2002_Butterworth Et Al_Molecular Systematics Tribe Cacteae

257

Systematic Botany (2002), 27(2): pp. 257270q Copyright 2002 by the American Society of Plant Taxonomists

Molecular Systematics of Tribe Cacteae (Cactaceae: Cactoideae):A Phylogeny Based on rpl16 Intron Sequence Variation

CHARLES A. BUTTERWORTH,1,3 J. HUGO COTA-SANCHEZ,2 and ROBERT S. WALLACE1

1Department of Botany, Iowa State University, Ames, Iowa 50011-1020;2Department of Biology, University of Saskatchewan, 112 Science Place, Saskatoon, SK

3Author for Correspondence ([email protected])

Communicating Editor: Thomas G. Lammers

ABSTRACT. Parsimony analysis of plastid rpl16 sequences from 62 members of Tribe Cacteae, and four outgroup taxayielded 1296 equally parsimonious trees of length 666. Strict consensus evaluation of these trees established a highly pectinatetopology, which delimited clades within the tribe that correspond to several previously considered generic groups. Aztekiumand Geohintonia, which manifest ribs in their stem morphology were shown to represent an early divergence in the tribe,forming a sister group to remaining members of the tribe. Clades containing other genera having ribbed stems also arebasal to those that develop tubercles. The most derived clade forms a distinct group of typically small stemmed specieswith tubercular stem morphology. Within Mammillaria, species formerly placed in the genus Cochemiea and members of theSeries Ancistracanthae formed a well-supported, sister clade to the remaining members of Mammillaria. Length variation ofthe intron in two members of Mammillaria series Stylothelae was also observed.

Buxbaum (1958b) rst described the tribe Cacteae(as the Echinocacteae) as a `clear-cut phylogenetic unit'in which he included all of the short-columnar or glo-bose cacti with spineless owers native to NorthAmerica, with the notable exception of the genus As-trophytum Lemaire, which he considered part of theNotocacteae. The tribe is in considerable taxonomicux, and poor generic delineation means that the exactnumber of genera is uncertain. Twenty-three generaare recognized in the CITES Cactaceae checklist (Hunt1999), although at least 34 other genera have been de-scribed. Hunt (1999) accepts 314 species (plus 224 pro-visionally accepted species) in the tribe, whereas An-derson (2001) recognizes 26 genera and 384 species.The geographic range of the Cacteae extends fromwestern Canada (Escobaria vivipara (Nuttall) Buxbaum)to Colombia, Venezuela, and the Caribbean (Mammil-laria colombiana Salm-Dyck and M. mammillaris (L.)Karsten), with maximal diversity in Mexico.

Characterized as globular or depressed to short co-lumnar cacti, members of the Cacteae range in sizefrom dwarf (Turbinicarpus (Backeberg) Backeberg andBuxbaum and some Mammillaria Haworth) to gigantic(Ferocactus Britton and Rose and Echinocactus Link andOtto). Stems may be either ribbed, as in Echinocactus,or tuberculate as in Coryphantha (Engelmann) Lemaire.Zimmerman (1985) states that ribs and tubercles aremutually exclusive terms, although a number of inter-mediates are found. He recommends the use of theterm podarium, suggesting that in reality ribs are se-ries of podaria joined end-to-end. Tubercles, however,represent free or distinct podaria. This terminology al-lows for intermediacy between ribs and tubercles. Thesize and shape of tubercles range from long and leaf-like (as in Leuchtenbergia Hooker, Obregonia Fric, andsome species of Ariocarpus Scheidweiler) to broad withshallow axils, as in Turbinicarpus.

Areoles may be borne on the ribs, or in the case ofthe tuberculate members, may occur at or near the tu-bercle apex, or form a groove on the upper surface asin Coryphantha and some species of Escobaria Brittonand Rose. In a number of genera the tubercles are dif-ferentiated into spine-bearing areoles at the tubercleapex and oriferous or vegetative areoles in the axilsof the tubercles. Buxbaum (1958b) suggested that spe-cies with differentiated tubercles such as Mammillariaare derived within the tribe. Actinomorphic or, rarely,zygomorphic (Mammillaria subgenus Cochemiea Bran-degee) diurnal owers arise from the areoles. The per-icarpel (in cacti dened as ovary wall plus stem tissueexternal to the ovary wall) ranges from scaly andwoolly to petaloid. Fruits in the Cacteae are eshy tojuicy berries, and the seeds are borne on short, simplefuniculi. Since the 1920`s, a number of researchers haverevised the Cacteae, variously interpreting its classi-cation based on differing concepts of broadly-denedor narrowly circumscribed genera. Table 1 lists generaaccepted in a number of key treatments of the Cacteae.

Britton and Rose (19191923) did not recognize theCacteae as a discrete entity. Within their tribe Cereeae(equals subfamily Cactoideae), they divided the barrelcacti into two subtribes, Echinocactinae and Cory-phanthinae. Their subtribe Echinocactinae included allribbed barrel cacti from both North and South Amer-ica, which manifest a generally low growing, globularhabit. The North American barrel cacti that share thecharacter of possessing tubercles were placed into thesubtribe Coryphanthinae, although some taxa withtrue tubercles or modied tubercles, such as PediocactusBritton and Rose, Ariocarpus, and Lophophora Coulterwere placed within the ribbed subtribe Echinocactinae.It is evident that Britton and Rose (19191923) realizedthat mutually exclusive suites of morphological char-

258 [Volume 27SYSTEMATIC BOTANY

TAB

LE

1.C

ompa

riso

nsof

prev

ious

trea

tmen

tsfo

rm

embe

rsof

the

Cac

teae

.

Thi

spa

per

And

erso

n(2

001)

Hun

t(1

999)

Bar

thlo

tt&

Hun

t(1

993)

Bac

kebe

rg(1

970)

Bux

baum

(195

8b)

Bri

tton

&R

ose

(191

919

23)

Ach

arag

ma

Esc

obar

iaE

scob

aria

Esc

obar

iaE

scob

aria

,Gym

noca

ctus

Esc

obar

iaN

otD

escr

ibed

Ari

ocar

pus

Ari

ocar

pus

Ari

ocar

pus

Ari

ocar

pus

Ari

ocar

pus,

Neo

gom

esia

,Ro-

seoc

actu

sA

rioc

arpu

sA

rioc

arpu

s

Ast

roph

ytum

Ast

roph

ytum

Ast

roph

ytum

Ast

roph

ytum

Ast

roph

ytum

Ast

roph

ytum

Ast

roph

ytum

Azt

ekiu

mA

ztek

ium

Azt

ekiu

mA

ztek

ium

Azt

ekiu

mA

ztek

ium

Not

Des

crib

edC

oryp

hant

haC

oryp

hant

haC

oryp

hant

haC

oryp

hant

haC

oryp

hant

ha,C

umar

inia

,Le

pido

cory

phan

tha

Cor

ypha

ntha

,Cum

arin

iaC

oryp

hant

ha

Ech

inoc

actu

sE

chin

ocac

tus

Ech

inoc

actu

sE

chin

ocac

tus

Ech

inoc

actu

sE

chin

ocac

tus

Ech

inoc

actu

sE

ncep

halo

carp

usPe

lecy

phor

aPe

lecy

phor

aPe

lecy

phor

aE

ncep

halo

carp

usE

ncep

halo

carp

usPe

lecy

phor

aEp

ithe

lant

haEp

ithe

lant

haEp

ithe

lant

haEp

ithe

lant

haEp

ithe

lant

haEp

ithe

lant

haEp

ithe

lant

haE

scob

aria

Esc

obar

iaE

scob

aria

Esc

obar

iaE

scob

aria

,Neo

bess

eya

Esc

obar

ia,N

eobe

ssey

aE

scob

aria

Fero

cact

usG

eohi

nton

iaFe

roca

ctus

Geo

hint

onia

Fero

cact

usG

eohi

nton

iaFe

roca

ctus

Geo

hint

onia

Fero

cact

usN

otD

escr

ibed

Fero

cact

usN

otD

escr

ibed

Fero

cact

usN

otD

escr

ibed

Gla

ndul

icac

tus

Scle

roca

ctus

Scle

roca

ctus

Scle

roca

ctus

Gla

ndul

icac

tus

Ham

atoc

actu

sH

amat

ocac

tus

Hom

aloc

epha

laE

chin

ocac

tus

Ech

inoc

actu

sE

chin

ocac

tus

Hom

aloc

epha

laH

omal

ocep

hala

Hom

aloc

epha

laLe

ucht

enbe

rgia

Leuc

hten

berg

iaLe

ucht

enbe

rgia

Leuc

hten

berg

iaLe

ucht

enbe

rgia

Leuc

hten

berg

iaLe

ucht

enbe

rgia

Loph

opho

raLo

phop

hora

Loph

opho

raLo

phop

hora

Loph

opho

raLo

phop

hora

Loph

opho

raM

amm

illar

iaC

oche

mie

a,M

amm

il-la

ria,

Mam

mill

oydi

aM

amm

illar

ia,M

am-

mill

oydi

aM

amm

illar

ia,M

am-

mill

oydi

aB

arts

chel

la,C

oche

mie

a,D

ol-

icot

hele,

Kra

inzi

a,M

amil-

lops

is,M

amm

illar

ia,P

hel-

losp

erm

a,So

lisia

Coc

hem

iea,

Dol

icho

thel

e,Le

p-to

clad

odia

,Mam

illop

sis,

Mam

mill

aria

,Mam

mill

oy-

dia,

Oeh

mea

,Pse

udom

am-

mill

aria

Bar

tsch

ella

,Coc

hem

iea,

Dol

-ic

hoth

ele,

Mam

illop

sis,

Neo

mam

mill

aria

,Phe

los-

perm

a,So

lisia

Neo

lloyd

iaN

eollo

ydia

Neo

lloyd

iaN

eollo

ydia

Neo

lloyd

iaN

eollo

ydia

Neo

lloyd

iaO

breg

onia

Obr

egon

iaO

breg

onia

Obr

egon

iaO

breg

onia

Obr

egon

iaN

otD

escr

ibed

Ort

egoc

actu

sO

rteg

ocac

tus

Ort

egoc

actu

sO

rteg

ocac

tus

Ort

egoc

actu

sN

otD

escr

ibed

Not

Des

crib

edPe

dioc

actu

sPe

dioc

actu

sPe

dioc

actu

sPe

dioc

actu

sN

avaj

oa,P

edio

cact

us,P

ilo-

cant

hus,

Uta

hia

Pedi

ocac

tus,

Uta

hia

Pedi

ocac

tus,

Uta

hia

Pele

cyph

ora

Pele

cyph

ora

Pele

cyph

ora

Pele

cyph

ora

Pele

cyph

ora

Pele

cyph

ora

Pele

cyph

ora

Scle

roca

ctus

Scle

roca

ctus

Scle

roca

ctus

Scle

roca

ctus

Col

orad

oa,E

chin

omas

tus,

Gym

noca

ctus

,Scl

eroc

ac-

tus,

Toum

eya

Anc

istr

ocac

tus,

Col

orad

oa,

Ech

inom

astu

s,Sc

lero

cact

usA

ncis

troc

actu

s,Sc

lero

cact

us,

Toum

eya

Sten

ocac

tus

Sten

ocac

tus

Sten

ocac

tus

Sten

ocac

tus

Ech

inof

ossu

loca

ctus

Ech

inof

ossu

loca

ctus

Ech

inof

ossu

loca

ctus

Stro

mbo

cact

usSt

rom

boca

ctus

Stro

mbo

cact

usSt

rom

boca

ctus

Stro

mbo

cact

usSt

rom

boca

ctus

Stro

mbo

cact

usT

helo

cact

usT

helo

cact

usT

helo

cact

usT

helo

cact

usE

chin

omas

tus,

The

loca

ctus

The

loca

ctus

Ech

inom

astu

s,T

helo

cact

usTu

rbin

icar

pus

Turb

inic

arpu

sTu

rbin

icar

pus

Neo

lloyd

iaG

ymno

cact

us,T

urbi

nica

rpus

Rap

icac

tus,

Toum

eya

Neo

lloyd

ia

2002] 259BUTTERWORTH ET AL.: PHYLOGENY OF TRIBE CACTEAE

acters could not be used to delineate subtribes withintheir tribe Cereeae, accepting that boundaries betweensubtribes Echinocereanae, Echinocactinae and Cory-phanthinae were not clearly dened.

Using an underlying principle of determining tax-onomic groups based on inferred phylogenetic relat-edness, Buxbaum (1958b) described the North Amer-ican barrel cacti (with minor exceptions) at the rank oftribe (Cacteae), and dened this group by bringing to-gether Schumann's earlier tribe Echinocacteae (Schu-mann 1899) and Britton and Rose's subtribes Echino-cactanae and Coryphanthanae. With the exception ofAstrophytum (which he placed into the tribe Notocac-teae), Buxbaum (1958b) recognized 36 genera in thetribe and regarded this group of North American bar-rel cacti as a distinct phylogenetic unit. Within his tribeCacteae, four subtribes were dened based upon seedmorphology: 1) the Echinocactinae, with a smooth,hard, black testa with conspicuous perisperm; 2) theThelocactinae, with a verrucose, mostly black testa be-coming secondarily smooth or `spotted'; 3) the Fero-cactinae, with a pitted or reticulate testa; and 4) theCoryphanthinae, with a smooth, brown testa.

In contrast to Buxbaum's phylogenetically-basedclassication, Backeberg's (1970) classication of thecacti used a complex system of infrafamilial ranks in-cluding semitribes, subtribes, groups, and subgroups.This classication was never intended to be phyloge-netic. Britton and Rose's tribe Cereeae was split, large-ly based on geographic origins of the plants, into theNorth and South American semitribes Boreocereeaeand Austrocereeae, respectively. Ignoring Buxbaum's(1958b) tribe Cacteae, Backeberg created the subtribeBoreocactinae to accommodate the North Americanbarrel cacti, which was further divided into twogroups based on ower position: 1. The Boreoechino-cacti has owers borne from undifferentiated (vegeta-tive vs. owering) areoles (the Boreoechinocacti werestill further divided into two subgroups, the Eubor-eoechinocacti and the Mediocoryphanthae) and 2. TheMammillariae, which has differentiated areoles (e.g.owers borne in tubercle axils), with three subgroups,Coryphanthiae, Mediomammillariae, and Eumammil-lariae. In total, Backeberg's subtribe Boreocactinae in-cluded 48 genera, consistent with his philosophy ofrecognizing many genera with few species in each. Inmodern taxonomic treatments, many of these ``micro-genera'' have been united into more broadly denedgroups; for example Ariocarpus was expanded by An-derson (1960, 1962) to include Roseocactus Berger andNeogomesia Castaneda, and the genera Porria Bodeck-er, Krainzia Backeberg, Phellosperma Britton and Rose,Dolichothele (Schumann) Britton and Rose, BartschellaBritton and Rose, Mamillopsis Morren ex Britton andRose, and Cochemiea (Brandegee) Walton were sub-

sumed into the genus Mammillaria by Hunt (1971,1977a, b; 1981).

Besides the treatment by Buxbaum (1958b), therehas been only one other attempt at a phylogenetic eval-uation of the tribe Cacteae. In his unpublished Ph.D.thesis, Zimmerman (1985) presented a cladistic studyof the tribe based on an analysis of morphologicalcharacters, the majority of which are derived from thestudy of oral structures. Zimmerman suggested thatthe Pachycereeae and the Notocacteae probably rep-resent the closest outgroups to tribe Cacteae, and thatthe tribe likely had its origins in South America, shar-ing a sister-group relationship with the Notocacteae.Inuenced by Buxbaum's (1958b) treatment, bothBarthlott (1977) and Zimmerman (1985) questioned theplacement of Astrophytum in the Cacteae, noting sig-nicant differences in seed morphology. Zimmermanconcluded that, with the possible exception of this ge-nus, the Cacteae formed a monophyletic unit. Further-more, Zimmerman (1985) placed Astrophytum within aclade together with Echinocactus and HomalocephalaBritton and Rose that shows a sister-group relationshipto other members of the tribe. Despite problems as-sociated with morphological plasticity in the tribe,Zimmerman made a number of insightful conclusions,for example that Escobaria, Ortegocactus, Mammillaria,and Coryphantha sensu stricto are derived from a Mam-millaria-like rather than a Ferocactus-like ancestor.

In their treatment of the genera of the Cactaceae,Barthlott and Hunt (1993) united a number of Cacteaegenera, recognizing 22 genera in total. Homalocephalawas included within Echinocactus, and the genera Oeh-mea Buxbaum, Cochemiea, Dolichothele, and Mamillopsiswere subsumed within Mammillaria. Hunt (pers.comm.) doubts that the Cacteae are monophyletic, rea-soning that because the globular growth form has aris-en independently in several cactus lineages in SouthAmerica, it has likely also arisen in North Americanlineages independently.

The primary goals of this investigation were to testmonophyly of the tribe, resolve intergeneric relation-ships in the Cacteae, and to assess monophyly in pre-viously proposed Cacteae genera using chloroplastrpl16 intron sequence data. Further, we wished to as-certain relevant outgroup taxa for an ongoing study ofthe genus Mammillaria.

MATERIALS AND METHODS

Taxonomic Sampling. A total of 66 taxa were sampled (Table2), including 62 representative taxa from the tribe Cacteae. Twospecies from tribe Notocacteae and one species from tribe Pachy-cereeae were also included with members of tribe Cacteae as theingroup. Calymmanthium substerile (tribe Browningieae) was usedas the outgroup based upon its basal position within the subfamilyCactoideae (Wallace 2001). Additional phylogenetic analyses ofchloroplast DNA variation (Butterworth and Wallace, unpub-lished) were conducted in which representative taxa were exam-ined from throughout the subfamily, and demonstrated that tribe


TABLE 2. Species sampled for rpl16 study. CANTE 5 CANTE Botanic Garden, Mexico; UCONN 5 University of Connecticut; DES5 Desert Botanic Garden, Arizona; ISC 5 Ada Hayden Herbarium, Iowa State University; HNT 5 Huntington Botanic Garden, Cali-fornia; HUMO 5 Universidad Autonoma del Estado de Morelos, Mexico; and UNAM 5 Universidad Autonoma de Mexico, MexicoCity.

Taxon Source/Voucher GenBank No.

Tribe CacteaeAcharagma aguirreana (Glass & Foster) GlassAcharagma roseana (Boed.) GlassAriocarpus agavoides (Castaneda) AndersonAriocarpus retusus Scheidw.Astrophytum capricorne (Dietrich) Br. & R.Astrophytum myriostigma Lem.Aztekium hintoni Glass & Fitz MauriceAztekium ritteri (Boed.) Boed.Coryphantha pallida Br. & R.Echinocactus grusonii Hildm.Echinocactus horizonthalonius Lem.Echinocactus ingens Zucc.

Mesa GardenISCDES 1990-0791-0201ISCC. Glass 6889CANTEC. Glass 6923CANTEHNT 69033ISCHNT 69032ISCC. Glass 6647CANTEC. Staples s.n.ISCH. Cota 8050HUMOR. Wallace s.n.UCONNM. Mendes 186CANTEHNT 59498ISC

AF267915AF267916AF267918AF267919AF267920AF267921AF267922AF267923AF267926AF267927AF267928AF267929

Encephalocarpus strobiliformis (Werderm.) BergerEpithelantha bokei L. D. BensonEscobaria zilziana (Boed.) Backeb.Ferocactus cylindraceus (Engelm.) OrcuttFerocactus avovirens (Scheidw.) Br. & R.Ferocactus glaucescens (DC.) Br. & R.Ferocactus histrix (DC.) LindsayFerocactus latispinus (Haw.) Br. & R.Ferocactus robustus (Link & Otto) Br. & R.

HNT 60211ISCDES 1993-0717-0101ISCDES 1989-0137-0102DESEcker (Slausson) 110ISCH. Cota 8051HUMOHNT 10339ISCH. Cota 8037CANTEH. Cota 8039CANTEH. Cota 8045HUMO

AF267930AF267931AF267932AF267933AF267934AF267979AF267935AF267936AF267974

Ferocactus wislizenii (Engelm.) Br. & R.Geohintonia mexicana Glass & Fitz MauriceGlandulicactus crassihamatus (Weber) Backeb.Glandulicactus uncinatus (Galeotti) Backeb.Homalocephala texensis (Hoppfer) Br. & R.Leuchtenbergia principis Hook.Lophophora diffusa (Croizat) BravoLophophora williamsii (Lem.) J. M. Coult.

L Slauson 112ISCC. Glass 6648CANTEC. Glass 5201CANTEC. Glass 6846CANTEHNT 67080ISCHNT s.n.ISCMesa GardenISCD. Martinez s.n.HUMO

AF267937AF267938AF267939AF267917AF267940AF267941AF267942AF267943

Mammillaria beneckei Ehrenb.Mammillaria candida Schweidw.Mammillaria decipiens Schweidw.Mammillaria glassii FosterMammillaria haageana PfeifferMammillaria halei BrandegeeMammillaria jaliscana (Br. & R.) Boed.Mammillaria karwinskiana Mart.Mammillaria longimamma DC.

DES 1993-0550-0101DESDES 1957-5907-0101ISCHNT 68830ISCHNT 60162ISCH. Cota 8053HUMOHNT 72646ISCLau 1050ISCH. Cota s.n.ISCDES 1992-0049-0203DES


Mammillaria magnica BuchenauMammillaria plumosa WeberMammillaria poselgeri Hildm.Mammillaria senilis Salm-DyckMammillaria voburnensis ScheerMammillaria yaquensis CraigNeolloydia conoidea (DC.) Br. & R.Obregonia denegrii Fric

HTNISCHTN 28166ISCDES 1983-0746-1018ISCMesa GardenISCLippold s.n.UCONNHNT 7715ISCLippold s.n.ISCR. Wallace s.n.ISC

AF267951AF267954AF267955AF267956AF267957AF267958AF267959AF267960

Ortegocactus macdougallii AlexanderPediocactus simpsonii (Engelm.) Br. & R.Pelecyphora aselliformis Ehrenb.Sclerocactus brevihamatus (Engelm.) D. R. HuntSclerocactus spinosior (Engelm.) Woodruff & L. BensonSclerocactus whipplei (Engelm. & Bigelow) Br. & R.Stenocactus crispatus BergerStenocactus lloydii BergerStenocactus vaupelianus (Werdem.) F. M. Knuth

R. Wallace s.n.ISCC. Butterworth 60ISCDES 1961-6848-0101DESDES 1989-0315-0101DESHughes 2ISCDES 1993-0925-0103DESHNT 46450HNTex Hort. UCONNUCONNDES 1948-1289-0101DES


Strombocactus disciformis (DC.) Br. & R.Thelocactus conothelos (Reg. & Klein) F. M. KnuthThelocactus hastifer (Werderm. & Boed.) F. M. KnuthThelocactus macdowellii (Rebut ex Quehl) C. GlassTurbinicarpus gielsdoranus (Werdermann) John & Riha

H. Sanchez-Mejorada 3603UNAMLippold s.n. (ex Hort)UCONNPeter Sharp s.n.HNT s.n.ISCHNT 50008ISC

AF267967AF267968AF267973AF267969AF267970


TABLE 2. Continued.

Taxon Source/Voucher GenBank No.

Turbinicarpus pseudomacrochele (Backeb.) F. Buxb. & Backeb.Turbinicarpus schmiedickianus var. schwartzii (Shurly)

Glass & Foster

Brach's NurseryISC

Ex Martiny s.n.ISC

AF267971

AF267972

Tribe BrowningieaeCalymmanthium substerile Ritter HNT 46555ISC AF267924

Tribe NotocacteaeCorryocactus brachypetalus (Vaupel) Br. & R.Parodia haselbergii (Haage ex Ru mpler) Brandt

HNT 18015ISCex Hort.UCONN

AF267925AF267975

Tribe PachycereeaeBergerocactus emoryi (Engelm.) Br. & R. HNT 16514ISC AF267976

Cacteae was well supported as a monophyletic group. Specimenswere obtained from a number of sources and maintained in thegreenhouse prior to DNA extraction. Institutions in which voucherspecimens are deposited are also listed in Table 2.

DNA Extraction and Purication. Total genomic DNA of rep-resentative Cacteae samples was isolated using one of two meth-ods:

1. Modied organelle pellet method suitable for mucilaginousmaterial. Genomic DNA samples were prepared using previouslypublished methods (Wallace 1995; Wallace and Cota 1996) for ex-traction of nucleic acids from highly mucilaginous plants, brieysummarized as follows: fresh, chlorenchymatous stem tissue washomogenized in 0.35M sorbitol buffer, ltered through Mira-clothy (Calbiochem). The organelles were pelleted, supernatantremoved, and pellets were then suspended in 2x CTAB (Doyle andDoyle 1987) for 1 h at 608C. After partitioning against CHCl3:oc-tanol, 24:1. DNA was isopropanol-precipitated and resuspendedfor further purication using isopycnic ultracentrifugation in ce-sium chloride/ethidium bromide gradients, followed by dialysisagainst TE.

2. Nucleon Phytopurey plant and fungal DNA extraction kit for1g samples (Amersham Life Science). DNA was extracted fromliving stem tissue according to the manufacturer's recommenda-tions and stored at 2208C in TE buffer.

Amplication and Sequencing. Polymerase chain reaction(PCR) amplication of the rpl16 intron was conducted in 100 m1reactions using GeneAmpy PCR Core Reagents (Perkin Elmer),and the amplication primers RP71F and RP1661R (Applequistand Wallace 2000). Each reaction included 20 ng of each primerand 5 m1 of unquantied DNA template. The PCR reactions wereconducted in a MJ Research PTC-100 thermal cycler using the fol-lowing temperature cycling parameters: l) initial melting at 958Cfor 5 min; 2) 24 cycles of the following protocol: 958C melt for 2min, 508C annealing for 1 min, ramp temperature increase of 158Cat 0.1258C per sec, 658C extension for 4 min; and a nal extensionstep at 658C for 10 min.

Agarose electrophoresis in TAE was used to conrm the pres-ence of 1.1 kb to 1.3 kb PCR amplication products. The ampliconswere cleaned and concentrated in Microcon 100 spin microcon-centrators (Amicon Inc.) following the manufacturer's directions.The products were then quantied in an ultraviolet spectropho-tometer and diluted to 50mg/ml for use in sequencing reactions.

Sequence data were obtained using the sequencing primersRP1516R and RP637R (Applequist and Wallace 2000) at concen-trations of 5 pmol in chain-termination reactions using the ABIPrism Big Dyey Terminator Cycle Sequencing Ready Reaction Kit(Perkin Elmer). We found that dilutions of 1:4 of terminator readyreaction solution gave acceptable reads.

Electrophoresis and automated sequence reading were conduct-ed using Perkin Elmer/Applied Biosystems automatic sequencingunits (ABI Prism 377) at the Iowa State University Nucleic AcidDNA Facility. Sequences typically were 650 or more nucleotides inlength. In a small number of taxa, a poly-T region approximately

400bp from the RP1516R priming site caused extremely poorreads upstream of the RP637R priming site. To overcome thisproblem, a new primer, RP543F, (59-TCAAGAAGCGATGGGAAC-GATGG-39) was designed to run forward from just downstreamof the RP637R priming site, overlapping the unreadable section ofsequence. Due to extensive poly-A and poly-T regions in DomainI at the 59 end, 150200bp of the intron sequence could not beobtained using the automated method. Kelchner and Clark (1997)demonstrated low levels of sequence divergence in this region andbecause it is of limited phylogenetic usefulness, further attemptsat obtaining a full length intron sequence were discontinued.

Phylogenetic Analysis. Sequence alignment was carried outusing AutoAssembler (Applied Biosystems 1995) and Se-Al (Ram-baut 1995). Following an initial Clustal W alignment, sequenceswere further aligned manually (e.g., Golenberg et al. 1993). Inser-tions/deletions considered to be phylogenetically informativewere coded in binary (presence/absence) and added to the end ofthe data matrix. There were two regions (totalling 61 nucleotides)where alignments were of doubtful homology. These regions wereexcluded from the analyses. All analyses were carried out usingPAUP* 4.0b2 (Swofford 1999). To test the rpl16 intron dataset forphylogenetic signal the g-statistic for 10,000 random trees was cal-culated. According to Hillis and Huelsenbeck (1992) the distribu-tion of lengths of random trees for all topologies provides a `sen-sitive' measure of phylogenetic signal within the dataset. Matricesthat contain a strong phylogenetic signal show distributions thatapproach a left-skewed gamma distribution as opposed to a morenormal distribution for matrices containing random noise.

Parsimony analyses were done using the heuristic search option.All substitutions and indels were equally weighted. An initial heu-ristic search using TBR branch swapping saving multiple parsi-monious trees (MULTREES ON) was conducted. Random additionsearches of 1,000 replicates, saving 100 most parsimonious trees ateach step, were undertaken to search for islands of shorter trees.Estimates of decay (Bremer 1988) were obtained using converseconstraint trees as implemented using Autodecay (Eriksson andWikstrom 1995). Bootstrap values were estimated using the `fastbootstrap' method for 1,000 replicates. A neighbor-joining analysiswas also undertaken using the F81 substitution model.

RESULTS

Sequence length ranged from 650bp in Mammillariaglassii and 673bp in M. magnica and M. haageana to935bp in Ferocactus glaucescens. Aligned sequencelength for the rpl16 dataset was 1057bp. The full da-taset (including binary-coded indels) totaled 1069 char-acters. After exclusion of indels and the two regions ofdoubtful homology, the dataset was 953 characterslong, of which 177 were parsimony-informative. The g-


statistic for the rpl16 dataset is 0.506. This value fallswell within the 99% condence interval (C.I.) for da-tasets of over 25 taxa and 500 characters (Hillis andHuelsenbeck 1992) and therefore indicates signicantphylogenetic structure within the dataset. The datamatrix of aligned rpl16 sequences is available from theauthors.

A heuristic search using PAUP* found 1296 mostparsimonious trees with length of 666 steps. There ap-pears to be considerable homoplasy in the rpl16 datasetwith a C.I. of 0.632 (0.494 excluding uninformativecharacters). However, a low C.I. may be expected, duein part to the nature of large datasets, thus the reten-tion index (R.I.) gives a more suitable indication ofsupport. In the case of the rpl16 dataset, the R.I. (ex-cluding uniformative characters) is 0.699. A randomaddition search of 100 replicates did not nd any is-lands of shorter trees. The strict consensus tree (Figure1) supports monophyly of the Cacteae, with a decayvalue of 9 and a bootstrap value of 100%.

Within the Cacteae, the general tree topology re-solves a number of clades nested pectinately withineach other: 1. ` A`ztekium Clade'' consisting of AztekiumBodecker and Geohintonia Glass and Fitz Maurice(bootstrap 100%, decay 7); 2. ``Echinocactus Clade''Astrophytum, Echinocactus, and Homalocephala (boot-strap 62%, decay 2); 3. Sclerocactus Britton and Rose(bootstrap 89%, decay 4); 4. ``Lophophora Clade''Acharagma (Taylor) Glass, Lophophora, and Obregonia(bootstrap 87%, decay 5); 5. Strombocactus disciformisforms a single lineage; 6. ` A`TEP Clade''A weaklysupported clade (bootstrap ,50%, decay 1) unitesAriocarpus, Turbinicarpus, Epithelantha, and Pediocactus;7. ``Ferocactus Clade''consisting of Ferocactus, Ancis-trocactus Britton and Rose, Leuchtenbergia, Echinocactusgrusonii, Thelocactus (Schumann) Britton and Rose, andGlandulicactus Backeberg (this clade is poorly resolvedand poorly supported with bootstrap ,50% and decay1); 8. Stenocactus (Schumann) Hill (bootstrap 100%, de-cay 4); 9. ``Mammilloid Clade'' including PelecyphoraEhrenberg, Encephalocarpus Berger, Escobaria, Coryphan-tha, Neolloydia Britton and Rose, Ortegocactus, andMammillaria (this terminal clade is well-supportedwith bootstrap 60%, and decay 3).

Analysis of the rpl16 data using a neighbor-joiningalgorithm with the F81 substitution model resulted inan initial tree that was topologically quite congruentwith the maximum parsimony tree. There were, how-ever a number of exceptions. Mammillaria glassii formsa sister-group to all of the remaining members of theCacteae in the neighbor-joining tree. This incongruenceis caused by differences in sequence length of Mam-millaria glassii of only 650bp due to a large deletionspanning the region with most informative characters.Other topological differences between the maximum-parsimony and neighbor-joining trees were observed

in the placement of members of the Lophophora andEchinocactus clades of the maximum parsimony tree,which form a single clade in the neighbor-joining tree.

DISCUSSION

Phylogenetic Relationships in the Cacteae. MONO-PHYLY OF THE CACTEAE. The phylogeny presented inthis paper supports a monophyletic origin for mem-bers of the Cacteae as currently circumscribed; no di-rect relationship was shown with Bergerocactus emoryi(Pachycereeae), and Parodia haselbergii and Corryocactusbrachypetalus, members of the morphologically similarSouth American tribe Notocacteae. A number of syn-apomorphic substitutions resulted in a decay value of9 with 100% bootstrap support, providing robust sup-port for the Cacteae clade. The monophyly of tribe Cac-teae was further tested using rpl16 intron sequences inwhich Austrocylindropuntia Backeberg (subfamilyOpuntioideae) and Maihuenia poepegii (Otto ex Pfeiffer)Philippi ex Schumann (subfamily Maihuenioideae)were used as outgroups for comparisons with each ge-nus of the Cacteae used in this study, together withrepresentatives of all other tribes in the subfamily Cac-toideae (Butterworth and Wallace, unpublished data).Support for monophyly of the tribe Cacteae was alsovery strong (98% bootstrap) with this test.

AZTEKIUM CLADE. The Aztekium Clade forms thesister-group to the remaining taxa of the Cacteae.Plants in this clade typically are globose to subglobose,rarely short columnar reaching 20 cm by 10 cm in size.Stem morphology is ribbed, the ribs in Aztekium hav-ing characteristic transverse wrinkles. Spines are no-table by their absence from mature areoles; even inyoung areoles they are highly reduced and very brittle.Aztekium and Geohintonia presently are restricted to asmall area of eastern Nuevo Leon in NE Mexico. Huntand Taylor (1992) suggested that Geohintonia may rep-resent an intergeneric hybrid involving Aztekium andpossibly Echinocactus horizonthalonius. Corriveau andColeman (1988) demonstrated biparental inheritance ofchloroplast DNA in Rhipsalidopsis Britton and Rose,and Zygocactus Schumann but maternal inheritance inEchinocereus Engelmann and Opuntia Miller. If Geohin-tonia is descended from an ancient intergeneric hybrid,and its plastid organelles are maternally inherited,then the maternal parent of the ancient hybrid wasclosely related to Aztekium, probably A. hintonii whichis sympatric with Geohintonia. However, if the chloro-plasts of Geohintonia show biparental inheritance thenAztekium could represent the descendant of either thepollen or ovule donor. The relationship between Aztek-ium and Strombocactus Britton and Rose has been causefor discussion. In a preliminary list of accepted generaby the working party of the International Organizationfor Succulent Plant Study (IOS) (Hunt and Taylor 1986)and a follow-up report (Hunt and Taylor 1990), the


FIG. 1. Strict consensus of 1296 most parsimonious trees for rpl16 intron sequences. Length 5 666 steps, C.I. 5 0.632, C.I.(excluding uninformative characters) 5 0.494, R.I. (excluding uninformative characters) 5 0.699. Bootstrap values over 50% for1000 replicates are given above the branches. Decay values are shown below the branches. * 5 switch from ribbed to tubercularstems. l 5 switch from tubercular to ribbed stems. Boxes indicate clades with dimorphic areoles.


FIG. 2. Neighbor-joining tree for the tribe Cacteae based on F81 distances.

generic status of Aztekium was accepted, althoughamong members of the working party, opinion was di-vided as to whether Aztekium and Strombocactus werecongeneric or convergent. Anderson and Skillman(1984) concluded that Aztekium and Strombocactusshould each be recognized at the generic level, citing

a number of differences in vegetative, oral, pollen andseed morphology. The phylogeny presented in this pa-per strongly supports (bootstrap 100%, decay 7) aclade containing Aztekium and Geohintonia and doesnot support a close relationship between Aztekium andStrombocactus.


ECHINOCACTUS CLADE. The clade comprising As-trophytum, Echinocactus horizonthalonius, E. ingens andHomalocephala are globose to shortly columnar cactiwith ribbed stems. Areoles are large, and in some spe-cies of Astrophytum, spines are lacking. Flowers areshortly funnelform to campanulate, the wooly pericar-pel having numerous spine-tipped bracts. These cactiare distributed throughout Mexico and SW UnitedStates. A close relationship between genera of thisclade is also supported by chloroplast restriction-sitedata (Wallace 1995). Previous authors (Bravo-Hollisand Sanchez-Mejorada 1991; Ferguson 1992; Barthlottand Hunt 1993) have considered Homalocephala as con-generic with Echinocactus. The rpl16 data (this paper)does not fully resolve the relationships between Astro-phytum, Echinocactus, and Homalocephala that wereshown by Wallace (1995), instead displaying a trichot-omy, such that with the inclusion of Homalocephala, thegenus Echinocactus may be paraphyletic. These dataalso corroborate the conclusion of Cota and Wallace(1997) that Echinocactus grusonii is more closely relatedto members of the genus Ferocactus than to other spe-cies in the Echinocactus clade.

SCLEROCACTUS CLADE. Porter et al. (2000) attempt-ed to dene generic boundaries for the morphologi-cally diverse genus Sclerocactus using chloroplast trnL-trnF sequence data. Although sampling from othergenera of the tribe was not as broad as in the studypresented here, sampling from within the genus Scler-ocactus clearly contradicted the hypothesized close re-lationship between Sclerocactus and Pediocactus, sug-gested by previous authors (Arp 1972 as cited in Porteret al. 2000; Benson 1982). Our phylogeny places mem-bers of Sclerocactus in a well-supported clade (boot-strap 89%, decay 4) and shows no afnity betweenSclerocactus and the genus Pediocactus; thus our resultsare consistent with those of Porter et al. (2000). Theworking party of the IOS (Hunt and Taylor 1986, 1990)and Barthlott and Hunt (1993) treat the genus Glan-dulicactus as a synonym of Sclerocactus. Ferguson (1991and pers. comm.) disagrees with the placement ofGlandulicactus within Sclerocactus, based on vegetativeand oral morphology. Our data support Ferguson'sview (see section on Ferocactus Clade).

LOPHOPHORA CLADE. Although there is strongsupport for this clade (87% bootstrap, 5 decay steps),few morphological features unite this clade. All mem-bers have napiform or carrotlike tap-root systems, al-though these features are also found in other membersof the tribe.

The two species of Acharagma have been a source oftaxonomic confusion. Described originally in the genusEchinocactus, E. roseanus was transferred into the genusGymnocactus Backeberg by Glass and Foster (1970),who later also described G. aguirreanus (Glass and Fos-ter 1972). However, Anderson and Ralston (1978) felt

that these two species were better placed in the genusTurbinicarpus, contrary to the views of Glass and Foster(1977), who felt that despite high degrees of similarityin distribution, appearance, and ower, fruit and seedmorphology, the larger size and generally heavier spi-nation of species of Gymnocactus warranted recognitionas a separate genus. In a review of Escobaria, Taylor(1986) placed G. roseanus and G. aguirreanus as solemembers of the section Acharagma of Escobaria. Unlikeother members of the genus, the axillary areole andtubercular groove is absent in these two species. Fur-thermore, the owers are borne in a zone adjacent tothe spine-bearing areoles, in contrast to the more typ-ical position in the axils of the tubercles. Glass (1998)elevated Taylor's section Acharagma to the rank of ge-nus following Zimmerman's provisional generic treat-ment in which Acharagma was placed in a large cladecontaining Ferocactus, Coryphantha, Mammillaria, Orte-gocactus, and Escobaria, mainly based on foveolateseeds (Zimmerman 1985). However, Zimmerman(1985) acknowledged that Acharagma only has weaklyderived character states and so his placement of thegenus was uncertain. The rpl16 intron data suggest theremoval of these two species from Escobaria, placingthem in a well-supported (bootstrap 87%, decay 5)clade containing Obregonia and Lophophora, the lattershown to be polyphyletic based on this topology.

STROMBOCACTUS. This monotypic genus from thestates of Queretero and Hidalgo in central Mexicoforms a sister lineage to the ` A`TEP'', Ferocactus, Sten-ocactus and ``Mammilloid'' clades according to thephylogeny presented in this paper. On the basis ofseed morphology, Buxbaum (1958a) suggested that thegenus Strombocactus ought to include the then mono-typic genus Aztekium. This was in spite of the tuber-culate stem anatomy of Strombocactus which contraststhe ribbed anatomy of Aztekium. Buxbaum (1958a) ex-plained this by suggesting a progression from thehardened tubercles of Strombocactus to the formation ofribs in Aztekium. Anderson and Skillman (1984), usingmorphological and anatomical data, concluded thatStrombocactus and Aztekium each deserved recognitionat the genus level. No direct relationship betweenStrombocactus and Aztekium is demonstrated in ourrpl16 phylogeny.

` A`TEP'' CLADE. This clade's acronym-based nameis derived from its included generaAriocarpus, Tur-binicarpus, Epithelantha Weber ex Britton and Rose, andPediocactus, and has poor support in our phylogeny(bootstrap ,50%, decay 1 step). Stem morphologiesare tuberculate, and in Ariocarpus dimorphic areolesare present, this feature an example of convergencewith members of the ``Mammilloid'' clade. Turbinicar-pus is a genus of around sixteen species of small, in-conspicuous cacti from north-central Mexico. Due topoor seed dispersal mechanisms, species of Turbinicar-


pus are highly localized (Glass and Foster 1977). Anumber of species of Turbinicarpus have been allied orsubsumed into other genera such as Gymnocactus(Backeberg 1970) and Neolloydia (Anderson 1986; Huntand Taylor 1990; Barthlott and Hunt 1993). However,in the CITES Cactaceae checklist, Hunt chose to acceptgeneric status for species of Turbinicarpus leaving onlytwo species in Neolloydia (Hunt 1992; 1999). The phy-logeny presented here supports the exclusion of Tur-binicarpus from Neolloydia s. str. as N. conoidea (type spe-cies for the genus) is strongly positioned within the``Mammilloid'' clade.

FEROCACTUS CLADE. This clade contains a numberof seemingly disparate genera with few morphologicalafnities (such as conspicuous pericarpel scales) thatunite the entire clade. Members of the genus Ferocactuspossess a number of morphological synapomorphiesincluding nectar-secreting areolar glands and a ring ofhairs that separate the stamens from the tepals. Al-though morphologically striking due to elongate, glau-cous tubercles, the single species of LeuchtenbergiaL.principis (the ` A`gave Cactus'') is placed in the Ferocac-tus clade. Barthlott and Hunt (1993) describe the ow-ers of this species as similar to those of Ferocactus, andthe fruit as being typical of those in subgenus Ferocac-tusdry, globose to oblong with thick-walls and de-hiscing at the base. A close afnity between Ferocactusand Leuchtenbergia is also demonstrated by the easewith which these genera hybridize. The phylogeny pre-sented here, as well as chloroplast restriction site data(Cota 1997; Cota and Wallace 1997), shows that Echin-ocactus grusonii is more closely related to members ofFerocactus (particularly F. histrix and F. glaucescens) thanit is to the remaining species of Echinocactus sampled.These species share a number of distinct charactertraits, including straight or slightly curved, terete cen-tral spines as opposed to hooked spines with at cross-sections that are more typical of Ferocactus. In our phy-logeny, however, F. histrix is positioned outside the Fer-ocactus clade. If F. histrix is moved and placed sister toF. glaucescens and E. grusonii, tree-length increases byonly three steps. The slight change in tree-length andlow decay value (decay 5 1) for the branch separatingF. histrix from the Ferocactus clade implies possible ho-moplasy in our dataset. Previous molecular studies ofFerocactus by Cota and Wallace (1997) and Cota (1997)were only able to partially resolve species relationshipsbetween Ferocactus and its allies, but did recover sim-ilar clades within the genus Ferocactus.

Glandulicactus uncinatus and G. crassihamatus cur-rently are recognized as Sclerocactus uncinatus and S.uncinatus ssp. crassihamatus, respectively, by a numberof authors (Hunt 1992; Barthlott and Hunt 1993; Hunt1999; Anderson 2001). However, Ferguson (1991) ar-gued that this genus did not belong in Sclerocactus, cit-ing a number of morphological differences. Instead, he

allied members of this genus with Ferocactus, Thelocac-tus, and Leuchtenbergia based on vegetative and oralmorphology. Although the phylogeny presented in thispaper does not necessarily support Ferguson's view-point that Glandulicactus should be recognized at genuslevel, it does corroborate his conclusions that the mem-bers of this genus are more closely related to Ferocactusand Thelocactus than to Sclerocactus.

STENOCACTUS CLADE. Comprising about 10 spe-cies, Stenocactus tends to be separable from the relatedFerocactus clade by two morphological characters: 1)narrow, n-like ribs as opposed to wide ribs, and 2)areoles in which the large spines are subtended by thesmaller spines as opposed to areoles in which the larg-er spines subtend the smaller spines in Ferocactus.However, Taylor (1983) argued that despite these mor-phological differences, ower, fruit and seed morphol-ogy required a broader generic concept that includedthe members of Stenocactus in the genus Ferocactus. Ourrpl16 phylogeny suggests that the Stenocactus clade(bootstrap 100%, decay 4) is distinct from the Ferocac-tus clade.

``MAMMILLOID'' CLADE. Although support for the``Mammilloid'' clade is not particularly strong (boot-strap 60%, decay 3), members share the morphologicalsynapomorphies of tuberculate stem anatomy and di-morphic areoles (the spine-bearing areoles being apicaland the owering areoles being axillary to the tuber-cles). Within the clade, generic delimitations have tra-ditionally been confused. Pelecyphora and Encephalocar-pus form a well-supported clade (bootstrap 100%, de-cay 8). These genera have been treated as congeneric(Pelecyphora) by some previous authors (Anderson andBoke 1969; Barthlott and Hunt 1993; Anderson 2001),and are recognized as such in the CITES CactaceaeChecklist (Hunt 1992, 1999). Relationships between Es-cobaria and Coryphantha are controversial. Berger (1929cited in Zimmerman 1985) subsumed Escobaria intoCoryphantha. Taylor (1986) cites a number of charactertraits that distinguish the two genera, including pittedseeds and ciliate outer perianth segments in Escobariaversus non-pitted seeds and non-ciliate outer perianthsegments in Coryphantha. Taylor (1986) suggests thatEscobaria is more closely related to Mammillaria than itis to Coryphantha. Indeed, a sister relationship betweenEscobaria and Coryphantha is suggested by rpl16 introndata, although additional data and more intense sam-pling are required in order to evaluate more robustlytheir relationships to Mammillaria.

Based on rpl16 sequences, Mammillaria is not mono-phyletic as currently circumscribed due to the place-ment of Neolloydia conoidea and Ortegocactus macdougal-lii. The working party of the IOS (Hunt and Taylor1986, 1990) chose to include the genus Turbinicarpusand Gymnocactus within Neolloydia. Barthlott and Hunt(1993) followed the same treatment but suggested that


the dimorphic areoles and hence axillary owers of thetype species of Neolloydia (N. conoidea) were sufcientto justify recognition of Turbinicarpus at genus levelwhile continuing to recognize the genus Neolloydia.Zimmerman (1985) concludes that Neolloydia is distinctfrom Coryphantha and its allies (being more closely re-lated to Ariocarpus, Obregonia, Lophophora, Strombocac-tus, and Aztekium), and that the tubercular groove, inthis case, is non-homologous with the areolar groovein Escobaria and Coryphantha. However, Zimmermanfailed to cite which species of Neolloydia were used inhis study, so it is unsure if he used the typeN. con-oidea or other species referable to Turbinicarpus. Ourdata supports the view of Barthlott and Hunt (1993)that Neolloydia in the strict sense (N. conoidea and N.matahuelensis Backeberg) are not closely related to ei-ther Turbinicarpus or Gymnocactus.

Ortegocactus is a monotypic genus, known only fromthe state of Oaxaca, Mexico. Although it shares manymorphological features with members of the ``Mam-milloid'' clade, taxonomists have had difculty assess-ing relationships of this species to other members ofthe clade. Bravo-Hollis and Sanchez-Mejorada (1991)placed Ortegocactus in the genus Neobesseya Britton andRose. Zimmerman (1985) concluded that Ortegocactuswas a member of a phylogenetically distinct clade con-taining Coryphantha, Escobaria and Mammillaria. Therpl16 phylogeny suggests that Ortegocactus is moreclosely related to Mammillaria than to other membersof the ``Mammilloid'' clade, and shows no direct re-lationship with Coryphantha or Escobaria.

The clade containing M. halei, M. poselgeri, and M.yaquensis is phylogenetically distinct (bootstrap 83%,decay 1) from the remaining species sampled in Mam-millaria sensu stricto. Two of these three species (M.halei and M. poselgeri) are referable to the genus Coch-emiea. Mammillaria yaquensis (series AncistracanthaeSchumann) has been allied to other members of Coch-emiea by Lu thy (1995). That Cochemiea (represented byM. halei and M. poselgeri in this study) are found onlyin Baja California raises the question of their originand dispersal from mainland Mexico. The series An-cistracanthae are distributed in western Mexico andBaja California reaching as far north as the southernUSA (California Arizona and New Mexico). A reason-able hypothesis suggests that ancestral members of theAncistracanthae migrated northwards in mainland Mex-ico, before migrating south through Baja California.The geologic history of Baja California seems unclear,but there is evidence that the Gulf of California beganto separate around 4.5 million yr ago (Atwater 1970)or 5.5 million yr ago according to Riddle et al. (1997).However, the gulf may have been in existence for thelast 12 million yr (Gastil et al. 1983). Assuming anorth-south migration of ancestral Ancistracanthae, thephylogeny presented here suggests a more recent ori-

gin for Cochemiea. Hunt (pers. comm.) disputes recog-nizing Cochemiea as distinct, stating that ornithophilousowers (found only in Cochemiea) are derived and con-tradict a sister-group relationship to other members ofMammillaria. Our phylogeny does support ornithophi-ly as being derived and suggests that Cochemiea arosefrom an Ancistracanthae-like ancestor.

Within the main clade of Mammillaria, there are anumber of species whose inclusion in the genus hasbeen disputed by various cactus taxonomists. Mam-millaria beneckei was considered by Buxbaum (1951, inHunt 1977a) as a distinct genus (Oehmea) and arguedthat this was an example of morphological conver-gence with Mammillaria, but that it was actually de-rived from a Thelocactus-type ancestor. Hunt (1977a)subsumed the genus Oehmea within Mammillaria, giv-ing it subgeneric status based upon the rugose/pittedseed testa, which is also found in other members ofthe genus. In their work on the Cactaceae, Britton andRose (19191923) gave separate generic status to Mam-millaria senilis by describing it within the genus Mam-illopsis. Their justications for this were based on anumber of oral traits that they considered sufcientlydifferent from Mammillaria to warrant its generic sta-tus. However, Hunt (1971) concluded that vegetative,oral, and seed morphology when taken in the sum oftheir characters did not support generic status of Mam-illopsis and that it should only be retained at the sub-generic level within Mammillaria. Britton and Rose(19191923) also elevated Schumann's (1899) subgenusDolichothele (represented in this study by M. longimam-ma DC) to genus level, separating it from other speciesof Mammillaria due to its very elongate tubercles. Aswith Mamillopsis, Hunt (1971) believed that there wereinsufcient differences to justify Dolichothele at the rankof genus, instead accepting it as a subgenus of Mam-millaria. Our rpl16 phylogeny does not support theview of Buxbaum (1951, in Hunt 1977a) regarding thegeneric status of Oehmea because no direct relationshipbetween Oehmea and Thelocactus is demonstrated. Al-though sampling from the genus Mammillaria is lim-ited, our data also suggest that recognition of Mamil-lopsis and Dolichothele at the generic rank is unwar-ranted as they are nested within other Mammillariaspecies. Further sampling from Mammillaria is re-quired and for the present, Oehmea, Mamillopsis, andDolichothele should be retained in Mammillaria.

The relationship of Mammillaria candida to othermembers of the genus has also been a source of pastdebate. Schumann (1899) considered this species to bewithin his subgenus Eumamillaria (true mammillarias).Buxbaum (1951, in Hunt 1977a) elevated this speciesto genus statusMammilloydia candida based soley ona tuberculate seed testa morphology. Riha and Riha(1975) disputed Buxbaum's observations, going so faras to suggest that Buxbaum had accidentally observed


seed material that was not from Mammillaria candida.Hunt (1977) argued in support of Buxbaum, suggest-ing that Mammilloydia candida was the product of a sep-arate evolutionary lineage than that of the remainingspecies of Mammillaria. However, he considered the re-tention of subgenus Mammilloydia a taxonomic com-promise. The International Cactaceae SystematicsGroup recently has accepted that M. candida merits rec-ognition at the generic rank as Mammilloydia (Hunt1999) and it is treated as such in Anderson (2001). Se-quence analysis of the rpl16 intron presented here, andfrom a more broad sampling of the ``Mammilloid''clade (Butterworth 2000) indicate that recognition ofM. candida at the rank of genus would render Mam-millaria paraphyletic. Further studies on the ``Mam-milloid'' clade are in progress to resolve these issues.

Morphological Evolution. EVOLUTION OF TUBERCLESIN THE CACTEAE. Buxbaum (1958a) presented a num-ber of different scenarios in which tubercular stemmorphologies may have arisen in the Cactaceae. With-in the tribe Cacteae, he described the convergent evo-lution (in a number of lineages within the tribe) oftransversely-arranged tubercles formed from ribs inwhich the basal portions of the podaria have becomeenlarged to form tubercles, implying that a ribbedstem morphology represents the primitive conditionfor the tribe. Gibson and Nobel (1986) also suggest thatthe primitive condition for the subfamily Cactoideae islikely based on ribbed-stem morphology. From ourstudies using rpl16 intron sequence data, it appearsthat in the Cacteae tubercular stem morphologies rep-resent a derived condition. The question of multipleorigins of tubercles or reversals to ribbed stems is de-batable. The most parsimonious explanation based onthe phylogeny presented in this paper is that tuber-cular stems have arisen independently in a number ofclades, once following the divergence between theEchinocactus clade and the remaining Cacteae. A rever-sal to ribbed stems is implicated in Ferocactus histrix,Stenocactus, and the Ferocactus clade, with secondarilyderived tubercular stem morphologies representing azone of transition between ribs and tubercles. The ge-nus Ferocactus has ribbed stems, while Leuchtenbergiahas very distinct elongate tubercles. Glandulicactus hasdeeply notched ribs and may represent the interme-diate condition, and in Thelocactus, both ribs and tu-bercles are presentThelocactus hastifer with distinct,spiraling ribs divided into tubercles, and the sisterclade of T. conothelos and T. macdowellii having indistinctribs and pronounced tubercles (Anderson 1987). Thereis also a switch from ribbed to tubercular stems in the``Mammilloid'' Clade, whose members also share thesynapomorphy of having dimorphic areoles (see bel-low).

EVOLUTION OF DIMORPHIC AREOLES. The majorityof genera in the Cacteae produce owers from the

spine-forming areoles. However, a number of taxa inthe tribe have dimorphic areoles in which spines andowers are borne from different regions or even fromseparate areoles. To an extent this correlates with tu-bercular stem morphologies where spines are pro-duced from apical areoles (the axillary areoles becom-ing reproductive). Buxbaum (1958a) proposed that theevolution of dimorphic areoles in the tribe Cacteae oc-curred along two distinct lines. The rst lineage showsa succession from Leuchtenbergia, which has elongatedtubercles tipped by undifferentiated areoles, to Roseo-cactus (Ariocarpus ssuratus, A. kotschoubeyanus) with ar-eoles forming an elongated furrow along the length ofthe tubercle with separate oral and spine-bearing re-gions (Anderson 1960), to Ariocarpus with separate o-ral and spine-bearing areoles and some species lackingspine-bearing areoles altogether (Anderson 1960). Bux-baum's second evolutionary lineage occurred from anon-differentiated ``Thelocactus-type'' areole in whichgrowth occurs below the areole causing it to be forcedto the tip of the tubercle. In a number of species,lengthening growth divides the growing point forcingthe spine producing part towards the tubercle tip, theower producing region remains in the tubercle axil.Species with this form of dimorphic areole may havea groove running along the adaxial length of the tu-bercle connecting the vegetative and reproductive ar-eoles. In Mammillaria, the groove is absent due to di-vision of the growing point at a very early stage indevelopment. According to our phylogeny, undiffer-entiated areoles represent the plesiomorphic conditionfor the Cacteae with the evolution of dimorphic areolesoccurring independently in Ariocarpus and the ``Mam-milloid'' clade.

In summary, the pattern of evolution that we presentin the tribe Cacteae, as inferred from the rpl16 intronphylogeny, suggests that the Aztekium-Geohintoniaclade, represents a relictual, yet highly specialized lin-eage. The remaining members of this North Americanbarrel cactus tribe have undergone diversication intoseveral clades, the more derived clades manifesting ashift from plesiomorphic ribbed stems to those that aretuberculate, concomitantly undergoing a general re-duction in plant size. Shifts in oral position and ar-eole are also inferred from the phylogeny. Thesechanges occurred in parallel within the Cacteae, fur-ther adding to the systematic confusion experienced bymany earlier cactologists. Although the inferred evo-lutionary relationships we present are based on datafrom a single molecular marker of the plastid genome,the resulting tree topology and clades dened are tell-ing as to the broad-scale, intergeneric relationshipswithin the tribe that have heretofore only received``support'' through speculative conclusions, accompa-nied by little empirical analyses.

Here we broadly sample representative members of


the tribe Cacteae in a uniformly comparative fashion,and evaluate the group to determine its primary line-ages. The intergeneric relationships inferred now lendthemselves to further testing with additional markersand more intensive sampling for the more species-richgenera (e.g., Mammillaria, Coryphantha, Escobaria, Fero-cactus), as well as reexamining those clades that werenot well supported (e.g. the A`TEP' Clade). These in-vestigations are ongoing at present and will extend thevalue of the present study through the use of its con-clusions in prudent outgroup sampling, identicationof morphological evolutionary trends among the taxa,and by establishing a baseline phylogeny for integra-tion with other similar studies being conducted on oth-er tribes in the Cactaceae.

ACKNOWLEDGEMENTS. We would like to thank all the follow-ing persons and institutions for supplying plant material for thisstudy. Steven Brack, W.A and Betty Fitz Maurice, Charles Glass,Lee Hughes, Mario Mendez, Liz Slauson, El Charco Botanic Gar-den (San Miguel de Allende, Guanajuato, Mexico), The Desert Bo-tanic Garden, Phoenix, AZ), Mesa Garden (Belen, NM), and TheHuntington Botanic Garden (San Marino, CA). Helpful discussionsregarding intergeneric relationships and morphological variationwithin tribe Cacteae were provided by Edward F. Anderson, Wil-helm Barthlott, Arthur Gibson, Anton Hofer, David Hunt and Ni-gel Taylor. We would also like to thank Jonathan F. Wendel andLynn G. Clark for their comments and suggestions. This study wassupported by grants from the National Science Foundation (NSFDEB 9527884) and from the Cactus and Succulent Society ofAmerica Research Fund to R.S.W. We also acknowledge logisticsupport from the National Geographic Society (Grant Number547395) and the State University of Morelos, Mexico, which pro-vided support for eld studies by R.S.W. and J.H.C. in Mexico.

LITERATURE CITED

ANDERSON, E. F. 1960. A revision of Ariocarpus (Cactaceae) I. Thestatus of the proposed genus Roseocactus. American Journalof Botany 47: 582589.

. 1962. A revision of Ariocarpus (Cactaceae) II. The status ofthe proposed genus Neogomesia. American Journal of Botany49: 615622.

. 1986. A revision of the genus Neolloydia B. & R. (Cacta-ceae). Bradleya 4: 128.

. 1987. A revision of the genus Thelocactus B. & R (Cacta-ceae). Bradleya 5: 4976.

. 2001. The cactus family. Timber Press, Portland, Oregon., and N. H. BOKE. 1969. The genus Pelecyphora (Cactaceae):

Resolution of a controversy. American Journal of Botany 56:314326.

and M. E. RALSTON. 1978. A study of Thelocactus (Cacta-ceae) I.: The status of the proposed genus Gymnocactus. Cac-tus and Succulent Journal (U.S.) 50: 216224.

and SKILLMAN, S. M. 1984. A comparison of Aztekium andStrombocactus (Cactaceae), Systematic Botany 9: 4249.

APPLEQUIST, W. L. and R. S. WALLACE. 2000. Phylogeny of theMadagascan endemic family Didiereaceae. Plant Systematicsand Evolution 221: 157166.

ATWATER, T. 1970. Implications of plate tectonics for the Cenozoictectonic evolution of western North America. Geological So-ciety of America Bulletin 81: 35133536.

BACKEBERG, C. 1970. Das Kakteenlexicon. Jena: VEB Gustav FischerVerlag.

BARTHLOTT, W. 1977. Cacti. Botanical aspects. Descriptions and culti-vation. Cheltenham, England: Stanley Thornes Ltd.

and D. HUNT. 1993. Cactaceae. Pp. 161197 in The familiesand genera of vascular plants, vol. II ed. K. Kubitzki. Berlin:Springer-Verlag.

BENSON, L. 1982. The cacti of the United States and Canada. StanfordCalifornia: Stanford Univ. Press.

BRAVO-HOLLIS, H. and H. SA NCHEZ-MEJORADA. 1991. Las Cactaceasde Mexico, Vol. II and III. Mexico City: Universidad NacionalAutonoma de Mexico.

BREMER K. 1988. The limits of amino acid sequence data in angio-sperm phylogenetic reconstruction. Evolution 42: 795803.

BRITTON, N. L. and J. N. ROSE. 19191923. The Cactaceae. Four Vol-umes. Washington: Carnegie Institution.

BUTTERWORTH, C. A. 2000. Phylogenetic studies of the cactus ge-nus Mammillaria Haw. using chloroplast DNA sequence data.American Journal of Botany 87(6, Supplement): 116117.

BUXBAUM, F. 1958a. Morphology of cacti. Pasadena: Abbey GardenPress.

. 1958b. The phylogenetic division of the subfamily Cereo-ideae, Cactaceae. Madrono 14: 177216.

CORRIVEAU, J. L. and A. W. COLEMAN. 1988. Rapid screeningmethod to detect potential biparental inheritance of plastidDNA and results for over 200 Angiosperm species. AmericanJournal of Botany 75(10): 14431458.

COTA, J. H. 1997. A phylogenetic study of Ferocactus Britton andRose (Cactaceae: Cactoideae), PhD. thesis, Iowa State Univer-sity, Ames.

and R. S. WALLACE. 1997. Chloroplast DNA evidence fordivergence in Ferocactus and its relationships to North Amer-ican columnar cacti (Cactaceae: Cactoideae). Systematic Bot-any 22: 529542.

DOYLE, J. J. and J. L. DOYLE. 1987. A rapid DNA isolation proce-dure for small quantities of fresh leaf tissue. PhytochemicalBulletin 19: 1115.

ERIKSSON, T. and N. WIKSTRO M. 1995. Autodecay version 3.0. http://www.bergianska.se/personal/TrostenE/

FERGUSON, D. J. 1991. In defense of the genus Glandulicactus Back-eb. Cactus and Succulent Journal (U.S.) 63: 8791

. 1992. The genus Echinocactus Link & Otto, subgenus Hom-alocephala (Br. & R) stat. nov. Cactus and Succulent Journal(U.S.) 64: 169172.

GASTIL, G., J. MINCH, and R. PHILLIPS. 1983. The geology and agesof the islands. Pp. 1325 in Island biogeography in the Sea ofCortez, eds. T. Case and M. Cody. Berkeley: University of Cal-ifornia.

GIBSON, A. C. and P. S. NOBEL. 1986. The cactus primer. Cambridge,Massachusetts: Harvard University Press.

GLASS, C. 1998. Acharagma. Guide to the threatened cacti of Mexico,ed. C. Glass. San Miguel de Allende, Guanajuato, Mexico:Cante AC.

and R. FOSTER. 1970. Mexico logbook. Cactus and Succu-lent Journal (USA) 42: 229234.

and . 1972. Gymnocactus aguirreanus. A new speciesfrom southern Coahuila, Mexico. Cactus and Succulent Jour-nal (USA) 44: 8081.

and . 1977. A revision of the genus Turbinicarpus.Cactus and Succulent Journal (USA) 49: 161176.

GOLENBERG, E. H., M. T. CLEGG, M. L. DURBIN, J. DOEBLEY, andD. P. MA. 1993. Evolution of a non-coding region of the chlo-roplast genome. Molecular Phylogenetics and Evolution 2:5264.

HILLIS, D. M and J. P. HUELSENBECK. 1992. Signal, noise, and reli-ability in molecular phylogenetic analyses. Journal of Hered-ity 83: 189195.

HUNT, D. 1971. Schumann and Buxbaum Reconciled. The Cactusand Succulent Journal of Great Britain 33: 5372.


. 1977a. Schumann and Buxbaum Recompiled (1). The Cac-tus and Succulent Journal of Great Britain 39(2): 3740.

. 1977b. Schumann and Buxbaum Recompiled (2). The Cac-tus and Succulent Journal of Great Britain 39(3): 7174.

. 1981. Revised classied list of the genus Mammillaria. TheCactus and Succulent Journal of Great Britain 43: 4148.

. 1992. CITES Cactaceae Checklist. Kew: Royal Botanic Gar-dens.

. 1999. CITES Cactaceae Checklist. 2nd Ed. Kew: Royal Bo-tanic Gardens.

and N. P. TAYLOR. 1986. The genera of the Cactaceae: to-wards a new consensus. Bradleya 4: 6578.

and . 1990. The Genera of Cactaceae: progress to-wards consensus. Bradleya 8: 85107.

and . 1992. Notes on miscellaneous genera of Cac-taceae (2). Bradleya 10: 1732.

KELCHNER S. A. and L. G. CLARK. 1997. Molecular evolution andphylogenetic utility of the chloroplast rpl16 intron in Chusqueaand the Bambusoideae (Poaceae). Molecular Phylogeneticsand Evolution 8: 385397.

LUTHY, J. M 1995. Taxonomische Untersuchung der Gattung Mam-millaria Haw. Ph.D. thesis, Universitat Bern, Switzerland.

PORTER J. M., M. S. KINNEY, and K. D. HEIL. 2000. Relationshipsbetween Sclerocactus and Toumeya (Cactaceae) based on chlo-roplast trnL-trnF sequences. Haseltonia 7: 823.

RAMBAUT, A. 1995. Sequence alignment program. Oxford: Univer-sity of Oxford.

RIDDLE, B. R., D. J. HAFNER, L. F. ALEXANDER and J. R. JAEGER.1997. Cryptic vicariance in the historical assembly of a BajaCalifornia Peninsular Desert biota. Proceedings of the Na-tional Academy of Science 97: 1443814443.

RIHA, J. and J. J. RIHA. 1975. The genus Mammilloydia Buxbaum.Cactus and Succulent Journal (U.S.) 47: 195197.

SCHUMANN, K. 1899. Gesamtbeschreibung der Kakteen (MonographiaCactearum), Neudamm: J. Neumann.

SWOFFORD, D. C. 1999. Paup*. Phylogenetic analysis using parsi-mony (* and other methods). Sunderland: Sinauer Associates.

TAYLOR, N. P. 1983. Comments on proposal 673 to conserve 5408Stenocactus (Schumann) Berger (1929) over various genericnames (Cactaceae). Taxon 32: 641643.

. 1986. The identication of Escobarias (Cactaceae). BritishCactus and Succulent Journal 4: 3644.

WALLACE, R S. 1995. Molecular systematic study of the Cactaceae:using chloroplast DNA variation to elucidate cactus phylog-eny. Bradleya 13: 112.

. 2001. An overview of columnar cactus evolution and sys-tematics. in Evolution, ecology and conservation of columnar cactiand their mutualists, eds. T. H. Fleming and A. Valiente Ban-uet. Tucson: University of Arizona Press. In Press.

, and J. H. COTA. 1996. An intron loss in the chloroplastgene rpoC1 supports a monophyletic origin for the subfamilyCactoideae of the Cactaceae. Current Genetics 29: 275281.

ZIMMERMAN, A. D. 1985. Systematics of the genus Coryphantha(Cactaceae). PhD. thesis, University of Texas, Austin.

2002_Butterworth Et Al_Molecular Systematics Tribe Cacteae

Documents

Transcript of 2002_Butterworth Et Al_Molecular Systematics Tribe Cacteae