-
1827
American Journal of Botany 97(11): 18271847. 2010.
American Journal of Botany 97(11): 18271847, 2010;
http://www.amjbot.org/ 2010 Botanical Society of America
The suborder Cactineae of order Caryophyllales (sensu Thorne and
Reveal, 2007 ) comprises ca. 2000 species in 130 genera and eight
families (estimated from Kubitzki et al., 1993 ), distributed
mainly in the Americas, Africa, and Australia. This group (also
known as Portulacineae) includes families Ba-sellaceae, Cactaceae,
Didiereaceae, Halophytaceae, and Portu-
lacaceae ( Thorne and Reveal, 2007 ). Recently, it has been
proposed that Portulacaceae should be split into four families (
Nyffeler and Eggli, 2010 ), resulting in a monogeneric
Portu-lacaceae ( Portulaca ) and Anacampserotaceae, Montiaceae, and
Talinaceae. Molecular data have provided important insights into
the relationships within the strongly supported, monophyletic
Cactineae ( Applequist and Wallace, 2001 ; M ller and Borsch, 2005
; Applequist et al., 2006 ; Nyffeler, 2007 ; Brockington et al.,
2009 ; Nyffeler and Eggli, 2010 ) and have led to a new classifi
cation of families, particularly involving members of Portulacaceae
s.l. Hershkovitz and Zimmer (1997) , using ribo-somal internal
transcribed spacer (ITS) sequences, found that Portulacaceae were
not monophyletic. In particular, Cactaceae were nested within the
family, and Ceraria and Portulacaria , considered genera of
Portulacaceae at the time, were more closely related to
Didiereaceae and Basellaceae. These relation-ships were confi rmed
later by Applequist and Wallace (2001) using sequences from the
chloroplast gene ndhF , which led to the transfer of Ceraria and
Portulacaria , along with Calyp-trotheca , to Didiereaceae (
Applequist and Wallace, 2003 ). In addition, these authors
uncovered a clade that grouped Cacta-ceae with Talinum + Talinella
, Portulaca , and tribe Anacamp-seroteae of Portulacaceae, although
without internal resolution. Subsequent studies using other
chloroplast and mitochondrial regions ( Applequist et al., 2006 ;
Nyffeler, 2007 ) have recov-ered this monophyletic group, known as
the ACPT clade ( Nyffeler, 2007 ), with strong support, but
internal relationships
1 Manuscript received 22 June 2010; revision accepted 14
September 2010.
The authors thank the following persons and institutions who
kindly provided plant material for this study: Jennifer
Cruse-Sanders, Urs Eggli, Holly Forbes, Naomi Fraga, Patricia
Jaramillo, Sean Lahmeyer, James Matthews, Clifford Morden, Robert
Nicholson, Mark Porter, Ernesto Sandoval, John Trager, National
Tropical Botanical Garden, Pretoria National Herbarium (PRE), and
the Waimea Arboretum Foundation. Lucinda McDade, Wendy Applequist,
Mark Simmons, and two anonymous reviewers provided helpful comments
on the manuscript. Elena Voznesenskaya kindly provided the carbon
isotope ratio value for Portulaca elatior . The study was fi nanced
by Rancho Santa Ana Botanic Garden and the Botanical Society of
America. Financial support to G. O. was provided by The Fletcher
Jones Foundation, Comisi n Nacional de Ciencia y Tecnolog a
(Mexico), Fundaci n Prywer (Mexico), and the Instituto de Ecolog a,
A. C. (Mexico).
2 Author for correspondence (e-mail:
[email protected]);
present address: California Academy of Sciences, Botany
Department, 55 Music Concourse Drive, Golden Gate Park, San
Francisco, CA 94118 USA
doi:10.3732/ajb.1000227
MOLECULAR PHYLOGENETICS OF SUBORDER CACTINEAE (CARYOPHYLLALES),
INCLUDING INSIGHTS INTO
PHOTOSYNTHETIC DIVERSIFICATION AND HISTORICAL BIOGEOGRAPHY 1
Gilberto Ocampo 2 and J. Travis Columbus
Rancho Santa Ana Botanic Garden and Claremont Graduate
University, 1500 North College Avenue, Claremont, California
91711-3157 USA
Premise of the study: Phylogenetic relationships were
investigated among the eight families (Anacampserotaceae,
Basellaceae, Cactaceae, Didiereaceae, Halophytaceae, Montiaceae,
Portulacaceae, Talinaceae) that form suborder Cactineae (=
Portu-lacineae) of the Caryophyllales. In addition, photosynthesis
diversifi cation and historical biogeography were addressed.
Methods: Chloroplast DNA sequences, mostly noncoding, were used
to estimate the phylogeny. Divergence times were cali-brated using
two Hawaiian Portulaca species, due to the lack of an unequivocal
fossil record for Cactineae. Photosynthetic pathways were
determined from carbon isotope ratios ( 13 C) and leaf anatomy.
Key results: Maximum likelihood and Bayesian analyses were
consistent with previous studies in that the suborder, almost all
families, and the ACPT clade (Anacampserotaceae, Cactaceae,
Portulacaceae, Talinaceae) were strongly supported as
mono-phyletic; however, relationships among families remain
uncertain. The age of Cactineae was estimated to be 18.8 Myr. Leaf
anatomy and 13 C and were congruent in most cases, and
inconsistencies between these pointed to photosynthetic
intermedi-ates. Reconstruction of photosynthesis diversifi cation
showed C 3 to be the ancestral pathway, a shift to C 4 in
Portulacaceae, and fi ve independent origins of Crassulacean acid
metabolism (CAM). Cactineae were inferred to have originated in the
New World.
Conclusions: Although the C 3 pathway is inferred as the
ancestral state in Cactineae, some CAM activity has been reported
in the literature in almost every family of the suborder, leaving
open the possibility that CAM may have one origin in the group.
Incongruence among loci could be due to internal short branches,
which possibly represent rapid radiations in response to
in-creasing aridity in the Miocene.
Key words: C 4 photosynthesis; Cactineae; Cactaceae;
Caryophyllales; Crassulacean acid metabolism divergence times;
his-torical biogeography; photosynthesis diversifi cation;
Portulacaceae; Portulacineae.
-
1828 American Journal of Botany [Vol. 97
criminates less against the 13 C isotope than Rubisco ( Lajtha
and Marshall, 1994 ; Winter and Holtum, 2002 ); thus, carbon
iso-tope ratios ( 13 C) can be used to distinguish plants that have
C 4 and CAM photosynthesis from those that use the C 3 pathway.
Although the use of 13 C can serve as the fi rst step in
determin-ing the photosynthetic pathway in a group, additional
evidence may be needed to discriminate C 4 from CAM photosynthesis,
because their 13 C values overlap ( O Leary, 1988 ; Winter and
Holtum, 2002 ; Sage et al., 2007 ), or to detect photosynthetic
intermediates (e.g., C 3 -C 4 ; Monson et al., 1984 ; Rawsthorne
and Bauwe, 1998 ). Other sources of evidence include stem and leaf
anatomy and biochemical assays (e.g., Ku et al., 1983 ; Rajendrudu
et al., 1986 ; Brown and Hattersley, 1989 ; Sage et al., 2007
).
The overarching aim of this study was to obtain a more ro-bust
phylogenetic estimate of relationships within Cactineae by using
different, noncoding chloroplast markers from those used in
previous studies. Specifi cally, the rpl14 rps8 infA rpl36 re-gion,
atpI atpH intergenic spacer, and ndhA intron ( Shaw et al., 2007 )
were used to explore their utility for resolving relation-ships
among the families of Cactineae. Using the phylogenetic estimate,
we studied the diversifi cation of the photosynthetic pathway
(inferred from 13 C values and leaf anatomy) and his-torical
biogeography of the group. In addition, age estimates of the major
groups were calculated based on a relaxed molecular clock model and
indirect calibration methods using as a refer-ence the age of
specifi c Hawaiian Islands inhabited by endemic Portulaca species.
These age estimates were used to address questions of place and
time of origin of the families of Cactineae.
MATERIALS AND METHODS
Taxon sampling Fifty-one species were sampled from all families
recog-nized in Cactineae by Nyffeler and Eggli (2010) and
corresponding to the major clades recovered in recent studies (
Applequist and Wallace, 2001 ; Applequist et al., 2006 ; Nyffeler,
2007 ; Brockington et al., 2009 ; Nyffeler and Eggli, 2010 )
(Appendix 1). Early-diverging species in all families (see Edwards
et al., 2005 ; Applequist et al., 2006 ; Nyffeler, 2007 ) were
included in the study, although relationships within Basellaceae
are not known, and only one genus (of four) was sampled. Species of
Aizoaceae, Molluginaceae, Nyctaginaceae, and Phy-tolaccaceae were
selected for rooting the phylogenies because they are known to be
close relatives of the suborder ( Rettig et al., 1992 ; Downie and
Palmer, 1994 ; Applequist and Wallace, 2001 ; Cu noud et al., 2002
; Applequist et al., 2006 ; Brockington et al., 2009 ).
Unfortunately, attempts to obtain sequences from Mollugo
(Molluginaceae) failed, so the outgroup taxa employed were from the
remaining three families (Appendix 1).
DNA extraction and sequencing Sources of DNA included leaves
taken directly from live plants, leaves dried in silica gel,
herbarium specimens, and in one instance ( Portulaca sclerocarpa ),
a DNA aliquot (Appendix 1). Total ge-nomic DNA was extracted from
10 mg of dried material or 20 mg of fresh tissue using the modifi
ed CTAB method of Doyle and Doyle (1987) or DNeasy kits (Qiagen,
Valencia, California, USA). In general, quality of the genomic DNA
obtained by these two methods was suffi cient for performing the
polymerase chain reaction (PCR) with the selected markers, although
the DNeasy kits out-performed the CTAB method when samples were
highly mucilaginous. DNA extracted using the CTAB method was
quantifi ed and diluted to a concentration of 10 ng/ L, whereas the
concentration of DNA extracted using the DNeasy kits was not
measured because typically a concentration of 10 ng/ L is obtained
by this method.
Preliminary analyses of a data matrix comprising sequences from
GenBank representing 10 loci and more than 100 Cactineae taxa
showed that a superma-trix approach does not improve internal nodal
support within the suborder. Therefore, we conducted a study using
the chloroplast rpl14 rps8 infA rpl36 region (comprising coding and
spacer sequences), atpI atpH intergenic spacer,
have low support values. Other relationships within Cactineae
are not known with certainty, because the branching order is
ambiguous or poorly supported ( Applequist and Wallace, 2001 ;
Applequist et al., 2006 ; Nyffeler, 2007 ; Nyffeler and Eggli, 2010
). Halophytaceae, a monotypic family from Patagonia and
traditionally considered a member of Chenopodiaceae Vent. (e.g.,
Cronquist, 1981 ), have been shown in molecular phyloge-netic
studies to be part of Cactineae ( Cu noud et al., 2002 ; M ller and
Borsch, 2005 ; Applequist et al., 2006 ; Brockington et al., 2009 ;
Nyffeler and Eggli, 2010 ), but its placement within the suborder
is unclear.
Cactineae tend to be better represented in the southern
hemi-sphere, although Portulaca has a worldwide distribution,
mainly in tropical and subtropical regions of both hemispheres.
Cacta-ceae and Montiaceae have important centers of diversity in
North America ( Barthlott and Hunt, 1993 ; Hershkovitz, 1993 ).
This distribution pattern suggests that the origin of the suborder
may have been in South America ( Applequist and Wallace, 2001 ),
but the temporal scale is unknown. The lack of an unam-biguous
fossil record for Cactineae has been a limiting factor in studying
its origins (see Hershkovitz and Zimmer, 2000 ). Age estimates for
cacti range from ca. 100 million years (Myr; Croizat, 1952 ;
Mauseth, 1990 ; Wallace and Gibson, 2002 ) to ca. 30 Myr (
Hershkovitz and Zimmer, 1997 ), while Montiaceae are esti-mated to
be 8 16 Myr (as western American Portulacaceae; Hershkovitz and
Zimmer, 2000 ), but to date there are no hy-potheses of the age of
the suborder, which impedes understand-ing of its evolution and
historical biogeography.
A fascinating aspect of evolution within Cactineae is
adapta-tion to xeric habitats, especially in plant morphology but
also extending to the physiological level in terms of
photosynthetic pathways. All three primary photosynthetic variants
C 3 , C 4 , and Crassulacean acid metabolism (CAM) are present in
the suborder (e.g., Winter, 1979 ; Nobel and Hartsock, 1986 ; Sage
et al., 1999 ; Guralnick and Jackson, 2001 ; Sayed, 2001 ;
Guralnick et al., 2008 ). In C 3 plants, 1,5-bisphosphate
carboxylase/oxygenase (Rubisco) catalyzes the reaction where
1,5-bisphosphate (RuBP) reacts with atmospheric CO 2 as the fi rst
step of the photosyn-thetic reaction ( Ehleringer and Monson, 1993
). Atmospheric O 2 and CO 2 are competitive substrates for Rubisco,
but the enzyme has more specifi city for the latter. However, the
concentration of CO 2 is signifi cantly reduced as it diffuses from
the atmo-sphere into the photosynthetic tissue, favoring the
oxygenation of RuBP. This condition, in addition to an eventual
liberation of the CO 2 , is termed photorespiration, which reduces
the overall photosynthetic effi ciency ca. 33% in C 3 plants (
Ehleringer and Monson, 1993 ). In C 4 and CAM plants,
phosphoenolpyruvate carboxylase (PEP) catalyzes the fi rst step of
the photosynthetic reaction instead of Rubisco. These
photosynthetic pathways de-pend upon structural (C 4 ) or temporal
(CAM) separation of PEP and Rubisco activity to concentrate CO 2
for the Calvin cycle while reducing photorespiration (C 4 ) and
water loss (CAM). In C 4 photosynthesis, fi xed carbon is
transported in the form of malate or aspartate to special cells
that form the vascular bundle sheath, where it is released as CO 2
and enters the Calvin cycle ( Kanai and Edwards, 1999 ). In the CAM
pathway, CO 2 is fi xed as malate in the mesophyll at night, when
the stomata are open, and stored inside the cell vacuoles. The
malate is subsequently decarboxylated during light hours, when the
stomata are closed, and enters the Calvin cycle in the mesophyll (
Winter and Smith, 1996 ; Nelson and Sage, 2008 ). In this case, CO
2 concentration remains high because it cannot escape through the
closed sto-mata ( Ehleringer and Monson, 1993 ). PEP carboxylase
dis-
-
1829November 2010] Ocampo and Columbus Phylogenetics of
Cactineae
Estimation of divergence times Estimation of divergence times
was con-ducted using a series of programs that are part of the
Bayesian Evolutionary Analysis Sampling Trees package (BEAST)
version 1.5.1 ( Drummond and Rambaut, 2007 ). The XML fi le for
BEAST was prepared for the combined data matrix in the Bayesian
Evolutionary Analysis Utility (BEAUti), and by manu-ally assigning
the best-fi t model of evolution suggested by MODELTEST for each
locus. As the fossil record for the suborder is uncertain (see
Chaney [1944] for a putative cactus fossil from the Eocene that was
refuted by Brown [1959] ; Muller [1981] and Ravn [1987] cite fossil
pollen records for Portulacaceae and Montiaceae from the upper
Miocene to the Pliocene, although Hershkovitz and Zimmer [2000]
question their correct identifi cation), an indirect approach using
estimations of geological events was undertaken. The approach
relies on the dates of geological events in the Hawaiian Islands,
as used by other researchers (e.g., Chac n et al., 2006 ;
VanderWerf et al., 2010 ). The age of particular is-lands or groups
of islands was taken as the age of Portulaca species endemic to
those islands. However, the results should be viewed with caution,
because dat-ing nodes with these volcanic hotspots overlooks the
possibility that the se-lected species existed before the islands
on which they presently occur arose (see Heads, 2005 ). Portulaca
molokiniensis is a narrow endemic found in the Maui volcanic
islands complex ( Naughton et al., 1980 ) of Molokini, Puukoae
Islet, and Kahoolawe ( Wagner et al., 1999 ). The ages of these
islands range from 0.148 to 1.03 Myr according to the K Ar method (
Naughton et al., 1980 ; Sherrod et al., 2003 ). The closest
relative of P. molokinensi s is P. howellii , en-demic to the Gal
pagos Islands ( Wiggins et al., 1971 ), thus the divergence
be-tween these two species was set at 1.03 0.18 million years ago
(Ma) ( Naughton et al., 1980 ), which is the age of Kahoolawe, the
oldest island where P. moloki-niensis is found. Calibration using
the Hawaiian Islands ( P. molokiniensis ) was preferred over the
Gal pagos ( P. howellii ) because the latter species is
distrib-uted throughout the Gal pagos Archipelago ( Wiggins et al.,
1971 ), and the ages of these islands differ greatly ( Bailey, 1976
), making it diffi cult to choose one calibration date. Another
Hawaiian endemic is P. sclerocarpa . The node for P. sclerocarpa
and P. villosa was calibrated to 0.43 0.02 Myr ( McDougall and
Swanson, 1972 ), the oldest hypothesized age for the island of
Hawaii (Kohala volcano) where P. sclerocarpa is endemic ( Wagner et
al., 1999 ). Although one Portulaca specimen in Po opo o matches
the description of P. sclerocarpa , Wagner et al. (1999) suggest
that it may be a recent dispersal from Hawaii or that the capsule
characteristics in this specimen have converged with P. sclerocarpa
.
To estimate divergence times, we used a relaxed clock
(uncorrelated lognor-mal; Drummond et al., 2006 ) and a Yule prior
on birth rate of new lineages ( Drummond and Rambaut, 2007 ),
enforcing Mirabilis , Rivina , and Sesuvium as the outgroup.
Fourteen independent analyses were run for 10 000 000 genera-tions
at Cornell University s Computational Biology Service Unit
(http://cbsua-pps.tc.cornell.edu/beast.aspx), saving every 1000th
tree. Trace fi les were loaded into the program Tracer version
1.4.1 ( Rambaut and Drummond, 2007 ) looking for an effective
sample size (ESS) > 200 for all parameters sampled from the
MCMC. Trees from the 14 independent analyses were combined in the
program LogCombiner, and the resulting tree fi le was run in the
program TreeAnnotator to summarize tree information in a maximum
clade credibility tree (the tree with the highest product of all
the posterior clade probabilities), discarding the fi rst 14 000
trees. The program FigTree version 1.2.3 ( Rambaut, 2009 ) was used
for visualizing results on divergence dates.
Historical biogeography Analysis of potential ancestral
distribution areas for taxa of Cactineae used a Bayesian approach
to dispersal vicariance analysis (DIVA; Ronquist, 1997 ), following
the method of Nylander et al. (2008) as implemented in the program
S-DIVA version 1.5c ( Yu et al., 2010 ), which ac-counts for
uncertainty in the phylogenetic estimate. The Bayes DIVA analysis
was done using 1000 random trees after the burn-in period from the
BEAST run and using the topology of the maximum clade credibility
tree, allowing the re-construction of four maximum ancestral areas
at each node. Distribution areas were considered at the continental
level. Widespread species were coded as present in multiple
regions, and only the natural distributions were taken into account
(Appendix 1). Because the outgroup is small in this study for
biogeo-graphical reconstruction, simulations were run to evaluate
the impact of out-group distribution on the results. The
distributions of the outgroup species were modifi ed to restrict
them to the southern hemisphere, where apparently the
Caryophyllales have their origin ( Raven and Axelrod, 1974 ),
specifi cally: (1) South America, (2) Africa, and (3) both
continents.
Carbon isotope ratio tests and leaf anatomy Determination of
photosyn-thetic pathways in Cactineae was accomplished by 13 C data
complemented by leaf anatomy, an approach that has been used by
Sage et al. (2007) to discriminate
and ndhA intron to explore their utility for resolving
relationships among fami-lies of Cactineae. These loci are among
the 12 most variable chloroplast mark-ers recommended by Shaw et
al. (2007) . Primers for amplifi cation via PCR were as in Shaw et
al. (2007) . Amplifi cations were performed in 25 L reac-tions with
0.62 units of Taq DNA polymerase (Promega, Madison, Wisconsin,
USA), 2.5 L of (NH 4 ) 2 SO 4 buffer, 0.5 pM of forward and reverse
primers, 0.25 mM MgCl 2 , 0.25 mM dNTPs, and 0.25 L of BSA 100 for
a fi nal con-centration of 1%, plus 1 L of genomic DNA in a
Robocycler 96 or RoboCy-cler Gradient 96 thermal cycler
(Stratagene, La Jolla, California, USA). PCR cycles were as
follows: (1) initial denaturation at 94 C for 4 min; (2) 35 cycles
of denaturation at 94 C for 1 min, primer annealing at 50 54 C for
1 min, and primer extension for 1 min 30 s at 72 C; and (3) fi nal
elongation for 7 min at 72 C. PCR products were purifi ed by the
PEG precipitation protocol ( Johnson and Soltis, 1995 ).
Alternatively, amplifi cation products were cleaned by adding 3 L
of a solution containing 0.2 L each of Antarctic phosphatase and
exonu-clease I (New England Biolabs, Ipswich, Massachusetts, USA)
and incubating for 30 min at 37 C then for 20 min at 80 C. Cycle
sequencing was carried out with ABI Prism Big Dye Terminator
solution (Applied Biosystems, Foster City, California, USA) using
reactions half the volume recommended by the manufacturer. Internal
sequencing primers were designed for the rpl14 rps8 infA rpl36
region (rps8F: 5 GYR AGA AAA CAT CAA GAA AGA AA 3 ; rps8R: 5 TCC
CGA TCH GTC ATT ATA CC 3 ) and atpI atpH intergenic spacer (atpIF1:
5 ATG GRC RGT TTA CGT TAT GGA 3 ). Products were cleaned using
Sephadex G-50 columns (GE Healthcare, Anaheim, California, USA) and
read on an ABI Prism automated sequencer 3130xl (Applied
Biosys-tems, Foster City, California, USA). Sequences were contiged
and edited using the program Sequencher 4.2.2 (Genes Codes Corp.,
Ann Arbor, Michigan, USA) and deposited in GenBank.
Sequence alignment and phylogenetic analyses DNA sequences were
aligned using the program MUSCLE version 3.7 ( Edgar, 2004 ),
followed by manual alignment with the program Se-Al version 2.0a11
( Rambaut, 2002 ) fol-lowing methods discussed by Morrison (2006) .
To assess congruence among genetic markers, we performed the
incongruence length difference test (ILD; Farris et al., 1994 ) as
implemented in the program PAUP* version 4.0b10 ( Swofford, 2002 ),
with 1000 replicates and 10 random addition sequences. The ILD test
results indicated that rpl14 rps8 infA rpl36 was incongruent with
atpI atpH and the ndhA intron ( P = 0.014 and P = 0.001,
respectively, thus rejecting the null hypothesis of congruent data
at the 0.05 confi dence level). Therefore, each marker was analyzed
separately to assess incongruence in tree topology. Phylogenetic
analyses of individual markers yielded incongruent re-lationships
among families of Cactineae, although with low nodal support ( <
75% bootstrap; < 0.95 posterior probability); thus, a data
matrix including combined sequences of the three regions was
prepared (archived in TreeBASE, study accession S10818 and matrix
accession M6498, http://treebase.org). Indi-vidual markers and the
combined data matrix were analyzed using maximum likelihood (ML;
Felsenstein, 1973 ) in the program Garli version 0.951 ( Zwickl,
2006 ) and Bayesian inference under Markov chain Monte Carlo (MCMC;
Yang and Rannala, 1997 ) in the program MrBayes version 3.1.2 (
Ronquist and Huelsenbeck, 2003 ). ML analyses used the model of
molecular evolution esti-mated by the program MODELTEST version 3.7
( Posada and Crandall, 1998 ), following the recommendation
provided by the Akaike information criterion (AIC; Akaike, 1974 ).
The best-fi t model for rpl14 rps8 infA rpl36 was a transversional
model (TVM) plus a gamma-distributed rate variation (G; Yang, 1993
); for atpI atpH and ndhA , it was general time reversible (GTR;
Tavar , 1986 ) plus G; and for the combined data set it was GTR
plus a proportion of invariant sites (I; Reeves, 1992 ). Bayesian
analyses were conducted using the best-fi t model of evolution
provided by MrModeltest version 2.3 ( Nylander, 2004 ), under the
AIC. The model selected for all data sets was GTR + I. Bayes-ian
analyses were run with two replicates of 10 000 000 generations;
trees were saved every 100th generation, unlinking data partitions
in the combined data matrix; log fi les were visually examined to
check convergence between runs, and the burnin value for obtaining
the majority rule consensus tree was set to ignore the fi rst 25%
of the trees to only include trees after stationarity was reached.
Clade support was determined using nonparametric bootstrapping (
Felsenstein, 1985 ) from 100 ML replicates and Bayesian posterior
probabili-ties ( Rannala and Yang, 1996 ; Li et al., 2000 ).
The Shimodaira Hasegawa (SH; Shimodaira and Hasegawa, 1999 )
test was performed to test selected alternative topologies from
different analyses. Con-straint topologies were prepared in the
program MacClade version 4 ( Maddison and Maddison, 2000 ) and
loaded into Garli. Estimated constrained and uncon-strained
topologies were then loaded into PAUP*, and their likelihood scores
were compared using the SH test with the RELL option.
-
1830 American Journal of Botany [Vol. 97
thetic pathway in the samples because 13 C values of different
photosynthesis types may overlap (see Winter and Holtum, 2002 ;
Guralnick et al., 2008 ).
Leaves of 42 of the 54 species in the study were sectioned
transversely and examined. Material of Opuntia vestita and
Portulaca sclerocarpa was not avail-able; Rhipsalis baccifera lacks
leaves; and leaf sections from nine species proved to be inadequate
for study. The central portion of mature leaves was cut into small
segments ca. 5 mm long and fi xed and stored in a solution of
formalin propionic acid alcohol (FPA; Ruzin, 1999 ). When fresh
material was not avail-able, leaf samples from herbarium specimens
or dried in silica gel for molecular study were treated in a
solution of 10% Aerosol OT or boiled in water for 10 min for
rehydration; however, these samples displayed tissue expansion and
were inadequate for anatomical characterization. The remaining leaf
samples were de-hydrated and embedded in paraffi n via the
following steps: 70% ethanol (EtOH), 2 h; 90% EtOH, 2 h; 95% EtOH,
2 h; 100% EtOH with 1% safranin, overnight; 100% EtOH, 2 h; 2 : 1
100% EtOH : xylene, 2 h; 1 : 2 100% EtOH : xylene, 2 h; xylene, 2
h; xylene, 2 h; 2 : 1 xylene : paraffi n oil, 2 h; 1 : 2 xylene :
paraffi n oil, 2 h; in 58 C oven: paraffi n step 1, 6 h, and
paraffi n step 2, 6 h, followed by fi nal paraffi n embedding.
Thick sections (10 m) were cut using an American Optical 820 rotary
microtome (American Optical, now part of Carl Zeiss Vision, San
Diego, California, USA). The staining schedule, based on Sharman
(1943) , was as follows: xylene, 10 min; xylene, 10 min; 1 : 1 100%
EtOH : xylene, 5 min; 100% EtOH, 5 min; 95% EtOH, 2 min; 90% EtOH,
2 min; 70% EtOH, 2 min; 50% EtOH, 2 min; 30% EtOH, 2 min; H 2 O, 2
min; 2% aqueous ZnCl 2 , 1 min; H 2 O, 5 s; 1 : 25 000 aqueous
safranin O, 5 min; H 2 O, 5 s; 10 g orange G + 25 g tannic acid +
20 drops HCl + 0.2 g thymol + 500 mL H 2 O, 1 min; H 2 O, 5 s; 25 g
tannic
C 4 from C 3 and CAM. Leaf material was used for photosynthetic
pathway de-termination by 13 C analysis except for Opuntia vestita
, for which leaf material was not available, and Rhipsalis
baccifera , which does not have leaves; there-fore, stem material
was used for these two species. In addition, leaves were not
available for Portulaca sclerocarpa ; however, as explained
already, the species was important for calibrating the phylogeny;
thus, it was included only in the estimation of divergence times.
Samples for 13 C analysis were prepared for all species except
Portulaca elatior by drying plant material in an oven at ca. 50 C
for 24 h, grinding ca. 1 mg of dried sample, and placing it in a 5
9 mm tin capsule (Costech Analytical Technologies, Valencia,
California, USA). Sam-ples were sent to the University of
California at Davis Stable Isotope Facility, which uses a PDZ
Europa ANCA-GSL elemental analyzer interfaced to a PDZ Europa 20-20
isotope ratio mass spectrometer (Sercon, Cheshire, UK). Leaf
material of P. elatior was processed at Washington State
University, where it was dried at 80 C for 24 h and analyzed in a
EuroVector elemental analyzer (EuroVector S.p.A., Milan, Italy).
The three photosynthetic pathways discrimi-nate in different
proportions against the isotope 13 C (isotope fractionation; O
Leary, 1988 ). The following scale was used to identify the
photosynthetic pathway. C 3 : typically 25 per mil ( ; O Leary,
1988 ; Raven et al., 2008 ; Guralnick et al., 2008 ), although it
can be as high as 20 due to some CAM activity ( Winter and Holtum,
2002 ); C 4 : 10 to 16 ( O Leary, 1988 ; Sage et al., 2007 ); CAM:
9 to 20 ( O Leary, 1988 ; Winter and Holtum 2002 ).
The carbon isotope discrimination ratios were compared to leaf
anatomy (when samples were available), which likewise is predictive
of photosynthetic pathway. This approach was used to more confi
dently determine the photosyn-
Fig. 1. Bayesian allcompat tree of Cactineae, using a combined
data matrix of the rpl14 rps8 infA rpl36 region, atpI atpH
intergenic spacer, and ndhA intron. The ML topology is identical to
the Bayesian estimate. p.p. = posterior probability.
-
1831November 2010] Ocampo and Columbus Phylogenetics of
Cactineae
ity) support in each phylogeny. However, besides the ACPT clade,
relationships among the families are equivocal and have low support
(Bayesian allcompat trees of each individual marker are shown in
online Appendices S1 S3, wherein the to-pological differences with
the ML estimates are indicated). The ML and Bayesian analyses of
the combined data set (3652 bp) yielded identical topologies ( Fig.
1 ), although relationships among families have low support.
Cactineae are strongly sup-ported, as are all the families except
for Didiereaceae, which has low bootstrap support. Montiaceae are
the fi rst diverging member of the order, and Basellaceae are
sister to Didiereaceae. The combined analysis shows Halophytaceae
as sister to the ACPT clade. Relationships inside the ACPT clade,
with Tali-naceae basal and Cactaceae sister to Portulacaceae, are
weakly supported. The SH test could not reject alternative
hypotheses of relationships involving placement of Halophytaceae, a
basal position of Didiereaceae, and the relationships within the
ACPT clade ( Table 1 ).
Divergence times A chronogram obtained using BEAST is shown in
Fig. 2 and has an identical topology to the ML tree and Bayesian
allcompat consensus tree of the combined data matrix, except for
the branching pattern among Pereskia acu-leata , P. lychnidifl ora
, and P. sacharosa (Cactaceae). The mean values for the age of the
most recent common ancestor (MRCA) and the maximum and minimum
values for the 95% highest posterior density interval (HPD) for
selected nodes are pre-sented in Table 2 . According to the
analysis, the age of the sub-order is 18.8 (6.7 33.7) Myr, which
corresponds to early Miocene ( IUGS, 2009 ).
Historical biogeography The optimal reconstruction from the
Bayes DIVA analysis showed 47 dispersals. The MRCA of Cactineae was
recovered as widely distributed from South to North America (69.3%
probability; Fig. 2 ). Although the phy-logenetic relationships
among the families of Cactineae are not highly supported, the
ancestral distribution for each family ex-cept Portulacaceae had
high probability values ( Fig. 2 ). Monti-aceae are North American
in origin, but dispersed to South America and Australia.
Didiereaceae are inferred to have Afri-can Madagascan origin.
Anacampserotaceae were found to have originated in the Americas and
Cactaceae in South Amer-ica and the Caribbean region. The rest of
the taxa in the subor-der have a South American origin (although
equivocal in Portulacaceae), with multiple dispersals to other
continents.
Two simulation analyses of species with hypothetical
distri-butions partially supported the origin of Cactineae in the
Amer-icas (48.19 56.12% probability). The exception was when the
three outgroup species were restricted to South America, which
resulted in an ancestor distributed in the New World, Africa, and
Madagascar (52.09%).
Photosynthetic pathway determination Representative leaf anatomy
micrographs of the examined species are shown in Figs. 3 5 . Leaf
anatomy corresponded to the photosynthetic pathways suggested by 13
C values ( Table 3 ) with some excep-tions. Grahamia bracteata
(Anacampserotaceae), Quiabentia verticillata (Cactaceae), and
Decarya madagascariensis , Di-dierea madagascariensis , and D.
trolli (Didiereaceae), have CAM-like 13 C values, but the leaves
were considered to have C 3 anatomy. Therefore, the photosynthetic
pathway for these species was scored as facultative CAM for
character recon-struction. CAM photosynthesis (including
facultative CAM)
acid + 0.2 g thymol + 500 mL H 2 O, 5 min; H 2 O, 1 3 s; 1%
aqueous iron alum, 2 min; H 2 O, 15 s; 30% EtOH, 5 s; 50% EtOH, 5
s; 70% EtOH, 5 s; 90% EtOH, 5 s; 95% EtOH, 5 s; 100% EtOH, 5 s;
100% EtOH, 10 s; 3 : 1 xylene : methyl sali-cylate, 2 min; xylene,
2 min; xylene, until permanently coverslipped using Cytoseal
(Richard Allan Scientifi c, Kalamazoo, Minnesota, USA).
Slides were examined with a light microscope and images recorded
with a SPOT digital camera (Diagnostic Instruments, Sterling
Heights, Minnesota, USA). Resulting images were edited in Photoshop
CS3 (Adobe Systems, San Jose, California, USA), specifi cally for
background subtraction and image lev-els adjustment; scale bars
were added to the fi nal image using ImageJ software ( Rasband,
1997 ). A set of slides is deposited at RSA, and original digital
image fi les are available upon request from the fi rst author.
C 3 leaf anatomy is distinguished by the presence of palisade
mesophyll and usually intercellular spaces between spongy mesophyll
cells ( Cutler et al., 2008 ). Kranz anatomy, which is correlated
with C 4 photosynthesis ( Guti rrez et al., 1974 ; Furbank, 1998 ;
Dengler and Nelson, 1999 ), is characterized by a sheath of large
cells surrounding each vascular bundle (= bundle sheath), each cell
containing large and abundant chloroplasts; outside the bundle
sheath is a layer of mesophyll cells, each cell usually radially
elongated (= radiate meso-phyll) ( Furbank, 1998 ; Kanai and
Edwards, 1999 ). CAM photosynthesis is cor-related with leaves
having a thick cuticle, large cell vacuoles, and minimal
intercellular space between mesophyll cells ( Cushman, 2001 ;
Nelson et al., 2005 ; Nelson and Sage, 2008 ).
Character evolution Evolution of photosynthetic pathways was
traced us-ing the program Mesquite version 2.72 ( Maddison and
Maddison, 2009 ), esti-mated by ML (Markov k -state 1 parameter
model, which corresponds to Lewis s [2001] Mk model) over the ML
tree from the combined data matrix.
RESULTS
Phylogenetic relationships of major clades Aligned lengths for
the loci were: rpl14 rps8 infA rpl36 region, 1351 bp; atpI atpH
intergenic spacer, 906 bp; and the ndhA intron, 1395 bp. Sequences
were unambiguously aligned except for rpl14 rps8 , although
exploratory ML analyses using different alignments of this region
yielded the same topology. Analyses of individual loci showed the
suborder to be highly supported as monophyletic, with a non-ACPT
group comprising Ba-sellaceae, Didiereaceae, Hallophytaceae, and
Montiaceae, and a strongly supported ACPT clade. Families of
Cactineae were resolved as monophyletic (except Didiereaceae in
rpl14 rps8 infA rpl36 ; see Appendix S1 at
http://www.amjbot.org/cgi/content/full/ajb.1000227/DC1) and have
moderate (75 89% bootstrap) to strong ( 90% bootstrap; 0.95
posterior probabil-
Table 1. Evaluation of alternative hypotheses regarding the
placement of select taxa of the suborder Cactineae. The difference
between the ln likelihood score of the most likely tree and the
constrained topology is reported along with results of the RELL
test ( Shimodaira and Hasegawa, 1999 ).
Hypothesis Outcome
Halophytum sister to Ceraria + Portulacaria
Cannot reject (Diff ln L = 8.58767, P = 0.189)
Halophytum part of Montiaceae Cannot reject (Diff ln L =
3.77526, P = 0.243)
Halophytum basal within Cactineae Cannot reject (Diff ln L =
0.43672, P = 0.366)
Halophytum sister to Basellaceae Cannot reject (Diff ln L =
0.98831, P = 0.421)
Didiereaceae basal within Cactineae Cannot reject (Diff ln L =
3.21613, P = 0.330)
Cactaceae sister to Portulacaceae + Talinaceae +
Anacampseroteae
Cannot reject (Diff ln L = 6.38097, P = 0.120)
Portulacaceae sister to Anacampseroteae Cannot reject (Diff ln L
= 0.85346, P = 0.318)
-
1832 American Journal of Botany [Vol. 97
Portulaca is the only member of Cactineae with C 4
photosyn-thesis, with a shift to C 3 in P. cryptopetala.
DISCUSSION
Relationships within Cactineae All analyses of the chloro-plast
data show suborder Cactineae and, in nearly every case, each of its
eight families (Anacampserotaceae, Basellaceae, Cactaceae,
Didiereaceae, Halophytaceae, Montiaceae, Portu-lacaceae, and
Talinaceae) to be monophyletic, most with strong support. In
addition, a clade comprising Anacampserotaceae, Cactaceae,
Portulacaceae, and Talinaceae (ACPT clade) is strongly supported.
However, owing to topological confl ict and low clade support,
relationships among the families, except for the grouping of the
ACPT families, are incongruent among
was inferred only for species of Anacampserotaceae, Cacta-ceae,
and Didiereaceae. Leaf anatomy and 13 C values for Por-tulaca
cryptopetala (Portulacaceae) both indicate that it undergoes C 3
photosynthesis ( Fig. 4K ), unlike all other species of Portulaca
examined, which were C 4 .
Table 4 shows the 13 C values obtained in this study. C 3
val-ues ranged between 20.44 and 33.52 ; for C 4 , between 10.41
and 15.56 ; and for CAM, from 15.05 to 19.48 (including facultative
CAM taxa).
Evolution of photosynthetic pathways Reconstruction of the
diversifi cation of photosynthetic pathways in Cactineae is shown
in Fig. 6 . ML reconstruction recovered the C 3 pathway as
ancestral to the suborder. CAM photosynthesis is inferred to have
evolved independently fi ve times, including facultative CAM in
Anacampserotaceae, Cactaceae, and Didiereaceae.
Fig. 2. Chronogram and biogeographical analysis of Cactineae.
Maximum clade credibility tree from a BEAST ( Drummond and Rambaut,
2007 ) analysis of the combined data matrix. Topology is identical
to the trees obtained from ML and Bayesian analyses using MrBayes (
Ronquist and Huelsen-beck, 2003 ), except for the relationships
among Pereskia aculeata , P. lychnidifl ora , and P. sacharosa
(Cactaceae). Dates in millions of years. Arrows indi-cate
calibration points. Information for selected nodes (black boxes) is
provided in Table 2 . Biogeographical reconstructions are displayed
in the form of a pie chart at each node, representing the
probability for each alternative ancestral area derived from the
dispersal-vicariance analysis (DIVA; Ronquist, 1997 ) optimizations
over 1000 trees randomly sampled from the BEAST run, as implemented
in the program S-DIVA ( Yu et al., 2010 ). Black portions of the
pie charts represent fi ve or more reconstructed ancestral ranges
with similar probability values. Letters after each taxon name
represent the distribution of the species.
-
1833November 2010] Ocampo and Columbus Phylogenetics of
Cactineae
proposed to be closely related to Chenopodiaceae ( Spegazzini,
1902 ; Eckardt, 1976 ; Blackwell, 1977 ; Cronquist, 1981 ; Rodman
et al., 1984 ; Rodman, 1990 ), to Aizoaceae or Phytolaccaceae (
Gibson, 1978 ), to Amaranthaceae Juss. or Chenopodiaceae ( Skvarla
and Nowicke, 1976 ), and to families of suborder Cactineae (
Behnke, 1994 ). Molecular studies have confi rmed the family to be
a member of Cactineae and have recovered Halophytaceae as sister to
Basellaceae ( Savolainen et al., 2000 ; Hilu et al., 2003 ), sister
to Basellaceae + Didiereaceae ( Brockington et al., 2009 ), or
sister to the ACPT clade (although no non-ACPT families were
sampled; M ller and Borsch, 2005 ). Results here are incongruent
but weakly supported among the different markers and methods, yet
the placement of the family is always among the non-ACPT families.
Morphological traits that may indicate affi nities of this family
with other taxa of the suborder include its cube-shaped pollen (
Skvarla and Nowicke, 1976 ), which is shared with Basellaceae. More
studies are needed to further clarify its evolutionary
relationships within Cactineae.
Other molecular studies have recovered the ACPT clade, usually
with moderate to strong statistical support ( Hershkovitz and
Zimmer, 1997 ; Applequist and Wallace, 2001 ; Edwards et al., 2005
; Applequist et al., 2006 ; Nyffeler, 2007 ; Nyffeler and Eggli,
2010 ). Interestingly, there are no known morphological or
anatomical synapomorphies for the ACPT clade ( Ogburn and Edwards,
2009 ). The monophyly of each family is strongly supported in
almost every analysis (except atpI atpH , where Talinaceae have a
bootstrap value of 32% and a 0.73 posterior probability and
Cactaceae have 71% bootstrap support; see on-line Appendix S2 for
the Bayesian tree), but the relationships lack strong support. Most
analyses show Talinaceae as basal within the ACPT clade, although
ML analysis of the ndhA in-tron recovers Cactaceae as basal (online
Appendix S3) in con-cordance with Hershkovitz and Zimmer (1997) .
The clade comprising Anacampserotaceae, Cactaceae, and
Portulacaceae has a potential synapomorphy of nodal trichomes and
bristles ( Ogburn and Edwards, 2009 ). The ML and Bayesian analyses
of the combined data matrix resolved Portulacaceae as sister to
Cactaceae, as in the phyC phylogeny of Edwards et al. (2005) and in
a combined chloroplast, mitochondrial, and nuclear loci tree of
Butterworth and Edwards (2008) . Morphological and anatomical
traits that may serve as indicators of relationships among these
three families are homoplastic ( Ogburn and Edwards, 2009 ); thus
further study of the group is clearly needed.
As we have detailed, there have been a number of efforts to
clarify relationships within suborder Cactineae, but consider-able
uncertainty remains about its evolutionary history. Famil-ial
relationships (besides the ACPT grouping) remain uncertain despite
the use of a variety of molecular markers for phyloge-netic
reconstruction. This uncertainty is associated with short branch
lengths. Internal short branches have been related to rapid
radiations ( Whitfi eld and Lockhart, 2007 ), where ances-tral
polymorphisms can persist and lead to incongruence among loci (
Maddison, 1997 ; Degnan and Rosenberg, 2006 , 2009 ). Phylogenetic
reconstruction involving short branches can also be infl uenced by
a few homoplastic characters, which are suf-fi cient to mask the
signal because relationships are supported by a limited number of
characters ( Rokas and Carroll, 2006 ; Whitfi eld and Lockhart,
2007 ). Because short branch lengths are manifest in Cactineae
phylogenies using different loci ( Hershkovitz and Zimmer, 1997 ;
Applequist et al., 2006 ; Nyffeler, 2007 ; Brockington et al., 2009
; Fig. 1 ; Appendices S1 S3), it could be hypothesized that these
are due to rapid radiations, as
analyses of individual markers (online Appendices S1 S3) and
weakly supported in the analysis of the combined data matrix using
both ML and Bayesian reconstruction methods ( Fig. 1 ). These
results are consistent with Hershkovitz and Zimmer (1997 ; ITS),
Applequist and Wallace (2001 ; ndhF ), Edwards et al. (2005 ;
combined analysis of phyC , psbA trnH , trnK matK , rbcL , and cox3
), Applequist et al. (2006 ; ndhF ), Nyffeler (2007 ; combined
analysis of matK , ndhF , and nad1 ), and Nyffeler and Eggli (2010
; combined analysis of matK and ndhF ) in which monophyly of
Cactineae, the ACPT group, and all families are statistically
supported, but the relationships among families are unresolved or
have low support.
Different hypotheses of relationships were evaluated with the SH
test, in particular relationships of Halophytaceae and
Didiereaceae, as well as the relationships inside the ACPT clade,
but none could be rejected ( Table 1 ). Montiaceae are the
earliest-diverging lineage within the suborder in analyses of the
combined data matrix, which agrees with Brockington et al. (2009)
based on a combined analysis of the chloroplast inverted repeat,
nine chloroplast and two nuclear regions, and the results of
Nyffeler and Eggli (2010) ; however, neither of these studies show
statistical support for this relationship. Cu noud et al. (2002) ,
in their combined rbcL and matK analysis, and Nyffeler (2007)
recovered Basellaceae as basal within Cactineae. The combined
analysis reveals Basellaceae as sister to Didiereaceae, in
agreement with Brockington et al. (2009) . Although a sister
relationship of Basellaceae and Didiereaceae has low statistical
support, the relationship between these families is supported by
the presence of a solitary basal ovule in members of both fami-lies
(except Calyptrotheca Gilg [Didiereaceae], which has up to six
ovules; Nyffeler and Eggli, 2010 ), a unique trait in Cactineae and
putative synapomorphy. However, Nyffeler s (2007) re-sults showed
Didiereaceae as sister to the ACPT clade (although with low
support), in agreement with Ogburn and Edwards (2009) , who
suggested that Didiereaceae share parallelocytic leaf stomata,
tannin cells, and pericyclic sclereids with the ACPT clade.
The position of Halophytaceae has been controversial for a long
time. Its single species was described in Aizoaceae ( Spegazzini,
1899 ), but based on different characters it has been
Table 2. Estimated ages for the most recent common ancestor
(MRCA) of select taxa of Cactineae, expressed in millions of years.
Nodes labeled in Fig. 2 .
Node MRCA of
95% Highest posterior density
Mean Lower Upper
1 Cactineae 18.8 6.7 33.72 Montiaceae 13 3.4 25.43 Basellaceae
3.8 0.4 94 Didiereaceae 12.1 2.4 24.45 Basellaceae + Didiereaceae
14.9 3.9 28.56 Cactineae except Montiaceae 17.6 6.5 31.97
Halophytaceae + ACPT clade 17.1 6.1 31.48 ACPT clade 15.2 5.4 27.89
Talinaceae 9.1 2 18.310 Anacampserotaceae 11.4 3.2 22.611
Anacampserotaceae + Cactaceae +
Portulacaceae14.3 5.1 26.6
12 Cactaceae 10 3.1 19.113 Portulacaceae 9.6 3.0 18.514
Cactaceae + Portulacaceae 13.9 4.9 26.5
-
1834 American Journal of Botany [Vol. 97
-
1835November 2010] Ocampo and Columbus Phylogenetics of
Cactineae
characterizing CAM activity using leaf anatomical traits, but
their criteria could not predict the correct photosynthetic
path-way in all cases. These studies show that predicting
photosyn-thetic pathways based solely on leaf anatomy cannot be
achieved with confi dence in some cases. Our primary basis for
differen-tiating C 3 leaf anatomy from CAM in this study was the
pres-ence of palisade parenchyma, which means a certain degree of
shape differentiation among cells inside the leaf.
For those Cactineae examined in this study, leaf anatomy and 13
C values are concordant in most samples, and the predicted
photosynthetic pathways partially agree with previous studies that
employed data besides leaf anatomy alone ( Nobel and Hartsock, 1986
; Ziegler, 1996 ; Martin and Wallace, 2000 ; Guralnick et al., 2008
). The inconsistencies with other studies rep-resent photosynthetic
variants that are diffi cult to identify with 13 C and leaf
anatomical data in tandem, such as facultative CAM and CAM-cycling.
In our study, fi ve facultative CAM taxa were detected, which are
distributed in Anacampserota-ceae, Cactaceae, and Didiereaceae (
Fig. 6 ). The results for these species are consistent with other
studies ( Ziegler, 1996 ; Guralnick et al., 2008 ), except
Quiabentia verticillata has been reported as CAM-cycling ( Martin
and Wallace, 2000 ). On the other hand, Alluaudia ascendens , A.
humbertii , and Portulacaria afra (Didiereaceae) are species
considered to be CAM in this study, but are known to behave as
facultative CAM ( Kluge and Ting, 1978 ; Guralnick and Ting, 1987
). Other species coded here as C 3 have been found to have some CAM
activity as well ( Fig. 6 ; Rayder and Ting, 1981 ; Winter and
Smith, 1996 ; Martin and Wallace, 2000 ; Guralnick and Jackson,
2001 ; Guralnick et al., 2008 ). C 4 13 C values for Portulaca
species with Kranz anat-omy are consistent with reported values for
C 4 photosynthesis ( 10 to 16 ; O Leary, 1988 ; Sage et al., 2007
). Only P. cryp-topetala has leaf anatomy and a 13 C value
corresponding to C 3 photosynthesis; however, the species has been
shown to be a C 3 C 4 intermediate based on biochemical and gas
exchange data ( Voznesenskaya et al., 2010 ).
Uncertainty about relationships within the ACPT clade and among
the non-ACPT families, as well as the widespread oc-currence of
various degrees of CAM activity within Cactineae, limit our ability
to reconstruct photosynthetic pathway evolu-tion. Despite that, our
results show that the C 3 pathway is well distributed in the
outgroup and Cactineae ( Fig. 6 ), and other possible topologies do
not seem to affect the inference of mul-tiple origins of CAM
photosynthesis. Character reconstruction analysis recovers the C 3
pathway as ancestral for the suborder. C 4 , CAM, and facultative
CAM are all derived from the C 3 pathway, in agreement with the
hypothesis that C 3 is the ances-tral type from which the other
pathways evolved ( Monson, 1989 ; Ehleringer and Monson, 1993 ).
However, some CAM activity has been reported for most of the
families in Cactineae (Anacampserotaceae [ Guralnick and Jackson,
2001 ; Guralnick et al., 2008 ], Cactaceae [ Rayder and Ting, 1981
; Nobel and Hartsock, 1986 ; Guralnick and Ting, 1987 ; Gibson,
1996 ; Martin and Wallace, 2000 ], Didiereaceae [ Kluge and Ting,
1978 ; Winter and Smith, 1996 ; Ziegler, 1996 ; Guralnick and
Jackson,
earlier suggested by Hershkovitz and Zimmer (2000) for
Mon-tiaceae (as western American Portulacaceae). The study by
Arakaki et al. (2010a) using low copy nuclear markers
(phyto-chromes B and C) and a number of chloroplast loci is of
great interest because some relationships are apparently being
re-solved within the suborder, although that work is at a
prelimi-nary stage. These results will allow analysis of a
supermatrix comprising additional chloroplast, nuclear, and
mitochondrial data to better understand the relationships among
Cactineae, although preliminary analyses show that missing data may
be a problem in resolving relationships within Cactaceae ( Arakaki
et al., 2010a ), and this may apply to the suborder as well.
Evolution of photosynthetic pathways in Cactineae Previ-ous
studies have shown that all three major photosynthetic pathways C 3
, C 4 , and CAM are represented in the suborder (e.g., Winter, 1979
; Sage et al., 1999 ; Guralnick and Jackson, 2001 ; Sayed, 2001 ;
Guralnick et al., 2008 ). In addition, there are taxa that use more
than one photosynthetic pathway in the same or different organs
(e.g., Portulaca and Quiabentia , re-spectively), or switch
pathways depending on the environmen-tal conditions ( Koch and
Kennedy, 1980 , 1982 ; Ku et al., 1981 ; Nobel and Hartsock, 1986 ;
Kraybill and Martin, 1996 ; Martin and Wallace, 2000 ; Guralnick
and Jackson, 2001 ; Guralnick et al., 2002 , 2008 ). There is
evidence of some degree of CAM activity for all families of
Cactineae except Basellaceae and Halophytaceae (photosynthetic
pathway data for the latter family are here reported for the fi rst
time) (e.g., Guralnick and Jackson, 2001 ; Guralnick et al., 2002 ,
2008 ), including facultative CAM (the plants can switch to CAM or
C 3 depending on water availability; they use CAM when
water-stressed and C 3 photo-synthesis when water is abundant;
Cushman, 2001 ) and CAM-cycling (during the day, the plants do not
completely close their stomata, and they fi x atmospheric CO 2 ; at
night, the stomata are closed and the plants fi x respiratory CO 2
; Cushman, 2001 ). In this study, leaf anatomy was important in
helping to predict the photosynthetic pathway with more accuracy
than with 13 C data alone, although it is evident that biochemical
data are needed to detect photosynthetic variants. Therefore, the
results presented herein may provide an incomplete view of the
distri-bution of the photosynthetic pathways in the suborder, and
fur-ther biochemical characterization could yield additional
insights.
Determination of photosynthetic pathways from leaf anat-omy
alone was challenging in some cases, in particular discrim-inating
C 3 from CAM, which was also borne out in previous studies.
Nyananyo (1988) concluded that Portulacaria afra , Talinopsis
frutescens , and Talinum paniculatum undergo C 3 photosynthesis
based on anatomy, whereas Landrum (2002) considered them to undergo
CAM; here we coded the fi rst spe-cies as CAM and the other two as
C 3 using leaf anatomy. Nyananyo (1988) also described the leaf
anatomy of Portulaca cryptopetala as C 4 , while Voznesenskaya et
al. (2010) and we in the present study showed that the species has
C 3 leaf anat-omy. Nelson et al. (2005) proposed a quantitative
method for
Fig. 3. Light micrographs of transectional leaf anatomy in
Cactineae. Photosynthesis type inferred from anatomy is indicated
within parentheses. (A) Alluaudia ascendens (CAM; Didiereaceae).
(B) A. humbertii (CAM). (C) Anredera cordifolia (C 3 ;
Basellaceae). (D) A. ramosa (C 3 ). (E) Calandrinia caespitosa (C 3
; Montiaceae). (F) Calyptridium parryi (C 3 ; Montiaceae). (G)
Ceraria namaquensis (C 3 ; Didiereaceae). (H) Claytonia parvifl ora
(C 3 ; Mon-tiaceae). (I) Decarya madagascariensis (C 3 ;
Didiereaceae). (J) Didierea madagascariensis (C 3 ; Didiereaceae).
(K) D. trolli (C 3 ). (L) Grahamia bracteata (C 3 ;
Anacampserotaceae). (M) Lewisia rediviva (C 3 ; Montiaceae). (N)
Maihuenia patagonica (C 3 ; Cactaceae). (O) Mirabilis sanguinea (C
3 ; Nyctag-inaceae). PM = palisade mesophyll. Black scale bar =
0.75 mm; gray scale bar = 0.25 mm.
-
1836 American Journal of Botany [Vol. 97
hand, Klak et al. (2004) showed that ca. 85% (more than 1500
species) of the diversity of the Aizoaceae originated ca. 5 Ma,
which was hypothesized to be the result of key innovations
as-sociated with adaptations to arid environments (e.g., wide-band
tracheids, which are also found in some species of Cactineae;
Mauseth, 2004 ; Landrum, 2002 , 2006 ). This suggests that late
radiations may not be unusual within Caryophyllales, although this
case involves the intrafamiliar level.
The estimated age of the MRCA of Cactineae from this study is
younger than the age proposed for a putative Cactaceae fossil
record from the Eocene ( Chaney, 1944 ) and Montia -like fossil
pollen from the Late Cretaceous (Montiaceae; Ravn, 1987 ). Muller s
(1981) list of fossil pollen records includes Monti-aceae ( Hedlund
and Engelhardt, 1970 ; Martin, 1973 ) and Por-tulacaceae ( Van
Campo, 1976 ) from the Miocene-Pliocene period (ca. 6 4 Myr; IUGS,
2009 ), which is younger than the estimated ages of the MRCA of
those two families in our study (13 and 9.6 Myr, respectively).
Further studies are required to clarify the interpretation of the
putative fossil pollen record of taxa of Cactineae, whose pollen
may be confused with that of other families in the Caryophyllales
(e.g., Nyctaginaceae, Po-lygonaceae Juss.; Erdtman, 1952 ). If
these fossil pollen records can be confi dently attributed to the
suborder and unequivocally assigned to families, they can be used
to calibrate deeper nodes of the phylogeny, which may provide more
reliable dating and reduce the 95% HPD intervals recovered in the
divergence times estimate (see Table 2 ).
Biogeographical reconstruction using the Bayes DIVA ap-proach
and a limited outgroup sample places the origin of Cactineae in the
Americas. The results of the simulation analy-ses showed that
alternate distribution areas assigned to the out-group do impact
the reconstruction of the ancestral distribution of the suborder,
in particular increasing its range. DIVA optimi-zations become less
reliable as the root node is approached ( Ronquist, 1996 ) and have
the tendency to yield wide distribu-tions that include all analyzed
areas, as is the case in one simu-lation restricting the outgroup
to South America. With a larger outgroup sample, Applequist and
Wallace (2001) recovered Cactineae as South American in origin, but
they mentioned a potential bias due to the wide distribution of
outgroup species, though the suborder is well represented in the
southern hemi-sphere. A more accurate biogeographical
reconstruction of the suborder may be obtained in a
Caryophyllales-wide study, which would place Cactineae farther from
the root node.
The recovered dates indicate that taxa of Cactineae were not
present in the Americas until well after the separation of South
America and Africa between 84 Ma and 106 Ma ( Pitman et al., 1993 )
and after the separation of South America and Antarctica ca. 45 Ma
( Raven and Axelrod, 1974 ). The proximity of the latter two
continents may have permitted dispersal of some plants to
Australia, but this may have been restricted to
cool-temperate-adapted taxa ( Raven and Axelrod, 1974 ). Therefore,
intercontinental disjunctions in Cactineae are better explained by
long-distance dispersal, in agreement with Raven and Axelrod (1974)
and Hershkovitz and Zimmer (1997) , and also supported
2001 ; Veste et al., 2001 ], Montiaceae [ Martin et al., 1988 ;
Harris and Martin, 1991 ; Guralnick and Jackson, 2001 ; Guralnick
et al., 2001 ], Portulacaceae [ Koch and Kennedy, 1980 , 1982 ;
Kraybill and Martin, 1996 ; Guralnick and Jackson, 2001 ; Guralnick
et al., 2002 ], and Talinaceae [ Herrera et al., 1991 ; G erere et
al., 1996 ; Guralnick and Jackson, 2001 ]), which may suggest that
the CAM pathway has only one origin within the suborder. More
biochemical studies are needed to determine the extent of the CAM
pathway, which might enhance our understanding of the origin and
age of this photosynthetic type in Cactineae.
The non-ACPT families of Cactineae have C 3 and CAM
pho-tosynthesis, while the ACPT clade has in addition the C 4
path-way. It is interesting to note that all leafy Cactaceae
sampled here (except Opuntia vestita , for which only stem material
was available) have C 3 leaf anatomy. This may represent a
symple-siomorphy, although the photosynthetic functions of the leaf
may vary (see Nobel and Hartsock, 1986 ; Martin and Wallace, 2000
). The C 4 pathway has evolved only once in Cactineae, specifi
cally in Portulacaceae ( Portulaca ); the MRCA of the family is
reconstructed as having C 4 photosynthesis ( Fig. 6 ), and the
divergence times analysis estimates that it evolved ca. 9.5 Ma (
Fig. 2 ). This is concordant with the hypothesis that decreasing
atmospheric CO 2 concentrations from the Oligocene into the
Pliocene were a critical factor for the evolution of C 4
photosynthesis ( Ehleringer and Monson, 1993 ; Sage, 2005 ). A
transition to an intermediate C 3 C 4 pathway from C 4 is found in
P. cryptopetala ( Voznesenskaya et al., 2010 ), which can be
classifi ed as type I intermediacy, characterized by an absence of
C 4 cycle activity and enhanced reassimilation of photorespired CO
2 ( Edwards and Ku, 1987 ). Fixation of CO 2 is accomplished
exclusively by Rubisco, which explains a 13 C value within the C 3
range ( Sage et al., 2007 ). Portulaca cryptopetala is distrib-uted
in central South America (Bolivia to Argentina and Brazil) and is
generally associated with rivers and streams. What drives the shift
from C 4 to C 3 C 4 intermediacy is not understood, al-though it
has been hypothesized that it may represent a physi-ological
response to environmental selection pressure ( Duvall et al., 2003
; McKown et al., 2005 ).
Historical biogeography of Cactineae Our study is the fi rst to
employ calibration points inside Cactineae, using two endemic
Hawaiian Portulaca species, which yields an estimate of 18.8 (6.7
33.7) Myr (early Miocene) for the MRCA of the suborder. This
implies a recent radiation in the Caryophyllales, whose MRCA is
estimated at ca. 100 Myr ( Wikstr m et al., 2001 , 2004 ). This
calibration approach has been criticized be-cause it overlooks the
possibility that the taxon may be older than the strata to which it
is endemic ( Heads, 2005 ), as shown by Rassmann (1997) in a study
of iguanas in the Gal pagos Is-lands. Even more, some researchers
contend that calibrating phylogenies using this method may result
in signifi cant (tens of millions of years) age underestimations (
Heads, 2009 ). Prelimi-nary results by Arakaki et al. (2010b) yield
an age of the MRCA of Cactaceae of ca. 30 Myr when calibrating
their phylogeny with fossils distributed across the angiosperms. On
the other
Fig. 4. Light micrographs of transectional leaf anatomy in
Cactineae. Photosynthesis type inferred from anatomy is indicated
within parentheses. (A) Montiopsis andicola (C 3 ; Montiaceae). (B)
Parakeelya pleiopetala (C 3 ; Montiaceae). (C) Pereskia aculeata (C
3 ; Cactaceae). (D) P. grandifolia (C 3 ). (E) P. lychnidifl ora (C
3 ). (F) P. quisqueyana (C 3 ). (G) P. sacharosa (C 3 ). (H)
Phemeranthus multifl orus (C 3 ; Montiaceae). (I) Portulaca amilis
(C 4 ; Portulacaceae). (J) P. bicolor (C 4 ). (K) P. cryptopetala
(C 3 ). (L) P. echinosperma (C 4 ). (M) P. elatior (C 4 ). (N) P.
guanajuatensis (C 4 ). (O) P. molokiniensis (C 4 ). PM = pali-sade
mesophyll. Black arrows indicate bundle sheaths (with abundant
chloroplasts) surrounded by radiate mesophyll, characteristic of
Kranz anatomy. Black scale bar = 0.75 mm; gray scale bar = 0.25
mm.
-
1837November 2010] Ocampo and Columbus Phylogenetics of
Cactineae
-
1838 American Journal of Botany [Vol. 97
by the high number of dispersal events inferred from the Bayes
DIVA analysis. It is not clear how ancestral taxa may have
dis-persed successfully around the world. Dispersal mechanisms
known in the suborder include zoochory ( Ridley, 1930 ; Barth-lott
and Hunt, 1993 ; Hershkovitz and Zimmer, 2000 ), anemo-chory (
Barthlott and Hunt, 1993 ; Carolin, 1993 ; Kubitzki, 1993 ;
Fig. 5. Light micrographs of transectional leaf anatomy in
Cactineae. Photosynthesis type inferred from anatomy is indicated
within parentheses. (A) P. pilosa (C 4 ; Portulacaceae). (B) P.
umbraticola subsp. lanceolata (C 4 ). (C) Portulacaria afra (CAM;
Didiereaceae). (D) Quiabentia verticillata (C 3 ; Cactaceae). (E)
Rivina humilis (C 3 ; Phytolaccaceae). (F) Sesuvium portulacastrum
(C 3 ; Aizoaceae). (G) Talinopsis frutescens (C 3 ;
Anacampserotaceae). (H) Talinum caffrum (C 3 ; Talinaceae). (I) T.
fruticosum (C 3 ). (J) T. lineare (C 3 ). (K) T. paniculatum (C 3
). (L) T. polygaloides (C 3 ). PM = palisade mesophyll. Black
arrows indicate bundle sheaths (with abundant chloroplasts)
surrounded by radiate mesophyll, characteristic of Kranz anatomy.
Black scale bar = 0.75 mm; gray scale bar = 0.25 mm; white scale
bar with black background = 0.1 mm.
Sperling and Bittrich, 1993 ), hydrochory ( Ridley, 1930 ; Danin
et al., 1978 ; Barthlott and Hunt, 1993 ), and voluntary or
invol-untary dispersal by humans (e.g., weedy species of Portulaca
and Talinum and species of Pereskia escaped from cultivation), but
the exact mechanisms for successful dispersal of these plants
throughout the world are not known.
-
1839November 2010] Ocampo and Columbus Phylogenetics of
Cactineae
Applequist and Wallace (2001) showed that the place of ori-gin
of Montiaceae is uncertain, but it is here recovered as North
America, with a MRCA age of 13 Myr, consistent with Hersh-kovitz
and Zimmer s (2000) estimate of 8 16 Myr. Phemeran-thus , a genus
with most species in North America and one disjunct species in
Argentina [ P. punae (R. E. Fr.) Eggli & Nyffeler], is basal.
Although only seven of the 15 recognized genera in the family were
sampled ( Nyffeler and Eggli, 2010 ), at least two early,
independent long-distance dispersal events to South America and one
to Australia are inferred. Hectorella , which is not sampled here,
is endemic to New Zealand, and ac-cording to the phylogeny in
Applequist et al. (2006) it may rep-resent another independent
long-distance dispersal event to the islands of the Pacifi c.
The biogeographical analysis postulates Didiereaceae as Af-rican
Malagasy in origin (fi ve of seven genera included in this study;
Applequist and Wallace, 2003 ), with a MRCA age of ca. 12 Myr.
Applequist and Wallace (2001) showed that the fami-ly s Old World
distribution is likely the result of an early long-distance
dispersal from South America to the Old World. These authors (
Applequist and Wallace, 2000 , 2001 ) provided evi-dence that
Calyptrotheca , a genus of Didiereaceae (Calyptroth-ecoideae Pax
& Gilg) endemic to east tropical Africa (not sampled here), is
the closest relative to Didiereoideae Appleq. & R. S. Wallace
(including Alluaudia , Decarya , and Didierea
Table 3. Photosynthetic pathways inferred from 13 C values and
leaf anatomy data derived from this study.
Species 13 C ( )Pathway
based on 13 C
Leaf anatomy
Pathway for sample
Anacampserotaceae Anacampseros
vulcanensis 24.53 C 3 C 3
Grahamia bracteata 19.48 CAM C 3 Fac. CAM Talinopsis frutescens
27.10 C 3 C 3 C 3 Basellaceae Anredera cordifolia 30.16 C 3 C 3 C 3
A. ramosa 26.31 C 3 C 3 C 3 Cactaceae Maihuenia patagonica 25.10 C
3 C 3 C 3 Opuntia vestita 16.87 CAM CAM Pereskia aculeata 27.56 C 3
C 3 C 3 P. grandifolia 30.18 C 3 C 3 C 3 P. lychnidifl ora 24.52 C
3 C 3 C 3 P. quisqueyana 29.97 C 3 C 3 C 3 P. sacharosa 24.16 C 3 C
3 C 3 Quiabentia verticillata 16.25 CAM C 3 Fac. CAM Rhipsalis
baccifera 17.08 CAM NA CAMDidiereaceae Alluaudia ascendens 17.16
CAM CAM CAM A. humbertii 15.05 CAM CAM CAM Ceraria namaquensis
20.44 C 3 C 3 C 3 Decarya
madagascariensis 15.52 CAM C 3 Fac. CAM
Didierea madagascariensis
19.37 CAM C 3 Fac. CAM
D. trollii 18.38 CAM C 3 Fac. CAM Portulacaria afra 18.35 CAM
CAM CAMHalophytaceae Halophytum ameghinoi 24.75 C 3 C 3 Montiaceae
Calandrinia caespitosa 25.67 C 3 C 3 C 3 Calyptridium parryi 25.08
C 3 C 3 C 3 Claytonia parvifl ora 33.52 C 3 C 3 C 3 Lewisia
rediviva 31.27 C 3 C 3 C 3
Table 4. 13 C statistics for Cactineae and outgroup taxa.
Statistic C 3 C 4 CAM
No. species 30 13 10Average ( ) 26.78 13.5015 17.35Standard
deviation ( ) 2.69 1.44912 1.52Minimum ( ) 33.52 15.56 19.48Maximum
( ) 20.44 10.41 15.05
Notes: Photosynthetic pathways were assigned using 13 C values
and leaf anatomical data, when available. CAM includes facultative
CAM taxa as scored in this study.
in our study). On the basis of this and the low sequence
diver-gence between the two subfamilies, they concluded that it is
more plausible that the clade was introduced via dispersal from
Africa to Madagascar.
Although only one genus was sampled for Basellaceae, the
analysis shows it originated in South America, in agreement with
Raven and Axelrod (1974) , with a MRCA age of ca. 4 Myr for
Anredera . The family has four genera ( Nyffeler and Eggli, 2010 ),
but only one with representatives in the Old World, which suggests
a New World origin. Halophytaceae include a single species from the
southern part of Argentina; the age of the MRCA of the
Halophytaceae + ACPT clade is ca. 17 Myr.
Notes: Fac. = facultative; NA = Not applicable; = leaf anatomy
not available for the sample.
Species 13 C ( )Pathway
based on 13 C
Leaf anatomy
Pathway for sample
Montiopsis andicola 27.52 C 3 C 3 C 3 Parakeelya pleiopetala
27.42 C 3 C 3 C 3 Phemeranthus multifl orus 27.39 C 3 C 3 C 3
Portulacaceae P. amilis 11.96 C 4 C 4 C 4 P. bicolor 13.75 C 4 C 4
C 4 P. californica 13.15 C 4 C 4 C 4 P. cryptopetala 26.55 C 3 C 3
C 3 P. echinosperma 10.41 C 4 C 4 C 4 P. elatior 12.74 C 4 C 4 C 4
P. guanajuatensis 14.25 C 4 C 4 C 4 P. howellii 12.27 C 4 C 4 P.
massaica 15.21 C 4 C 4 P. molokiniensis 15.56 C 4 C 4 C 4 P. pilosa
14.05 C 4 C 4 P. quadrifi da 13.03 C 4 C 4 P. umbraticola
subsp.
lanceolata 14.09 C 4 C 4 C 4
P. villosa 15.05 C 4 C 4 Talinaceae Talinum arnottii 24.96 C 3 C
3 T. caffrum 23.29 C 3 C 3 C 3 T. fruticosum 29.34 C 3 C 3 C 3 T.
lineare 26.28 C 3 C 3 C 3 T. paniculatum 28.22 C 3 C 3 C 3 T.
polygaloides 26.70 C 3 C 3 C 3 T. tenuissimum 23.69 C 3 C 3
Outgroups Mirabilis sanguinea
(Nyctaginaceae) 27.02 C 3 C 3 C 3
Rivina humilis (Phytolaccaceae)
29.33 C 3 C 3 C 3
Sesuvium portulacastrum (Aizoaceae)
25.62 C 3 C 3 C 3
-
1840 American Journal of Botany [Vol. 97
Fig. 6. Maximum likelihood (ML) ancestral character
reconstruction for photosynthetic pathways in Cactineae. Analysis
based on the ML tree from the combined data matrix. Proportional
likelihoods in the form of a pie chart are shown at each node of
the ML reconstruction. Superscript capital letters by some taxa
names are references to other studies reporting different
photosynthesis pathways than found here. Facultative CAM or
CAM-cycling activity: A = Kluge and Ting, 1978; B = Rayder and
Ting, 1981; C = Nobel and Hartsock, 1986; D = Guralnick and Ting,
1987; E = Herrera et al., 1991; F = G erere et al., 1996; G =
Kraybill and Martin, 1996; H = Winter and Smith, 1996; I = Ziegler,
1996; J = Martin and Wallace, 2000; K = Guralnick and Jackson,
2001; L = Guralnick et al., 2008. C 3 -C 4 intermediate: M =
Voznesenskaya et al., 2010. *As Talinum triangulare (Jacq.) Willd.;
**as Quiabentia chacoensis Backeb.; ***as Portulaca mundula I. M.
Johnst.
-
1841November 2010] Ocampo and Columbus Phylogenetics of
Cactineae
conjectured that this location would explain the near absence of
Cactaceae in the Old World and would facilitate the early
dis-persal of the leafy Pereskia and other subfamilies to North
America and the Caribbean. In contrast, Wallace and Gibson (2002)
hypothesized that the family originated in the central Andes based
on the presence there of early-diverging lineages.
Portulacaceae, reduced to the single genus Portulaca , have an
uncertain place of origin in the Bayes DIVA analysis, though
re-stricted to southern hemisphere continents, with a MRCA age of
ca. 9.5 Myr. Preliminary divergence dates analysis using a wider
sampling of the family and a different set of molecular markers
(except for the ndhA intron) estimates an older age for the MRCA of
Portulacaceae (ca. 15 Myr; G. Ocampo and J. T. Columbus,
unpublished data). Differences in divergence estimates have been
observed using different loci for the same set of samples (e.g.,
Rodr guez-Trelles et al., 2004 ), attributable to differences in
rates of evolution of the loci; in addition, it has been shown that
increased taxon sampling may produce older estimates ( Pirie et
al., 2005 ). Therefore, results obtained in each study should be
considered with caution, although in general terms it can be
concluded that the MRCA of the family coincided with the Miocene.
Results indicate an early split of the genus into an Old World
Australian clade ( P. bicolor and P. quadrifi da ) with opposite
leaves and a primarily New World clade with alternate to
subopposite leaves. The biogeo-graphical analysis is not clear
about the origin of Portulaca , but Applequist and Wallace (2001)
indicate that it is of South Ameri-can origin and has successfully
colonized all continents except Antarctica. In a separate study
with a wider sampling of the genus (G. Ocampo and J. T. Columbus,
unpublished data), it is evident that taxa of different subclades
have independently dispersed to other continents (e.g., Africa,
Australia) and to the islands of the Pacifi c Ocean (e.g., Gal
pagos, Hawaiian Islands) from South and North America. Although
there is no clear dispersal mechanism, there is some evidence of
long-distance dispersal via birds ( Ridley, 1930 ), by fl oating
across bodies of water ( Danin et al., 1978 ), and by tropical
storms ( Matthews et al., 1991 ). Like the other families of
Cactineae, more studies are desirable to better understand the
distribution of Portulaca.
This study adds to the knowledge of the evolution of Cactineae,
but a strongly supported hypothesis of relationships within the
group remains elusive. Phylogenetic studies of the suborder have
yielded incongruent gene trees, in particular within the ACPT and
non-ACPT clades, likely owing to short branches that obscure
evo-lutionary relationships, including when loci are analyzed in
combi-nation ( Degnan and Rosenberg, 2006 ). Although the branching
pattern is not clearly known, it seems plausible that the short
branches may represent rapid radiations as a response of the
organ-isms to increased aridity. These processes may have occurred
above all in the species of the ACPT clade, which is South American
in origin and whose MRCA coincides with an arid environment
al-ready present across the central Andes ( Hartley, 2003 ).
Population-level studies and complete chloroplast genome sequence
analysis may contribute to a better understanding of the
evolutionary his-tory of the suborder by providing greater insights
into adaptations to aridity, as well as the mechanisms that
triggered those adapta-tions in this charismatic lineage of fl
owering plants.
LITERATURE CITED
Akaike , H. 1974 . A new look at the statistical model identifi
cation. IEEE Transactions on Automatic Control 19 : 716 723 .
Anderson , E. F. 2001 . The cactus family. Timber Press,
Portland, Oregon, USA.
The ACPT clade, which has high statistical support in almost all
analyses ( Fig. 1 ; Appendices S1 S3), has its origin in South
America, in agreement with Applequist and Wallace (2001) , with a
MRCA date of ca. 15 Myr. The clade has Talinaceae as basal with a
MRCA age of ca. 9 Myr. Applequist and Wallace s analysis (2001)
showed the place of origin of the family as equivocal, while here
it is recovered as South America. The family comprises three genera
(one sampled here; Nyffeler and Eggli, 2010 ) and has been very
successful in terms of dispersal, colonizing almost all continents.
The early-divergent clade comprising T. fruticosum and T.
paniculatum is widely distrib-uted in the Americas and is
naturalized in the Old World (e.g., Phillips, 2002 ; Dequan and
Gilbert, 2003 ). The analysis shows that T. lineare , a species
endemic to the North American deserts, is sister to the South
American T. polygaloides and is derived from a vicariance event
involving a widely distributed ancestor. It is also interesting
that the African taxa form a clade, suggesting a single dispersal
event to the continent.
The origin of Anacampserotaceae is shown to be in the Americas,
with the MRCA age of ca. 11 Myr. The analysis, which includes all
three genera recognized in the family ( Nyffeler and Eggli, 2010 ),
shows an early vicariance event, resulting in the North American
endemic Talinopsis frutescens and the South American Anacampseros
vulcanensis and Gra-hamia bracteata . Other members of the family,
such as the Old World Anacampseros , as well as the Australian
endemic A. australiana J. M. Black, are derived taxa of
Anacampserota-ceae ( Nyffeler, 2007 ); thus, it can be hypothesized
that they are result of long-distance dispersals from South
America.
Our results show that the MRCA of Cactaceae was distrib-uted in
South America and the Caribbean region, but other stud-ies have
suggested an exclusively South American origin (e.g., Hershkovitz
and Zimmer, 1997 ; Applequist and Wallace, 2001 ; Edwards et al.,
2005 ; Nyffeler, 2007 ). Croizat (1952) com-mented on a possible
Old World origin of Cactaceae because of the presence of Rhipsalis
in tropical Africa, Madagascar, and Sri Lanka. Results from our
analysis (fi ve of 126 recognized genera sampled; Anderson, 2001 )
are in concordance with oth-ers (e.g., Nyffeler, 2002 ; Wallace and
Gibson, 2002 ; Crozier, 2005 ), in which the genus is derived and
likely its distribution is the result of long-distance dispersal,
probably via birds, as suggested by Barthlott and Hunt (1993) . On
the other hand, there is a controversy about the age of the family.
Croizat (1952) proposed the origin of Cactaceae before the
separation of Af-rica and South America in the early Cretaceous,
while others suggest it was soon after the separation of the
continents ca. 100 Ma ( Mauseth, 1990 ; Leuenberger, 1986 ; Wallace
and Gibson, 2002 ) to explain the absence of cacti in the Old World
(with the exception of Rhipsalis ). These dates imply that the age
of the family is similar to, or even older, than the estimated age
of the order that it belongs to [Caryophyllales, with an estimated
MRCA of 111 85 Myr ( Wikstr m et al., 2001 , 2004 )]. Hersh-kovitz
and Zimmer (1997) suggested a mid-Tertiary (Oligo-cene) origin ca.
30 Ma for Cactaceae, which has been adopted by Nyffeler (2002) and
Edwards et al. (2005) , and supported by the preliminary results of
Arakaki et al. (2010b) . In our study, an age of ca. 10 (3.1 19.1)
Myr for the MRCA of Cactaceae supports Hershkovitz and Zimmer s
(1997) hypothesis that the family is relatively young. Leuenberger
(1986) speculated a probable origin of the family in northwestern
South America in the late Cretaceous, distant from the closest
point where the continent was in contact with Africa (presently
northeast Brazil and Gabon, respectively; Raven and Axelrod, 1974
); the author
-
1842 American Journal of Botany [Vol. 97
Cutler , D. F. , T. Botha , and D. W. Stevenson . 2008 . Plant
anatomy: An applied approach. Blackwell Publishing, Malden,
Massachusetts, USA.
Danin , A. , I. Baker , and H. G. Baker . 1978 . Cytogeography
and taxon-omy of the Portulaca oleracea L. polyploid complex.
Israel Journal of Botany 27 : 177 211 .
Degnan , J. H. , and N. A. Rosenberg . 2006 . Discordance of
species trees with their most likely gene trees. PLoS Genetics 2 :
e68 .
Degnan , J. H. , and N. A. Rosenberg . 2009 . Gene tree
discordance, phylogenetic inference and the multispecies
coalescent. Trends in Ecology & Evolution 24 : 332 340 .
Dengler , N. G. , and T. Nelson . 1999 . Leaf structure and
development in C 4 plants. In R. F. Sage and R. K. Monson [eds.], C
4 plant biology, 133 172. Academic Press, San Diego, California,
USA.
Dequan , L. , and M. G. Gilbert . 2003 . Portulacaceae. In Z.
Wu, P. H. Raven, and D. Y. Hong [eds.], Flora of China, vol. 5, 442
444. Missouri Botanical Garden Press, St. Louis, Missouri, USA.
Downie , S. R. , and J. D. Palmer . 1994 . A chloroplast DNA
phylogeny of the Caryophyllales based on structural and inverted
repeat restric-tion site variation. Systematic Botany 19 : 236 252
.
Doyle , J. J. , and J. L. Doyle . 1987 . A rapid DNA isolation
procedure for small quantities of fresh leaf tissue. Phytochemical
Bulletin 19 : 11 15 .
Drummond , A. J. , S. Y. W. Ho , M. J. Phillips , and A. Rambaut
. 2006 . Relaxed phylogenetics and dating with confi dence. PLoS
Biology 4 : e88 .
Drummond , A. J. , and A. Rambaut . 2007 . BEAST: Bayesian
evolution-ary analysis by sampling trees. BMC Evolutionary Biology
7 : 214 .
Duvall , M. R. , D. E. Saar , W. S. Grayburn , and G. P.
Holbrook . 2003 . Complex transitions between C 3 and C 4
photosynthesis dur-ing the evolution of Paniceae: A phylogenetic
case study emphasizing the position of Steinchisma hians (Poaceae),
a C 3 C 4 intermediate. International Journal of Plant Sciences 164
: 949 958 .
Eckardt , T. 1976 . Classical morphological features of
centrospermous families. Plant Systematics and Evolution 126 : 5 25
.
Edgar , R. C. 2004 . MUSCLE: Multiple sequence alignment with
high ac-curacy and high throughput. Nucleic Acids Research 32 :
1792 1797 .
Edwards , E. J. , R. Nyffeler , and M. J. Donoghue . 2005 .
Basal cac-tus phylogeny: Implications of Pereskia (Cactaceae)
paraphyly for the transition to the cactus life form. American
Journal of Botany 92 : 1177 1188 .
Edwards , G. E. , and M. S. B. Ku . 1987 . The biochemistry of C
3 C 4 inter-mediates. In M. D. Hatch and N. K. Boardman [eds.], The
biochemis-try of plants, vol. 10, Photosynthesis, 275 325. Academic
Press, New York, New York USA.
Ehleringer , J. R. , and R. K. Monson . 1993 . Evolutionary and
ecologi-cal aspects of photosynthetic pathway variation. Annual
Review of Ecology and Systematics 24 : 411 439 .
Erdtman , G. 1952 . Pollen morphology and plant taxonomy.
Angiosperms: An introduction to palynology. I. Almqvist and
Wiksell, Stockholm, Sweden.
Farris , J. S. , M. K llersj , A. G. Kluge , and C. Bult . 1994
. Testing signifi cance of incongruence. Cladistics 10 : 315 319
.
Felsenstein , J. 1973 . Maximum likelihood and minimum-steps
meth-ods for estimating evolutionary trees from data on discrete
characters. Systematic Zoology 22 : 240 249 .
Felsenstein , J. 1985 . Confi dence limits on phylogenies: An
approach us-ing the bootstrap. Evolution 39 : 783 791 .
Furbank , R. T. 1998 . C 4 pathway. In A. S. Raghavendra [ed.],
Photosynthesis, a comprehensive treatise, 123 135. Cambridge
University Press, Cambridge, UK.
Gibson , A. C. 1978 . Rayless secondary xylem of Halophytum.
Bulletin of the Torrey Botanical Club 105 : 39 44 .
Gibson , A. C. 1996 . Structure function relations of warm
desert plants. Springer-Verlag, Berlin, Germany.
G erere , I. , W. Tezara , C. Herrera , M. D. Fern ndez , and A.
Herrera . 1996 . Recycling of CO 2 during induction of CAM by
drought in Talinum paniculatum (Portulacaceae). Physiologia
Plantarum 98 : 471 476 .
Applequist , W. L. , and R. S. Wallace . 2000 . Phylogeny of the
Madagascan endemic family Didiereaceae. Plant Systematics and
Evolution 221 : 157 166 .
Applequist , W. L. , and R. S. Wallace . 2001 . Phylogeny of the
Portulacaceous cohort based on ndhF sequence data. Systematic
Botany 26 : 406 419 .
Applequist , W. L. , and R. S. Wallace . 2003 . Expanded
circumscrip-tion of Didiereaceae and its division into three
subfamilies. Adansonia 25 : 13 16 .
Applequist , W. L. , W. L. Wagner , E. A. Zimmer , and M.
Nepokroeff . 2006 . Molecular evidence resolving the systematic
position of Hectorella (Portulacaceae). Systematic Botany 31 : 310
319 .
Arakaki , M. , R. Nyffeler , U. Eggli , M. R. Matthew , E.
Spriggs , and E. J. Edwards . 2010a . Phylogenetic relationships in
the Portulacineae (Caryophyllales), combining phylogenomic and
di-rected PCR approaches. In Proceedings of Botany 2010,
Providence, Rhode Island, abstract 392, website
http://2010.botanyconference.org/engine/search/index.php?func=detail
& aid=392 [accessed 23 August 2010].
Arakaki , M. , R. Nyffeler , M. Moore , and E. J. Edwards .
2010b . Using plastid genomic data to date the origin of the cacti.
In Proceedings of Botany 2010, Providence, Rhode Island, USA.
Abstract 398,
http://2010.botanyconference.org/engine/search/index.php?func=detail
& aid=398 [accessed 23 August 2010].
Bailey , K. 1976 . Potassium argon ages from the Galapagos
Islands. Science 192 : 465 467 .
Barthlott , W. , and D. R. Hunt . 1993 . Cactaceae. In K.
Kubitzki, J. G. Rohwer, and V. Bittrich [eds.], The families and
genera of fl owering plants, vol. 2, 161 196. Springer-Verlag,
Berlin, Germany.
Behnke , H. D. 1994 . Sieve-element plastids: Their signifi
cance for the evolution and systematics of the order. In H. D.
Behnke and T. J. Mabry [eds.], Caryophyllales, evolution and
systematics, 87 121. Springer-Verlag, Berlin, Germany.
Blackwell , W. H. 1977 . The subfamilies of the Chenopodiaceae.
Taxon 26 : 395 397 .
Brockington , S. F. , R. Alexandre , J. Ramdial , M. J. Moore ,
S. Crawley , A. Dhingra , K. Hilu , et al . 2009 . Phylogeny of the
Caryophyllales sensu lato: Revisiting hypotheses on pollination
biology and perianth differentiation in the core Caryophyllales.
International Journal of Plant Sciences 170 : 627 643 .
Brown , H. R. , and P. W. Hattersley . 1989 . Leaf anatomy of C
3 C 4 species as related to evolution of C 4 photosynthesis. Plant
Physiology 91 : 1543 1550 .
Brown , R. W. 1959 . Some paleobotanical problematica. Journal
of Paleontology 33 : 120 124 .
Butterworth , C. , and E. J. Edwards . 2008 . Investigating
Pereskia and the earliest divergences in Cactaceae. Haseltonia 14 :
46 53 .
Carolin , R. C. 1993 . Portulacaceae. In K. Kubitzki, J. G.
Rohwer, and V. Bittrich [eds.], The families and genera of fl
owering plants, vol. 2, 544 555. Springer-Verlag, Berlin,
Germany.
Chac n , J. , S. Madri n , M. K. Chase , and J. J. Bruhl . 2006
. Molecular phylogenetics of Oreobolus (Cyperaceae) and the origin
and diversifi cation of American species. Taxon 55 : 359 366 .
Chaney , R. W. 1944 . A fossil cactus from the Eocene of Utah.
American Journal of Botany 31 : 507 528 .
Croizat , L. 1952 . Manual of phytogeography. W. Junk, The
Hague, Netherlands.
Cronquist , A. 1981 . An integrated system of classifi cation of
fl owering plants. Columbia University Press, New York, New York,
USA.
Crozier , B. S. 2005 . Systematics of Cactaceae Juss.:
Phylogeny, cpDNA evolution, and classifi cation, with emphasis on
the genus Mammillaria Haw. Ph.D. dissertation, University of Texas,
Austin, Texas, USA.
Cu noud , P. , V. Savolainen , L. W. Chatrou , M. Powell , R. J.
Grayer , and M. W. Chase . 2002 . Molecular phylogenetics of
Caryophyllales based on nuclear 18S rDNA and plastid rbcL, atpB ,
and matK DNA sequences. American Journal of Botany 89 : 132 144
.
Cushman , J. C. 2001 . Crassulacean acid metabolism: A plastic
pho-tosynthetic adaptation to arid environments. Plant Physiology
127 : 1439 1448 .