCGCGGGGTGTGAC-GTGAT-CTA-GCGCA GAACGTGCCAAGGAACTCGA ATAAGAGAAG CGCGGTCTCGGCACCCCCGTTCGCGGGGTGTGAC-GTGAT-CTA-GCGCA GAACGTGCCAAGGAACTCGA ATAAGAGAAG CGCGGTCTCGGCACCCCCGTTCGCGGGACGTGGC-GTGGC-CAA-GCGCA GAACGTGCCAAGGAACTCGA CGCGGTCTCGGCACCCCCGTTCGCGGGGCGTGAC-GCGATCCAA-GCGCA ATAAGAGAAG GAACGTGCCAAGGAACTCGA ATAAGAGAAG CGCGTTCTCGGCACCCCCGTTCGCGGGGCGTGGC-GTGAT-CCA-GCGCA GAACGTGCCAAGGAACTCGA CGCGGTCTCGGCACCCCCGTTCGCGGGJCGTGAC-GCGAT-CCA-GCGCA ATAAGAGAAG GAACGTGCCAAGGAACTCGA CKCGGTCTCGGCACCCCCGTTCGCGGGGCGTGACTGTGAT-CCA-GCGCA ATAAGAKAAG GAACGTGCCAAGGAACTCGA ATAAGAGAAG CGCGGTCTCGGCACCCCCGTTCGCGGGGCGTGAC-GTGAT-CCA-ACGCA GAACGTGCCAAGGAACTCGA ATAAGAGAAG CGCGGTCTCGGCACCCCCGTTTGCGGGGTGTGAC-GTGAT-CCA-ACGCA GAACGTGCCAAGGAACTCGA CGGTGTCTCGGCACCCCCGTTCGCGGGGCGTGGC-GTGAT-CTA-GCGCA AGGAACTCGA ATAAGAGAAG GAACGTGCCA CGGTGTCCCGGCACCCCCGTTCGCGGGJCGTGGC-GTGAT-CTA-GCGCA ATAAGAGAAG GAACGTGCCAAGGAACTCGA

This content downloaded from 128.104.46.196 on Mon, 02 Mar 2015 14:42:38 UTC All use subject to JSTOR Terms and Conditions

386

1.

APPENDIX

Indels: Gayophytum Clarkia E. dodonei E. angustifolium E. latifolium E. luteum E. cdiatum E. obcordatum E. rigidum E. siskiyouense E. canumcanum E. canumlatifolium E. septentrionale E. nevadense E. suffruticosum E. brachycarpum E. concinnum E. torreyi E. pallidum E. densiforum E. pygmaeum E. cleistogamum E. foliosum E. ninutum

250

22

260

23

Gayophytum Clarkia E. dodone: E. angustifolium E. latifolium E.luteum E. ciliatum E. obcordatum E. rigidum E. siskiyouense E. canumcanum E. canumlatifolium E. septentrionale E. nevadense E. suffruticosum E. brachycarpum E. concinnum E. torrey: E. pallidum E. densiflorum E.pygmaeum E. cleistogamum

E.folhosum

E. minutum

270

Continued. 280

290

300

310

320

ATSTTTTCTA -TCAATATCAAAA-CGACTCTCGGCAACGG ATATCTCGNNNNNNGCATCG ATGAAGAACG TAGCGAAATG TAGCGAAATG ATCTTTTCTA-TCAATATCAAAA-CGACTCTCGGCAACGG ATATCTCGGCTCTCGCATCGATGAAGAACG

ATCTTTTCTA ATCTTTTCTA

-TCAATATCG -TAAATATCG

AAA-CGACTC TCGGCAACGG ATATCTCGGC AAAACGACTC TCGGCAACGG ATATCTCGGC

TCTCGCATCG TCTCGCATCG

ATGAAGAACG TAGCGAAATG ATGAAGAACG TAGCGAAATG

ATCTTTTCTA ATCTTTTCTA

-TCATTATCA -NCAATATCA

TAA-CGACTC TAN-CGACTC

TCGGCAACGG ATATCTCGGC TCGGCAACNG ATATCTCGGC

TCTCGCATCG TCTCGNNTCG

ATGAAGAACG TAGCGAAATG ATGAAGAACG TAGCGAAATG

ATCTTTTCTA ATCTTTTCTA ATCTTTTCTA ATCTTTTCTA ATCTTTTCTA ATCTTTTCTA ATCTTTTCTA ATCTTTTCTA ATCTTTTCTA ATCTTTTCTA ATCTTTTSTA ATCTTTTCTA ATCTTTTCTA

-TCAATATCA -TCAATATCA -TCAATATCA -TCAATATCA -TCAATACCA -TCAATATCA -TCAATATCA -TCAATATCA ATCAATATCA -TCAATATCA -TCAATATCA -TCAATATCA -TCAATATCA

TAA-CGACTC TAA-CGACTC TAA-CGACTC TAA-CGACTC TAA-CGACTC TAA-CGACTC CAA-CGACTC TAA-CGACTC TAA-CGACTC TAA-CGACTC TAA-CGACTN TAA-CGACTC TAA-CGACTY

TCGGCAACGG TCGGCAACGG TCGGCAACGG TCGGCAACGG TCGGCAACGG TCGGCAACGG TCGGCAACGG TCGGCAACGG TCGGCAACGG TCGGCAACGG TCGGCAACGG TCGGCAACGG TCGGCAACGG

TCTCGCATCG TCTCGCATCG TCTCGCATCG TCTCGCATCG TCTCGCATCG TCTCGCATCG TCTCGCATCG TCTCGCATCG TCTCGCATCG TCTCGCATCG TCTCGCATCG TCTCGCATCG TYTCGCATCG

ATGAAGAACG ATGAAGAACG ATGAAGAACG ATGNAGAACG ATGAAGAACG ATGAAGAACG ATGAAGAACG ATGAAGAACG ATGAAGAACG ATGAAGAACG ATGAAGAACG ATGAAGAACG ATGAAGAACG

ATCTTTTCTA -TAAATATCGAAA-CGACTCTCGGCAACGG ATATCTCGGCTCTCGCATCGATGAAGAACG TAGCGAAATG ATCTTTTCTA-TCAATATCATAA-CGACTCTCGGCAACGG ATATCTCGGCTCTCGCATCGATGAAGAACG TAGCGAAATG ATCTTTTCTA-TCAATATCATAA-CGACTCTCGGCAACGG ATATCTCGGCTCTCGCATCGATGAAGAACG TAGCGAAATG ATATCTCGGCTCTCGCATCGATGAAGAACG ATCTTTTCTA-TCAATATCATAA-CGACTCTCGGCAACGG TAGCGAAATG ATCTTTTCTA-TCAATATCATAA-CGACTCTCGGCAACGG ATATCTCGGCTCTCGCATCGATGAAGAACG TAGCGAAATG

APPENDIX Indels:

[Volume 19

SYSTEMATIC BOTANY

330

340

1.

350

ATATCTCGGC ATATCTCGGC ATATCTCGGC ATATCTCGGC ATATCTCGGC ATATCTCGGC ATATCTCGGC ATATCTCGGC ATATCTCGGC ATATCTCGGC ATATCTCGGC ATATCTCGGC ATATCTCGGC

TAGCGAAATG TAGCGAAATG TAGCGAAATG TAGCGAAATG TAGCGAAATG TAGCGAAATG TAGCGAAATG TAGCGTAATG TAGCGAAATG TAGCGAAATG TAGCGAAATG TAGCGAAATG TAGCGAAATG

Continued. 360

370

380

390

400

CGATACTTGGTGTGAATTGCAGAATCCCGTGAACCATCGAGTCTTTGAACGCAAGTTGCGCCCTAAGCCATTTGGCCGAG CGATACTTGGTGTGAATTGCAGAATCCCGTGAACCATCGAGTCTTTGAACGCAAGTTGCGCCCTAAGCCATTTGGCCGAG CGATACTTGGTGTGAATTGCAGAATCCCGTGAACCATCGAGTCTTTGAACGCAAGTTGCGCCCTAAGCCATTCGGCTGAG CGATACTTGGTGTGAATTGCAGAATCCCGTGAACCATCGAGTCTTTGAACGCAAGTTGCGCCCTAAGCCATTTGGCCGAG CGATACTTGGTGTGAATTGCAGAATCCCGTGAACCATCGAGTCTTTGAACGCAAGTTGCGCCCTAAGCSATTTGGCCGAG CGATACTTGGTGTGAATTGCAGAATCCNGTGAACCATCGAGTCTTTGAACGCAAGTTGCGCCCTAAGCCATTTGGCCGAG CGATACTTGGTGTGAATTGCAGAATCCCGTGAACCATCGAGTCTTTGAACGCAAGTTGCGCCCTAAGCCATTTGGCCGAG CSATACTTGGTGNNAATTGCAGAATCNNNN NNNCCATCGAGTCTTTGAACGCAAGTTGSSCCCTAAGCCATTTGGCCGAG CGATACTTGGTGTGAATTGCAGAATCCCGTGAACCATCGAGTCTTTGAACGCAAGTTGCGCCCTAAGCCATYTGGCCGAG CGATACTTGGTGTGAATTGCAGAATCCCGTGAACCATCGAGTCTTTGAACGCAAGTTGCGCCCTAAGCCATTTGGCCGAG CGATACTTGGTGTGAATTGNAGAATCCCGTGAACCATCGAGTCTTTGAACGCAAGTTGCGCCCTAAGCCATTTGGCCGAN CGATACTTGGTGTGAATTGCAGAATCCNGTGAACCATCGAGTCTTTGAACGCAAGTTGCGCCCTAAGCCATTTGGCCGAN CGATACTTGGTGTGAATTGCAGAATCCNGTGAACCATCGAGTCTTTGAACGCAAGTTGCGCCCTAAGCCATTTGGCCGAG CGATACTTGGTGTSAATTGCAGAATYCCGTGAACCATCGAGTCTTTGAACGCAAGTTGCGCCCTAAGCCATTTGGCCGAG CGATACTTGGTGTGAATTGCAGAATCCCGTGANNNNTCGA GTCTTTGAACGCAAGTTGCGCCCTAAGCCATTTGGCCGAG CGATACTTGGTGTGAATTGCAGAATCCNGTGAACCATCGAGTCTTTGAACGCAAGTTGCGCCCTAAGCCATTTGGCCGAG CGATACTTGGTGTGAATTGCAGAATCCCGTGAACCATCGAGTCTTTGAACGCAAGTTGCGCCCTAAGCCATTTGGNCGAG CGATACTTGGTGTGAATTGCAGAATCCCGTGAACCATCGAGTCTTTGAACGCAAGTTGCGCCCTAAGCCATTTGGCCGAG CGATACTTGGTGTGAATTGCAGAATCCCGTGAACCATCGAGTCTTTGAACGCAAGTTGCGCCCTAAGCCATTTGGCCGAS CNATACTTGGTGTGAATTGCAGAATCCCGTGAACCATCGAGTCTTTGAACGCAAGTTGCGCCCTAAGCCATTTGGCCGAG CGATACTTGGTGTGAATTGCAGAATCCCGTGAACCATCGAGTCTTTGAACGCAAGTTGCGCCCTAAGCCATTTGGCCGAG CGATACTTGGTGTGAATTGCAGAATCCCGTGAACCATCGAGTCTTTGAACGCAAGTTGCGCCCTAAGCCATTTGGCCGAG CGATACTTGGTGTGAATTGCAGAATCCCGTGAACCATCGAGTCTTTGAACGCAAGTTGCGCCCTAAGCCATTTGGCCGAG CNATACTTGGTGTGAATTGCAGAATCCCGTGAACCATCGAGTCTTTGAANGCAAGTTGCGCCCTAAGNCATTTGGCNGAG

This content downloaded from 128.104.46.196 on Mon, 02 Mar 2015 14:42:38 UTC All use subject to JSTOR Terms and Conditions

1994]

APPENDIX Indels: Gayophytum Clarkia E. dodonei E. angustifolum E. latifolum E. luteum E. ciliatum E. obcordatum E. rigidum E. siskiyouense E. canumcanum E canumlatifolum E septentrionale E nevadense E. suffruticosum E. brachycarpum E. concinnum E. torreyi E. pallidum E. densiflorum E. pygmaeum E. cleistogamum E. foliosum E. mrnutum

410

420

Gayophytum Clarkia E. dodonei E. angustifocium E. latifolum E luteum E ciiatum E obcordatum E. rigidum E.siskiyouense E. canumcanum E. canumlatifocium E. septentrionale E. nevadense E. suffruticosum E. brachycarpum E. concinnum E. torreyi E. palldum E. densflorum E. pygmaeum E. cleistogamum E. foliosum E minutum

1.

430

Continued. 440

450

460

470

480

GGCACGTCTGCCTGGGCGTCAATCATCTATTCGTCACCCAACCTCTTCTC CCCTAAAAGGAGCTCAGGTCCT-GGGTACG GGCACGTCTGCCTGGGCGTCAATCATCTATTCGTCACCCAACCTCTTCTC CCCCAAAAGGAGCTCGGGTCCT-GGGTACG GGCACGCCTGCCTGGGCGTCAATCATCTATTCGTCACCCAACCTCTGCTC CCCATTAAGGAGCTCGGGTCTT-GGGTACG GGCACGCCTGCCTGGGCGTCAATCATCTATTCGTCACCYARCCTCTGCTC CCCATAAAGGAGCTCGGGTCCT-GGTTACG GGCASGCCTGCCTGGGCGTCAATCATCTATTCGTCACCCAACCTCTGCTC CCCATATAGGAGCTCGGGTCCT-GGTTACG GGCACGTCTGCCTGGGCGTCAATCATCTATTCGTCACCCAACCTCCGCTC CCCGAAAAGG AGCTTTGGTCCC-GGGTACG GGCACGTCTGCCTGGGCGTCAATCATCTATTCGTCACCCAACCTCCGCTC CCCGAAAAGG AGCTTTGGTCCC-GGGTACG GGCACGTCTGCCTGGGCGTCAATCATCTATTCGTCACCCAACCTCCGCTC CCCGAAAAGG AGCTTTGGTCCC-GGGTACG GGCACGTCTGCCTGGGCGTCAATCATCTATTCGTCACCCAACCTCCGCTC CCCGAAAAGG AGCTTTGGTCCC-GGGTACG GGCACGTCTGCCTGGGCGTCAATCATCTATTCGTCACCCAACCTCCGCTC CCCGAAAAGG AGCTTTGGTCCC-GGGTACG GGCACGTCTGCCTGGGCGTCAATCATCTATTCGTCACCCAACCTACGCTCCCCGAAAGGGTGCTGTGGTCYCCGGGTACG GGCACGTCTGCCTGGGCGTCAATCATCTATTCGTCACCCAACCTACGCTCCCCGAAAGGG TGCTGTGGTCCCCGGGTACG GGCACGTCTGCCTGGGCGTCAATCATCTATTCGTCACCCAACCTACGCTCCCCGAAAGGG TGCTGTGGTCCCCGGGTACG GGCACGTCTGCCTGGGCGTCAATCATCTATTCGTCASCCAACQKCCGCTCCCCGARAGGGTGCTTTGGTCCC-GGGTACG GGCACGTCTGCCTGGGCGTCAATCACCTATTCGTCACCCAACCTCCGCTCCCCAAAAGGGTGCTTTGGTCCC-GGGTACG GGCACGTCTGCCTGGGCGTCAATCATCTATTCGTCACCCAACCTCCGCTC CCCGAAAGGG WGCTTTGGTCAC-GGGTACG GGCACGTCTGCCTGGGCGTCAATCATCCATTCGTCACCCAACCTCCGCCCCCCGAAAGGGTGCTGTGGTCCCCGGGTACG GGCACGTCTGCCTGGGCGTCAATCATCTATTCGTCACCCAACSTCCGCTC CCCGAAAGGGTGCTGTGGTCCCCGGGTACG GGCACGTCTGCCTGGGCGTCAATCATCTATTCGTCACCCAACCTCCGCCCCMCGAAAGGG TGCTGTGGTCCCCGGGTACG GGCACGTCTGCCTGGGCGTCAATCATCTATTCGTCACCCAACCTTCGCTC CCCGAAAGGGTGCTGTGGTCCCCGGGTACG GGCACGTCTGCCTGGGCGTCAATCATCTATTCGTCACCCAACCTCCGCTC (CCGAAAGGGTGCTGTGGTCCCCGGGTACG GGCACGTCTGCCTGGGCGTCAATCATCTATTCGTCACCCAACCTCCGCTC CCCAAAAGGGTGCTGTGGTCCCCGGGTACG GGCACGTCTGCCTGGGCGTCAATCATCTATTCGTCACCCAACCTCCGCTC CCCGAAAGGG TGCTTTGGTCCC-GGGTACG GGCACGTCTGCCTGGGCGTCAATCATCTATTCGCCACCCAACCTCCGCTC CCCGAAAGGG TGCTTTGGTCCC-GGGTACG

APPENDIX Indels:

387

BAUM ET AL.: EPILOBIUM

490

500

25

1.

26 510

Continued. .520

530

540

550

560

GAAGTTGGCCTCCCGTGCTAT-TGATG-CG CGGCTGGTCTAAAATCGAGCATCGGACTGATGATCTCCGAGGCACGCGGT ATCGGACTGATGATCTCCGAGGCACGCGGT GAAGTTGACCTCCCGTGCTCT-TGAAG-CG CGGCTGGTCTAAAATCGAGC GAAGTTGGCCTCCCGTGGTCT-TGAAG-TG CGGCTGGCCTAAAATCGAGCATCGGATTGATGATCTCCGAGGCACGCGGT GAAGTTGGCCTCCCGTGGTCTACGAAG-YGCGGCTGGCCTAAAAYYGAGC ATCGGGTTGGTGATCTCCGAGGCACGCGGT GAAGTTGGCCTCCCGTGGTCT-CGAAG-CG CGGCTGGCCTAAAATTGAGCATCGGGTTGGTGATCTCCGAGGCACGCGGT GAAGTTGGCCTCCCGTGCTCY-TGAAG-CG CGGCTGGMCT AAAACTGAGCATCGGACTGATGATCTCCGAGGCACGCGGT GAAGTTGGCCTCCCGTGCTCT-TGAAG-CG CGGCTGGCCTAAAACTGAGCATCGGACTGATGATCTCCGAGGCACGCGGT GATGTTGGCCTCCCGTGCTCT-TGAAG-CG CGGCTGGCCTAAAACTGAGCATCGGACTGATGATCTCCGAGGCACGCGGT GAAGTTGGCCTCCCGTGCTCT-TGAAGGCGCGGCTGGCCTAAAACTGAGCATCGGACTGATGATCTCCGAGGCACGCGGT GAAGTTGGCCTCCCGTGCTCT-TGAAG-CG CGGCTGGCCTAAAACTGAGCATCGGACTGATGATCTCCGAGGCACGCGGT GAAGTTGGCCTCCCGTGCTCT-CGAAG-CG CGGYTGGCCTAAAACTGAGCRTCGGACTGATGATCTCCGAGGCACGCGGT GAAGTTGGCCTCCCGTGCTCT-CGAAG-CG CGGTTGGCCTAAAACTGAGC GTCGGACTGATGATCTCCGAGGCACGCGGT GAAGTTGGCCTCCCKTGCTCT-CGAAG-CG CGGCTGGCCTAAAACTGAGC GTCGGACTGATGATCTCCGAGGCACGCGGT GAAGTTGGCCTCCCGTGCTCT-CGAAG-CG CGGCTGGCCTAAAACTGAGC ATCGGACTGATGATCTCCGAGGCACGCGGT GAAGTTGGCCTCCCGTGCTCT-CGAAG-CG CGGCTGGCCTAAATCTGAGCATCGGACTGATGATCTCCGAGGCACGCGGT GAAGCTGGCCTCCCGTGCTCT-CGAAG-CG CGGCTGGCCTAAAACTGAGCATCGGACTGATGATCTCCGAGGCACGCGGT GAAGTTGGCCTCCCGTGCTCT-CGAAG-CG CGGCTGGCCTAAAACTGAGCATCGGACTGATGATCTCCGAGGCACGCGGT GAAGTTGGCCTCCCGTGCTCT-CGAAG-CG CGGCTGGCCTAAAACTGAGCATCGGACTGATGATCTCCGAGGCACGCGGT GAAGTTGGCCTCCCGTGCTCT-CGAAG-CG CGGCTGGCCTAAAACTGAGC ATCGGACTGATGATCTCCGAGGCACGCGGT TCCCGTGCTCT-CGAAG-CG CGGCTGGCCTAAAACTGAGC ATCGGACTGATGATCTCCGAGGCACGCGGT GAWGTTGGCC GAAGTTGGCCTCCCGTGCTCT-CGAAG-CG CGGCTGGCCTAAAACTGAGCATCGGACTGATGATCTCCGAGGCACGCGGT GAAGTTGGCCTCCCGTGCTCT-CAAAG-CG CGGCTGGCCTAAATCTGAGCATCGGACTGATGATCTCCGAGGCACGCGGT GAAGTTGGCCTCCCGTGCTCT-CGAAG-CG CGGCTGGCCTAAAACTGAGCATCGGACTGACGATCTCCGAGGCACGCGGT GAAGTTGGCCTCCCGTGCTCC-CGAGG-CG CGGCTGGCCTAAAACTGAGC ATCGGACTGACGATCTCCGAGGCACGCGGT

This content downloaded from 128.104.46.196 on Mon, 02 Mar 2015 14:42:38 UTC All use subject to JSTOR Terms and Conditions

388

APPENDIX Indels: Gayophytum Clarkia E. dodonei E. angustifolum E. latifolum E. luteum E.cnhatum E. obcordatum E.rigidum E. siskiyouense E. canumcanum E. canumlatifolium E. septentrionale E. nevadense E. suffruticosum E. brachycarpum E. concinnum E. torreyi E. pallidum E. densiflorum E. pygmaeum E. cleistogamum E. foliosum E. minutum

570

27

580

Gayophytum Clarkia E. dodonei E. angustifolium E.latifolium E.luteum E. ciliatum E. obcordatum E. rigidum E. siskiyouense E. canumcanum E. canumlatifolium E. septentrionale E. nevadense E. suffruticosum E. brachycarpum E. concinnum E. torreyi E. pallidum E. densiflorum E. pygmaeum E. cleistogamum E. foliosum E. minutum

1.

Continued.

590

600

610

620

630

28

640

GGTTGTTCATTC-TTACCTC GTGATGTTGCCCCGGAGCATCTCCCATACGAAGCTCCACGACCCTAGATTCA----TAAC GGTTGTTAATTC-TTACCTC GTGATGTTGCCACGGAGCCTCTTCCATACGGAGCTCCACGACCCTAGATTCA----TAAC GGTTGTTAATTC-TTACCTC GTGATGTTGCCCCGGAGCCACTTCCATGTGGAGCTCCACGACCCTAGATATA----TATC GGTTGTTCATTCATTACCTC GTGATGTTGCCCCGGRGCATCTTCCACAAGAAGCTCCACGACCCTAGATACA----TATC GGTTGTTCATTC-TTACCTC GTGATGTTGCCCCGGAGCATCTTCCACAAGAAGCTCCATGACCCTAGATACA----TATC GGTTGTTCATTC-CTACCTC GTGACGTTGCCAAGGAGCGT CTCCCGTACGAAGCTCCACGACCCTAGATTTA---CTATC GGTTGTTCATTC-CTACCTC GTGACGTTGCCAAGGAGCGT CTCCCGTACGAAGCTCCACGACCCTAGATTTA---CTATC GGTTGTTCATTC-CTACCTC GTGACGTTGCCAAGGAGCGT CTCCCGTACGAAGCTCCACGACCCTAGATTTA---CTATC GGTTGTTCATTC-TTACCTC GTGACGTTGCCAAGGAGCGT CTCCCGTACGAAGCTCCACGACCCTAGATTTT---CTATC GGTTGTTCATTC-CTACCTC GTGACGTTGCCAAGGAGCGY CTCCCGTACGAAGCTCCACGACCCTAGATTTA---CTATC GGTTGTTCATTC-TTACCTC GTGACGTTGCCAAGGAGCGT CTCCCGTGCGAAGCTCCACGACCCTAGATTAT---CTAEC GGTTGTTCATTC-TTACCTC GTGACGTTGCCAAGGAGCGT CTCCCGTGCGAAGCTCCACGACCCTAGATTAT---CTAAC GGTTGTTCATTC-TTACCTC GTGACGTTGCCAAGGAGYGT CTCCCGTGCGAAGCTCCACGACCCTAGATTAT---CTATC GGTTGTTCATTCETTACCYC GTGACGTTGCCAAGGAGCGT CTCCCGTGCGAAGCTCCACGRCCCTAGATTAT---CTATC GGTTGTTCATTC-TTACCTC GTGACGTTGCCAAGGAGCGT CTCCCGTGCGAAGCTCCACGACCCTAGATTAN---CTATC GGTTGTTCATTC-ATACCTC GTGATGTTGCCAAGGAGCGT CTCCCGTGCGAAGCTCCACGACCCTAGATTAT---CTATC GGTTGTTCATTC-TTACCTC GTGACGTTGCCAAGGAGCGT CTCCCGTGCGAAGCTCCACGACCCTAGATTAT---CTATC GGTTGTTCATTC-TTACCTC GTGACGTTGCCAAGGAGCGT CTCCCGTGCGAAGCTCCACGACCCTAGATTATTATCTATC GGTTGTTCATTC-TTACCTC GTGACGTTGCCAAGGAGCGT CTCCCGTGCGAAGCTCCACGACCCTAGATTAT---CTGTC GGTTGTTCATTC-TTACCTC GTGACGTTGCCAAGGAGCGT CTCCCGTGCGAAGCTCCACGACCCTAGATTAT---NNNNN GGTTGTTCATTC-TTACCTC GTGACGTTGCCAAGGAGCGT CTCCCGTGCGAAGCTCCACGACCCTAGATTAT---CTATC GGTTGTTCATTC-TTACCTC GTGACGTTGCCAAGGAGCGT CTCCCGTGCGAAGCTCCACGACCCTAGATTAT---CTATC GGTTGTTCATTC-TTACCTC GTGACGTTGCCAAGGAGCGT CTCCTGTGCGAAGCTCCACGACCCTAGATTAT---CTATC GGTTGTTCATTC-TTACCTC GTGACGTTGCCAAGGAGCGT CTCCCGTGCGAAGCTCCACGACCCTAGATNAT---CTATC

APPENDIX Indels:

[Volume 19

SYSTEMATIC BOTANY

6643 GAT GAT GAT GAT GAT GAT GAT GAT GAT GAT GAT GAT GAT GAT GAT GAT GAT GAT GAT NNN GAT GAT GAT GAT

10

Ildels 0110000111 0110000110 0100001003 0101010003 1100010003 0300100012 0000100012 0300100012 0300100012 0300100012 0300000011 0300000011 0300000011 0300000011 0300000011 0300000011 0300000011 0300000011 0300000011 0300000011 0300000011 0300000011 0300000011 0200200011

Continued.

1.

20 0000111111 1000110111 0200110110 0200110110 0200110110 0200010110 0200010110 0200010110 0201010110 0200010110 0210010110 0210010110 0210010110 0210010110 0210010110 0210010110 0110010100 0210010110 0210010110 0210010010 0110010110 0110010110 0210000110 0210010110

28 11111112 11111112 11111112 11011102 11110112 01111111 01111111 01111111 01111011 01111111 01101111 01101111 01101111 011111?1 01111111 01111111 01101111 01101110 01101111 00101111 0110111? 01101111 01111111 01111111

GOD Accession

Number

L28022 L28016 L28020 L28011 L28023 L28024 L28015 L28027 L28030 L28032 L28013 L28014 L28031 L28026 L28033 L28012 L28018 L28034 L28028 L28019 L28029 L28017 L28021 L28025

This content downloaded from 128.104.46.196 on Mon, 02 Mar 2015 14:42:38 UTC All use subject to JSTOR Terms and Conditions