Polyphyly of the Genus Dentaria (Brassicaceae): Evidence from trnL Intron and ndhF Sequence Data

Polyphyly of the Genus Dentaria (Brassicaceae): Evidence from trnL Intron and ndhF Sequence Data Patrick W. Sweeney; Robert A. Price Systematic Botany, Vol. 25, No. 3. (Jul. - Sep., 2000), pp. 468-478. Stable URL: http://links.jstor.org/sici?sici=0363-6445%28200007%2F09%2925%3A3%3C468%3APOTGD%28%3E2.0.CO%3B2-%23 Systematic Botany is currently published by American Society of Plant Taxonomists.

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/about/terms.html. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/journals/aspt.html. Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission.

The JSTOR Archive is a trusted digital repository providing for long-term preservation and access to leading academic journals and scholarly literature from around the world. The Archive is supported by libraries, scholarly societies, publishers, and foundations. It is an initiative of JSTOR, a not-for-profit organization with a mission to help the scholarly community take advantage of advances in technology. For more information regarding JSTOR, please contact [email protected].

http://www.jstor.org Wed Oct 17 10:35:17 2007

Systemntic Botnrly (2000), 25(3): pp. 468478 C Copyright 2000 by the Amencan Society of Plant Taxonom~sts

Polyphyly of the Genus Dentaria (Brassicaceae):Evidence from trnL

Intron and ndhF Sequence Data

PATRICK W. SWEENEY~ and ROBERT A. PRICE Department of Botany, University of Georgia, Athens, Georgia 30602

Present address: Department of Biology, University of Missouri, St. Louis, Missouri 63121

Communicating Editor: James R. Manlzart ABSTRACT. Members of the traditional genus Dentarin (Brassicaceae) comprise about twenty species disjunctly distributed in three biogeographical areas: eastern North America, western North America, and Eurasia. The group has variously been treated as a distinct genus, as a subgenus of Cnrdnaline, or as several sections within Cnrdnnzine. Using chloroplast DNA sequence data from the tr11L intron and izdhF gene we examine the evolutionary relationships of traditional Dentaria species to Cardatnine sensu stricto and the validity of Schulz's sectional treatment of Cflrdatnit~e(including Deiztnria). Sequence data were obtained from nine Dei~tariaspecies, representing all three disjunct biogeographical groups and two of Schulz's sections, and six other Cardnmine species representing a range of morphological variation within the genus and three of Schulz's sections. Results from maximum parsimony analysis show multiple origins for De~ztariawithin Cnrdamir~eand do not correspond well to Schulz's sectional classification. Cardnilline including Dentarin appears to form a monophyletic group with ;vhsturtiunl as its sister-group. We also find strong support for monophyly of the eastern North American species group previously placed in the genus Dentarin. Further, we find that C . clematitis and C . fingelliJern, considered by some to be conspecific, are distinct and not closely related.

The generic and sectional relationships of the toothworts (Dentaria L.), a prominent group of woodland herbs in the Brassicaceae, have long been controversial. Some have followed ~ i n n a e u iand treat Dentaria as a distinct genus (e.g.,Detling 1936, 1937; Bush 1939; Fernald 1950; Spooner 1984; Wofford 1989; Turrill et al. 1994); however, many have opted to recognize Dentaria as part of the genus Cardamine L. (e.g., Schulz 1903, 1936; Oi 1965; Radford et al. 1968; Al-Shehbaz 1988b; Gleason and Cronquist 1991; Jones and Akeroyd 1993; Rollins 1993a, 1993b).Linnaeus first described Cardamine in Species Plantarum (1753) and distinguished the group from Dentaria and other members of the mustard family by the presence of elastically dehiscent fruit valves in Cardamine. Soon afterwards, Crantz (1769) noted that this unusual type of fruit dehiscence was also present in Dentaria, and united the two genera under the name Cardamine. When recognized as a distinct genus, Dentaria is purportedly separated from ~ardamineby a limited number of morphological characters including the possession of an enlarged and fleshy rhizome, ternately or palmately compound leaves, few stem leaves, a specialized rhizomal leaf, and large flowers (Detling 1936; Harriman 1965; Al-Shehbaz 1988~1,1988b).However, when both Cardamine sensu stricto and Dentaria are examined on a worldwide basis, many of these characters do not com-

pletely distinguish the two genera (Table 1). For example, leaf shape is highly variable within Cardamine sensu stricto (Fig. 1) and species are found with simple leaves (e.g., C. constancei Detling and C , bellidifolia L.), pinnately divided leaves (e.g., C. kirsuta L. and many other species within the genus), palmately divided leaves (e.g., C, rupicola (0. Schulz) C. Hitchc.), and ternately divided leaves (e.g., C. angulata Hook. and C. flagellifera 0.Schulz). This same argument can be made for all of the other characters used to separate Dentaria from Cardamine sensu stricto (Table l), and it is on this basis that many chose to treat the two as a single genus (Al-Shehbaz 1988a, 1988b). About twenty species consistently have been recognized as belonging to Dentaria or to equivalent groups within Cardamine. These taxa are distributed disjunctly in eastern North America, western North America, and Eurasia. Dentaria species are rhizomatous uerennials of deciduous forest habitats and frequently are prominent members of their respective spring floras. Some Dentaria species are cytologically and reproductively very interesting. For example, within the eastern North American Dentaria, high levels of polyploidy and apparently frequent secondary aneuploidy have been reported, o Cardamine concatenata (Michaux) and h ~ members, Schwarz and C. diphylla (Michaux) Alph. Wood, have the highest reported chromosome number in

20001

469

SWEENEY 8~ PRICE: POLYPHYLY OF DENTARIA

TABLE1. Distribution of character states for characters typically used to separate Cnrdnrnine sensu stricto from Dentnrin. The sister genus Nnsturtium is included to suggest apparent apomorphic and plesiomorphic states within Carda1i1il~e.Measurements are taken from Detling (1936), Komarov and Bush (1939), Oi (1965), Rollins (1993a), Jones and Akeroyd (1993), and Schulz (1903, 1936). Nflstiirti~irn

Cnrdfl~~iirie

Eastern N. A. De~itnrln

Leaf type (see text for specific examples)

Pinnately compound

Pinnately com-

pound or ter-

nately com-

pound

Many

Ternately compound or appearing palmately compound

Few

Simple, pinnately compound, or ternately compound

Number of cauline leaves (few = 25; many = >5) Rhizome leaves (i.e. arising from rhizome) Large, fleshy rhizome (1-10 mm thick in Deiztnrin) Toothed or scaly rhizome Large flowers (petals 6-25 mm long)

Simple, pinnately compound, palmately compound, or ternately compound Few to many

Few

Few

In some species e.g., C , rupicoln

Yes

Yes

Yes

No

In some species e.g., C. bzilbosn & C. rnacrophylla

Yes

Yes

Yes

No

In some species e.g., C . trfolin L. In some species e.g., C. nngulnta

Yes

No

Yes

Yes

Yes

Yes

Yes

the Brassicaceae, 2n = 256, which is 32-ploid in the genus (Harriman 1965; Al-Shehbaz 1988b). All of the European dentarias studied to date are also polyploid, with numbers ranging from 2n = 48 (hexaploid within the genus) to 2n = 96 (Jones and ~ e n t a i i a scharacteristically repro~ k e r o ~1993). d duce vegetatively by rhizome, and some species (e.g., C. diphjlla and the European C. bulbifera (L.) Crantz) can also reproduce by production of plantlets or bulbils in the axils of leaves (Harriman 1965; Al-Shehbaz 198813; P. Sweeney pers. obs.). The relative frequency of asexual and sexual reproduction in the group is a matter of dispute, and a number of cases of suspected interspecific hybridization also have been reported from eastern North America (Montgomery 1955; Harriman 1965; Al-Shehbaz 198813) and from Europe (Jones and Akeroyd 1993). In the last worldwide treatments of Cardamine, Schulz (1903,1936) included Dentaria as part of Cardamine. He placed Dentaria species into four different sections, two of these containing exclusively Dentaria species. He circumscribed his sections morphologically using rhizome characters, funiculus characters, number of ovules per silique, and cotyledon arrangement. Al-Shehbaz (1988b) argued that many of these morphological characters fail to

Western N. 4 De~itariil

Eurasian

De~itilriil

fully distinguish the sections and that many of Schulz's sections may not be natural. This was further supported by Rashid and Ohba (1993) who, in a morphological examination of C , loxostemonoides 0. Schulz, concluded that section Macrocarpus 0 . Schulz was not monophyletic. A recent molecular phylogenetic study by Franzke et al. (1998), focusing primarily on relationships among Old World groups of Cardamine sensu stricto, has compared approximately 20 taxa in the genus from Eurasia, Australia and New Guinea, using sequence comparisons of the nuclear internal transcribed spacer regions (ITS) and the chloroplast trnT-L (transfer RNA threonine and leucine) intergenic spacer gene region and the nearby trnL intron. The relationships among groups of species were only partially resolved in the 50% bootstrap consensus trees obtained by Franzke et al., yielding a large polychotomy among clades of Eurasian Cardamine and the genus Nasturtium R. Br., but they provide strong indications that several of Schulz's sectional groups are unlikely to be monophyletic. Also of particular interest to our current study, they found that the two species of Dentaria included in their analyses (Cardamine pentaplzyllos (L.) Crantz and C, bulbifera, both from central Europe) appear

SYSTEklATIC BOTANY

[Volume 25

1 crn

I

FIG. 1. Outline drawings of cauline leaves from selected Cnrlininiize and traditional Dentnria species: a. C. angulntn (drawn from R. R. Hnlse 1522, GA) b. C.Jlngellijern (drawn from W H. Duncan 10858, GA), c. C.lzirs~itn(drawn from S. B. Jones 20594, GA), d. C.constnr~cei(drawn from M . Oiunbey 2054, GA), e. C.concntenntn (drawn from P Siueeizey 0089, GA), f. C.itznsinzn (drawn from P Sicleeney 0107, GA), and g. C. dipl~ylln(drawn from P Si~leeeeney0047, GA).

to be nested within groups of Cardamine sensu stric- 1998), we have chosen the closely related genera to and are apparently not closely related to one an- Rorippa Scop, and Barbarea R. Br. as outgroups for our study. other. In the following study, we use sequence comparisons of the rapidly evolving chloroplast gene ndhF (thought to encode a subunit of NADH dehydrogenase; Sugiura 1992; Olmstead and Sweere 1994; The taxa used in this study, authorities for their Kim and Jansen 1995), supplemented by sequences names, and voucher information are listed in Table from the chloroplast trnL intron (a group I intron), 2. Nine Dentaria species, representing all three disin order to assess the relationships among the east- junct geographical groups and two of Schulz's secern North American, western North American, and tions, are included. Among the Dentaria species is Eurasian groups traditionally placed in the genus the type for the genus, C. pentaphyllos (D. pentaDentaria, and their placement in relation to repre- phylla). The six Cardamine sensu stricto species used sentative groups of Cardamine sensu stricto. We also in this study were selected to reflect a range of moruse our data set to assess the relationship between phological diversity within the genus. Three of Cardamine and Dentaria and the putatively closely Schulz's sections are represented among the Carrelated genus Nasturtium, which was not resolved damine sensu stricto taxa. in the comparisons of Franzke et al. (1998), but has Total genomic DNA was extracted from silica-gel been suggested as a likely sister group to Cardamine dried leaves or fresh leaf material using the CTAB by Les (1994) and Al-Shehbaz and Price (1998). extraction method of Doyle and Doyle (1987).DNA Based upon rbcL sequence comparisons of Les extracts were further purified using High Pure fil(1994) and a broader set of ndlzF sequence compar- ters (Boehringer Mannheim Corp.). The trnL-F reisons (Galloway et al. 1998; Price and Sweeney gion was amplified using the forward primer c and

ZOOO]

471

SLYEENEY& PRICE: POLYPHYLYOF DENTARIA

TABLE2. Taxa sequenced for the trnL intron and 11dl1F. The synonymous name under Delltauin is also given for species traditionally treated in that genus. Species

Ingroup: Cnrdnniii~enngustatn 0 . Schulz = Deiltaria ileteropl~yllnNutt. Cnrdntiliile bellid(fo1ia L. Cnrdarniile brezoeui Watson Cnrdniizii~ebzilbosa (Schreber ex Muhlenb.) BSP Cardntniile cnliforilicn (Nutt.) Greene = Dentnrin califor~zicnNutt. Cardnmiile cleinntitis Shuttlew Cardanline coilcntenatn (Michaux) Schwarz = D e w taria lnciilnta Muhlenb. ex M7illd. Cardninii~ediplylln (Michaux) Alph. Wood = Dentarin diyhylln Mid~aux Cardamiile dissecta (Leavenw) Al-Shehbaz = Deiztnria ~iliiltijidaMuhlenb. ex Elliot Cnrdanzii~eflagellifern 0 . Schulz Cnrdnmiize X iizcisa (Eames) Schumann = Delltarin incisifolia Eames ex Britton Cnrdninine inncrophylln Willd. Cardanlii~emnxiinn (Nutt.) Alph. Wood = Deiztaria nzaxinzn Nutt. Cardainiile peiltnphyllos (L.) Crantz = Deiltnrin pent y l ~ y l l nL. Cnrdarnine zcaldsteinii Dyer = Dentnrin trifolia Waldst. and Kit. Outgroups: Bnrbarea olilgnris Aiton ,VRst~irtitimganlbelii (S. Watson) 0 . Schulz hksttirtilin~oficiilnle R.Br. Rorippn curoipes Greene

Voucher or source

Gen Bank accession (trnL1 rld11F)

P. Sweeney 0026 (GA) C. Brochmann 94-536 (TRH) R. Price 1359 (GA) R. Price 1368 (GA)

D. Kelch s.n. (GA) P. Sweeney 012 (GA) P. Sweeney 0149 (GA) P. Sweeney 0089 (GA) P. Sweeney 0002 (GA) R. Price s.n. (GA) P. Sweeney 0138 (GA) I. A. Al-Shehbaz 9312 (MO) P. Su7eeney0106 (GA) Heronswood Nursery P. Sweeney 0278 (GA)

R. R. R. R.

Price Price Price Price

reverse primer of Taberlet et al. (1991) and the trnL intron was sequenced using the forward primer c. The portion of the ndhF gene from bases 600 through 2123 (approximately 70% of the total length of the gene) was amplified in two steps. The middle portion of the gene was amplified using primers ndhF599 [5'-TAGGTCTTTATTGGATAAC3'1 and ndhFl354R [5'-AAATGTCCTTCAAAAGTAAG-3'1 (designed from the consensus of Sisymbrium irio L. and Erysimum cheiri (L.) Crantz sequences, R. Price unpublished data). The adjacent 3' portion of the gene was amplified and sequenced using primers ndhF1318 and ndhF2110R of Olmstead and Sweere (1994) which are based on sequences from Nicotiana tabacum L, and differ slightly in numbering from their positions in Brassicaceae due to length changes in the molecule. The resulting PCR products were purified using High Pure filters (Boehringer Mannheim) and then direct-

1197 (GA) C34 (GA) 1362 (GA) 1357 (GA)

ly sequenced using an ABI version 3703 or 3777 automated sequencer at the Molecular Genetics Instrumentation Facility of the University of Georgia. The resulting sequences have been submitted to GenBank and accession numbers are given in Table 2. No insertions or deletions (indels) were found for the ndhF gene and only a single 4 bp indel was found in the trnL intron, allowing the sequences to be unambiguously aligned by hand. The 4 bp indel was coded as a gap where absent (treated as missing data), and then added to the sequence data set as an individual binary character. Maximum parsimony analysis was performed on the data set using the Branch and Bound algorithm of PAUP* version 4.0b2a (Swofford 1999).Strength of support for individual branches on the cladogram was assessed using 100 bootstrap replications (Felsenstein 1985) and parsimony jacknife replications (Farris et al. 1996).

472

SYSTE~VIATIC B OTANY

For the maximum parsimony analysis utilizing ndhF; 1523 bp of sequence were compared, of which 108 (7.1%) were variable and 56 (3.7%) were phylogenetically informative. A total of 30 minimum length trees of 127 steps were produced with a consistency index (CI) excluding uninformative characters of 0.913 and a retention index (RI) of 0.885. One of the 30 minimum length trees with branch lengths and bootstrap support values is shown in Fig. 2. This analysis shows Nasturtium as sister to Cardamitze, with bootstrap support of 100%. All species of Cardamine and Dentaria included in the analysis formed a single clade supported by a bootstrap value of 70%. Two unique base substitutions are consistently found within all Cardamine and Dentaria examined. Within the Cardamine and Dentaria clade, several groups are well supported. The two European Dentaria species examined, Cardamine pentaphyllos and C , waldsteinii, form a strongly supported monophyletic group (with 100°h bootstrap support) and are sister (with 92% bootstrap support) to a moderately supported clade (with 70% bootstrap support) containing C, brezueri, C , bulbosa, C. californica (a western North American Dentaria), and C. flagellifera. Also strongly supported is a clade containing C. bellidifolia and C. clematitis (with 100% bootstrap support). Two clades receiving moderate bootstrap support are a clade comprising C. calijornica and C. flagellifera (with 71% bootstrap support) and a clade comprising the eastern North American Dentaria (with 70% bootstrap support). A grouping of Cardamine brezueri and C. bulbosa is weakly supported as monophyletic (with only 52% support). No other clades are apparent, leaving the backbone of the Cardamine and Dentaria clade unresolved. Five-hundred and ten bp of sequence were compared in the trnL intron, of which 31 (6.1%) were variable and 16 (3.1%)were phylogenetically informative. Maximum parsimony analysis of the trnL intron yielded two minimum length trees of 49 steps with a CI excluding uninformative characters of 0.959 and a RI of 0.951. Within the trnL intron there was a four bp insertion (ACAT) restricted to five species of the Cardaminel Dentaria clade, including t6ree North American species of Cardamine sensu stricto ( C , breuieri, C , bulbosa, and C. flagellifera) and the two European Dentaria species. The analysis of the trnL intron gave only four moderately-to strongly supported clades: a C. brew eri and C. flagellijera clade; a European Dentaria

[Volume 25

clade of C. pentaphyllos and C , waldsteinii; a clade containing C , brezueri, C. bulbosa, and C. flagellifera; and a clade containing all eastern North American Dentaria. These clades and all other Cardamine and Dentaria taxa form a large polychotomy. The resolved portions of the two trees produced by the analysis of the tr,nL intron and the strict consensus tree produced by the ndhF analysis were largely congruent and differed only in areas supported by single characters in the trnL data set. Of note is the exclusion of C. calijornica in the trnL analysis from the clade containing the European Detztaria and C , brezueri, C , bulbosa, and C , flagellifera. However, only the indel character in the trnL analysis supports the exclusion of C. californica from the group of European Dentaria and the North American Cardamine, whereas C. calijornica is strongly supported as part of this clade by four non-homoplasious characters in the ndhF analysis and a bootstrap value of 92% (Fig. 2). Given the high degree of congruence of the well supported portions of the trees and the inheritance of the chloroplast genome as a single unit, the data sets were combined and re-analyzed. Maximum parsimony analysis of the combined data sets yielded 30 minimum length trees of 178 steps with a CI excluding uninformative characters of 0.916 and an RI of 0.891. The strict consensus tree with bootstrap replication and jacknife values for corresponding branches along with Schulz's sectional placement of each taxon is shown in Fig. 3, and one of the thirty most parsimonious trees with proportional branch lengths is shown in Fig. 4. The topologies of the strict consensus trees from ndhF and the combined data set (Figs. 2,3) are identical except for the weakly supported C. brmerilC. bulbosa clade being collapsed in the combined analysis, but the bootstrap values are raised significantly for some subgroups of the Cardamine/ Dentaria clade. In particular, bootstrap support for the lineage comprising the eastern North American Detztaria clade increased from 70% to 98%.

Comparisons using the quickly changing chloroplast gene ndlzF provide support for a sistergroup relationship between Nasturtium and Cardamine sensu lato, which was unable to be resolved by the ITS and trnT-L comparisons of Franzke et al. (1998). This is consistent with the results of Les (1994) based on rbcL sequence comparisons, which indicated that Nasturtium officinale is distinct from

20001

SWEENEY & PRICE: POLYPHYLY OF DENTARIA

r-

Barbarea vulgaris

Rorippa cum-pes

IF

Cardamine angustara

69 2

-1 Cardamine dissecta

-1 Cardamine conca tena ta x i

Cardamine X incisa 2

7

Cardamine diphylla 1

Cardamine maxima

-

Cardamine bellidifolia

3

100 8

Cardamine clema titis

11

7 Cardamine

70

-

52 2

2

breweri

1

2

71 -

4

2

92

4

4

Cardamine bulbosa

11

70

Cardamine californica Cardamine flagellifera

100 9

Cardamine pen taphyllos

L -

3

Cardamine waldsteinii

Cardamine macrophylla

8

100 15

3 2

Nasturtium officinale Nasturtium gambelii

FIG. 2. One of 30 most parsimonious trees of 127 steps produced by maximum parsimony analysis of ndhE Bootstrap replication values (>5056 out of 100 replicates) are given above corresponding branches and numbers of substitutions are given belo~v.

474

[Volume 25

SYSTEMATIC BOTANY

I I Rorippa curvipes

Cardamine angustata - Dentaria

Cardamine djssecta - Den taria Cardamine concatenam - Dentaria Cardamine diphylla

- Den taria

Eastern North American Dentaria

-

Cardamine X incisa Dentaria

I

99

98

-1

Caniamine bellidifolia - Cardarninella Cardarnhe clemati tis Cardamine breweri

- Cardamhe

- Cardamine

Cardamine bulbosa - Carciamine Western Cardamine caljfornica - Eutreptophyllum m North American Dentaria

91

I

I

Cardamine flagellifeta - Macrophyllum

Cardamine macrophylla

- Macrophyllum

FIG. 3. Strict consensus tree of 30 most parsimonious trees of 178 steps produced by maximum parsimony analysis of combined trnL intron and rld1zF data. Bootstrap replication values are given above and jackknife values are given below corresponding branches. The location of the indel event is indicated by solid (gain) and open (loss) bars. Schulz's (1936) sectional classification is given to the right of each taxon. Traditional Delltarin species are indicated.

Rorippa and most similar to a representative species of Caudamine. Our finding that Cardamine sensu lato is monophyletic and separate from Nasturtium is also consistent with morphological evidence, since Nasturtium lacks the unusual elastic fruit valve uncoiling mechanism and winged replum character-

istic of Cardamine and Dentaria (Al-Shehbaz and Price 1998).It is of interest, however, that only two shared derived characters in our sequence data set were found to consistently support the monophyly of Cardamine sensu lato relative to Nastuutium, perhaps implying an early radiation of groups within

20001

-1

I f -

SWEENEY & PRICE: POLYPHYLY OF DENTARIA

Barbarea vulgaris Rorippa curvipes Cardamine angustata

P

Cardamine dissecta

Cardamine concatenata

Cardamine X incisa

I

Eastern North American Dentaria

Cardamine diphylla Cardamine maxima

k 1

Cardamine bellidifolia

I

Cardamine clema titis

Cardamine breweri

d

Cardamine bulbosa Cardamine californica

Western North American Dentaria

Cardamine flagellifera

Cardamine macrophyIla

I

1,

p-

'

Nasturtium oftkinale

Nasturtium gambelit

FIG. 4. One of 30 most parsimonious trees produced by maximum parsimony analysis of combined trnL intron and ndhF data. Numbers of base substitutions are given above corresponding branches. The location of the indel event is indicated by solid (gain) and open (loss) bars.

476

SYSTEMATIC BOTANY

Cardamine shortly after it separated from the Nasturtium lineage. The results-of this study do not support the treatment of Dentaria as a distinct genus, as our phylogenies strongly suggest that the group is polyphyletic and has apparently arisen at least three separate times from within Cardamine (Figs. 3,4) to give rise to the Eurasian, eastern North American, and western North American groups. Our finding that the the type species of Dentaria (the European C. pentaplyllos) is placed within a group of perennial Cavdamine is consistent with the results of the ITS comparisons of Franzke et al. (1998) which utilized a completely different set of species of Cardamine sensu stricto from those in our analyses and also found C , pentapl~yllos to be nested within a subgroup of Cardamine. The broader circumscription of Cardamine including Dentaria has been adopted by several recent floras (e.g., Jones and Akeroyd (1993) for Europe and Rollins (1993a) for North America), and only those working with restricted biogeographical subsets of Cardamine and Dentaria have sometimes chosen to separate the two. This follows from the fact that it is difficult to cleanly separate Cavdamine and Dentavia on morphological grounds when all of the variation in Cavdamine is accounted for (Table 1; Fig. 1; Al-Shehbaz 1988a, 1988b).Cardamine including Dentavia can easily be recognized by its elastically uncoiling fruit valves, and with the recent reduction of the small Himalayan genus Loxostemon Hook, f. & Thomson to synonymy in Cardamine (Al-Shehbaz and Guang 1998) this feature is unique to Cardamine within the family Within the genus Cardamine, our results do not provide support for Schulz's (1903, 1936) sectional classification. None of the three of Schulz's sections (Cardamine 0. Schulz, Dentaria (L.) 0. Schulz, and Macrophyllum 0.Schulz) for which we sampled two or more species are supported as monophyletic by our sequence comparisons (Fig. 3). This is consistent with the results of Franzke et al. (1998) whose ITS and trnT-L comparisons of Old World Cardamine also suggest that sections Cardamine, Dentavia, and Cardaminella Prantl are not monophyletic as currently delimited. These results are in accord with the view that Schulz used too few morphological characters when delimiting sections (see AiShehbaz 1988a, 198813; Khatri 1989; Rashid and Ohba 1994). Broader sampling of species using larger areas of rapidly evolving sequence will be needed to assess more thoroughly sectional relationships within the genus, which is cosmopolitan

[Volume 25

in distribution and has been estimated to include at least 200 species (Al-Shehbaz 1988b). At a lower taxonomic level, the six taxa of eastern North American dentarias are strongly supported as a monophyletic group by our comparisons, with 98% bootstrap and 97% jackknife support in our combined analysis (Fig. 3). This is consistent with the strong morphological resemblance among the species and reported hybridization among some of them (Al-Shehbaz 198810). Since these species do not fall into the same clade as the type species of Dentavia and the current infrageneric classification seems poorly workable, we treat the eastern North American dentaria group informally as the Cardamine concatenata alliance. A detailed study of species delimitation in this alliance based on multivariate analysis of morphology is presented in Sweeney and Price (submitted). The two species of European dentarias included in our analyses (C. pentnphyllos and C. zualdsteinii) form a well-supported clade with 100% bootstrap and jackknife support (Fig. 3), which is consistent with their morphological similarity. These species have been placed in subgenus Dentaria within Cardamine by Jones and Akeroyd (1993), who recognize eight European species in this group. The European dentaria group is distinguished from the other European Cardamine species by the presence of thickened subterranean rhizomes with scale leaves. In contrast to the morphological classification of Jones and Akeroyd and our results, the monophyly of the European dentarias was not supported by the results of Franzke et al. (1998). Their sequence analyses did not place C. bulbifera in a clade with C. pentaphjllos, but instead suggested that C. bulbifera is closely related to the extremely morphologically dissimilar C, impatiens L., a small-flowered biennial species with pinnately compound cauline leaves and without fleshy rhizomes. As noted by Franzke et al. (1998), it would be useful to further study this species relationship, since C. bulbifera is 12-ploid (2n = 96) and C. impatiens is diploid (2n = 16), suggesting the possibility that C bulbifera is an allopolyploid involving both Cardamine sensu stricto and dentaria species as parents. The Pacific North American dentaria C. califovnica is placed by our ndhF and combined analyses in a well-supported clade including the western North American C , breweri and the eastern North American C, bulbosa and C, j7agellifera. All of these species are rhizomatous perennials and often possess simple cordate basal leaves (Rollins 1993a, 1993b). Among these four species, C. californica and C. bul-

20001

SWEENEY & PRICE: POLYPHYLY OF DESTARIA

bosa are similar in having short tuberlike rhizomes, while C. bvewevi and C,flagellifeva have very narrow elongated rhizomes. Interestingly, C , bulbosa and its close relative C. douglasii Britton have almost always been placed in Cardamine sensu strict0 despite their thickened rhizomes (Hart and Eshbaugh 1976; Al-Shehbaz 1988b),perhaps because they have simple and almost unlobed basal and cauline leaves. In contrast, C , californica and several putatively closely related Pacific coastal taxa, e.g. C. nuttallii E. Greene, typically have 3 to 5-foliate cauline (and sometimes basal) leaves and thus have often been placed in Dentaria (Detling 1936; Munz 1959).Species delimitation in the western North American dentaria group has been controversial (see Detling 1936; Rollins 1993a, 1993b) and it would be useful to explore in detail using molecular comparisons. Cardamine flagellifera and C. clematitis, both endemic to the Appalachian region of the eastern United States, have sometimes been treated as conspecific (Radford et al. 1968), although they are clearly distinguished by the pubescent lower stems and nonauriculate petioles of the former versus the glabrous stems and auriculate petioles of the latter (Al-Shehbaz 198813). Our sequence comparisons suggest that despite superficial similarity in morphology, these two species fall into quite separate groups within the genus. Cardamine clematitis forms a well supported clade with the perennial arcticalpine species C. bellidifolia in our ndhF and combined trees, with bootstrap support of 100% and 99%, respectively. These two species are quite dissimilar in overall morphology, with C. bellidifolia being a caespitose and subscapose plant with noncordate diamond-shaped to ovate leaf blades (Rollins 1993a, b). The immediate grouping of these two species may simply reflect insufficient sampling of other related perennial species from Eurasia or North America, but the complete lack of support for a close relationship between the morphologically similar C. clematitis and C. flagellifera provides further indication of the lack of reliability of some of the morphological characters used to determine sectional and interspecific relationships within the genus. Because hybridization has been reported within Cardamine and greater than 60% of the species studied are polyploid (Al-Shehbaz 1988b), some caution should be used in interpreting the results produced from chloroplast DNA data alone as it will reflect only one parent in taxa of hybrid origin. However, it is useful to note that the ITS sequence analysis of Franzke et al. (1998)is consistent with our results

477

in indicating that the European dentaria species C . pentaplyllos and C . bulbifera are nested within Cardamine. Based upon our sequence comparisions, the morphological characters traditionally used to circumscribe the genus Dentavia (i.e., trifoliate leaves, few stem leaves, and fleshy rhizome) appear to be subject to convergence among Dentaria lineages that apparently have originated separately in Eurasia, eastern North America, and western North America. Perhaps the morphological similarity between species traditionally placed in Dentaria is a result of independent adaptation to similar but geographically disjunct deciduous forest habitats. This would be consistent with the findings of van Groenendael et al. (1996) that clonal species are more frequent in low-light environments across a wide variety of plant groups. ACKROWLEDGEMEKTS. We are grateful to the Botany Department of the University of Georgia and to the Highlands Biological Research Station for grants-in-aid in support of this research (to PWS). We thank James Hamrick, David Giannasi, James Manhart, and two anonymous reviewers for critical reading of the manuscript. We further thank Ihsan Al-Shehbaz, Christian Brochmann, Dean Kelch, and the Munich Botanical Garden for providing plant material used in this study. And finally a special thanks goes out to Hildegard Mboten of the University of Georgia Herbarium for translating selected portions of Schulz (1936) from German to English.

AL-SHEHBAZ, I. A. 1988a. Cavdniizine dissectn, a new combination replacing Deiztnrin inultifidn (Cruciferae). Journal of the Arnold Arboretum 69: 81-84. -1988b. The genera of Arabideae (Cruciferae; Brassicaceae) in the Southeastern United States. Journal of the Arnold Arboretum 69: 87-166. -and Y.GUANG.1998. Notes on Chinese Cnrdnmiize (Brassicaceae). Harvard Papers in Botany 3: 73-78. R. A. PRICE.1998. Delimitation of Nasturtiun~ -and (Brassicaceae). Novon 8: 124-126. BUSH,N. A. 1939. Cnvdniizine and Deiztavin. Pp. 110-129 in Florn of the US.S.R. Vol. VIII, English version trans. by R. Lavoott and ed. by U. Plitma. Jerusalem: Keter Press. CRANTZ,H. J. N. 1769. Clnssis crucifoviniuin emeizdnta. Leipzig: Ioannis Pauli Kraus. DETLING,L. E. 1936. The Genus Deiztarin in the Pacific States. American Journal of Botany 23: 570-576. -1937. The Pacific Coast species of Cnvdninine. American Journal of Botany 24: 70-76. DOYLE,J. J. and J. L. DOYLE.1987. A rapid DNA isolation procedure from small quantities of fresh leaf tissue. Phytochemical Bulletin 19: 11-15.

478

SYSTEMATIC BOTANY

FARRIS,J . S. 1996. Parsimony jackknifing outperforms neighbor-joining. Cladistics 12: 99-124. FELSENSTEIN, J. 1985. Confidence limits on phylogeny: an approach using the bootstrap. Evolution 39: 783-791. FERNALD, M. L. 1950. G r q 9 inniz~lalof botany. 8th ed. New York: American Book Company. n! BLEEKER,R. KOHRT,and FRANZKE,A,, K. POLLMANN, H. HURKA.1998. klolecular systematics of Cnrdnirzine and allied genera (Brassicaceae): ITS and non-coding chloroplast DNA. Folia Geobotanica 33: 225-240. GALLOWAY, G. L., R. L. MALMBERG, and R. A. PRICE.1998. Phylogenetic utility of the nuclear gene arginine decarboxylase: an example from the Brassicaceae. Molecular Biology and Evolution 15: 1312-1320. GLEASON,H. A. and A. CRONQUIST. 1991. Mai?unl of rnsculnr plnnts of izortizenstern United States nnd ndjncent Canada. 2nd ed. brew York: New York Botanical Garden. HARRIMAN, N. A. 1965. The genus Derztnrin L, (CruciJerne) i n enstern North Aiizericn. Ph.D. dissertation, Vanderbilt University, Nashville, Tennessee. 1976. The biosystematHART,T. W. and M! H. ESHBAUGH. ics of Cnvdninine bulbosn (Muhl.) B. S. P. & C . Douglnssii Britt. Rhodora 78: 329419. JANCHEN,E. 1942. Das System der Cruciferen. ~ s t e r r e i chische Botanische Zeitschrift 91: 1-28. JONES,B. M. G. and J. R. AKEROYD. 1993. Cardanzine. Pp. 285-289 in Florn Europnen, 2nd ed., Vol. I, ed. T. G. Tutin. Cambridge, New York: Cambridge University Press. KHATRI,K. S. 1989. Cardninine L. sect. Cnrdaininella Prantl (Cruciferae) in the Asian territory of Soviet Union. Feddes Repertorium 101: 81-96. 1995. NdhF sequence evolution KIM,K-J. and R. K. JANSEN. and the major clades in the sunflower family. Proceedings of the National Academy of Sciences of the United States of America 92: 10379-10383. LES, D. H. 1994. Molecular systematics and taxonomy of lake cress (Neobeckin aquntica; Brassicaceae), an imperiled aquatic mustard. Aquatic Botany 49: 149-165. LINNAEUS, C. 1753. Species Plniztnrlinz. Stockholm: Impensis Laurentii Salvii. F. H. 1955. Preliminary studies in the genus MONTGOIVIERY, Dentnvin in eastern North America. Rhodora 57: 161173. MUNZ,P. A. 1959. A Califorizinflorn. Berkeley: University of California Press. 01, J. 1965. Flora of Jqniz. Washington D. C.: Smithsonian Institution. R. G. and J. A. SWEERE. OLMSTEAD, 1994. Combining data in phylogenetic systematics: an empirical approach

[Volume 25

using three molecular data sets in the Solanaceae. Systematic Biology 43: 467481. PRICE,R. A. and P. \V. SWEENEY. 1998. Relationships in the tribe Cardamineae (Brassicaceae) from trnL-F and ndhF sequence comparisons. 85 (6, Supplement): 152. RADFORD, A. E., H. E. AHLES,and C. R. BELL.1968. Maizunl of the vnsculnr flora of the Cnrolinas. Chapel Hill: The University of North Carolina Press. RASHID,A. H. and H. OHBA.1993. A revision of Cnvdninine losostenlonoides 0 . E. Schulz (Cruciferae). Journal of Japanese Botany 68: 199-208. ROLLINS,R. C. 1993a. The Cruciferae of contiizental North America. Stanford: Stanford University Press. -1993b. Cnvdnmine. Pp. 407410 in The jepson nznnunl, ed. J. C. Hickman. Berkeley: University of California Press. SCHULZ,0 . E. 1903. Monographie der Gattung Cnrdniniize. Botanische Jahrbucher fiir Systematik, Planzengeschichte und Planzengeographie 32: 280-623. -1936. Cvuciferne. Pp. 227-658 in Die nntiirlichen Plnilzenfnmilieil, Vol. 17., eds. A. Engler and K. Prantl. Leipzig: Engelmann. SPOONER,D. M. 1984. Reproductive features of Deiztnrin lnciiziatn and D. diphylln (Cruciferae), and the implications in the taxonomy of the eastern North American Deiztnvia complex. American Journal of Botany 71: 999-1005. SUGIURA, M. 1992. The chloroplast genome. Plant Molecular Biology 19: 149-168. P. \V. and R. A. PRICE.(in press). Mutivariate SWEENEY, morphological analysis of the Cnrdnnzine concntennta alliance (Brassicaceae). Brittonia. SWOFFORD, D. L. 1999. PAUP*: Plzylogeizetic analysis using parsirizorzy (* and other methods). Version 4. Sinauer Associates, Sunderland, Massachusetts. I? L., L. GIELLY,G. PAUTOU,and J. BOLVET. TABERLET, 1991. Universal primers for the amplification of three non-coding regions of choroplast DNA. Plant Molecular Biology 17: 1105-1109. N. L., D. K. EVANS,and F. S. GILLIAM. 1994. IdenTURRILL, tification of West Virginia members of the Deiztnvia complex [D. dipizylln Michx., D. Izeteroplzylln Nutt., and D. lncinintn Muhl. ex Willd. (Brassicaceae)] using above-ground morphological characters. Castanea 59: 22-30. VANGROENENDAEL, J. M., L. KLIIVIES, J. KLIMESOVA, and R. J. J. HENDRIKS. 1996. Comparitive ecology of clonal plants. Philosophical Transactions of the Royal Society of London. B. 1331-1339. WOFFORD, E. G. 1989. Guide to the vascular plaizts of the Blue Ridge. Athens, Georgia: University of Georgia Press.