Molecular Phylogenetics of Core Brassicales, Placement of Orphan Genera Emblingia, Forchhammeria,...

654

Systematic Botany (2004) 29(3) pp 654ndash669q Copyright 2004 by the American Society of Plant Taxonomists

Molecular Phylogenetics of Core Brassicales Placement of Orphan GeneraEmblingia Forchhammeria Tirania and Character Evolution

JOCELYN C HALL12 HUGH H ILTIS and KENNETH J SYTSMA

Department of Botany University of Wisconsin Madison Wisconsin 537061Current Address Arnold Arboretum Dept OEB Harvard University 22 Divinity Ave

Cambridge MA 021382Author for Correspondence (email jhalloebharvardedu)

Communicating Editor Alan Meerow

ABSTRACT Many genera previously placed in the traditionally circumscribed Capparaceae are either unrelated or morecommonly isolated lineages in the order Brassicales This study examines the relationships of three of these orphan generaEmblingia Forchhammeria and Tirania in the context of a focused analysis of the core Brassicales In order to assess rela-tionships of these genera analyses were conducted across Brassicales using chloroplast rbcL ndhF and matK sequence in-formation Both maximum parsimony and maximum likelihood analyses place all three genera in the well-supported coreBrassicales (Brassicaceae Capparaceae Cleomaceae Gyrostemonaceae Pentadiplandraceae Resedaceae and Tovariaceae)The Asiatic Tirania and New World tropical Forchhammeria are closely related to two small families the pan-temperateResedaceae and the Australian Gyrostemonaceae These analyses also indicate a novel placement of Emblingia as sister toall remaining members of core Brassicales Although there is strong support for the relationships among most of these taxarelationships of Pentadiplandraceae and Tovariaceae are weakly resolved Thus the core Brassicales is a biogeographicallydispersed lineage that is comprised of many small and morphologically distinct clades plus the large crown group Brassi-caceae s lat Patterns of morphological evolution appear complex especially in floral merosity and carpel and locule numberLikewise the evolution of breeding systems within this lineage involves recurrent shifts towards monoecy or dioecy andpossible reversals to bisexuality Further sampling of Capparaceae tribe Stixeae is critical for any taxonomic recommendationof familial status for these orphan genera

The pantropical family Capparaceae Jussieu withup to 45 genera and 900 species has long presentedenormous difficulties related to monophyly and gener-ic placements Recent molecular and morphologicalphylogenetic analyses of Capparaceae s lat and close-ly related Brassicaceae in the order Brassicales indicatethat Capparaceae s lat as traditionally circumscribedare paraphyletic (Rodman et al 1993 1996 1998 Juddet al 1994 Hall et al 2002 Capparaceae s lat indicatestraditional paraphyletic familial delimitation) Cap-paraceae s lat comprise two monophyletic groups cor-responding to the two major subfamilies Cleomoideaeand Capparoideae with the Cleomoideae more closelyrelated to Brassicaceae than to Capparoideae (Hall etal 2002) These three taxa have been either combinedinto one family Brassicaceae s lat (Judd et al 1994APG 1998 2003) or elevated to three separate familiesCapparaceae s str (equivalent to Capparoideae) Cleo-maceae (Airy Shaw 1965) and Brassicaceae (Hall et al2002) In this paper we use the later classificationbased on both morphological and molecular grounds(Hall et al 2002) In addition to challenges of mono-phyly of Capparaceae s lat and issues of familial rankthere have been numerous challenges based on bothmorphological and molecular evidence for placementof genera within Capparaceae s lat (Table 1) Al-though some of these genera are now removed fromBrassicales entirely (eg Calyptrotheca and Physena toCaryophyllales) most questionable genera remain inBrassicales but outside Capparaceae or Cleomaceae(eg Pentadiplandra Villiers 1973 Setchellanthus Karol

et al 1999) Two prime examples include Pentadiplandraand Tovaria monotypic genera placed in Capparaceaes lat that have been elevated to familial status and areseemingly isolated members of Brassicales (Rodman etal 1996 1998)

Recent studies indicate two genera placed in Cap-paraceae s lat Emblingia and Forchhammeria representgenera that are closely related to but not part of Cap-paraceae s lat Emblingia is an enigmatic monotypicgenus that was placed in Capparaceae (Pax and Hoff-man 1936) but has been recently aligned with Sapin-daceae (Leins in Erdtman et al 1969 Thorne 1992)Goodeniaceae (Melville and Metcalfe in Erdtman et al1969) or Polygalaceae (Erdtman in Erdtman et al1969 Cronquist 1981) Based only on rbcL sequencesEmblingia was tentatively linked with Resedaceae inthe core Brassicales (Chandler and Bayer 2000) a cladein the order comprising Brassicaceae CapparaceaeCleomaceae Gyrostemonaceae PentadiplandraceaeResedaceae and Tovariaceae (Rodman et al 19931996 1998 Hall et al 2002) Chloroplast sequences ofndhF and trnL-trnF placed the largely Central Ameri-can Forchhammeria within the core Brassicales and assister to either Resedaceae or Gyrostemonaceae (Hallet al 2002) The ten species of Forchhammeria are rela-tively nondescript dioecious shrubs that have been as-signed to numerous families by a variety of taxono-mists and their placement in Capparaceae s lat hasoften been regarded as provisional (Hansen 1977) Paxand Hoffmann (1936) placed Forchhammeria in the tribeStixeae with Neothorelia Stixis and Tirania Based on

2004] 655HALL ET AL PHYLOGENETICS OF CORE BRASSICALES

TAB

LE

1Pr

oble

mat

icge

nera

plac

edin

Cap

para

ceae

sla

t(5

Cap

para

ceae

sst

ran

dC

leom

acea

e)by

Pax

and

Hof

fman

n(1

936)

and

Bran

dege

e(1

909)

1In

Erdt

man

etal

(19

69)

each

ofth

efo

urau

thor

shy

poth

esiz

esa

diff

eren

ttax

onom

icpl

acem

ent

ofEm

blin

gia

Gen

ust

rad

itio

nally

ofC

appa

race

aes

lat

Subf

amily

(Tri

be)

ofPa

xan

dH

offm

ann

(193

6)O

ther

sugg

este

dre

latio

nshi

psC

urre

ntfa

mili

alcl

assi

ficat

ion

Ord

inal

Cla

ssifi

catio

n(A

PG19

98)

Buh

sia

Ben

geB

uhsi

oide

aeU

ncer

tain

Cal

yptr

othe

caG

ilg

Cal

yptr

othe

coid

eae

Port

ulac

acea

e(N

yana

nyo

1986

)N

ear

Did

iere

acea

e(A

pple

quis

tan

dW

al-

lace

2001

)C

aryo

phyl

lale

s

Em

blin

gia

FM

uell

Em

blin

gioi

deae

Poly

gala

ceae

affi

nity

wit

hG

oode

niac

eae

oraf

finit

yto

Sapi

ndac

eae

(Erd

tman

etal

19

691 )

Em

blin

giac

eae

(APG

1998

Cha

ndle

ran

dB

ayer

2000

)B

rass

ical

es

Forc

hham

mer

iaL

iebm

C

appa

roid

eae

(Sti

xeae

)E

upho

rbia

ceae

Sap

inda

ceae

Ole

acea

eM

al-

vace

ae(H

anse

n19

77)

Nea

rR

esed

acea

ean

dG

yros

tem

onac

eae

this

stud

yB

rass

ical

es

Koe

berl

inia

Zuc

cC

appa

roid

eae

(Koe

berl

i-ni

eae)

Koe

berl

inia

ceae

(Rod

man

etal

199

6)B

rass

ical

es

Neo

thor

elia

Gag

nep

Cap

paro

idea

e(S

tixe

ae)

Unc

erta

in(K

ers

2003

)Pe

ntad

ipla

ndra

Bai

llPe

ntad

ipla

ndro

idea

eN

ear

Cel

estr

acea

e(H

utch

inso

n19

73)

Pent

adip

land

race

ae(V

illie

rs19

73)

Bra

ssic

ales

Phy

sena

Nor

inha

exT

houa

rsC

appa

roid

eae

(Sti

xeae

)Pa

ssifl

orac

eae

orFl

acou

rtia

ceae

(rev

iew

edin

Dic

kiso

nan

dM

iller

1993

)Ph

ysen

acea

e(M

orto

nK

arol

and

Cha

se19

97)

Car

yoph

ylla

les

Setc

hella

nthu

sB

rand

egee

N

ASe

tche

llant

hace

ae(I

ltis

1999

)B

rass

ical

esSt

efan

ina

Chi

oven

daB

uhsi

oide

aeR

esed

acea

e(H

utch

inso

n19

73n

otac

cept

-ed

byK

ubit

zki

2003

f)B

rass

ical

es

Stix

isL

our

Cap

paro

idea

e(S

tixe

ae)

Unc

erta

in(K

ers

2003

)T

iran

iaPi

erre

Cap

paro

idea

e(S

tixe

ae)

Nea

rR

esed

acea

ean

dG

yros

tem

onac

eae

this

stud

yB

rass

ical

es

656 [Volume 29SYSTEMATIC BOTANY

FIG 1 Distributions of lsquoorphanrsquo genera of Capparaceae s lat and their putative relatives Emblingia Forchhammeria Gyros-temonaceae Pentadiplandraceae Resedaceae Tirania and Tovariaceae

differences in perianth number and number of loculesfrom Capparaceae s lat Kers (2003) excluded allmembers of Stixeae from his recent generic level treat-ment of Capparaceae s lat Further phylogenetic stud-ies within the core Brassicales are thus warranted asonly one species of Forchhammeria was sampled andputative relatives of Forchhammeria such as Tirania inthe tribe Stixeae have not been sampled

Perhaps more importantly from evolutionary andbiogeographical perspectives the isolated genera andsmall families of the core Brassicales exhibit remark-able variation in habit (annual perennial herb orshrub climbing shrub) floral structure (carpel loculestamen and perianth number) breeding systems (bi-sexual monoecious dioecious polygamous) and dis-junct areas of endemism (Fig 1) Emblingia is a smallshrub possessing five sepals two petals a gynophoreand a multi-carpellate fruit with replum The uniquepollen of Emblingia has short colpi with rounded endsand exine that thickens close to the apertures such thatthe apertural areas bulge out (Erdtman et al 1969 Ku-bitzki 2003a) The ovary in Forchhammeria flowers isbicarpellate and bilocular with each locule usually con-taining two seeds early in development However onlyone of the seeds usually develops giving the appear-ance of being unilocular (Hansen 1977) The petals ofTirania have a ligulate rim similar to Pentadiplandra(Kers 2003) Pentadiplandra is a monoecious climbingshrub possessing pentamerous flowers with more thanthree carpels and locules Although some authors in-clude Tovaria in Capparaceae s lat based on leaf and

wood anatomy (Carlquist 1985 Thorne 1992) thiscoarse shrub is distinct in its typically octmerous flow-ers with plurilocular ovaries with axile placentationand developed endosperm (Cronquist 1981 Appel andBayer 2003) a distinction supported by molecular data(Rodman et al 1993 1996 1998) Gyrostemonaceaeand Resedaceae represent the remaining members ofthe core Brassicales Gyrostemonaceae comprise fivegenera and 18 species of dioecious shrubs Resedaceaecomprise approximately six genera and 70ndash75 speciesof herbs or small shrubs with variable breeding sys-tems Examination of habit morphological charactersand breeding systems of these problematic genera us-ing an explicit phylogenetic framework should provideinsight into their evolution and biogeography

Members of the core Brassicales other than Bras-sicaceae s str Cleomaceae and Capparaceae s strexhibit widely disparate biogeographical distributions(Fig 1) although many are found or centered in drierhabitats (eg Mediterranean biome) The distributionof Forchhammeria is limited to Mexico Central Americaand the West Indies where seven of the ten species arerare and local mostly in either dry or seasonally dryareas (Hansen 1977) Other members of tribe Stixeaeputatively related to Forchhammeria are restricted tosoutheast Asia (eg Tirania) Like Emblingia Gyroste-monaceae are endemic to Australia where most speciesare fire-opportunists or plants of disturbed areas(George 1982) The monotypic Pentadiplandraceae andTovariaceae (questionably ditypic) are restricted totropical Africa and America respectively Whereas To-


varia is found in cloud forests (Appel and Bayer 2003)Pentadiplandra grows in savannas rain forests and sec-ondary growth (Ronse Decraene 2002 Bayer and Ap-pel 2003a) Only Resedaceae have a widespread dis-tribution restricted to the more arid parts of the North-ern Hemisphere mostly in the Old World but with afew species in southwestern United States and north-ern Mexico A more detailed phylogenetic analysis ofthe core Brassicales may provide insights into under-standing their peculiar biogeographical patterns

Presented here is a phylogenetic analysis of theBrassicales with emphasis on the core Brassicales us-ing chloroplast sequence information from three cod-ing regions ndhF (Alverson et al 1999 Hall et al 2002Sytsma et al 2002) rbcL (Alverson et al 1998 Conti etal 1996 1997 2002 Rodman et al 1996 1998 Sytsmaet al 2002) and matK (Koch et al 2001 Sytsma et al2003 Wilson et al 2001) have been shown to be infor-mative at the infrafamilial level in Brassicales and re-lated orders Malvales Sapindales and Myrtales In or-der to address questions on relationships among fam-ilies of the core Brassicales and the placement of Em-blingia Forchhammeria and Tirania we addedapproximately 60 new sequences to the Rodman et al(1996 1998) rbcL Koch et al (2001) matK and our own(Hall et al 2002) ndhF data sets With increased sam-pling of taxa within the core Brassicales and combinedsequence information from three gene regions weshould be able to improve phylogenetic resolutionThis study has three major goals (1) determine rela-tionships of lineages within the core Brassicales spe-cifically relationships among Gyrostemonaceae Rese-daceae Tovariaceae and Pentadiplandraceae (2) deter-mine the placement of three of the unusual genera ofCapparaceae s lat Forchhammeria Tirania and Emblin-gia within the core Brassicales and (3) elucidate pat-terns based on these molecular phylogenetic esti-mates of evolution in floral morphology and breedingsystems within core Brassicales

MATERIALS AND METHODS

Taxon Sampling We sampled widely within the Brassicales(sensu Rodman et al 1998 Appendix 1) and covered 13 of the 16families in the order excluding Limnanthaceae Setchellanthaceaeand Akaniaceae (which has been recently combined with Bret-schneidera into one family [Bayer and Appel 2003b]) Rodman etal (1998) did not sample Emblingia so this taxon was not includedas a member of the order until APG (1998) and Chandler andBayer (2000) placed it there Our sampling includes all families ofcore Brassicales Although Capparaceae Cleomaceae and Brassi-caceae represent the three largest clades of the core Brassicaleswe sampled only four to five representatives from each familybecause all three are well supported as monophyletic and taxaselected represent both the smaller early diverging lineages andthe larger subclades within each family (Hall et al 2002) A singlerepresentative each from Pentadiplandraceae (monotypic) and To-variaceae (monotypic or questionably ditypic [Appel and Bayer2003]) were available Resedaceae and Gyrostemonaceae had twoand three representatives respectively Five species of Forchham-meria were sampled (for both rbcL and matK) representing all three

subgenera recognized by Hansen (1977) The monotypic Tiraniaan additional member of the tribe Stixeae and Emblingia were in-cluded in the ndhF and rbcL data sets Outside of the core Brassi-cales sequences were obtained for six additional families (sevenin the rbcL analysis that also samples Salvadoraceae) with onespecies representing each family Bretschneideraceae Koeberlini-aceae and Setchellanthaceae are monotypic and Caricaceae Mor-ingaceae and Tropaeolaceae are each monophyletic based on mo-lecular data (Gadek et al 1992 Andersson and Andersson 2000Olson 2002a b) Tropaeolum and Bretschneidera have been identifiedas sister to the remaining Brassicales (Rodman et al 1993 19941996 1998) and were designated as a monophyletic outgroup Asuite of floral and embryological characters supports Tropaeolumand Bretschneidera as a monophyletic group (Ronse Decraene et al2002) For the majority of the 38 taxa sampled all three genes wereobtained from the same DNAs or species (Appendix 1) In rarecases the sequence was obtained from closely related species (egCapsella in matK)

Extraction Amplification and Sequencing Total genomic DNAwas extracted from fresh frozen silica-dried or herbarium sam-ples using a modified CTAB method (Doyle and Doyle 1987 Smithet al 1991) or Dneasy Plant Mini Kits (Qiagen Valencia Califor-nia USA) Standard polymerase chain reaction (PCR) and cyclesequencing techniques were used to amplify and sequence doublestranded DNA (eg Hall et al 2002) The 39 end of the ndhF genewas amplified using forward primer 972F and reverse primer2110R from Olmstead et al (1993) or slightly modified based onArabidopsis sequences (Hall et al 2002) Four primers were used tosequence both strands of the ndhF gene 972F 1603R 1318F and2110R The rbcL gene was either amplified in one reaction using59 and 39 primers or in two reactions using two primer pairs 59674R and 523F39 (see Conti et al 1997 for details) Four sequenc-ing reactions were conducted using primers 59 39 523F 674R al-lowing for verification of both strands Occasionally other primers(1020F 1088R 895R) were used for sequence verification The matKregion was amplified using trnK-710F and matK-1495R primers(Koch et al 2001) A combination of primers was used to sequencethis region 1F 16F 495F 495R 1010R 1010F 1088R 1412F 1412Rand 1495R (Koch et al 2001) In almost every instance bothstrands of each of the three genes were obtained In the few ex-ceptions where this was not possible manual proofreading of se-quences was done in order to detect any misreads

Sequences were aligned using Sequencher v30 (Gene CodesCorporation Ann Arbor Michigan USA) and alignment was fur-ther refined using Se-Al v20a6 (Rambaut 2001) or MacCladev405 (Maddison and Maddison 2002) All regions were codonaligned using the known Arabidopsis sequences Whenever possi-ble rbcL sequences were also checked for an extension past thetypical termination point which has been demonstrated to becharacteristic of core Brassicales and near relatives (Rodman et al1994 1996 Karol et al 1999) Previous phylogenetic studies havenoted indel events both in ndhF (Hall et al 2002 Sytsma et al2002) and matK (Koch et al 2001) that can be phylogeneticallyinformative Indels that were potentially parsimony informative(eg shared by two or more but not all taxa) were scored andadded to the end of the data set as presenceabsence charactersfollowing the guidelines of Baum et al (1994)

Phylogenetic Analyses Variation in DNA sequences was usedto reconstruct phylogenetic relationships using maximum parsi-mony (MP) and maximum likelihood (ML) in PAUP (v 40b10Swofford 2002) All three data sets were analyzed individually andin combination Phylogenetic data sets have been deposited inTreeBASE (study accession number S1000) To explore the possi-bility of multiple islands of most parsimonious trees 1000 randomaddition replicates with Multrees (save multiple trees) holding fivetrees at each step and TBR branch swapping were used All char-acters were equally weighted and treated as unordered (Fitch1971) In analyses of ndhF and matK these data were analyzedwith and without scored indels In addition to standard measuresof fit of characters to the trees produced (ie consistency indexretention index) the strength of support for individual brancheswas estimated using the bootstrap (Felstenstein 1985) Bootstrap


analyses used 1000 replicates (simple addition saving up to 1000trees per replicate TBR branch swapping multrees holding 1 treeper step)

Fifty-six maximum likelihood models were explored in individ-ual and combined data using Modeltest v306 (Posada and Cran-dall 1998) To establish which model of DNA substitution best fitsthe data this program compares these ML models in a hierarchicaltesting framework by calculating the likelihood ratio statistic be-tween different models Likelihood ratio tests and their associatedp-value compare alternative models of sequence evolution and im-provement in fit with increasing model complexity A heuristic MLsearch with TBR branch swapping was then conducted using pa-rameters determined for the best model of sequence evolution

Combined analyses were conducted with two different sam-pling schemes The first included only the 31 taxa for which se-quence information existed for all three genes The one exceptionis Aethionema which has a critical relationship of being sister toall other Brassicaceae s str (Galloway et al 1998) Aethionema wasrepresented as a composite terminal taxon based on Aethionemasaxatile and Aethionema grandiflora This approach was justified bythe strong sister relationship of the two taxa species with the ndhFanalyses However this reduced 31 taxon sample excluded somecritical taxa specifically Emblingia one species of Forchhammeriaand Tirania for which we were not able to obtain sequence datafor matK To evaluate the relationships of these taxa we conductedparsimony and maximum likelihood analyses on an expandedcombined analysis with 35 taxa to include all taxa for which therewas missing data (Aethionema Emblingia Forchhammeria pallida andTirania)

The incongruence length difference (ILD) test (Farris et al 19941995) implemented in PAUP as the partition homogeneity testwas conducted to measure conflict between the three data setsAlthough aspects of the ILD test have been criticized (Yoder et al2001 Barker and Lutzoni 2002 Darlu and Lecointre 2002) Hippet al (2004) have demonstrated that some of these criticisms areinvalid and argue that the ILD has merit as a first estimate ofcombinability of data sets A three-way test on parsimony-infor-mative characters of 1000 replicates (simple addition TBR branchswapping) was performed saving only 1000 trees per replicatePair-wise ILD tests were also conducted using the same searchstrategy above on each pair of data sets

Alternative a priori taxonomic hypotheses were explored by en-forcing topological constraints using 100 random addition repli-cates and calculating the number of additional steps required Al-ternative relationships suggested by individual but not the com-bined analyses were also examined in similar fashion As pres-ently no parsimony based statistical test of such tree differences(a priori vs a posteriori derived trees) are generally accepted(Goldman et al 2000) we implemented ML based tests of treedifferences using the Shimodaira and Hasegawa (1999) parametrictest (SH test) Following guidelines for tree selection in Buckley etal (2001 2002) a tree set including a number of the a priori tax-onomic hypotheses previously explored plus a posteriori ML treesfrom individual and combined analyses were evaluated with theSH test in PAUP with RELL optimization using ML settings forthe combined data set

Character Evolution Patterns of morphological evolution wereassessed in MacClade by mapping characters of interest onto themost parsimonious trees obtained from the combined parsimonyanalysis The following six characters were evaluated 1 habit (aannual herb b perennial herb c woody unordered) 2 carpelnumber (a two b three c 3 both ordered and unordered) 3locule number (a two b three c 3 both ordered and unor-dered) 4 sepal number (a four b four c five d six e irregular[unevenly lobed] unordered) 5 petal number (a four b fourc five d six e irregular [unevenly lobed] f absent unordered)and 6 breeding system (a hermaphroditic b monoecious c di-oecious d polygamous unordered) Morphological data were de-termined using field studies herbarium sheets and literature(Hansen 1977 Cronquist 1981 George 1982 2003 Rodman et al1991 Hufford 1996 Olson 2002a b Ronse Decraene 2002 RonseDecraene et al 1998 2002 Appel and Al-Shehbaz 2003 Appel and

Bayer 2003 Bayer and Appel 2003a-c Kubitzki 2003andashg) Charac-ter states were scored for all species present in the combined anal-yses of 35 taxa and then mapped onto the maximum likelihoodtree from that analysis using the TRACE CHARACTERS functionin MacClade which explores both Deltran and Acctran resolvingoptions The topologies of the maximum likelihood and maximumparsimony trees differed only with respect to the position of Pen-tadiplandra so both positions were explored when mapping char-acters In rare instances taxa were polymorphic but were scoredas having one state (see discussion for details)

RESULTS

Phylogenetic Relationships Although there areminor differences between individual and combinedanalyses there was very strong congruence in topol-ogies (Figs 2 3) All analyses support the core Brass-icales as a monophyletic clade Koeberlinia and Batisform a sister group to the core Brassicales (with theexception in rbcL analysis) and Caricaceae and Mor-ingaceae are sister to these two clades Within the coreBrassicales all families and genera with more than onespecies sampled (Capparaceae Cleomaceae Gyroste-monaceae Forchhammeria and Resedaceae) weremonophyletic Brassicaceae s lat form a monophyleticlineage with Cleomaceae and Brassicaceae s str as sis-ter groups Forchhammeria is consistently sister to Re-sedaceae with Gyrostemonaceae sister to these twoclades and is clearly outside the Capparaceae Al-though Tirania is most closely related to Resedaceaeand Forchhammeria the precise relationships amongthese three clades are unresolved For ease of discus-sion we refer to the lineage containing Gyrostemon-aceae Resedaceae Forchhammeria and Tirania as theGRFT clade Emblingia is sister to all other core Brass-icales in ndhF and rbcL analyses (not sampled formatK) Relationships of Tovariaceae and Pentadiplan-draceae within the core Brassicales vary among allanalyses with little support for alternative topologies

rbcL Analysis In parsimony analyses of the rbcLsequences of 35 taxa 48 most parsimonious trees oflength 817 on a single island were recovered (Table 2)The aligned length of the region is 1479 of which 242(164) characters were parsimony informative (Table2) and no indels were introduced during alignmentOnly partial sequences were obtained for Crateva pal-meri Polanisia dodecandra and Wislizenia refracta (withsequence lengths of 864 787 and 851 respectively)There are 142 cells missing in this data set All Gy-rostemonaceae share the same c-terminal amino acidsequence inferred from nucleotide sequences as pre-viously reported (Karol et al 1998) The inferred ami-no acid at positions 1426ndash1428 in the rbcL sequences ofEmblingia Forchhammeria watsonii F sessilifolia F spnov and Tirania is aspartic acid (rather than a stopcodon that is typical of many dicots) a synapomorphyfor the core Brassicales (Rodman et al 1994 1996 Ka-rol et al 1999) Unfortunately for most of these taxaour sequences ended at this codon position so we are


FIG 2 Phylogram results of parsimony analyses for individual and combined data sets showing only key relationshipswithin the core Brassicales Bootstrap values are above branches and for Forchhammeria Resedaceae and Gyrostemonaceae thevalues indicate monophyly for those clades (A) Consensus of the 48 trees based on analysis of 35 sequences of rbcL (B) Themost parsimonious tree in analyses of 35 ndhF sequences (C) Consensus of the matK data set of 31 taxa excluding Tirania andEmblingia (D) Eighty percent bootstrap consensus of expanded combined analysis (35 taxon sample) of rbcL ndhF and matKincluding taxa with missing sequences for matK (Tirania and Emblingia)

not able to examine the extent of the lsquotailrsquo The strictconsensus of the core Brassicales is shown with boot-strap values indicated (Fig 2A) There is low support(bootstrap 55) for the monophyly of the core Brass-icales including Emblingia and for relationships ofPentadiplandra Tovaria and Emblingia within the coreBrassicales (all 50 bootstrap) Emblingia is sister toall other core Brassicales whereas Tovaria and Pentadi-plandra are part of the GRFT clade Forchhammeria andTirania are strongly supported as closely related to Re-sedaceae (bootstrap 98) with the position of Tiraniaunresolved Gyrostemonaceae is weakly supported assister to Forchhammeria Tirania and Resedaceae (boot-strap 61) Among other members of Brassicales thereis more than one most parsimonious relationship hy-pothesized for Koeberlinia

Hierarchical likelihood ratio tests indicated that thebest model of substitution for the rbcL data is TrN 1I 1 G which allows one substitution rate for transver-sions two substitutions rates for transitions and

among site rate heterogeneity is modeled by allowingsome sites to be invariant (I) while the rest have ratesdrawn from a discrete approximation to a gamma dis-tribution with a single shape parameter alpha (G Ta-ble 2) The heuristic search converged on a single tree(ln L 52 667834464 tree not shown) that is nearlyidentical in topology to the consensus of the parsi-mony analysis In the ML tree Tirania is sister to Re-sedaceae and these two clades are sister to Forchham-meria Also within Cleomaceae Cleome pilosa and Cviridiflora are sister species whereas in the MP topologyC pilosa is sister to Podandrogyne The overall topologyof the ML tree is similar to analysis of rbcL by Karolet al (1999) with the exception of the sister relation-ship of Tovaria and Pentadiplandra recovered in Karol etal (1999)

ndhF Analysis One most parsimonious tree(length 1283) was recovered in parsimony analysis of35 ndhF sequences (Fig 2B) Six indels in lengths of 36 or 9 base pairs were introduced during sequence


FIG 3 The maximum likelihood tree of the 35 taxa combined rbcL ndhF and matK analysis including Tirania and Emblingiaunder the TVM 1 I 1 G model of DNA substitution The topology of this tree is almost identical to the maximum parsimonyanalyses with the exception of position of Pentadiplandra Branch lengths are proportional to the number of changes along eachbranch GRFT clade in shaded box

alignment The entire aligned data set is 1097 basepairs in length and 774 cells are scored as missing(Table 2) Aethionema saxatile is missing approximatelyone half of the sequence Excluding or including thescored indels even up to weight of 10 had no effect onresulting topology and only minor effect on relativebootstrap values The partition homogeneity test in-dicates rbcL and ndhF have similar phylogenetic struc-ture (P 5 0193) The following three relationships re-

covered in the ndhF analyses differ from rbcL (1) asister relationship of Maerua and Apophyllum (Cappar-aceae relationships not shown in figure) (2) Tovaria assister to Capparaceae Cleomaceae plus Brassicaceaeand (3) Tirania as sister to a well supported (bootstrap85) clade of Resedaceae plus Forchhammeria Emblin-gia is supported (bootstrap 87) as sister to remainingcore Brassicales The heuristic search under likelihoodsettings resulted in a single tree (ln L 52 804693736


TABLE 2 Characteristics of the three data sets and parsimony and maximum likelihood analyses 1 Analyses excluding indels 2

Model selected using hierarchical likelihood ratio tests in Modeltest (Posada and Crandall 1998)

rbcL ndhF1 matK1 Combined 1 Combined 2

Number of taxa 35 35 31 31 35Aligned length 1479 1097 1566 4142 4142

ParsimonyParsimony informative sites () 242 (164) 342 (312) 458 (292) 988 (239) 1031 (249)MP trees (length) 48 (817) 1(1283) 6 (1582) 2 (3494) 2 (3660)CI (RI) 0661 (0704) 0633 (0726) 0677 (0760) 0674 (0741) 0656 (0736)

Maximum likelihoodML model2 TrN 1 I 1 G GTR 1 G TVM 1 G TVM 1 I 1 G TVM 1 I 1 GLikelihood score (-ln L) 667834464 804693736 1049042846 2450801206 2539054641

Base frequenciesA 02845 03416 03272 03143 03147C 01886 01291 01463 01571 01577G 02397 01322 01453 01745 01723T 02872 03972 03812 03540 03553

Substitution ratesA-C 10000 14130 15828 14784 14329A-G 19635 21857 21531 21438 21309A-T 10000 02872 03107 04356 04261C-G 10000 21703 16964 14891 14609C-T 26116 18780 21531 21438 21309G-T 10000 10000 10000 10000 10000

Proportion invariant sites (I) 05418 0 0 02603 02906Gamma distribution shape

parameter (G) 07781 06765 11023 09474 10425

tree not shown) that is identical to the parsimony treeexcept the relationship of Tovaria is unresolved (thereis a hard polytomy) The best model of substitution forthe ndhF data set is GTR 1 G that allows for six dif-ferent substitution rates for all nucleotide pairs andamong site rate heterogeneity is approximated by agamma distribution (Table 2)

matK Analysis Six most parsimonious trees oflength 1582 resulted from analyses of 31 matK se-quences (Fig 2C Table 2) The aligned length of thedata set was 1566 base pairs with five indel events in-troduced For some taxa we were not able to obtaincomplete sequences and 544 of the data set is scoredas missing Including or excluding the scored indelsaltered topology and support of resulting trees Theinclusions of weighted indels dissolve the followingsister relationships (1) Moringa and Carica and (2) Ba-tis and Koeberlinia These relationships are well sup-ported in other analyses (Figs 2 3 Rodman et al1993 1998 Olson 2002a b) indicating the indels inmatK are homoplasious and justifiably excluded fromanalyses The relationships within the core Brassicalesand the placement of Forchhammeria are not alteredwith the inclusion of indels The resulting topology ofthe matK parsimony analysis is highly congruent withboth ndhF and rbcL analyses (Figs 2 3) Tovaria andPentadiplandra are both unresolved by matK analysesThe ILD test indicates there is similar phylogenetic

structure of matK with rbcL (P 5 0684) and with ndhF(P 5 0934) The best model of DNA substitution forthe matK data is the transversional model TVM 1 Gin which there are four different transversion rates onetransition rate and rate heterogeneity is approximatedusing the gamma distribution The topology of the sin-gle tree resulting from maximum likelihood analysesof the matK data shows a topology (ln L 521049042846 tree not shown) in which Pentadiplandra issister to the Capparaceae Cleomaceae and Brassica-ceae with Tovaria sister to these two clades

Combined Analysis The three-way partition ho-mogeneity test conducted on the smaller 31 taxonsample indicated all three data sets have similar phy-logenetic structure (P 5 0414) Parsimony analyses ontaxa for which sequences were available for all threeregions (with the exception of Aethionema treated as asingle taxon) resulted in two most-parsimonious trees(length 3494 tree not shown) which differ only in therelationships among species of Forchhammeria The to-pology is highly congruent with all individual analy-ses and bootstrap support for all branches is in-creased The combined analysis with 31 taxa suggestsrelationships of Tovaria and Pentadiplandra not indicatedin the individual analyses Tovaria is sister to all othercore Brassicales (bootstrap 71) whereas Pentadiplan-dra is sister to Capparaceae Cleomaceae and Brassi-caceae (bootstrap 60) In the expanded combined


analysis with 35 taxa two trees resulted in which theoverall topology is almost identical to the 31 taxonsample tree (Fig 2D) The two most parsimonioustrees differ with respect to Pentadiplandra One topol-ogy is identical to the 31 taxon sample tree topology(eg sister to Brassicaceae s lat) whereas the othertopology indicates Pentadiplandra is sister to the GRFTclade Tirania is placed in a polytomy with Resedaceaeand Forchhammeria and Emblingia is sister to remainingcore Brassicales (Fig 2D) Relationships of Tovaria andPentadiplandra are unresolved although there is weaksupport (bootstrap 60) for Tovaria as sister to all coreBrassicales other than Emblingia

The best model of DNA substitution for the com-bined data set is TVM 1 I 1 G (transversional model)in which there are four different transversion rates onetransition rate and among-site rate heterogeneity ismodeled by allowing some sites to be invariant whilethe rest have rates drawn from a discrete approxima-tion to a gamma distribution That model of evolutionis the same for both 31 and 35 taxa data sets (Table 2)and the 31 taxon tree is identical to the 35 taxon tree(Fig 3) Tirania and Emblingia have the same relation-ships as the rbcL ndhF and combined parsimony anal-yses The precise placement of Tirania is unresolved bylikelihood analyses

DISCUSSION

Data from more than 60 new rbcL ndhF and matKsequences individually and in combination resolvemany relationships within the core Brassicales andhelp place Emblingia Forchhammeria and Tirania Sev-eral results confirm relationships within Brassicalesand core Brassicales indicated in previous analyses(Rodman et al 1993 1994 1996 1998 Hall et al 2002)but with increased support Novel phylogenetic resultsthat emerge from these studies include (1) Emblingiais sister to remaining core Brassicales (2) Gyrostemon-aceae Resedaceae Forchhammeria and Tirania form awell-supported clade within the core Brassicales and(3) the relationships of Pentadiplandra and Tovaria with-in the core Brassicales remain unresolved Mapping ofmorphological traits onto phylogenetic trees indicatesthat many of the characters used to classify these taxaare prone to convergent evolution Although many pat-terns of morphological evolution remain elusive someinsight is gained on the evolution of floral merosityand breeding systems Two other orphan genera ofCapparaceae s lat Stixis and Neothorelia which havebeen placed in the tribe Stixeae clearly need to be sam-pled

Monophyly of the Core Brassicales The core Brass-icales are a strongly supported clade (bootstrap 100in combined analyses) comprising Brassicaceae Cap-paraceae Cleomaceae Emblingiaceae ForchhammeriaGyrostemonaceae Pentadiplandraceae Resedaceae

Tirania and Tovariaceae The GRFT clade of Gyroste-monaceae Resedaceae Forchhammeria and Tirania is anovel lineage established here with problematic affin-ities to Pentadiplandra and Tovaria Ronse Decraene(2002) suggested that racemose inflorescences (andro)gynophore extrastaminal nectary tendency to dissym-metry reduced stipules and imbricate sepals and pet-als unite the core Brassicales Other characters that arecommon but not necessarily synapomorphies includecampylotropous seed orientation (Hufford 1996) withthe exception of Forchhammeria (Hansen 1977) anom-ocytic stomates and curved embryos (also seen in Koe-berlinia)

Emblingia is Sister to Remainder of CoreBrassicales The studies presented here suggest anovel relationship of Emblingia as sister to remainingcore Brassicales A previous molecular study usingrbcL placed Emblingia in Brassicales as sister to Rese-daceae but with low support (bootstrap 50 Chan-dler and Bayer 2000) The lack of resolution in theirstudy was likely the result of a very broad samplingacross many orders that had been suggested to houseEmblingia (Table 1 Chandler and Bayer 2000) limitedsampling within core Brassicales and reliance solelyon rbcL sequences Although Emblingia has manyunique morphological features there are some featuresin common with other members of core Brassicalesandrogynophore (Erdtman et al 1969 Chandler andBayer 2000) curved or reniform seeds tricolporate pol-len stamens usually 4-merous (or 8) and zygomorphicflowers To our knowledge no studies have been con-ducted to examine the presence of mustard oils (glu-cosinolates) in Emblingia

Positions of Pentadiplandra and Tovaria are notResolved Although there were 1031 parsimony in-formative characters in the combined 35 taxon analy-sis some relationships within core Brassicales are stillnot resolved Every analysis suggested different rela-tionships of Tovaria and Pentadiplandra each with lowstatistical support (Figs 2 3) Since both families aremonotypic (Tovaria questionably ditypic) increasedsampling is not a viable approach for resolving theserelationships Both genera have been classified as Cap-paraceae in the broad sense by some taxonomists (Paxand Hoffmann 1936 Carlquist 1985 Thorne 1992) al-though Hutchinson (1973) placed Pentadiplandra nearthe Celastraceae Ronse Decraene (2002) suggested astrong affinity of Tovaria and Pentadiplandra based onthe presence of plesiomorphic characters within Brass-icales such as stipulate leaves pentamery diploste-mony and axile placentation However none of theanalyses presented here indicate a sister relationshipof these two genera (but see Rodman et al 1998 Karolet al 1999)

Forchhammeria and Tirania are Related to Gyros-temonaceae and Resedaceae The clade comprising


Gyrostemonaceae Resedaceae Forchhammeria and Tir-ania (GRFT clade) is moderately (bootstrap 61 inrbcL) to strongly supported (bootstrap 94ndash99 in otheranalyses) Gyrostemonaceae here represented by threespecies is sister to all remaining members of the cladeDespite the strong support for Forchhammeria Tiraniaand Resedaceae as a clade there is no consensus onhow these three lineages are related The combineddata presented here indicate a sister relationship ofForchhammeria and Resedaceae a relationship furthersupported by a unique 9bp matK deletion In a previ-ous analysis using only trnL-trnF data (Fig 3 in Hallet al 2002) Forchhammeria is placed sister to Reseda-ceae plus Gyrostemonaceae (bootstrap 83) Thisanalysis however was conducted with only one rep-resentative from each of the three lineages and limitedsampling may be an issue In order to evaluate thesignificance of these differences the ML based Shi-modaira-Hasegawa test was conducted on combinedand the trnL-trnF data sets (trnL-trnF data set was ex-panded to include all the same representatives of theGRFT clade J Hall unpublished data) under ML pa-rameters determined by ModelTest (Posada and Cran-dall 1998) Our a priori hypotheses included 15 topol-ogies representing all possible relationships among thefour members of the GRFT clade The combined dataset rejects all topologies (P 5 0001) except those inwhich Gyrostemonaceae are sister to all other mem-bers of the GRFT clade (P 5 0896ndash0914) In contrastthe trnL-trnF data set fails to reject any of the 15 to-pologies (P 5 0158ndash0569) These analyses indicatethat the sister relationship of Forchhammeria and Re-sedaceae suggested by combined matK rbcL and ndhFis more likely than that indicated by the previous trnL-trnF analysis based on fewer taxa

Although sampling is limited among some membersof the GRFT clade all genera or families with morethan one species sampled are strongly supported asmonophyletic with clear putative morphological syn-apomorphies Resedaceae have a syncarpous gynoeci-um that remains open at the distal end throughout de-velopment (Cronquist 1981 Hufford 1996 Kubitzki2003f) although the family exhibits variation in carpelnumber The massive embryos of Forchhammeria arepseudo-monocotyledonous (Hansen 1977) where onlythe outer cotyledon develops into a fat folded nutrientstorage structure completely enveloping the inner un-developed cotyledon and with the radicle highly re-duced (except Forchhammeria sp nov) Female flowersof Gyrostemonaceae are apopcarpic with carpels ad-nate to a central column that is often expanded at thetop (Cronquist 1981 Hufford 1996) In fact the unusu-al syncarpy of Gyrostemonaceae appears to be asso-ciated with the formation of a large sterile region inthe apex of flowers (Hufford 1996)

Despite the lack of a clear morphological synapo-

morphy for the GRFT clade members of the clade havesome morphological similarities or trends Gyroste-monaceae Forchhammeria and Ochradenus (Resedaceaeunsampled in these studies) all have uniseriate peri-anths grow in arid habitats and are anemophilous(Hansen 1977 Hufford 1996) The flowers of Forchham-meria like Gyrostemonaceae are much reduced with afused calyx that splits in an irregular fashion Gyros-temonaceae and Forchhammeria are dioecious withhighly reduced flowers presumably adapted for windpollination However many of these features may bethe result of convergent evolution due to dioecy andarid habitat Tirania and Resedaceae (excluding Ochra-denus) are the only members of the clade that have pet-als Of some interest is that both genera sampled fromthe tribe Stixeae of Capparaceae s lat (Pax and Hoff-mann 1936) Forchhammeria and Tirania are closely re-lated to one another and Resedaceae Members of thetribe Stixeae have two to multilocular ovaries axileplacentation three- to five-merous perianth and welldeveloped stigmas (Pax and Hoffmann 1936 Kers2003) which support their placement within the GRFTclade (Fig 4) Precisely how Forchhammeria and Tiraniaare related to one another is still unresolved based onrbcL and ndhF sequence data (Figs 2 3) Additionalsampling of remaining Stixeae (ie Neothorelia andStixis) is warranted based on the clear separation ofthese two genera from Capparaceae s lat and theirclose relationship

Morphological Evolution Mapping of characterstates revealed few clear patterns of morphologicalevolution within the Brassicales Figure 4 maps breed-ing systems and shows selected habit and floral char-acters (mapping of these characters can be obtainedfrom authors upon request) Moving Pentadiplandra assister to the GRFT clade does not change the recon-struction of any of the characters (typically the branchis equivocal in either relationship) with the exceptionof breeding systems Habit is the only morphologicalcharacter for which there are no equivocal branches inthe reconstruction Woodiness is plesiomorphic inBrassicales and based on current sampling the her-baceous habit has arisen four times (1) Cleomaceaeplus Brassicaceae (2) Oligomeris (Resedaceae) (3) Ter-sonia (Gyrostemonaceae) and (4) Tropaeolum (Tropaeo-laceae)

Merosity of both the gynoecium and perianth is verylabile within the Brassicales (Fig 4) When designatedas unordered the plesiomorphic state of locule andcarpel number is equivocal When ordered three car-pels are basal within the Brassicales but the base ofthe core Brassicales is still equivocal Within the coreBrassicales two carpels characterize the CapparaceaeCleomaceae and Brassicaceae clade in addition toForchhammeria and Gyrostemon tepperi Whether treatedas ordered or unordered the primitive locule number


FIG 4 Optimization mapping of breeding systems on the maximum likelihood tree from the combined 35 taxa analysisCharacter states for habit sepal number petal number carpel number and locule number are given W5 woody perennial P5 herbaceous perennial H 5 annual herbaceous I 5 irregular

for the core Brassicales is equivocal Within the coreBrassicales two locules is characteristic of both Bras-sicaceae sstr and Forchhammeria The evolution of peri-anth number is equally ambiguous although merositycharacterizes a few clades Within the core Brassicalesall dioecious plants lack petals and all but one haveirregular sepals There is no clear pattern of sepalnumber within the GRFT clade except that in generalthey have irregular splitting sepals The equivocal sta-tus of many of the characters at the base of the GRFTclade is due to the unresolved nature of Tirania

The ancestral condition of breeding systems for thecore Brassicales and Brassicales in the broad sense is

bisexual (Fig 4) Based on this sampling dioecyevolved unambiguously from the hermaphroditic con-dition three times (Carica Batis Apophyllum) and mon-oecy or polygamy evolved twice (Crateva Pentadiplan-dra) also unambiguously The evolution of dioecywithin the GRFT clade is ambiguous as a result of thelack of resolution among Forchhammeria Tirania andResedaceae Alternative topologies of these three taxawere explored If Forchhammeria is sister to Tirania plusResedaceae and Pentadiplandra is sister to the GRFTclade then monoecy dioecy and hermaphroditism areall possible ancestral conditions In all other arrange-ments of Forchhammeria Tirania and Resedaceae (re-


gardless of placement of Pentadiplandra) the ancestralcondition is hermaphroditism with two independentorigins of dioecy

Biogeographical Relationships Many members ofthe core Brassicales and the GRFT clade in particularhave disjunct distributions (Fig 1) There is a strongAustral-Asian component in the clade as well as a pro-pensity for Mediterranean habitats The majority ofGyrostemonaceae are endemic to southwestern Aus-tralia occur in eastern Australia (Codonocarpus attenu-atus) or throughout central Australia (Codonocarpus co-tinifolius) (George 1982 2003) Resedaceae are primar-ily in arid regions of the Mediterranean with outliersfrom Oligomeris in South Africa and southwesternNorth America (Kubitzki 2003f) Tirania is endemic tosouthern Vietnam (Hutchinson 1967 Kers 2003) Forch-hammeria is distributed in Mexico Central Americaand the West Indies (Hansen 1977) Hansen (1977) sug-gested that Forchhammeria was closely related to someAfrican genera of Capparaceae (Boscia and Maerua)and that the current distribution of the genus was theresult of long distance dispersal Given the relation-ships presented here using chloroplast sequence datathe biogeographic history of Forchhammeria needs to beexamined instead relative to the temperate ResedaceaeAustralian Gyrostemonaceae and Asiatic TiraniaThese biogeographical relationships are not clearly re-solved in the GRFT clade and become increasingly lessclear when the distributions of other core Brassicalesare considered Pentadiplandra of tropical west AfricaTovaria of tropical America Emblingia of Australia andthe worldwide distributed Brassicaceae Cleomaceaeand Capparaceae A preliminary molecular clock dat-ing of these lineages (Hall 2003) suggests that the coreBrassicales diversified during the late Cretaceous andearly Tertiary but more detailed analyses are neededto understand biogeographical relationships andevents within the core Brassicales in the context of con-tinental separations and contacts

Floral Divergence and Convergence in CoreBrassicales One result of using a molecular phylo-genetic framework to address patterns of morpholog-ical evolution in the Brassicales is the demonstration ofthe diversity and lability of floral merosity in the orderRecurring or convergent shifts in sepal petal carpeland locule number are common especially within thecore Brassicales (Fig 4) In assigning character statesto taxa some of this diversity was necessarily simpli-fied from even greater complexity in floral evolution ofBrassicales Resedaceae were scored as having greaterthan three carpels (plesiomorphic condition) althoughthe family varies in carpel number between two andseven Reseda has three to four whereas Oligomeris hasfour to five (Cronquist 1981 Kubitzki 2003f) Also al-though flowers of Resedaceae are clearly hermaphro-ditic (plesiomorphic as scored here) apparently de-

rived unisexual forms occur (Cronquist 1981 Kubitzki2003f) Tovaria was scored with a merosity number ofsix but merosity varies between six and eight (RonseDecraene 2002)

Patterns of morphological evolution are limited bytaxon sampling in the analyses Although Batis isscored as dioecious the family is ditypic and the otherspecies not sampled in these analyses is monoecious(Bayer and Appel 2003b) Gyrostemon sp (probably anew species see Rodman et al 1994) was scored basedon generic characters and the typical carpel numberfor most Gyrostemonaceae is six to eight (Hufford1996) Gyrostemon tepperi typically has one to two car-pels which likely represents an evolutionary reduction(Hufford 1996) Assuming that this assessment is cor-rect and the reduced carpel number in G tepperi isderived an alternative scoring of G tepperi was imple-mented with a high carpel number for the family Withthis scoring the GRFT clade (regardless of the positionof Pentadiplandra) has a plesiomorphic condition of highcarpel number with a reduction in Forchhammeria Infact the entire core Brassicales has a plesiomorphiccharacter state of greater than three carpels with tworeductions (1) Forchhammeria and (2) Brassicaceae plusCleomaceae plus Capparaceae

Floral dimorphism (dioecy monoecy or polygamy)evolved minimally eight times within Brassicalesbased on current sampling (Fig 4) This is likely anunderestimate of change in breeding systems becausethere are many unsampled species in Capparaceae andCleomaceae that are either monoecious or polygamousDioecy has evolved four to five times within Brassi-cales Carica papaya (Caricaceae) Batis maritima (Bata-ceae) Apophyllum (Capparaceae) Gyrostemonaceaeand Forchhammeria The transition to dioecy is from anhermaphroditic ancestor except for Forchhammeria andGyrostemonaceae where the ancestral state is equiv-ocal Because the sister relationship of Forchhammeriaand Gyrostemonaceae is rejected by the Shimodaira-Hasegawa test two independent origins of dioecyfrom hermaphroditism are likely The shifts to dioecyfrom hermaphroditism differ from the general pattern(Renner and Ricklefs 1995) but are similar to the pat-tern of breeding system shifts within monocots (Wei-blen et al 2000) Generally hypotheses of evolutionarypathways to dioecy include gynodioecious or monoe-cious intermediates (Charlesworth and Charlesworth1978 Sytsma et al 1991 Barrett 2002) The direct tran-sition between hermaphroditism and dioecy may bespurious as transitions can be missed when looking atphylogenetic reconstructions (Weiblen et al 2000)

Dioecy has been associated with many characteris-tics in flowering plants including wind pollination pe-rennial growth and fleshy fruits (Givnish 1980 Ren-ner and Ricklefs 1995) which can convolute homologyassessment A change to drier habitats can also result


in subsequent shifts to wind pollination due to loss ofpollinators (Sakai et al 1995) Initial loss of insect pol-lination may lead to selfing and inbreeding depressionwhich in turn may favor the spread of females (Char-lesworth and Charlesworth 1978) Four of the five di-oecious lineages sampled occur in dry or saline habi-tats suggesting a correlation with water stressed hab-itats (Fig 4 see also Sakai et al 1995) Nonethelesspresence in dry habitats may occur without changes inreproductive systems since the hermaphroditic Rese-daceae are distributed in arid regions Alternatively ifthe ancestral condition of the GRFT clade was dioe-cious then reversal to hermaphroditism occurs with-out changes to more mesic habitats Perianth reductionis often associated with wind pollination which is cen-tral to this hypothesis on the evolution of dioecy (Sakaiet al 1995) Renner and Ricklefs (1995) argued thatabiotic pollination favors unisexual flowers (less likelyto self pollinate) which facilitates the transition frommonoecy to dioecy Four of the dioecious groups (BatisApophyllum Forchhammeria and Gyrostemonaceae) lackpetals Developmental studies support independent or-igins of sexual systems which correspond to correla-tions of drier habitat and wind pollination Flowers ofPentadiplandra and Forchhammeria are unisexual as theresult of organ abortion (Hansen 1977 Ronse Decraene2002) whereas there is no indication of staminal ini-tiation in floral development studies of female flowersof Gyrostemon and Tersonia (Gyrostemonaceae Huf-ford 1996)

Taxonomic Implications As suggested by Kers(2003) Tirania and Forchhammeria should no longer beclassified as Capparaceae s lat Although we haveclearly demonstrated that these taxa belong to theGRFT clade not the Brassicaceae s lat lineage (Cap-paraceae Cleomaceae Brassicaceae) we hesitate tomake any formal taxonomic changes at this point intime First further sampling of members of tribe Stix-eae that are likely to be part of the GRFT clade basedon morphological features are needed Second the ex-act placement of Tirania relative to Forchhammeria hasnot been established As argued earlier (Chandler andBayer 2000) Emblingia is distinct and should be rec-ognized as a monotypic family Likewise Pentadiplan-dra and Tovaria are each morphologically distinct andisolated within the core Brassicales and merit familialstatus With additional recognition of a family forForchhammeria (with or without other Stixeae) the coreBrassicales would be composed of a high number ofsmall families Despite discouragement of such prac-tice (APG 1998) the core Brassicales is a clade thatmerits a number of small families as the lineages arephylogenetically isolated morphologically distinctand geographically disjunct (see Iltis 1999)

ACKNOWLEDGEMENTS The authors wish to thank the curatorsof Wisconsin State Herbarium (WIS) Missouri Botanical Garden

Herbarium (MO) and Gray Herbaria (GH) for access to herbariumleaf tissue and collections We also wish to thank J Rodman KKarol and C Quinn for sharing DNAs M Fishbein for providingleaf material L Kers for first suggesting Forchhammeria might beplaced outside of Capparaceae based on morphological evidenceK Elliot for graphical support S Smith and M S Heschel forconstructive comments the systematics section of the Botany De-partment for discussions and J Rodman and an anonymous re-viewer for their helpful comments We gratefully acknowledge thefinancial supports of NSF grant DEB-0073204 Garden Club ofAmerica Scholarship in Tropical Botany American Society of PlantTaxonomists graduate student award Society of Systematic Biol-ogists graduate student award and the Davis Fund from the Bot-any Department University of Wisconsin

LITERATURE CITED

AIRY SHAW H K 1965 Diagnoses of new families new namesetc for the seventh edition of Willisrsquos lsquoDictionaryrsquo Kew Bul-letin 18 249ndash273

ALVERSON W S KAROL K G D A BAUM M W CHASE S MSWENSEN R MCCOURT and K J SYTSMA 1998 Circumscrip-tion of the Malvales and relationships to other Rosidae evi-dence from rbcL sequence data American Journal of Botany 85876ndash887

mdashmdashmdash R NYFFELER B WHITLOCK C BAYER and D A BAUM1999 Phylogenetic analysis of the core Malvales based onsequences of ndhF American Journal of Botany 86 1474ndash1486

ANDERSSON L and S ANDERSSON 2000 A molecular phylogenyof Tropaeolaceae and its systematic implications Taxon 49721ndash736

APG (ANGIOSPERM PHYLOGENY GROUP) 1998 An ordinal classi-fication for the families of flowering plants Annals of the Mis-souri Botanical Garden 85 531ndash553

mdashmdashmdash 2003 An update of the Angiosperm Phylogeny Groupclassification for the orders and families of flowering plantsAPG II Botanical Journal of the Linnean Society 141 399ndash436

APPEL O and I A AL-SHEHBAZ 2003 Cruciferae Pp 75ndash174 inThe families and genera of vascular plants V Flowering plantsdicotyledons Malvales Capparales and non-betalain Caryophylla-les eds K Kubitzki and C Bayer Berlin Springer

mdashmdashmdash and C BAYER 2003 Tovariaceae Pp 397ndash399 in The familiesand genera of flowering plants V Flowering plants DicotyledonsMalvales Capparales and non-betalain Caryophyllales eds K Ku-bitzki and C Bayer Berlin Springer

APPELQUIST W L and R S WALLACE 2001 Phylogeny of theportulaceous cohort based on ndhF sequence data SystematicBotany 26 409ndash416

BARKER F K and F M LUTZONI 2002 The utility of the incon-gruence length difference test Systematic Biology 51 625ndash637

BARRETT S C H 2002 The evolution of plant sexual diversityNature Reviews Genetics 3 274ndash284

BAUM D A K J SYTSMA and P C HOCH 1994 A phylogeneticanalysis of Epilobium (Onagraceae) based on nuclear ribosom-al DNA sequences Systematic Botany 26 406ndash419

BAYER C and O APPEL 2003a Pentadiplandraceae Pp 329ndash331in The families and genera of vascular plants V Flowering plantsdicotyledons Malvales Capparales and non-betalain Caryophylla-les eds K Kubitzki and C Bayer Berlin Springer

mdashmdashmdash and mdashmdashmdash 2003b Akaniaceae Pp 21ndash24 in The familiesand genera of vascular plants V Flowering plants dicotyledonsMalvales Capparales and non-betalain Caryophyllales eds K Ku-bitzki and C Bayer Berlin Springer

mdashmdashmdash and mdashmdashmdash 2003c Bataceae Pp 30ndash32 in The families andgenera of vascular plants V Flowering plants dicotyledons Mal-vales Capparales and non-betalain Caryophyllales eds K Ku-bitzki and C Bayer Berlin Springer

BUCKLEY T R P ARENSBURGER C SIMON and G K CHAMBERS


2002 Combined data Bayesian phylogenetics and the originof the New Zealand cicada genera Systematic Biology 51 4ndash18

mdashmdashmdash C SIMON H SHIMODAIRA and G K CHAMBERS 2001Evaluating hypotheses on the origin and evolution of theNew Zealand alpine cicadas (maoricicada) using multiple-comparison tests of tree topology Molecular Biology and Evo-lution 18 223ndash234

CARLQUIST S 1985 Vegetative anatomy and familial placement ofTovaria Aliso 11 69ndash76

CHANDLER G T and R J BAYER 2000 Phylogenetic placement ofthe enigmatic Western Australian genus Emblingia based onrbcL sequences Plant Species Biology 15 67ndash72

CHARLESWORTH B and D CHARLESWORTH 1978 A model for theevolution of dioecy and gynodioecy American Naturalist 112975ndash997

CONTI E T ERIKSSON J SCHONENBERGER K J SYTSMA and DA BAUM 2002 Early Tertiary out-of-India dispersal of Cryp-teroniaceae evidence from phylogeny and molecular datingEvolution 56 1931ndash1942

mdashmdashmdash A LITT and K J SYTSMA 1996 Circumscription of Myr-tales and their relationships to other rosids evidence fromrbcL sequence data American Journal of Botany 83 221ndash233

mdashmdashmdash mdashmdashmdash P G WILSON S A GRAHAM B G BRIGGS L AS JOHNSON and K J SYTSMA 1997 Interfamilial relation-ships in Myrtales molecular phylogeny and patterns of mor-phological evolution Systematic Botany 22 629ndash647

CRONQUIST A 1981 An integrated system of classification of floweringplants Columbia University Press New York New York USA

DARLU P and G LECOINTRE 2002 When does the incongruencelength difference test fail Molecular Biology and Evolution 19432ndash437

DICKISON W C and R B MILLER 1993 Morphology and anatomyof the malagasy genus Physena (Physenaceae) with a discus-sion of the relationships of the genus Bulletin du MuseumNational drsquoHistoire Naturelle Section B Adansonia Botanique Phy-tochimie 15 85ndash106

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

ENDRESS P K 1992 Evolution and floral diversity the phyloge-netic surroundings of Arabidopsis and Antirrhinum Interna-tional Journal of Plant Sciences 153 S106ndashS122

ERDTMAN G P P LEINS R MELVILLE and C R METCALFE 1969On the relationships of Emblingia Botanical Journal of the Lin-nean Society 62 169ndash186

FARRIS J S M KALLERSJO A G KLUGE and C BULT 1994 Test-ing significance of incongruence Cladistics 10 315ndash319

mdashmdashmdash mdashmdashmdash mdashmdashmdash and mdashmdashmdash 1995 Constructing a signifi-cance test for incongruence Systematic Biology 44 570ndash572

FELSENSTEIN J 1985 Confidence limits on phylogenies an ap-proach using the bootstrap Evolution 39 783ndash791

FITCH W M 1971 Toward defining the course of evolution min-imum change for a specific tree topology Systematic Zoology20 406ndash416

GADEK P A C J QUINN J E RODMAN K G KAROL E CONTIR A PRICE and E S FERNANDO 1992 Affinities of the Aus-tralian endemic Akaniaceae new evidence from rbcL se-quences Australian Systematic Botany 5 717ndash724

GALLOWAY G L R L MALMBERG and R A PRICE 1998 Phy-logenetic utility of the nuclear gene arginine decarboxylasean example from Brassicaceae Molecular Biology and Evolution15 1312ndash1320

GEORGE A S 1982 Gyrostemonaceae Flora of Australia Lecythi-dales to Batales volume 8 Canberra Australian GovernmentPublishing Service

mdashmdashmdash 2003 Gyrostemonaceae Pp 213ndash217 in The families andgenera of vascular plants V Flowering plants dicotyledons Mal-

vales Capparales and non-betalain Caryophyllales eds K Ku-bitzki and C Bayer Berlin Springer

GIVNISH T G 1980 Ecological constraints on the evolution ofbreeding systems in seed plants dioecy and dispersal ingymnosperms Evolution 34 959ndash972

GOLDMAN N J P ANDERSON and A G RODRIGO 2000 Likeli-hood-based tests of topologies in phylogenetics SystematicBiology 49 652ndash670

HALL J C 2003 Systematics and floral evolution of Capparaceaeand other core Brassicales PhD Thesis University of Wis-consinmdashMadison

mdashmdashmdash K J SYTSMA and H H ILTIS 2002 Phylogeny of Cap-paraceae and Brassicaceae based on chloroplast sequencedata American Journal of Botany 89 1826ndash1842

HANSEN B F 1977 A monograph of Forchhammeria (Capparidaceae)MS Thesis University of WisconsinmdashMadison

HIPP A L J C HALL and K J SYTSMA 2004 Congruence versusphylogenetic accuracy revisiting the incongruence lengthdifference test Systematic Biology 53 81ndash89

HUFFORD L 1996 Developmental morphology of female flowersof Gyrostemon and Tersonia and floral evolution among Gy-rostemonaceae American Journal of Botany 83 1471ndash1487

HUTCHINSON J 1967 The genera of flowering plants Oxford Clar-endon Press

mdashmdashmdash 1973 The families of flowering plants Dicotyledons arrangedaccording to a system based on their probably phylogeny OxfordClarendon Press

ILTIS H H 1999 Setchellanthaceae (Capparales) a new family fora relictual glucosinolate-producing endemic of the Mexicandeserts Taxon 48 257ndash275

JUDD W S R W SANDERS and M J DONOGHUE 1994 Angio-sperm family pairs preliminary phylogenetic analyses Har-vard Papers in Botany 5 1ndash51

KAROL KG JE RODMAN E CONTI and K J SYTSMA 1999 Nu-cleotide sequence of rbcL and phylogenetic relationships ofSetchellanthus caeruleus (Setchellanthaceae) Taxon 48 303ndash315

KERS L E 2003 Capparaceae Pp 36ndash56 in The families and generaof vascular plants V Flowering plants dicotyledons Malvales Cap-parales and non-betalain Caryophyllales eds K Kubitzki and CBayer Berlin Springer

KOCH M B HAUBOLD and T MITCHELL-OLDS 2001 Molecularsystematics of the Brassicaceae evidence from plastidic matKand nuclear Chs sequences American Journal of Botany 88 534ndash544

KUBITZKI K 2003a Emblingiaceae Pp 206ndash207 in The families andgenera of vascular plants V Flowering plants dicotyledons Mal-vales Capparales and non-betalain Caryophyllales eds K Ku-bitzki and C Bayer Berlin Springer

mdashmdashmdash 2003b Introduction to Capparales Pp 7ndash11 in The familiesand genera of vascular plants V Flowering plants dicotyledonsMalvales Capparales and non-betalain Caryophyllales eds K Ku-bitzki and C Bayer Berlin Springer

mdashmdashmdash 2003c Caricaceae Pp 57ndash61 in The families and genera ofvascular plants V Flowering plants dicotyledons Malvales Cap-parales and non-betalain Caryophyllales eds K Kubitzki and CBayer Berlin Springer

mdashmdashmdash 2003d Koeberliniaceae Pp 218ndash219 in The families and gen-era of vascular plants V Flowering plants dicotyledons MalvalesCapparales and non-betalain Caryophyllales eds K Kubitzkiand C Bayer Berlin Springer

mdashmdashmdash 2003e Moringaceae Pp 312ndash314 in The families and generaof vascular plants V Flowering plants dicotyledons Malvales Cap-parales and non-betalain Caryophyllales eds K Kubitzki and CBayer Berlin Springer

mdashmdashmdash 2003f Resedaceae Pp 334ndash338 in The families and genera ofvascular plants V Flowering plants dicotyledons Malvales Cap-parales and non-betalain Caryophyllales eds K Kubitzki and CBayer Berlin Springer


mdashmdashmdash 2003g Setchellanthaceae Pp 353ndash354 in The families andgenera of vascular plants V Flowering plants dicotyledons Mal-vales Capparales and non-betalain Caryophyllales eds K Ku-bitzki and C Bayer Berlin Springer

LES D H 1994 Molecular systematics and taxonomy of lake cress(Neobeckia aquatica Brassicaceae) an imperiled aquatic mus-tard Aquatic Botany 49 149ndash165

MADDISON D R and W P MADDISON 2002 MacClade 4 analysisof phylogeny and character evolution Sunderland SinauerAssociates

MORTON CM KG KAROL and M W CHASE 1997 Taxonomicaffinities of Physena (Physenaceae) and Asteropeia (Theaceae)Botanical Review 63 231ndash239

OLMSTEAD R G J A SWEERE and K H WOLFE 1993 Ninetyextra nucleotides in the ndhF gene of tobacco chloroplastDNA a summary of revisions to the 1986 genome sequencePlant Molecular Biology 22 1191ndash1193

OLSON M E 2002a Combining data from DNA sequences andmorphology for a phylogeny of Moringaceae (Brassicales)Systematic Botany 27 55ndash73

mdashmdashmdash 2002b Intergeneric relationships within the Caricaceae-Moringaceae clade (Brassicales) and potential morphologicalsynapomorphies of the clades and its families InternationalJournal of Plant Sciences 163 51ndash65

PAX F and K HOFFMANN 1936 Capparidaceae Pp 146ndash233 inDie naturlichen pflanzenfamilien 17b eds A Engler and KPrantl Leipzig Engelmann

POSADA D and K A CRANDALL 1998 Modeltest testing themodel of DNA substitution Bioinformatics 14 817ndash818

PRICE R A and J D PALMER 1993 Phylogenetic relationships ofthe Geraniaceae and Geraniales from rbcL sequence compar-isons Annals of the Missouri Botanical Garden 80 661ndash671

RAMBAUT A 2001 Sequence alignment editor v20a6 httpevolvezoooxacuk

RENNER S S and R E RICKLEFS 1995 Dioecy and its correlatesin the flowering plants American Journal of Botany 82 596ndash606

RODMAN J E 1991 A taxonomic analysis of glucosinolate-pro-ducing plants part 2 cladistics Systematic Botany 16 619ndash629

mdashmdashmdash K G KAROL R A PRICE E CONTI and K J SYTSMA1994 Nucleotide sequences of rbcL confirm the Capparaleanaffinity of the Australian endemic Gyrostemonaceae Austra-lian Systematic Botany 7 57ndash69

mdashmdashmdash mdashmdashmdash mdashmdashmdash and K J SYTSMA 1996 Molecules mor-phology and Dahlgrenrsquos expanded order Capparales System-atic Botany 21 289ndash307

mdashmdashmdash R A PRICE K KAROL E CONTI K J SYTSMA and J DPALMER 1993 Nucleotide sequences of the rbcL gene indicatemonophyly of mustard oil plants Annals of Missouri BotanicalGarden 80 686ndash699

mdashmdashmdash P S SOLTIS D E SOLTIS K J SYTSMA and K G KAROL1998 Parallel evolution of glucosinolate biosynthesis inferredfrom congruent nuclear and plastid gene phylogenies Amer-ican Journal of Botany 85 997ndash1006

RONSE DE CRAENE L P 2002 Floral development and anatomyof Pentadiplandra (Pentadiplandraceae) a key genus in theidentification of floral morphological trends in the core Brass-icales Canadian Journal of Botany 80 443ndash459

mdashmdashmdash J DE LAET and E F SMETS 1998 Floral development andanatomy of Moringa oleifera (Moringaceae) what is the evi-dence for a Capparalean or Sapindalean affinity Annals ofBotany 82 273ndash284

mdashmdashmdash T Y A YANG P SCHOLS and E F SMETS 2002 Floralanatomy and systematics of Bretschneidera (Bretschneidera-ceae) Botanical Journal of the Linnean Society 139 29ndash45

SAKAI A K W L WAGNER D M FERGUSON and D R HERBST

1995 Biogeographical and ecological correlates of dioecy inthe Hawaiian flora Ecology 76 2530ndash2543

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

SMITH J F K J SYTSMA J S SHOEMAKER and R L SMITH 1991A qualitative comparison of total cellular DNA extractionsprotocols Phytochemical Bulletin 23 2ndash9

SWOFFORD D L 2002 PAUP Phylogenetic analysis using parsi-mony (and other methods) Version 4 Sunderland SinauerAssociates

SYTSMA K J J F SMITH and P E BERRY 1991 The use of chlo-roplast DNA to assess biogeography and evolution of mor-phology breeding systems and flavonoids in Fuchsia sectSkinnera (Onagraceae) Systematic Botany 16 257ndash269

mdashmdashmdash J MORAWETZ J C PIRES M NEPOKROEFF E CONTI MZJHRA J C HALL and M W CHASE 2002 Urticalean rosidscircumscription rosid ancestry and phylogenetics based onrbcL trnL-F and ndhF sequences American Journal of Botany89 1531ndash1546

mdashmdashmdash A LITT M L ZJHRA J C PIRES M NEPOKROEFF E CON-TI and P WILSON (Accepted pending revision) Cladesclocks and continents historical and biogeographical anal-ysis of Myrtaceae Vochysiaceae and relatives in the southernhemisphere International Journal of Plant Sciences

THORNE R F 1992 Classification and geography of the floweringplants Botanical Review 58 225ndash348

VILLIERS J F 1973 Pentadiplandraceae In A Audbreville and JF Leroy [eds] Flore du Gabon vol 15 pp 163ndash170 Paris Mu-seum National drsquoHistoire Naturelle

WEIBLEN G D R K OYAMA and M J DONOGHUE 2000 Phy-logenetic analysis of dioecy in monocotyledons AmericanNaturalist 155 46ndash58

WELLER S G and A K SAKAI 1999 Using phylogenetic ap-proaches for the analysis of plant breeding system evolutionAnnual Review of Ecology and Systematics 30 167ndash199

WILSON P G M M OrsquoBRIEN P A GADEK and C J QUINN 2001Myrtaceae revisited a reassessment of infrafamilial groupsAmerican Journal of Botany 88 2013ndash2025

YODER A D J A IRWIN and B A PAYSEUR 2001 Failure of theILD to determine data combinability for slow loris phyloge-ny Systematic Biology 50 408ndash424

ZHU G R G JENSEN and H J BOHNERT 1997 DNA sequenceof ribulose-15-biphosphate carboxylaseoxygenase largesubunit from Arabidopsis thaliana Plant Physiology 114 395

APPENDIX 1

Accession list of taxa and GenBank numbers used in phyloge-netic analyses The same vouchers and DNAs were used in thisstudy Hall et al 2002 and Rodman et al 1993 1994

BRASSICALES

Bataceae Batis maritima L (Iltis 30500 WIS rbcL [L22438 Rod-man et al 1994] ndhF [AY122403 Hall et al 2002] matK AY483219)

Bretschneideraceae Bretschneidera sinensis Hemsl (Leu amp Lin726 WIS rbcL [M95753 Rodman et al 1993] ndhF AY483247 matKAY483220)

Caricaceae Carica papaya L (WIS Botanical Garden rbcL[M95671 Rodman et al 1993] ndhF AY483248 matK AY483221)

Koeberliniaceae Koeberlinia spinosa Zucc (Al Shehbaz sn MOrbcL [L14600 Rodman et al 1993] ndhF AY483249 matKAY483222)

Moringaceae Moringa oleifera Lam (Iltis 30501 WIS rbcL[L11359 Rodman et al 1993] ndhF [AY122405 Hall et al 2002]matK AY483223)

Salvadoraceae Salvadora angustifolia Turrill (rbcL [U38532 Rod-man et al 1996])


Tropaeolaceae Tropaeolum majus L (rbcL [L14706 Price andPalmer 1993] Rodman 529 WIS ndhF [AY122408 Hall et al 2002]matK AY483224)

CORE BRASSICALES

Brassicaceae s str Aethionema grandiflora L (ndhF [AF064657Galloway et al 1998] matK [AF144354 Koch et al 2001]) Aethi-onema saxatile R Br (Moore sn WIS rbcL AY483262 ndhFAY483250) Arabidopsis thaliana (L) Heynh (rbcL [U91966 Zhu etal 1997] ndhF [AY122394 Hall et al 2002] matK [AF144348 Kochet al 2001]) Capsella bursa-pastoris (L) Medicus (rbcL [D88904 Tsu-kaya et al 1997] ndhF [AY122396 Hall et al 2002] matK C rubellaReuter [AF144334 Koch et al 2001]) Nasturtium officinale RBr(rbcL [AF020325 Les et al 1994] Stahmann 233 WIS ndhF[AY122399 Hall et al 2002] matK AY483225) Stanleya pinnata(Pursh) Britton (Hall 1 AZ rbcL AY483263 ndhF [AY122401 Hallet al 2002] matK AY483226)

Capparaceae s str Apophyllum anomalum F Muell (Covery12044 MO rbcL AY483264 ndhF [AY122356 Hall et al 2002] matKAY483227) Capparis hastata Jacq (Iltis 30330 WIS rbcL [M95754Rodman et al 1993] ndhF [AY122366 Hall et al 2002] matKAY483228) Crateva palmeri Rose (Hall 105 WIS rbcL AY483265ndhF [AY122370 Hall et al 2002] matK AY483229) Maerua kirkii(Oliv) F White (Hall 261 WIS rbcL AY483266 ndhF [AY122378Hall et al 2002] matK AY483230)

Cleomaceae Cleome pilosa Benth (Iltis 30585 WIS rbcLAY483267 ndhF [AY122385 Hall et al 2002] matK AY483231) Cle-ome viridiflora Schreb (Solomon sn MO rbcL AY483268 ndhF[AY122386 Hall et al 2002] matK AY483232) Podandrogyne chiri-quensis (Standl) Woodson (Nepokroeff 450 WIS rbcL AY483269ndhF [AY122393 Hall et al 2002] matK AY483233) Polanisia dode-candra DC (Grette 8603 WIS rbcL AY483270 ndhF AY483251 matK

AY483234) Wislizenia refracta Engelm (Vanderpool 1340 OKL rbcLAY483271 ndhF [AY122391 Hall et al 2002] matK AY483235)

Gyrostemonaceae Gyrostemon sp (Cranfield PERTH no02068672 rbcL [L22439 Rodman et al 1994] ndhF AY483252 matKAY483236) Gyrostemon tepperi (F Muell ex H Walter) A S George(Thompson 2243 MO rbcL [L22440 Rodman et al 1994] ndhFAY483253 matK AY483237) Tersonia cyathiflora (Fenzl) ASGeorge(Cranfield PERTH no02068682 rbcL [L22441 Rodman et al 1994]ndhF [AF122404 Hall et al 2002] matK AY483238)

Pentadiplandraceae Pentadiplandra brazzeana Baill (rbcL[U38533 Rodman et al 1996] Hall 263 WIS ndhF AY483254 matKAY483239)

Resedaceae Oligomeris linifolia MacBride (Ertter 5613 WIS rbcLAY483272 ndhF AY483255 matK AY483240) Reseda lutea L (Rod-man 535 WIS rbcL AY483273 ndhF [AY122406 Hall et al 2002]matK AY483241)

Tovariaceae Tovaria pendula Ruiz amp Pav (Smith and Smith 1834WIS rbcL [M95758 Rodman et al 1993] ndhF [AY122407 Hall etal 2002] matK AY483242)

ANOMALOUS GENERAEmblingia calceoliflora F Muell (rbcL [AF146014 Chandler and Bay-

er 2000] Quinn 288a ndhF AY483256)Forchhammeria pallida Liebmann (Iltis 29350a WIS rbcL AY483274

ndhF [AY122381 Hall et al 2002])Forchhammeria sessilifolia Standl (Cochrane 12967 WIS rbcL

AY483275 ndhF AY483257 matK AY483243)Forchhammeria sp nov (Iltis 30784 WIS rbcL AY483276 ndhF

AY483258 matK AY483244)Forchhammeria trifoliata Radlkofer (Hansen 3002 WIS rbcL

AY483277 ndhF AY483259 matK AY483245)Forchhammeria watsonii Rose (Fishbein 3070 WS rbcL AY483278

ndhF AY483260 matK AY483246)Tirania purpurea Pierre (Squires 874 GH rbcL AY483279 ndhF

AY483261)


TAB

LE

1Pr

oble

mat

icge

nera

plac

edin

Cap

para

ceae

sla

t(5

Cap

para

ceae

sst

ran

dC

leom

acea

e)by

Pax

and

Hof

fman

n(1

936)

and

Bran

dege

e(1

909)

1In

Erdt

man

etal

(19

69)

each

ofth

efo

urau

thor

shy

poth

esiz

esa

diff

eren

ttax

onom

icpl

acem

ent

ofEm

blin

gia

Gen

ust

rad

itio

nally

ofC

appa

race

aes

lat

Subf

amily

(Tri

be)

ofPa

xan

dH

offm

ann

(193

6)O

ther

sugg

este

dre

latio

nshi

psC

urre

ntfa

mili

alcl

assi

ficat

ion

Ord

inal

Cla

ssifi

catio

n(A

PG19

98)

Buh

sia

Ben

geB

uhsi

oide

aeU

ncer

tain

Cal

yptr

othe

caG

ilg

Cal

yptr

othe

coid

eae

Port

ulac

acea

e(N

yana

nyo

1986

)N

ear

Did

iere

acea

e(A

pple

quis

tan

dW

al-

lace

2001

)C

aryo

phyl

lale

s

Em

blin

gia

FM

uell

Em

blin

gioi

deae

Poly

gala

ceae

affi

nity

wit

hG

oode

niac

eae

oraf

finit

yto

Sapi

ndac

eae

(Erd

tman

etal

19

691 )

Em

blin

giac

eae

(APG

1998

Cha

ndle

ran

dB

ayer

2000

)B

rass

ical

es

Forc

hham

mer

iaL

iebm

C

appa

roid

eae

(Sti

xeae

)E

upho

rbia

ceae

Sap

inda

ceae

Ole

acea

eM

al-

vace

ae(H

anse

n19

77)

Nea

rR

esed

acea

ean

dG

yros

tem

onac

eae

this

stud

yB

rass

ical

es

Koe

berl

inia

Zuc

cC

appa

roid

eae

(Koe

berl

i-ni

eae)

Koe

berl

inia

ceae

(Rod

man

etal

199

6)B

rass

ical

es

Neo

thor

elia

Gag

nep

Cap

paro

idea

e(S

tixe

ae)

Unc

erta

in(K

ers

2003

)Pe

ntad

ipla

ndra

Bai

llPe

ntad

ipla

ndro

idea

eN

ear

Cel

estr

acea

e(H

utch

inso

n19

73)

Pent

adip

land

race

ae(V

illie

rs19

73)

Bra

ssic

ales

Phy

sena

Nor

inha

exT

houa

rsC

appa

roid

eae

(Sti

xeae

)Pa

ssifl

orac

eae

orFl

acou

rtia

ceae

(rev

iew

edin

Dic

kiso

nan

dM

iller

1993

)Ph

ysen

acea

e(M

orto

nK

arol

and

Cha

se19

97)

Car

yoph

ylla

les

Setc

hella

nthu

sB

rand

egee

N

ASe

tche

llant

hace

ae(I

ltis

1999

)B

rass

ical

esSt

efan

ina

Chi

oven

daB

uhsi

oide

aeR

esed

acea

e(H

utch

inso

n19

73n

otac

cept

-ed

byK

ubit

zki

2003

f)B

rass

ical

es

Stix

isL

our

Cap

paro

idea

e(S

tixe

ae)

Unc

erta

in(K

ers

2003

)T

iran

iaPi

erre

Cap

paro

idea

e(S

tixe

ae)

Nea

rR

esed

acea

ean

dG

yros

tem

onac

eae

this

stud

yB

rass

ical

es
























RESULTS























parameter (G) 07781 06765 11023 09474 10425








DISCUSSION
































LITERATURE CITED

































































































APPENDIX 1


BRASSICALES









CORE BRASSICALES
















AY483261)
























RESULTS























parameter (G) 07781 06765 11023 09474 10425








DISCUSSION
































LITERATURE CITED

































































































APPENDIX 1


BRASSICALES









CORE BRASSICALES
















AY483261)





















parameter (G) 07781 06765 11023 09474 10425








DISCUSSION
































LITERATURE CITED

































































































APPENDIX 1


BRASSICALES









CORE BRASSICALES
















AY483261)




DISCUSSION
































LITERATURE CITED

































































































APPENDIX 1


BRASSICALES









CORE BRASSICALES
















AY483261)


























LITERATURE CITED

































































































APPENDIX 1


BRASSICALES









CORE BRASSICALES
















AY483261)

















































































APPENDIX 1


BRASSICALES









CORE BRASSICALES
















AY483261)



CORE BRASSICALES
















AY483261)

Molecular Phylogenetics of Core Brassicales, Placement of Orphan Genera Emblingia, Forchhammeria,...

Documents

Transcript of Molecular Phylogenetics of Core Brassicales, Placement of Orphan Genera Emblingia, Forchhammeria,...