Content uploaded by Charles Butterworth
Author content
All content in this area was uploaded by Charles Butterworth on May 20, 2020
Content may be subject to copyright.
800
Systematic Botany (2005), 30(4): pp. 800–808
qCopyright 2005 by the American Society of Plant Taxonomists
Molecular Phylogenetics of the Leafy Cactus Genus Pereskia (Cactaceae)
C
HARLES
A. B
UTTERWORTH
1,3
and R
OBERT
S. W
ALLACE
2
1
Department of Botany, Iowa State University, Ames, Iowa 50011;
2
Department of Ecology, Evolution and Organismal Biology, Iowa State University, Ames, Iowa 50011;
3
Author for correspondence. Current Address: Desert Botanical Garden, 1201 North Galvin Parkway, Phoenix,
Arizona 85008 (cbutterworth@dbg.org)
Communicating Editor: Sara B. Hoot
A
BSTRACT
.Members of Pereskia exhibit some presumably plesiomorphic characters for the Cactaceae including shrubby
habit, non-succulent or partially succulent leaves, and in some species, nearly superior ovaries. In addition, the members
show a transition from perigynous flowers with half-inferior ovaries to those species having true receptacular epigyny (the
predominant condition in the Cactaceae). To examine interspecific relationships within Pereskia we utilized cpDNA restric-
tion-site data and sequences from two non-coding regions of the plastid genome—the psbA-trnH intergenic spacer and the
rpl16 intron. Maximum parsimony and Bayesian analyses identified three major clades: a clade containing the widespread
P. aculeata and the Andean species, a clade containing six species found primarily in southeastern Brazil, Paraguay, Uruguay,
Argentina, and Bolivia, and a third clade centered in southern Central America and the Caribbean. The relationship between
these three clades and the rest of the Cactaceae remains unresolved, but our data do suggest that Pereskia may be paraphyletic.
The sister taxon relationship for the yellow flowered species of Pereskia (P. aureiflora,P. guamacho) was also confirmed, despite
their widely disjunct distribution.
For many, a typical cactus is a green, leafless stem-
succulent plant with numerous spines. However, mem-
bers of the genus Pereskia Miller are broad-leaved trees
and shrubs. They are clearly members of the cactus
family due to the presence of spine-bearing areoles, a
floral cup with leaf-bearing nodes, and numerous peri-
anth segments. Unlike other members of the family, the
ovary in Pereskia ranges from superior to fully inferior.
This feature, coupled with aspects of habit, physiology,
and morphology have led some researchers to con-
clude that Pereskia species represent some of the most
primitive members of the cacti (Gibson and Nobel
1986). Species of Pereskia are distributed throughout the
northern two-thirds of South America (from northern
Argentina) to Mesoamerica and the Caribbean. Back-
eberg (1942) concluded that the distribution of Pereskia
indicates that the genus and the cactus family arose in
Mesoamerica and the Caribbean.
Pereskia was first described as Peireskia by Plumier
(1703) and Linnaeus (1753) used the name at species
rank as Cactus pereskia. However, in the following year,
Miller (1754) elevated the name to genus level in the
first valid use of Pereskia at that rank. Berger (1926)
believed that variation in the ovary position in Pereskia
was sufficiently significant to warrant the description
of subgenus Rhodocactus Berger, which was itself raised
to genus level by Backeberg and Knuth (1935). The re-
maining species in the genus Pereskia were divided be-
tween two subgenera by Backeberg (1956), who placed
the small-leaved Andean species in subgenus Neopei-
reskia Backeberg. More recently, authors such as Bravo-
Hollis (1978) and Leuenberger (1986) have disregarded
the genus Rhodocactus, preferring to recognize a more
widely circumscribed genus Pereskia. The CITES Cac-
taceae Checklist (Hunt 1999) and Anderson (2001) ac-
cept 17 species and two subspecies.
The only recent monograph of Pereskia is that of
Leuenberger (1986), in which he gives a detailed mor-
phological, anatomical, and developmental account of
the genus. He also presents an infrageneric treatment
of the genus in which he puts forward an evolutionary
and biogeographic hypothesis for the genus based
upon a number of anatomical and morphological char-
acters. Without being explicit, Leuenberger (1986) pre-
sents seven infrageneric groups based on a small suite
of morphological characters (summarized in Table 1).
The lack of clear-cut synapomorphies for Pereskia sug-
gest that this genus represents a grade of ‘‘basal’’ taxa,
and that an exploration of variation in the genus is
important to our understanding of early evolution in
cacti as a whole. This paper investigates evolutionary
relationships in Pereskia and informal infrageneric
groupings developed by Leuenberger (1986) by devel-
oping a phylogeny using a combination of sequence
data from two chloroplast regions, rpl16 intron and the
psbA-trnH intergenic spacer (IGS), and chloroplast
DNA (cpDNA) restriction site variation.
M
ATERIALS AND
M
ETHODS
We sampled 18 of the 19 taxa (17 spp. and 2 subsp.) of Pereskia
as currently recognized in the CITES Cactaceae Checklist (Hunt
1999). We also included taxa from all other subfamilies of the Cac-
taceae, including a representative of the genus Pereskiopsis Britton
and Rose, which is very similar in appearance to members of Per -
eskia due to the presence of persistent leaves. However, it is com-
monly included in the Opuntioideae due to the presence of glo-
chids, seeds with a bony aril, and typically ‘‘opuntioid’’ flowers
(Barthlott and Hunt 1993). Dickie (1996) also demonstrated the
position of Pereskiopsis in the Opuntioideae using chloroplast DNA
sequence data. For a non-cactus outgroup, we chose Talinum pan-
iculatum from the Portulacaceae. Hershkovitz and Zimmer (1997)
and Applequist and Wallace (2001) clearly demonstrated that
members of the Portulacaceae form the sister group to the Cac-
2005] 801BUTTERWORTH & WALLACE: PHYLOGENY OF PERESKIA
T
ABLE
1. Infrageneric groups, species names, and character correlations in Pereskia according to Leuenberger (1986). Notes:
1
Brachyblast leaves absent in P. quisqueyana;
2
Some multiseriate trichomes
present in P. quisqueyana;
3
Leuenberger (1986) did not observe fruit in P. quisqueyana;
4
Tuberous roots in P. guamacho, fibrous roots in P. aureiflora;
5
Fibrous roots in P. zinniiflora.
Group Sclereids Brachyblast
leaves Trichomes Pollen Fruit
umbilicus Roots Stem
stomata Periderm
formation
Group 1.
P. aculeata
simple-
fusiform
absent some
multiseriate
6–9 colpate small fibrous absent early
Group 2.
P. lychnidiflora
simple-
fusiform
present some
multiseriate
6–9 colpate small fibrous absent early
Group 3.
P. horrida
P. diaz-romeroana
P. weberiana
simple-
fusiform
absent all uniseriate 3 colpate small tuberous present retarded
Group 4.
P. bleo
aggregated-
fusiform
absent all uniseriate 9–12 colpate large fibrous absent early
Group 5.
P. stenantha
P. bahiensis
P. grandifolia
P. sacharosa
P. nemorosa
aggregated-
fusiform
present all uniseriate 12–15 colpate small fibrous absent to
present
retarded
Group 6.
P. guamacho
P. aureiflora
stone cells present some
multiseriate
12–15 colpate small tuberous &
fibrous
4
absent retarded
Group 7.
P. zinniiflora
P. portulacifolia
P. quisqueyana
stone cells present
1
all uniseriate
2
12–15 colpate medium
3
tuberous
5
absent early
802 [Volume 30SYSTEMATIC BOTANY
T
ABLE
2. List of taxa included in this study. B 5Berlin Botanical Garden, HNT 5Huntington Botanical Garden, ISU 5Iowa State
University. Data are in the following sequence: taxon name, botanic garden accession number or collector name and number, Genbank
accession number for rp116 intron, psbA-trnH IGS. Vouchers for all taxa are deposited in the herbarium of Iowa State University (ISC).
Subfamily Pereskioideae
Pereskia aculeata Miller, B 259-04-81-80, AY851589, AY851605; Pereskia aureiflora Ritter, B 166-54-83-20, AY851595, AY851569;
Pereskia bahiensis Guerke, B 166-86-83-10, AY851605, AY851579; Pereskia bleo (Kunth) De Candolle, B 277-01-80-80, AY851600,
AY851574; Pereskia diaz-romeroana Cardenas, ex Hort ISU, AY851592, AY851566; Pereskia grandifolia Haworth subsp. grandifolia,B
047-01-78-84, AY851603, AY851577; Pereskia grandifolia Haworth subsp. violacea (Leuenberger) Taylor & Zappi, B 036-01-77-30,
AY851604, AY851578; Pereskia guamacho Weber, B 001-16-74-70, AY851596, AY851570; Pereskia horrida (Kunth) De Candolle subsp.
horrida, B 039-04-77-30, AY851590, AY851564; Pereskia horrida (Kunth) De Candolle subsp. rauhii (Backeberg) Ostolaza, B 039-03-
77-30, AY851591, AY851565; Pereskia lychnidiflora De Candolle, B 003-12-78-10, AY851594, AY 851568; Pereskia nemorosa Rojas, B
039-05-77-30, AY851601, AY851575; Pereskia portulacifolia (L.) Haworth, B 376-01-86-10, AY851598, AY851572; Pereskia quisqueyana
Liogier, B 259-05-82-30, AY851599, AY851573; Pereskia sacharosa Grisebach, B 133-10-82-30, AY851602, AY851576; Pereskia stenantha
Ritter, B 166-81-83-20, AY851606, AY851580; Pereskia weberiana Schumann, B 037-01-77-30, AY851593, AY851567; Pereskia zinniiflora
De Candolle, B 200-01-80-30, AY851597, AY851571
Subfamily Opuntioideae
Opuntia polyacantha Haworth, J. F. Weedin 1790, AY851611, AY851585; Opuntia subulata (Muehlenpfordt) Englemann, R. S.
Wallace s.n., AY851612, AY851586; Pereskiopsis porteri (Brand. ex Weber) B & R., B 169-03-84-30, AY851607, AY851581; Pterocactus
kuntzei Schumann, F. Katterman 621 AY851613, AY851587
Subfamily Maihuenioideae
Maihuenia poeppigii (Pfeiffer) Shumann, F. Katterman s.n., AY851609, AY851583
Subfamily Cactoideae
Calymmanthium substerile Ritter, HNT 46555, AY851614, AY851588; Leptocereus quadricostatus (Bello) Britton & Rose, ex Hort
ISU, AY851608, AY851582
Outgroup—Portulacaceae
Talinum paniculatum (Jacq.) Willd., ex Hort ISU, AY851610, AY851584
taceae. Table 2 lists the sources for living material used in this
study as well as GenBank accession numbers for all sequences
generated. Voucher material is deposited in the herbarium at Iowa
State University (ISC).
Total genomic DNA was isolated using a modified organelle
pellet method suitable for mucilaginous material (Wallace 1995;
Wallace and Cota 1996), briefly summarized as follows: fresh,
chlorenchymatous plant tissue was homogenized in a 0.35M sor-
bitol buffer and filtered through Miraclothy(EMD Biosciences
Inc., San Diego, California). The organelles were pelleted, super-
natant removed, and pellets were then suspended in 2x CTAB
(Doyle and Doyle 1987) for 1 hr at 608C. After partitioning against
CHCl
3
:octanol, 24:1, DNA was isopropanol-precipitated and re-
suspended for further purification using isopycnic ultracentrifu-
gation in cesium chloride/ethidium bromide gradients, followed
by dialysis against TE.
With the exception of four taxa (Calymmanthium substerile,Opun-
tia polyacantha,O. subulata,andPterocactus kuntzei), all samples
were cut with a battery of 18 restriction endonucleases (Ava I,
BamHI, BanI, BanII, BclI, BglII, BstNI, ClaI, DraI, EcoO109, EcoRI,
EcoRV, HincII, HindIII, NciII, Nsi I, XbaI, and XmnI). The digested
DNA fragments were separated using agarose gel electrophoresis
in TAE buffer. Following electrophoresis, the DNA fragments were
bidirectionally transferred (Smith and Summers 1980) to nylon
membranes (Zetabind AMF-Cuno, Meridian, Connecticut). The
fragments were then hybridized with nick-translated [
32
P] plasmid
probes according to Jansen and Palmer (1987). Recombinant plas-
mid subclones for the chloroplast genome of Nicotiana tabacum L.
(Shinozaki et al. 1986) were used to assess restriction site variation
according to Palmer (1986). Restriction site variation wasidentified
relative to the condition observed in the outgroup taxon Talinum.
These were scored as binary characters (0, 1) for absence or pres-
ence. Cells with missing data were scored as ‘N’.
Polymerase chain reaction (PCR) amplification of the rpl16 in-
tron and the psbA-trnH intergenic spacer was conducted in 100 ml
reactions using GeneAmpyPCR Core Reagents (Perkin Elmer).
Primers and PCR reaction conditions used for amplification and
sequencing are detailed in Butterworth and Wallace (2004).
After agarose electrophoresis confirmation of amplification, the
amplicons were cleaned and concentrated in Microcon 100 spin
microconcentrators (Amicon Inc.) following the manufacturer’s di-
rections. The products were quantified using an ultraviolet spec-
trophotometer and diluted to 50 mgml
22
for use in sequencing
reactions.
Sequence data for the rpl16 intron and psbA-trnH IGS wer e ob -
tained in chain-termination reactions using the ABI Prism Big
DyeyTerminator Cycle Sequencing Ready Reaction Kit (Perkin
Elmer). We found that dilutions of 1:4 of Big Dye to terminator
ready reaction solution gave acceptable reads. Electrophoresis and
automated sequence readings were conducted using Perkin El-
mer/Applied Biosystems automatic sequencing units (ABI Prism
377) at the Iowa State University DNA Sequencing and Synthesis
Facility. For the rpl16 intron, sequences typically were 650 or more
nucleotides in length. Due to extensive poly-A and poly-T regions
in Domain I at the 59end, 150–200bp of the rpl16 intron sequence
could not be obtained using the automated method. Kelchner and
Clark (1997) demonstrated low levels of sequence divergence in
this region and because it is of limited phylogenetic usefulness,
further attempts at obtaining a full-length intron sequence were
discontinued. For a number of taxa, poly-A regions thwarted at-
tempts to sequence approximately 100 bp of the middle region of
the psbA-trnH IGS.
Sequence alignment was carried out using AutoAssembler (Ap-
plied Biosystems 1995) and Se-Al (Rambaut 1995). Sequences were
aligned manually. Insertions/deletions considered to be phyloge-
netically informative (Graham et al. 2000) were coded as binary
characters (presence/absence) and added to the end of the data
matrix. Areas where alignments were of doubtful homology were
excluded from the analyses. A total of 1,698 (3.6%) of the cells
were scored as missing data. The complete data matrix is available
from TreeBASE (study accession no. S1343; matrix accession no.
M2364) or from the corresponding author.
2005] 803BUTTERWORTH & WALLACE: PHYLOGENY OF PERESKIA
Incongruence Length Difference (ILD) tests (Farris et al. 1995)
were undertaken to assess congruence and hence combinability of
the three datasets (cpDNA restriction site, rpl16 intron and psbA-
trnH IGS). The ILD tests was conducted in PAUP* 4.0b2 (Swofford
1999) for 100 replicates, saving 1,000 most-parsimonious trees for
each replicate.
Parsimony analyses were done using the heuristic search option
in PAUP*. All substitutions and indels were equally weighted. An
initial heuristic search using TBR branch swapping saving multi-
ple parsimonious trees (MULTREES ON) was conducted. Random
addition searches of 1,000 replicates, saving 1,000 most-parsimo-
nious trees for each replicate, were undertaken to search for is-
lands of shorter trees. Estimates of decay (Bremer 1988) were ob-
tained using converse constraint trees as implemented using Au-
todecay (Eriksson 1998). Bootstrap values were estimated for 1,000
replicates using a heuristic search with the same parameters as
above.
A Bayesian analysis was also undertaken on the combined da-
taset using MrBayes version 3 (Huelsenbeck and Ronquist 2001;
Ronquist and Huelsenbeck 2003) in a 4 chain (three hot, one cold)
Markov chain Monte Carlo run for a million cycles with sampling
every 100
th
cycle. ModelTest (Posada and Crandall 1998) was used
to determine an appropriate model of sequence evolution and rec-
ommended the F81 (Felsenstein 1981) model plus a gamma dis-
tribution—F811G. Following the Bayesian analysis, tree posterior
probabilities were graphed to allow an estimate of the number of
trees to be discarded as ‘‘burn-in.’’ The majority-rule consensus
was created from the trees produced by the Bayesian analysisafter
the first 168 trees had been discarded as ‘‘burn-in.’’
R
ESULTS
The cpDNA restriction site data yielded 206 char-
acters and the chloroplast DNA sequence data contrib-
uted 992 and 586 characters for the rpl16 intron and
psbA-trnH IGS respectively. The total, aligned data ma-
trix, including binary coded indels, had 1,789 charac-
ters. After exclusion of 20 nucleotides from psbA-trnH
that were of doubtful homology, the dataset consisted
of 1,769 characters, of which 201 were parsimony in-
formative (RFLP 588, psbA-trnH IGS 563, rpl16 in-
tron 550). The g-statistics for the datasets were 20.96
and 21.45 for the cpDNA restriction site and sequence
data respectively, indicating strong phylogenetic sig-
nals (Hillis and Huelsenbeck 1992).
The results of the ILD tests clearly indicated that
there was sufficient congruence between all datasets to
justify combining data (RFLP vs. rpl16 50.05; RFLP
vs. psbA-trnH 50.34; rpl16 vs. psbA-trnH 50.63). Only
phylogenies derived from the combined dataset are
presented here.
The initial heuristic search in PAUP using the com-
bined data found six most parsimonious trees of 638
steps with a consistency index (CI) of 0.84 (rescaled CI
50.70), homoplasy index (HI) of 0.16 (rescaled HI 5
0.30), and retention index of 0.81. The strict consensus
of the six most-parsimonious trees is shown in Fig. 1.
In terms of the number of clades recovered by the anal-
yses, the strict consensus tree (Fig. 1) is 80% resolved
(i.e., resolution index 50.80; Butterworth and Wallace
2004).
The general topology of the strict consensus tree
from the parsimony analysis (Fig. 1) shows that the
Patagonian species Maihuenia poeppigii along with
members of subfamily Opuntioideae form a polytomy
with the remaining genera of the Cactaceae. Although
forming a polytomy, sampled members of subfamily
Opuntioideae (Pereskiopsis,Opuntia and Pterocactus) are
strongly supported as a single clade with a bootstrap
(BS) of 100% and a decay value of 20 steps.
Of the remaining members of the Cactaceae, a sin-
gle, albeit moderately supported, clade (BS 574%, de-
cay 51) contains all species of Pereskia sampled and
two genera of subfamily Cactoideae (Calymmanthium
and Leptocereus); Pereskia does not form a monophyletic
group in this phylogeny. A well-supported clade (BS
5100%, decay 511) contains the widespread P. acu-
leata and the Andean species (P. diaz-romeroana,P. w e -
beriana and both subspecies of P. horrida). Although sta-
tistical support is moderate (BS 574%, decay 50), a
single clade unites the Caribbean species of Pereskia (P.
zinniiflora,P. portulacifolia,P. quisqueyana) with species
distributed in Colombia (P. bleo) and Venezuela (P. gua-
macho). Also included in this clade is the yellow flow-
ered P. aureiflora from southeastern Brazil, which is sis-
ter to P. guamacho, the only other yellow-flowered spe-
cies of Pereskia (BS 578, decay 52). The remaining
species form two moderately to well-supported clades:
1) P. lychnidiflora,Leptocereus quadricostatus,andCalym-
manthium substerile, and 2) the Grandiflora Group, P.
bahiensis,P. stenantha,P. grandifolia,P. sacharosa,andP.
nemorosa.
Although the phylogeny recovered by the Bayesian
analysis (Fig. 1) is not as well resolved as that from
the parsimony analysis (due to the large polytomy to-
wards the base of the tree), it is largely congruent with
the most parsimonious tree (Fig. 1). In the Bayesian
tree, Maihuenia forms the sister species to all remaining
members of the Cactaceae with a posterior probability
(PP) of 0.84. Those clades recovered by the maximum
parsimony analysis but not recovered by the Bayesian
analysis all had little to no bootstrap support. Simi-
larly, most of the well supported (BS $90) clades from
the MP analysis were well supported by the Bayesian
analysis (PP $95%).
D
ISCUSSION
Intergeneric Relationships of Pereskia.Karl Schu-
mann (1898) placed Pereskia in the subfamily Peres-
kioideae, which also included the genus Maihuenia
(Philippi ex Weber) Schumann. A number of subse-
quent workers (Backeberg 1970; Barthlott and Hunt
1993; Gibson and Nobel 1986; Hunt and Taylor 1986)
continued this classification. However, Britton and
Rose (1920) placed Maihuenia in their tribe Opuntieae
rather than with Pereskia. Wallace (1995) reported that
preliminary cpDNA restriction site data for Pereskia in-
dicated that the placement of Maihuenia within sub-
804 [Volume 30SYSTEMATIC BOTANY
F
IG
. 1. Strict consensus (solid lines) tree of six most parsimonious trees for the combined cpDNA RFLP, rpl16 intron, and
psbA-trnH IGS data. Bootstrap values greater than 50% for 100 replicates are given above the branches followed by decay values.
The Bayesian tree is shown as dashed lines with Bayesian posterior probabilities shown below the branches. Infrageneric
groupings within Pereskia referred to in the text (following Leuenberger, 1986) are shown to the right of the cladogram.
2005] 805BUTTERWORTH & WALLACE: PHYLOGENY OF PERESKIA
family Pereskioideae needed reevaluation. Fearn (1996)
circumscribed subfamily Maihuenioideae to include
only the genus Maihuenia. The data presented here
moderately support (BS 574%, decay 51) the exclu-
sion of Maihuenia from both subfamilies Pereskioideae
and Opuntioideae.
Leuenberger (1986) believed that the presence of
leaves as primary photosynthetic organs and variable
ovary position indicated that Pereskia represents the
most ‘‘primitive’’ genus within Cactaceae. Gibson and
Nobel (1986) also referred to Pereskia as possessing the
most ‘‘primitive’’ features in the Cactus family. The
occurrence of Crassulacean Acid Metabolism (CAM)
cycling in species of Pereskia and Maihuenia (Martin and
Wallace 2000) rather than obligate CAM further sup-
ports this hypothesis. Furthermore, facultative CAM
metabolism has been observed in P. guamacho (E. Ed-
wards, pers. comm.). However, the phylogeny pre-
sented in this paper contradicts Leuenberger’s (1986)
hypothesis that Pereskia represents the most ‘‘primi-
tive’’ cacti, suggesting that the genus Maihuenia forms
the sister-group to all other cacti. If our phylogeny is
a good representation of evolution in Cactaceae, and if
the presence of persistent leaves in Pereskia is a ple-
siomorphic condition for the family, then the loss of
leaves and the acquisition of a stem succulent habit
(found in the subfamilies Opuntioideae and Cacto-
ideae) must have occurred independently in each of
these cactus lineages.
The phylogeny shown in Fig. 1 demonstrates that
subfamily Cactoideae (represented in this study by Ca-
lymmanthium substerile and Leptocereus quadricostatus)is
nested within Pereskia, rendering it paraphyletic. The
genus Pereskiopsis (subfamily Opuntioideae) is clearly
not a member of Pereskia. This genus, along with Quia-
bentia Britton and Rose (not sampled for this study) is
unusual within the Opuntioideae due to the presence
of persistent leaves. However, all members of the
Opuntioideae (including Pereskiopsis and Quiabentia)
have glochids and seeds possessing a bony aril—fea-
tures which are not found elsewhere in the Cactaceae
(Anderson 2001).
Infrageneric Relationships within Pereskia.Leuen-
berger’s (1986) subgeneric groupings within Pereskia
(outlined in Table 1) receive considerable support in
the MP phylogeny presented in Fig. 1. The clade con-
taining P. aculeata and the Horrida Group is well sup-
ported in our phylogeny (BS 5100%, decay 511).
Leuenberger (1986) treated P. aculeata by itself, whereas
the other members of this clade were placed in the
Horrida Group. Although the natural range of P. acu-
leata is not clear—it is found throughout eastern South
America and into the Caribbean (Leuenberger 1986),
the members of the Horrida Group have very distinct
distributions in the Andes of Peru and Bolivia. This
strictly Andean clade is well supported (BS 5100%,
decay 512; Fig. 1), and is morphologically and ana-
tomically characterized by reduced leaf size, tricolpate
pollen, tuberous roots, stem stomata, delayed periderm
formation, and lack of brachyblast leaves (Leuenberger
1986). Both subspecies of P. horrida (treated as P. hum-
boldtii by Leuenberger, 1986) are found in northern
Peru in the Rı´o Maran˜o´n drainage system (Leuenber-
ger 1986). Further south, Pereskia weberiana is found in
the dry open forests of the Rı´o Beni drainage area
(Leuenberger 1986). Pereskia diaz-romeroana is distrib-
uted in Central Bolivia. Pereskia weberiana and P. diaz-
romeroana form a well-supported clade (BS 5100%,
decay 510; Fig. 1). Leuenberger (1986) states that al-
though these species are very closely related, they can
be distinguished because P. weberiana lacks the long
hairs on the receptacular and fruit areoles. Pereskia
diaz-romeroana also has purplish-red stamens, as op-
posed to white-pink in P. weberiana.
The clade consisting of Guamacho, Bleo, and Zin-
niiflora Groups are from northern South America and
the Caribbean. All members of this clade (with the ex-
ception of P. bleo) possess stone cells (short, roughly
isodiametric sclereids). All other species of Pereskia (in-
cluding P. bleo) possess either simple or aggregated fu-
siform sclereids (Leuenberger 1986). Within this clade,
Pereskia aureiflora (Brazil) and P. guamacho (Colombia
and Venezuela) comprise the Guamacho Group which
corresponds to one of Leuenberger’s (1986) species
groups. This group is noteworthy as it contains the
only species with yellow flowers and provides possible
evidence of an independent, long-distance dispersal
event from northern South America into Brazil.
The Brazilian and Argentinean species of Pereskia
found in the Grandiflora Group (Fig. 1) were united
into a single group by Leuenberger (1986) and form a
well supported clade (BS 5100%, decay 516; Fig. 1).
These species have aggregated fusiform sclereids and
12–15 colpate pollen with the exception of P. grandifolia,
which has 9–15 colpi (Leuenberger 1986). Pereskia ne-
morosa and P. sacharosa have been confused in the past.
Pereskia nemorosa is distributed in southern Brazil, Par-
aguay, northeastern Argentina, and Uruguay whereas
P. sacharosa is generally found further west in south-
eastern Brazil, Bolivia, Paraguay, and northwestern Ar-
gentina. Pereskia nemorosa also has the largest flowers
in the genus, besides the presence of staminodal hairs
which are lacking in its closest relatives (Leuenberger
1986). The natural distribution of P. grandifolia is un-
certain—it is commonly planted throughout eastern
South America, the Caribbean, Central America, and
Florida. Specimens of P. grandifolia subsp. violacea were
misidentified as P. bahiensis by a number of authors
including Leuenberger (1976), Barthlott (1979), and
Rauh (1979). Subsequently Leuenberger (1986) formal-
ly described P. grandifolia var. violacea (as distinct from
P. bahiensis) noting that while herbarium specimens of
806 [Volume 30SYSTEMATIC BOTANY
var. violacea are almost impossible to distinguish from
var. grandifolia, there are notable differences in the col-
oration of bracts and flower buds in live material. Tay-
lor and Zappi (1997) changed the rank of this taxon to
subspecies. Our phylogeny (Fig. 1) clearly separates P.
bahiensis from P. grandifolia var. violacea, supported by
12 unique mutational differences in our dataset (nine
RFLP gains, three single point mutations in the psbA-
trnH IGS a nd rpl16 intron). Furthermore, only a single
unique RFLP gain supports the clade containing P.
grandifolia var. grandifolia,P. bahiensis,andP. stenantha.
Pereskia stenantha and P. bahiensis form a species pair
in our phylogeny (BS 5100%, decay 56), supported
by five unique RFLP gains and a single, unique nucle-
otide substitution in the rpl16 intron. Leuenberger
(1986) noted that these two species are almost identical
vegetatively. However, the flowers of P. stenantha are
unique in the genus due to their urceolate corolla and
larger nectary, indicative of hummingbird pollination;
Ritter (1979) suggested this may be due to rapid evo-
lution in this taxon. Pereskia bahiensis is very similar in
appearance to P. grandifolia but according to Leuenber-
ger (1986) it has shorter, fleshier leaves, smaller seeds
and inflorescences containing fewer flowers.
Biogeography of Pereskia.The distribution of Per-
eskia in Central and South America ranges from the
Mexican state of Guerrero through Central America,
along the eastern edges of the Andes into Argentina,
eastward into the West Indies and southward to Brazil,
Uruguay, and Paraguay. Leuenberger (1986) states that
the genus is doubtfully native in Florida and is absent
from the Pacific side of the Andes. There are a number
of areas of endemism—the Greater Antilles, Brazil,
and eastern Andes of Bolivia and Peru.
Based on the presence of assumed ‘‘primitive’’ char-
acters in the genus, Backeberg (1942) postulated that
ancestors of the cactus family would be found in the
region of the West Indies and Central America, fol-
lowed by dispersal of cacti to the arid zones of North
and South America. Leuenberger (1986) does not to-
tally accept the reasoning of Backeberg, preferring a
center of origin for the Cactaceae that is located on the
northwestern South American continent. He states that
the dispersal of the major groups of cacti to North
America and the Caribbean, and the fact that only one
species of cactus has dispersed to Africa, supports his
hypothesis. Applequist and Wallace (2001) investigat-
ed the biogeography of the ‘‘Portulacaceous Cohort’’
based upon chloroplast ndhF sequence data. Their data
showed that the majority of species in the sister-clade
to the Cactaceae have modern-day distributions in
North and South America, giving credence to a New
World origin for the Cactaceae.
Based upon our phylogeny (Figs. 1, 2), Leuenber-
ger’s (1986) hypothesis of the origin of Pereskia in
northwestern South America appears reasonable.
However, in the absence of fossil data for cacti and a
better understanding of divergence times for the fam-
ily and its subfamilies, a more detailed discussion of
the location of origin for the Cactaceae and Pereskia can
only be speculation. This is certainly the case consid-
ering the phylogenetic placement (albeit only moder-
ately supported, Fig. 1) of Maihuenia, whose extant
members with a geographic distribution in southern
South America are far from the hypothesized center of
origin for the cactus family. If Leuenberger’s theory of
the origin of Pereskia is accurate, then our data support
an early divergence of Pereskia with migration of one
lineage along the Andes to the region of present day
Bolivia and Peru (Fig. 2). This region currently harbors
P. horrida,P. diaz-romeroana,andP. weberiana, which,
along with the widespread P. aculeata, form the sister-
clade to all other species within the genus. A major
migration into Brazil, Paraguay, Bolivia, Uruguay and
Argentina is indicated by the clade containing P. gran-
difolia,P. stenantha,P. bahiensis,P. nemorosa and P. sa-
charosa (Fig. 2). The species from northern South
America (P. guamacho and P. bleo) are closely related to
the Caribbean species (P. quisqueyana,P. zinniiflora and
P. portulacifolia); they form a well-supported clade in-
dicative of a single migration into the Caribbean (Fig.
2).
Evolution of the Earliest Cactus Lineages. The
placement of Maihuenia in our phylogeny (Fig. 1) is
problematic with respect to the concept of Pereskia rep-
resenting the earliest lineages of the cacti. To date, re-
lationships among the basal lineages of the cacti have
not been fully resolved. Hershkovitz and Zimmer
(1997) placed the Cactaceae as a monophyletic assem-
blage within members of the Portulacaceae. Although
their taxon sampling for the Cactaceae was limited to
five species (one species of Pereskiopsis,twoofPereskia,
and two of Maihuenia), their phylogeny indicated that
Pereskia was paraphyletic due to the inclusion of Mai-
huenia and Pereskiopsis, the latter belonging to subfam-
ily Opuntioideae. Wallace (1995) presented a phylog-
eny of the Cactaceae based upon rbcL sequences in
which a ‘‘basal’’ polytomy within the family is formed
between members of the Opuntioideae, Pereskia,Mai-
huenia, and the Cactoideae. Nyffeler (2002) used trnK/
matK and trnL-trnF sequences for 70 members of the
Cactaceae. While his data further supported the con-
cept of monophyletic subfamilies Cactoideae and
Opuntioideae, members of Pereskia formed a basal
grade, and both species of Maihuenia formed a mono-
phyletic group in a weakly supported sister-group re-
lationship to the Opuntioideae. Work is currently in
progress to evaluate relationships between the ances-
tral lineages of the cacti through the use of DNA se-
quence data (E. Edwards, pers. comm.).
It must be noted that the data presented in this pa-
per do not resolve relationships robustly enough to al-
2005] 807BUTTERWORTH & WALLACE: PHYLOGENY OF PERESKIA
F
IG
. 2. Distribution and biogeography of Pereskia. The cladogram is adapted from the most parsimonious tree shown in
Fig. 1 and distributions are for assumed natural populations after Leuenberger (1986).
low confident taxonomic changes to be made. The use
of cpDNA restriction sites can be criticized because
homologies within the cpDNA restriction site data are
assumed. However, because of the use of hybridization
probes, false homologies within the cpDNA restriction
site dataset are highly unlikely. The markers chosen for
our study were also selected because they have been
shown to evolve at a rate rapid enough for study at
the species level (the original purpose of study). Fur-
ther morphological studies need to be undertaken and
combined with molecular data into a more thorough
synthesis of Pereskia, Pereskioideae, and Cactoideae.
A
CKNOWLEDGEMENTS
. The authors would like to thank the fol-
lowing people and institutions for their assistance in this study:
Beat Ernst Leuenberger (Botanische Garten und Botanisches Mu-
seum Berlin-Dahlem) for most of the living material used in this
study; Huntington Botanic Garden; Linda Prince and Mark Porter
808 [Volume 30SYSTEMATIC BOTANY
(both at Rancho Santa Ana Botanic Garden), and Erika Edwards
(Yale University) for their comments on the manuscript.
L
ITERATURE
C
ITED
A
NDERSON
, E. F. 2001. The cactus family. Portland: Timber Press.
A
PPLEQUIST
,W.L.andR.S.W
ALLACE
. 2001. Phylogeny of the
portulacaceous cohort based on ndhF sequence data. System-
atic Botany 26: 406–419.
B
ACKEBERG
, C. 1942. Zur Geschichte der Kakteen im Verlauf der
Entwicklung des amerikanischen Kontinentbildes. Cactaceae
1942: 1–72.
———. 1956. Descriptiones cactacearum novarum. Jena: G. Fisher.
———. 1970. Das kakteenlexicon. Jena: VEB Gustav Fischer Verlag.
———andF.M.K
NUTH
. 1935. Kaktus-ABC. Copenhagen: Nordisk
Forlag.
B
ARTHLOTT
, W. 1979. Cacti. Cheltenham: Stanley Thornes Ltd.
——— and D. H
UNT
. 1993. Cactaceae. Pages 161–197 in The fam-
ilies and genera of vascular plants, ed. K. Kubitzki. Berlin:
Springer-Verlag.
B
ERGER
, A. 1926. Die Entwicklungslinien der Kakteen. Jena: Gustav
Fisher Verlag.
B
RAVO
-H
OLLIS
, H. 1978. Las Cacta´ceas de Mexico. Mexico City: Univ-
ersidad Nacional Autonoma de Mexico.
B
REMER
, K. 1988. The limits of amino acid sequence data in an-
giosperm phylogenetic reconstruction. Evolution 42: 795–803.
B
RITTON
,N.L.andJ.N.R
OSE
. 1920. The Cactaceae. Washington
DC: Carnegie Institution.
B
UTTERWORTH
, C. A. and R. S. W
ALLACE
. 2004. Phylogenetic stud-
ies of Mammillaria (Cactaceae)—insights from chloroplast se-
quence variation and hypothesis testing using the parametric
bootstrap. American Journal of Botany 91: 1086–1098.
D
ICKIE
, S. L. 1996. Phylogeny and evolution in the subfamily Opun-
tioideae (Cactaceae): insights from rpl16 intron sequence variation.
M.S. Thesis, Iowa State University, Ames, IA.
D
OYLE
,J.J.andJ.L.D
OYLE
. 1987. A rapid DNA isolation proce-
dure for small quantities of fresh leaf tissue. Phytochemical
Bulletin 19: 11–15.
E
RIKSSON
, T. 1998. AutoDecay, version 4.0 (program distributed by
the author). Stockholm: Bergius Foundation, Royal Swedish
Academy of Sciences.
F
ARRIS
,J.S.,M.K
A
¨ELLERSJO
¨
,A.G.K
LUGE
, and C. B
ULT
. 1995.
Testing significance of incongruence. Cladistics 10: 315–319.
F
EARN
, P. J. 1996. A review of the origins of the cactus family and the
search for a system of classification. Wimborne: privately pub-
lished.
F
ELSENSTEIN
, J. 1981. Evolutionary trees from DNA sequences: a
maximum likelihood approach. Journal of Molecular Evolution
17: 368–376.
G
IBSON
,A.C.andP.S.N
OBEL
. 1986. The cactus primer. Cambridge:
Harvard University Press.
G
RAHAM
,S.W.,P.A.R
EEVES
,A.C.E.B
URNS
, and R. G. O
LMSTEAD
.
2000. Microstructural changes in noncoding chloroplast
DNA: interpretation, evolution, and utility of indels and in-
versions in basal angiosperm phylogenetic inference. Inter-
national Journal of Plant Sciences 161: S83–S96.
H
ERSHKOVITZ
,M.A.andE.A.Z
IMMER
. 1997. On the evolutionary
origins of the cacti. Tax on 46: 217–232.
H
ILLIS
,D.M.andJ.P.H
UELSENBECK
. 1992. Signal, noise, and re-
liability in molecular phylogenetic analyses. Journal of Hered-
ity 83: 189–195.
H
UELSENBECK
,J.P.andF.R
ONQUIST
. 2001. MrBayes: Bayesian in-
ference of phylogeny. Bioinformatics 17: 754–755.
H
UNT
, D. 1999. CITES Cactaceae checklist. Kew: Royal Botanic Gar-
dens.
——— and N. T
AYLOR
. 1986. The genera of the Cactaceae: towards
a new consensus. Bradleya 4: 65–78.
J
ANSEN
,R.K.andJ.D.P
ALMER
. 1987. A chloroplast DNAinversion
marks an evolutionary split in the sunflower family (Aster-
aceae). Proceedings of the National Academy of Science, USA 84:
5818–5822.
K
ELCHNER
,S.A.andL.G.C
LARK
. 1997. Molecular evolution and
phylogenetic utility of the chloroplast rpl16 intron in Chusquea
and the Bambusoideae (Poaceae). Molecular Phylogenetics and
Evolution 8: 385–397.
L
EUENBERGER
, B. E. 1976. Die pollenmorphologie der Cactaceae. Stutt-
gart: J. Cramer.
———. 1986. Pereskia (Cactaceae). Memoirs of The New York Botanical
Garden 41: 1–141.
L
INNAEUS
, C. 1753. Species plantarum. Stockholm: Laurentii Salvii.
M
ARTIN
,C.E.andR.S.W
ALLACE
. 2000. Photosynthetic pathway
variation in leafy members of two subfamilies of the Cacta-
ceae. International Journal of Plant Sciences 161: 639–650.
M
ILLER
, P. 1754. The gardener’s dictionary, 4th Ed. London: ‘‘printed
for the author’’.
N
YFFELER
, R. 2002. Phylogenetic relationships in the cactus family
(Cactaceae) based on evidence from trnK/matK and trnL-trnF
sequences. American Journal of Botany 89: 312–326.
P
ALMER
, J. D. 1986. Isolation and structural analysis of chloroplast
DNA. Methods in Enzymology 118: 167–186.
P
LUMIER
, C. H. 1703. Nova plantarum Americanarum genera. Paris: J.
Boudot.
P
OSADA
, D. and K. A. C
RANDALL
. 1998. Modeltest: testing the
model of DNA substitution. Bioinformatics 14: 817–818.
R
AMBAUT
, A. 1995. Sequence Alignment Program, version 1.d1.
Oxford: Oxford University.
R
AUH
, W. 1979. Kakteen an ihren standorten. Berlin: Parey.
R
ITTER
, F. 1979. Kakteen in Su¨damerika. Band 1, Brasilien/Uruguay/
Paraguay. Spangenberg: Friedrich Ritter Selbstverlag.
R
ONQUIST
,F.andJ.P.H
UELSENBECK
. 2003. MrBayes 3: Bayesian
phylogenetic inference under mixed models. Bioinformatics19:
1572–1574.
S
CHUMANN
, K. 1898. Gesamtbeschreibung der kakteen (monographia
cactearum). Neudamm: J. Neuman.
S
HINOZAKI
, K., M. O
HME
,M.T
ANAKA
,T.W
AKASUGI
,N.H
AYASH
-
IDA
,T.M
ATSUBAYASHI
,N.Z
AITA
,J.C
HUNWONGSE
,J.O
BOK
-
ATA
,K.Y
AMAGUCHI
-S
HINOZAKI
,C.O
HTO
,K.T
ORAZAWA
, B.-
Y. M
ENG
,M.S
UGITA
,H.D
ENO
,T.K
AMAGOSHIRA
,K.Y
AMA
-
DA
,J.K
USUDA
,F.T
AKAIWA
,A.K
ATO
,N.T
OHDOH
,H.S
HI
-
MADA
andM.S
UGIURA
. 1986. The complete nucleotide
sequence of the tobacco chloroplast genome: its gene orga-
nization and expression. European Molecular Biology Organi-
zation Journal 5: 2043–2049.
S
MITH
,G.E.andM.D.S
UMMERS
. 1980. The bidirectional transfer
of DNA and RNA to nitrocellulose or diazobenzyloxymethyl
paper. Analytical Biochemistry 109: 123–129.
S
WOFFORD
, D. C. 1999. PAUP*. Phylogenetic analysis using parsi-
mony (* and other methods), version 4.0b2. Sunderland: Sin-
auer Associates.
T
AYLOR
,N.andD.Z
APPI
. 1997. Nomenclatural adjustments and
novelties in Brazilian Cactaceae. Cactaceae Consensus Initiatives
3: 7–8.
W
ALLACE
, R. S. 1995. Molecular systematic study of the Cactaceae:
Using chloroplast DNA variation to elucidate cactus phylog-
eny. Bradleya 13: 1–12.
——— and J. H. C
OTA
. 1996. An intron loss in the chloroplast gene
rpoC1 supports a monophyletic origin for the subfamily Cac-
toideae. Current Genetics 29: 275–281.
