Reticulate evolution and sea-level ﬂuctuations together
drove species diversiﬁcation of slipper orchids
(Paphiopedilum) in South-East Asia
YAN-YAN GUO,*†‡ YI-BO LUO,* ZHONG-JIAN LIU†and XIAO-QUAN WANG*
*State Key Laboratory of Systematic and Evolutionary Botany, Institute of Botany, Chinese Academy of Sciences, Xiangshan,
Beijing 100093, China, †Shenzhen Key Laboratory for Orchid Conservation and Utilization, The National Orchid Conservation
Center of China and The Orchid Conservation and Research Center of Shenzhen, No. 889, Wangtong Road, Shenzhen 518114,
China, ‡Center for Biotechnology and BioMedicine, Graduate School at Shenzhen, Tsinghua University, Shenzhen 518055,
South-East Asia covers four of the world’s biodiversity hotspots, showing high species
diversity and endemism. Owing to the successive expansion and contraction of distri-
bution and the fragmentation by geographical barriers, the tropical ﬂora greatly diver-
siﬁed in this region during the Tertiary, but the evolutionary tempo and mode of
species diversity remain poorly investigated. Paphiopedilum, the largest genus of slip-
per orchids comprising nearly 100 species, is mainly distributed in South-East Asia,
providing an ideal system for exploring how plant species diversity was shaped in this
region. Here, we investigated the evolutionary history of this genus with eight cpDNA
regions and four low-copy nuclear genes. Discordance between gene trees and network
analysis indicates that reticulate evolution occurred in the genus. Ancestral area recon-
struction suggests that vicariance and long-distance dispersal together led to its current
distribution. Diversiﬁcation rate variation was detected and strongly correlated with
the species diversity in subg. Paphiopedilum (~80 species). The shift of speciation rate
in subg. Paphiopedilum was coincident with sea-level ﬂuctuations in the late Cenozoic,
which could have provided ecological opportunities for speciation and created bridges
or barriers for gene ﬂow. Moreover, some other factors (e.g. sympatric distribution,
incomplete reproductive barriers and clonal propagation) might also be advantageous
for the formation and reproduction of hybrid species. In conclusion, our study sug-
gests that the interplay of reticulate evolution and sea-level ﬂuctuations has promoted
the diversiﬁcation of the genus Paphiopedilum and sheds light into the evolution of
Orchidaceae and the historical processes of plant species diversiﬁcation in South-East
Keywords: biogeography, hybridization, molecular phylogeny, Orchidaceae, Paphiopedilum,
Received 4 July 2014; revision received 29 March 2015; accepted 31 March 2015
Islands are natural laboratories for evolutionary studies.
Several seminal works of evolutionary biology were
stimulated by observation and research of organisms
on islands and archipelagos (Darwin & Wallace 1858;
Darwin 1859; Wallace 1860; Carlquist 1974), such as
adaptive radiation of Darwin’s ﬁnches in Gal
(Grant & Grant 2002, 2008) and silversword plants in
Hawaii (Baldwin & Wagner 2010). The geologically
young and geographically isolated islands provide best
opportunities to elucidate the interplay of ecological
Correspondence: Dr. Xiao-Quan Wang, Fax: 86 10 62590843;
E-mail: firstname.lastname@example.org and Dr. Zhong-Jian Liu, Fax:
86 755 25711928; E-mail: email@example.com
©2015 John Wiley & Sons Ltd
Molecular Ecology (2015) 24, 2838–2855 doi: 10.1111/mec.13189
and evolutionary processes in generating biological
diversity (Losos & Ricklefs 2009).
South-East Asia covers four (Indo-Burma, Sundaland,
Wallacea and Philippines) of the 25 biodiversity hot-
spots (Myers et al. 2000), showing high species diversity
and endemism. Although it only occupies 4% of the
planet’s land area, South-East Asia holds 20–25% of the
species on earth (Myers et al. 2000; Mittermeier et al.
2005). In addition, this region was geographically more
complex and dynamic than the tropical regions of
Africa and South America due to some geological
events, such as the collision of the Indian and Asian
plates, and the northward subduction of the Australian
plate beneath Indonesia in the Tertiary. Currently,
South-East Asia has more than 20 000 islands, and over
50% of the land area is composed of islands and island
groups, but biodiversity of this region is poorly studied
(Sodhi & Brook 2006; Lohman et al. 2011). Owing to the
successive expansion and contraction of distribution
and the fragmentation by geographical barriers, the
tropical ﬂora continued diversifying in this region dur-
ing a major part of the Tertiary (Holloway & Hall 1998;
Morley 2000). Moreover, the land area of this region
varied twofold when sea levels ﬂuctuated up to 50 m
with each of the Pleistocene glacial cycles (for details,
see review by Woodruff 2010). For the shallow seas of
South-East Asia, ﬂuctuating sea levels periodically con-
verted mountains into geographically isolated islands,
which facilitated speciation (Sodhi & Brook 2006;
Woodruff 2010). The evolution of species endemic to
this region is a very important question for biodiversity
Previous studies indicated that South-East Asia’s geo-
logical history, complex origin of ﬂora and fauna, and
geography had great effects on species diversiﬁcation
and extinction (Mittermeier et al. 1999; Sodhi & Brook
2006). Particularly, molecular phylogenetic and biogeo-
graphical studies showed historical migration of plants
and animals among islands and also the impacts of sea-
level ﬂuctuations on speciation (see review by Lohman
et al. 2011). However, compared to the studies on ani-
mals (see review by Brown et al. 2013), the studies on
plants are relatively scarce (Cannon & Manos 2003;
anfer et al. 2006; Muellner et al. 2008; Su & Saunders
2009; Nauheimer et al. 2012; Thomas et al. 2012; Ohtani
et al. 2013; Grudinski et al. 2014). Moreover, to date,
there are no studies investigating the temporal variation
of plant species diversiﬁcation rates based on time-cali-
brated phylogenies, and the degree to which the geolog-
ical history has inﬂuenced the species diversiﬁcation is
still an open question for this region. Therefore, there is
an urgent need for more studies to unravel the mecha-
nisms underlying the origin and maintenance of the
extraordinary ﬂoristic diversity in South-East Asia using
comprehensive methods, including the combination of
phylogenetic reconstruction and divergence time esti-
mation (Ricklefs 2007).
Paphiopedilum Pﬁtzer (Venus slipper), the largest
genus of the subfamily Cypripedioideae (slipper orch-
ids) comprising 96 accepted species (data collected from
KBG, 01/2014), is native to the subtropical and tropical
regions of South-East Asia (Fig. 1) and is an ideal group
for investigating the evolution of island species diver-
sity. It is a monophyletic genus strongly supported by
previous studies (Cox et al. 1997; Chochai et al. 2012;
Guo et al. 2012). Cox et al. (1997) used nuclear ribo-
somal DNA internal transcribed spacers (ITS) to recon-
struct the phylogeny of Paphiopedilum, but the
intersectional relationships were poorly resolved. Morri-
son et al. (2005) further constructed an ITS phylogeny
for this genus based on more samples. Their results
were largely consistent with Cox et al. (1997), but many
species did not form monophyletic groups. Recently,
Chochai et al. (2012) studied phylogenetic relationships
in the genus with ITS and four cpDNA regions and got
a better resolution. However, the ITS and cpDNA trees
showed discordance in topology. In addition, G
et al. (2014) detected incongruent phylogenetic positions
of Paphiopedilum canhii using cytological and nuclear
gene markers. Another very interesting phenomenon is
that the species diversity of Paphiopedilum is almost
three times higher than that of its sister clade (96 vs. 29,
excluding hybrid species). So far, all previous studies
mainly focused on phylogeny and infrageneric classiﬁ-
cation of the genus, with less attention paid to the
tempo and underlying driving force of species diversiﬁ-
A number of previous studies on other plant groups
used ‘cytonuclear discordance’ and reticulation in phy-
logenetic networks as evidence of reticulate evolution,
and great progress has been achieved in studying retic-
ulate evolution of homoploids and polyploids based on
multiple low-copy nuclear genes (Peng & Wang 2008;
Frajman et al. 2009; Kelly et al. 2010, 2013; Sessa et al.
2012a,b; Yang et al. 2012). The combination of low-copy
nuclear gene and chloroplast DNA markers could also
improve the power of molecular data to test phyloge-
netic hypotheses in the orchid family. For instance, Rus-
sell et al. (2010) studied the phylogeny of Polystachya
and inferred reticulate evolution events in the genus
based on the discordance between cpDNA and low-
copy nuclear gene trees. Thus, the additional use of
low-copy nuclear genes would be helpful to test the dis-
crepancies among gene trees in Paphiopedilum.
This study was motivated by the desire to under-
stand the mechanisms underlying the great diversiﬁca-
tion of Venus slipper orchids in South-East Asia,
particularly the impacts of sea-level ﬂuctuations driven
©2015 John Wiley & Sons Ltd
EVOLUTION OF SLIPPER ORCHIDS IN SE ASIA 2839
by climatic oscillations in the late Cenozoic on the evo-
lutionary tempo and mode of island species diversity.
First, we used eight cpDNA regions and four unlinked
low-copy nuclear genes to reconstruct the phylogeny of
Paphiopedilum and to test whether reticulate evolution
occurred in the genus. Then, the driving forces of
Fig. 1 Distributions of the subgenera and sections of Paphiopedilum (a–g; modiﬁed from Cribb 1998) and a comparison of numbers of
mainland and island species between different groups (h).
©2015 John Wiley & Sons Ltd
2840 Y.-Y. GUO ET AL.
species diversiﬁcation and persistence were investigated
in the genus based on estimation of divergence times
and diversiﬁcation rates, lineage-through-time (LTT)
analysis, ancestral area reconstruction, paleoclimate of
South-East Asia and geological evidence.
Materials and Methods
A total of 109 samples, representing 77 species of
Paphiopedilum (including three hybrid species), were col-
lected from the Orchid Conservation & Research Center
of Shenzhen and Missouri Botanical Garden. All of the
species are accepted by the World Checklist of Mono-
cotyledons (http://www.kew.org/wcsp/monocots). For
convenience of discussion, we followed infrageneric
classiﬁcation of the genus by Cribb (1998), and Paphio-
pedilum canhii was placed in the subgenus Megastamino-
orniak et al. 2014). Out-groups were chosen
from the sister clade of Paphiopedilum (Cox et al. 1997;
Guo et al. 2012), including three Phragmipedium species
and Mexipedium xerophyticum. The origins of the materi-
als are listed in Table S1 (Supporting information). In
the analysis of nuclear gene data, the intron sequences
of ACO and LFY were unalignable between different
genera, and thus, the two basal-most lineages of Paphio-
pedilum (Sect. Parvisepalum or Sect. Concoloria) were cho-
sen as functional out-groups.
DNA extraction, PCR ampliﬁcation, cloning and
Silica gel-dried leaves were used for DNA extraction by
the CTAB method (Rogers & Bendich 1988) or Plant
Genomic DNA Kit (Tiangen Biotech Co.), and some
DNA samples were further puriﬁed with the Wizard
DNA Clean-Up System (Promega, Madison, WI, USA).
Eight cpDNA fragments (accD, matK, rbcL, rpoC2, ycf1,
atpF-atpH, atpI-atpH and trnS-trnfM) and four unlinked
low-copy nuclear genes (ACO,DEF4, LFY and RAD51)
were PCR ampliﬁed with the primers listed in Fig. S1
and Table S2 (Supporting information). The ACO and
LFY genes were ampliﬁed with the same primers as in
Guo et al. (2012), but, for some samples, LFY fragments
were ampliﬁed with the primers PaLFYE1fF and PaL-
FYE3cR that were further designed based on the
obtained sequences. Ampliﬁcation reactions were con-
ducted following the protocols of Guo et al. (2012), with
the exception that the ampliﬁcation of DEF4 used the
protocol for cpDNA and an annealing temperature of
55 °C. The puriﬁed PCR products of the chloroplast
genes were directly sequenced with the PCR primers
(Table S2, Supporting information). The PCR product
puriﬁcation, cloning and sequencing of nuclear genes
also followed the protocols of Guo et al. (2012). After
precipitation with 95% EtOH, 3 MNaAc and 125 mM
EDTA, the sequencing products were separated on an
ABI PRISM 3730xl DNA Analyzer (Applied Biosystems).
The sequences generated in this study are deposited in
GenBank under accession nos KP311695–KP313035.
Sequences from different primers were assembled with
the CONTIGEXPRESS program of the VECTOR NTI SUITE 6.0
(Informax Inc.). Sequence alignments were made with
CLUSTALW implemented in BIOEDIT 7.0 (Hall 1999) and
reﬁned manually. Chimeric sequences were excluded
based on sequencing multiple clones and alignment of
closely related species. Then, sequences with an occa-
sional single-nucleotide mutation were removed. We
also checked possible recombination events in aligned
sequences with RDP3 (Martin et al. 2010), using three
detection methods (RDP, GNEECONV and MaxChi)
under default settings. Only potential recombination
signals detected by at least two methods were taken in
Nucleotide diversity (Pi) was estimated using DNASP
version 5.0 (Librado & Rozas 2009). The incongruence
length difference test (ILD) (Farris et al. 1994), as imple-
mented in PAUP*4.0b10 (Swofford 2002), was used to
assess congruence between different data sets. This test
only included samples shared by all data sets. For each
nuclear gene, one allele was randomly chosen to repre-
sent an individual. The results showed that signiﬁcant
conﬂicts occurred between plastid and nuclear genes
and between different nuclear genes (all Pval-
ues =0.001). Therefore, separate phylogenetic analyses
were ﬁnally conducted for different data sets.
Phylogenetic analyses based on maximum parsimony
(MP) and Bayesian inference (BI) were performed with
PAUP*4.0b10 and MRBAYES 3.1.2 (Ronquist & Huelsenbeck
2003), respectively. The MP analysis used a heuristic
search with 1000 random addition sequence replicates,
tree-bisection–reconnection (TBR) and MULTREES off.
The MAXTREES was set to 10 000, and a maximum tree
limit for each addition sequence replicate was set to
1000 (nchuck =1000). Conﬁdence levels of the tree
topologies were evaluated by bootstrap analysis (Felsen-
stein 1985), with 1000 replicates using the same heuris-
tic search settings. The best evolutionary models for
different data sets used in the BI analyses were deter-
mined by MRMODELTEST v2.2 (Nylander 2004) under the
AIC criterion (Table S3, Supporting information). For
the Bayesian inference, one cold and three incremen-
tally heated Markov chain Monte Carlo (MCMC) chains
were run for 10 000 000 cycles and repeated twice to
©2015 John Wiley & Sons Ltd
EVOLUTION OF SLIPPER ORCHIDS IN SE ASIA 2841
avoid spurious results. One tree per 1000 generations
was sampled, with a burn-in of the ﬁrst 300 samples for
each run. TRACER v1.5 (Rambaut & Drummond 2009)
was used to assess chain convergence and ensure that
the effective sample sizes (ESS) are above 200 for all
parameters. Phylogenetic inferences were based on the
trees sampled after generation 300 000.
To visualize the conﬂicts among gene trees, the FIL-
TEREDSUPERNETWORK implemented in SPLITSTREE4 v4.13.1
(Huson & Bryant 2006) was used to generate a consen-
sus tree of the separate MP trees (1 cpDNA and 4
nuclear genes). The input trees were edited as follows:
multiple individuals in a monophyletic clade collapsed
to one terminal, and a single clone was preserved for
each species in the nuclear gene trees. Networks were
constructed using default settings. Two data sets were
performed in the analysis, one including and the other
excluding clones of the three species (P.glaucophyllum,
P.primulinum and P.9yingjiangense) that were nested
in sect. Paphiopedilum (see Results).
Divergence times were estimated with BEAST v1.8.0
(Drummond & Rambaut 2007) based only on cpDNA,
because signiﬁcant conﬂicts between different nuclear
genes were detected by the ILD test and phylogenetic
analysis, and many species were revealed to be non-
monophyletic in the nuclear gene trees due to incom-
plete lineage sorting or interspeciﬁc hybridization (see
Results). Two cpDNA data sets were used for molecular
dating. One included 104 samples (see cpDNA-1 in
Table 1). The other included 71 samples, in which each
species was represented by one individual that was ran-
domly chosen, and the putative hybrids were excluded.
The GTR+I+G model was used based on the MRMODEL-
TEST as mentioned above, with estimated base frequen-
cies, uncorrelated lognormal relaxed clock and a birth–
death speciation process. The BEAST analysis was
conducted in the CIPRES SCIENCE GATEWAY V. 3.3 (Miller
et al. 2010). We conducted three independent runs
(200 000 000 generations each) of MCMC simulations,
and we sampled every 20 000 generations, with a
burn-in of 1000 trees. TRACER v1.5 was used to ensure
that the ESS is above 200 for all parameters. The
remaining trees were combined using LOGCOMBINER
v1.8.0 and annotated with TREEANNOTATOR v1.8.0. The
phylogenetic chronogram was displayed by FIGTREE
v1.4.0 (http://tree.bio.ed.ac.uk/software/ﬁgtree/) . The
crown age of the genus Paphiopedilum was set with a log-
normal distribution and an offset of 18 Ma (mean =1,
SD =0.5) and that of the conduplicate-leaved slipper
orchids was set with a lognormal distribution and an off-
set of 33 Ma (mean =2, SD =0.5), which took into
account the ages of the groups estimated in Guo et al.
(2012). The lognormal priors consider the errors in the
original estimation and thus are appropriate for the sec-
ondary calibration points (Ho & Phillips 2009). Consider-
ing that the divergence times estimated from the two
cpDNA data sets are very close (see Results), we used the
data set of 71 samples in the following diversiﬁcation rate
estimation and ancestral area reconstruction.
The temporal dynamics of diversiﬁcation in Paphio-
pedilum was measured with the LTT plots using the APE
package in R(Paradis et al. 2004). Plots were produced
based on 100 random trees from the BEAST analysis. The
net diversiﬁcation rates (c) of the genus Paphiopedilum
and the subgenus Paphiopedilum were calculated with
the GEIGER package in R(Harmon et al. 2008) following
the equation (7) of Magall
on & Sanderson (2001) (with
no extinction Ɛ= 0 and high extinction rate Ɛ= 0.9). To
test whether the late Cenozoic sea-level ﬂuctuations had
promoted the development of species diversity on
islands, the net diversiﬁcation rates were also estimated
for the species of the mainland and the islands (includ-
ing Sunda Shelf, Philippines, Wallacea, and Sahul
Shelf), respectively. In addition, we used the LASER pack-
age in Rto test for temporal variation in diversiﬁcation
rates (Rabosky 2006a,b). The test compared the ﬁt of the
best constant rate model (AIC
) with the ﬁt of the best
rate variable model (AIC
), using the statistic
. The positive value of
indicates the variation of diversiﬁcation rate.
Moreover, we used the program BAMM 2.2.0 to explore
the diversiﬁcation rate heterogeneity (Rabosky 2014).
The analysis was run with 10 000 000 MCMC genera-
tions, which sampled every 5000 generations. The con-
vergence was tested based on the MCMC output using
the CODA package in R(Plummer et al. 2006), with a
Table 1 Sequence information of the genes used in this study
length (bp) Variable sites
cpDNA 109 7289–7889 8600 267 181 0.00986
cpDNA-1 104 7289–7873 8424 262 178 0.00971
ACO 95 1265–1344 1398 467 278 0.03438
DEF4 99 1062–1209 1256 307 192 0.02800
LFY 83 1459–3026 3864 476 288 0.04561
RAD51 99 794–887 1006 255 138 0.03295
©2015 John Wiley & Sons Ltd
2842 Y.-Y. GUO ET AL.
burn-in of 10%, and effective sample sizes were ensured
to be above 1000 for all estimated parameters. The
results were used to calculate diversiﬁcation rates with
the BAMMTOOLS 2.0.2 in Rpackage (Rabosky et al. 2014).
A phylorate plot was generated with the getEventData
and plot.bammdata functions.
The ancestral distribution of Paphiopedilum was recon-
structed with RASP 2.0 (Yu et al. 2010a,b). We used the
randomly sampled 1000 trees derived from the BEAST
analysis for ancestral area reconstruction. Based on the
present distribution of the genus (data collected from
KBG), we directly divided it into ﬁve geographical
areas, including mainland South-East Asia, Sunda Shelf,
Philippines, Wallacea and Sahul Shelf. Out-groups were
excluded due to their narrow distribution in Middle
and South America.
We also tried to discriminate between incomplete
lineage sorting and hybridization, two factors that could
be responsible for the topological incongruence among
the gene trees, following the methods of previous stud-
ies (Maureira-Butler et al. 2008; Blanco-Pastor et al. 2012;
Ramadugu et al. 2013). The ﬁve data sets (1 cpDNA
and 4 nuclear genes) for the ILD test mentioned earlier
were used in the analysis. We took 20 chronograms that
were generated from the BEAST analysis as input trees
(gene tree) for coalescent simulation. The divergence
time between subg. Brachypetalum and subg. Paphiopedi-
lum was used a calibration point for the chronograms.
For each gene tree, 20 trees were simulated using the
‘Coalescence Contained within Current Tree’ module in
Mesquite 3.01 (Maddison & Maddison 2014). Consider-
ing that no information is available about the effective
population size (N
) of the genus Paphiopedilum (It
should be small based on previous studies in other
orchid groups), the N
value was set to 10, 100, 1000
and 10 000, respectively. The tree-to-tree distances were
generated in PAUP. The pairwise distances between the
gene trees (observed distribution) were compared with
those between the gene trees and the simulate trees
(base-line distribution) to check whether the observed
distribution falls in the baseline distribution.
We obtained the sequences of all eight cpDNA regions
from all samples, except failure in rbcL of one species,
trnS-trnfM of two species and rpoC2 of three species.
These sequences were combined directly for phyloge-
netic analysis. However, we found that the phylogenetic
placements of ﬁve samples from four taxa (Paphiopedi-
lum canhii,P. canhii var. funingense,P. fairrieanum, and
P. hirsutissimum) were not resolved, and in particular,
the inclusion of them resulted in very low support val-
ues for several main clades (Fig. S2, Supporting infor-
mation). Therefore, the ﬁve samples were excluded
from further analyses (the data set of cpDNA-1 in
Table 1). Compared to the ampliﬁcation of cpDNA frag-
ments, the ampliﬁcation of low-copy nuclear genes
failed in more samples. For instance, it was not possible
to amplify the LFY gene from subg. Parvisepalum. For
each of the nuclear genes sequenced in this study, none
of the sampled individuals harboured more than two
distinct sequence types (clones). Conspeciﬁc clones
were generally very similar in sequence, but showed
great variation in a few species, such as the two ACO
clones (27 variable sites) and the two DEF4 clones (22
variable sites) from Paphiopedilum glaucophyllum as well
as the two DEF4 clones (22 variable sites) from P.prim-
ulinum. Finally, we obtained ﬁve DNA data sets
(Table 1), including 104 (cpDNA), 95 (ACO), 99 (DEF4),
83 (LFY) and 99 (RAD51) samples, respectively.
The cpDNA showed the least variability
(Pi =0.00986). For the nuclear genes, LFY showed the
greatest variability (Pi =0.04561) and length variation,
and DEF4 was the least variable (Pi =0.02800). How-
ever, variation in the least variable nuclear gene was
three times of that in the chloroplast genes. The parsi-
mony-informative sites for the four nuclear regions
(ACO,DEF4, LFY and RAD51) were 278, 192, 288 and
138, respectively (Table 1).
No signiﬁcant signals of recombination were detected
by the RDP3 analysis, and thus, no sequences were
removed from the alignments.
The generated cpDNA tree is shown in Fig. 2, which is
largely congruent with that reported in Chochai et al.
(2012). The subg. Parvisepalum diverged ﬁrst, followed
by the subg. Brachypetalum and ﬁnally the ﬁve sections
of subg. Paphiopedilum. In subg. Paphiopedilum, sect. Pa-
phiopedilum and sect. Barbata formed a clade sister to a
clade comprising the other three sections, and sect.
Coryopedilum was closely related to sect. Pardalopetalum.
Sect. Cochlopetalum was not monophyletic and formed
two subclades, one including P.glaucophyllum,P.victo-
ria-mariae and P.victoria-regina and the other comprising
P.liemianum and P.primulinum. In addition, nine
P.victoria-reginae) of the 19 species, each represented by
two or more individuals, did not form monophyletic
groups, respectively (Fig. 2).
©2015 John Wiley & Sons Ltd
EVOLUTION OF SLIPPER ORCHIDS IN SE ASIA 2843
Fig. 2 Majority-rule consensus tree obtained from the maximum parsimony analysis based on cpDNA. Numbers along branches indi-
cate bootstrap values ≥70 and Bayesian posterior probabilities ≥0.90. Barb, sect. Barbata; Brac, subg. Brachypetalum; Coch, sect. Coch-
lopetalum; Cory, sect. Coryopedilum; Outg, out-group; Pard, sect. Pardalopetalum; Paph, sect. Paphiopedilum; Parv, subg. Parvisepalum.
©2015 John Wiley & Sons Ltd
2844 Y.-Y. GUO ET AL.
The individual nuclear gene trees, as shown in Fig. S3
(Supporting information), were summarized in Fig. 3.
The topologies of the four nuclear gene trees were
highly congruent in revealing the relationships of the
subgenera, including the basal position of subg. Parvi-
sepalum and a sister relationship between subg. Brac-
hypetalum and subg. Paphiopedilum with ﬁve sections,
and all supported a close relationship between sect.
Coryopedilum and sect. Pardalopetalum. However, posi-
tions of the other three sections (Cochlopetalum,Paphio-
pedilum and Barbata) of subg. Paphiopedilum were
discordant between the nuclear gene trees. The ACO
tree contained three parallel clades, corresponding to
sect. Barbata, sect. Paphiopedilum and sect. Cochlopetalum-
sect. Coryopedilum-sect. Pardalopetalum, respectively. In
the DEF4 tree, subg. Paphiopedilum was separated into
four parallel clades, that is sect. Barbata, sect. Paphiopedi-
lum, sect. Cochlopetalum and sect. Coryopedilum-sect.
Pardalopetalum. In the LFY tree, subg. Paphiopedilum was
divided into two clades, one comprising sect. Barbata
and sect. Cochlopetalum and the other comprising sect.
Paphiopedilum and sect. Coryopedilum-sect. Pardalopeta-
lum, and the monophyly of sect. Paphiopedilum was not
strongly supported. In the RAD51 tree, sect. Cochlopeta-
lum had a basal position in subg. Paphiopedilum with
low support values, and relationships of the other four
sections were poorly resolved. Additionally, it is inter-
esting that a few alleles from P.glaucophyllum of sect.
Cochlopetalum were nested into the clade of sect. Paphio-
pedilum. Moreover, in the ACO,DEF4 and RAD51 trees,
the two alleles from P.9yingjiangense were clustered
into sect. Barbata and sect. Paphiopedilum, respectively.
In contrast, only one LFY allele was obtained from this
Fig. 3 Majority-rule consensus trees
obtained from the maximum parsimony
analysis based on nuclear genes (ACO,
DEF4, LFY and RAD51). The samples
marked with asterisks in sect. Paphiopedi-
lum are alleles from sect. Cochlopetalum.
Bootstrap values ≥70 and Bayesian pos-
terior probabilities ≥0.90 are shown in
bold lines. Barb, sect. Barbata; Brac, subg.
Brachypetalum; Coch, sect. Cochlopetalum;
Cory, sect. Coryopedilum; Outg, out-
group; Pard, sect. Pardalopetalum; Paph,
sect. Paphiopedilum; Parv, subg. Parvisep-
©2015 John Wiley & Sons Ltd
EVOLUTION OF SLIPPER ORCHIDS IN SE ASIA 2845
hybrid species, which was nested in sect. Barbata (Fig.
S3, Supporting information).
Within sections, the interspeciﬁc relationships were
poorly resolved, and conspeciﬁc alleles did not form
monophyletic groups for most species, such as the ACO
gene of P.concolor,P.charlesworthii,P.insigne and
P.villosum, the DEF4 gene of P.concolor, the LFY gene of
and P.wardii, and the RAD51 gene of P.barbigerum,
P.concolor,P.primulinum,P.tranlienianum and P.wardii.
The generated networks revealed inter- and intrasec-
tional reticulations in subg. Paphiopedilum (Fig. 4). The
three sections (Cochlopetalum,Paphiopedilum and Barbata)
that exhibited discordant phylogenetic positions in the
separate gene trees formed complex networks, suggest-
ing that hybridization events occurred between sect.
Cochlopetalum and sect. Paphiopedilum and between sect.
Paphiopedilum and sect. Barbata (Fig. 4a). When the three
species (P.glaucophyllum,P.primulinum and P.9yingj-
iangense) with highly diverged alleles were excluded,
the network became much simpler, but reticulation was
still observed (Fig. 4b).
Molecular dating, diversiﬁcation rate analysis and
ancestral area reconstruction
The divergence times estimated from the two data sets
are close for the main clades (Figs 5 and S4, Supporting
information). The crown ages of subg. Parvisepalum and
subg. Paphiopedilum are very close and were dated back
to the Upper Miocene (about 7.6 Ma), whereas that of
subg. Brachypetalum was dated to the Pliocene (about
2.7 Ma). In subg. Paphiopedilum, many species diverged
in the Pliocene (Fig. 5). In particular, the LTT analysis
suggested an increase in the diversiﬁcation rate of the
genus by recent radiation (Fig. 6). The net diversiﬁca-
tion rates of the genus Paphiopedilum were c=0.1894
sp/Myr (Ɛ= 0) and c=0.1125 sp/Myr (Ɛ= 0.9),
whereas those of subg. Paphiopedilum were c=0.4841
sp/Myr (Ɛ= 0) and c=0.2801 sp/Myr (Ɛ= 0.9), respec-
tively (Table 2). The nonconstant diversiﬁcation rate
was detected with DAIC
=3.227. The BAMM analysis
also revealed a high heterogeneity of diversiﬁcation rate
among lineages and in different geological periods. For
instance, the highest diversiﬁcation rate occurred in
sect. Paphiopedilum and sect. Barbata in the Late Pliocene
and Quaternary (Fig. 7). The dot plot of divergence
times showed that most of the extant species originated
in the last 2 Ma (Fig. 8). Moreover, we found that the
species diversiﬁcation rate was higher in island than in
mainland, especially for subg. Paphiopedilum (Table 2).
The ancestral area reconstruction suggested that the
genus Paphiopedilum originated in mainland SE Asia
and then dispersed to the islands at least two times.
Both vicariance and long-distance dispersal led to its
current distribution (Fig. 5a).
Test of incomplete lineage sorting vs. hybridization
The test found that the observed distribution of the dis-
tances between the species trees partially overlaps with
the baseline distribution of the distances between the
species trees and the simulated trees under different
settings of the N
value (Fig. S5, Supporting informa-
tion). This indicates that both hybridization and incom-
plete lineage sorting could be responsible for the
topological inconsistency among the gene trees.
Reticulate evolution in Paphiopedilum
Based on the theory of universal common descent, evo-
lutionary relationships at the species level and above
could be idealistically represented with bifurcating phy-
logenetic trees. However, some evolutionary events,
such as interspeciﬁc hybridization, introgression, gene
duplication and recombination, and horizontal gene
transfer will disrupt a bifurcating tree structure, and
therefore, phylogenetic networks are more suitable to
model the real relationships among species (Doolittle
1999; Linder & Rieseberg 2004; Vriesendorp & Bakker
2005; McBreen & Lockhart 2006; Frajman et al. 2009;
Russell et al. 2010; Yang et al. 2012). Many plant evolu-
tionary studies used ‘cytonuclear discordance’ as evi-
dence of reticulate evolution, and multiple unlinked
low-copy nuclear genes and network analysis have been
used successfully to resolve hybridization and polyploi-
dization events (Peng & Wang 2008; Frajman et al. 2009;
Kelly et al. 2010; Russell et al. 2010). In the present
study, we used eight cpDNA and four low-copy
nuclear genes to elucidate the evolutionary history of
Both cpDNA and nuclear gene trees indicate that
subg. Parvisepalum diverged ﬁrst, followed by subg.
Brachypetalum and ﬁnally the ﬁve sections of subg. Pa-
phiopedilum. However, with the exception that sect.
Coryopedilum is closely related to sect. Pardalopetalum in
all gene trees, positions of the other three sections of
subg. Paphiopedilum (Cochlopetalum,Paphiopedilum, and
Barbata) are discordant between chloroplast and nuclear
gene trees, even between different nuclear gene trees
(Figs 2, 3 and S3, Supporting information). In the
cpDNA tree, sect. Paphiopedilum and sect. Barbata form a
clade with high bootstrap support, while sect. Cochlope-
©2015 John Wiley & Sons Ltd
2846 Y.-Y. GUO ET AL.
Fig. 4 Filtered supernetworks constructed from separate cpDNA and nuclear gene trees. (a) All the species were included. (b) Three
species (P.glaucophyllum,P.primulinum and P.9yingjiangense) were excluded.
©2015 John Wiley & Sons Ltd
EVOLUTION OF SLIPPER ORCHIDS IN SE ASIA 2847
Fig. 5 Divergence time estimates and reconstructed ancestral areas for Paphiopedilum based on cpDNA (a) and the history of sea-level
ﬂuctuations (b). The island species are shown in bold, and the remaining species are from the mainland. The crown age of Paphioped-
ilum and conduplicate slipper orchids was set as calibration points for time estimation. The ﬁgure showing sea-level ﬂuctuations in
South-East Asia was redrawn from Miller et al. (2005).
©2015 John Wiley & Sons Ltd
2848 Y.-Y. GUO ET AL.
talum forms two paraphyletic subclades that are closely
related to the sect. Coryopedilum-sect. Pardalopetalum
lineage (Fig. 2). The cpDNA tree topology is largely
congruent with that reported in Chochai et al. (2012). In
contrast, in the LFY tree, Sect. Barbata is sister to sect.
Cochlopetalum, whereas sect. Paphiopedilum is sister to
the sect. Coryopedilum-sect. Pardalopetalum lineage. More-
over, unlike in the cpDNA tree, the species of sect.
Cochlopetalum are clustered together in the nuclear gene
trees, with the exception that a few alleles are nested in
sect. Paphiopedilum with high bootstrap values. Based
on the topological discordance among the gene trees,
we infer that reticulate evolution could have occurred
among the three sections of subg. Paphiopedilum (sect.
Barbata, sect. Cochlopetalum and sect. Paphiopedilum).
This inference is further corroborated by the complex
reticulation found in the ﬁltered supernetworks
Some species of sect. Cochlopetalum very likely origi-
nated from intersectional hybridization. In ACO,DEF4
and RAD51 gene trees, the two alleles of Paphiopedilum
glaucophyllum are placed in different clades with strong
statistic support, one in the main clade of sect. Cochlope-
talum and the other in sect. Paphiopedilum. Moreover, in
the DEF4 tree, the two alleles of P.primulinum are
placed in two clades, like P.glaucophyllum (Figs 3 and
S3, Supporting information). The above evidence and
network analysis suggest that some species of sect.
Cochlopetalum could have originated from hybridization
between species from sect. Paphiopedilum and sect. Coch-
lopetalum (Figs 3 and 4a). Given the fact that the two
species P.glaucophyllum and P.primulinum are closely
related to the other species of sect. Cochlopetalum in the
cpDNA tree (Fig. 2), their paternal donors were very
likely from sect. Paphiopedilum. Besides, in the ACO,
DEF4 and RAD51 gene trees (Fig. S3, Supporting infor-
mation), the two alleles of P.9yingjiangense are nested
into sect. Barbata and sect. Paphiopedilum, respectively,
indicate that this species possibly originated from
hybridization with parental donors from the two sec-
tions. One may argue that the placements of the conspe-
ciﬁc alleles in different clades could be attributed to
incomplete lineage sorting. However, the fact that each
of the ﬁve sections of subg. Paphiopedilum generally
forms its own clade (Figs 2, 3 and S3, Supporting infor-
mation) does not support the incomplete lineage sorting
hypothesis, although it is difﬁcult to discriminate
between hybridization and incomplete lineage sorting
by the test (Fig. S5, Supporting information). Actually,
natural interspeciﬁc hybridization occurs commonly in
Paphiopedilum owing to synchronous ﬂowering, sympat-
ric distribution and weak reproductive isolation of the
species (Cribb 1998; Liu et al. 2009). Notably, there have
been 29 natural hybrid species reported in the genus, of
which 7 are hybrids between sect. Barbata and sect.
Furthermore, thousands of artiﬁcial interspeciﬁc hybrids
have been produced and registered with the Royal
Horticultural Society (http://apps.rhs.org.uk/horti
absence of strong interspeciﬁc reproductive barriers and
hybrid zones were also documented in other orchid
groups, such as Dactylorhiza (Aagaard et al. 2005), Epi-
dendrum (Pinheiro et al. 2010; Marques et al. 2014), Oph-
okl et al. 2008; Cortis et al. 2009) and Orchis
(Bateman et al. 2008). Therefore, hybridization might
have played an important role in orchid speciation.
25 20 15 10 5 0
12 51020 50
Number of species (loge)
Time before present (Ma)
Fig. 6 Log lineage-through-time plot for the genus Paphiopedi-
Table 2 Net diversiﬁcation rates estimated for Paphiopedilum
of species c(Ɛ=0) c(Ɛ= 0.9)
Paphiopedilum 20.44 96 0.1894 0.1125
20.44 46 0.1534 0.0809
20.44 50 0.1575 0.0843
7.62 80 0.4841 0.2801
7.62 30 0.3554 0.1718
7.62 50 0.4224 0.2261
©2015 John Wiley & Sons Ltd
EVOLUTION OF SLIPPER ORCHIDS IN SE ASIA 2849
Another interesting ﬁnding is that, within the main
lineages of the genus Paphiopedilum, many species do
not form monophyletic groups, especially in the nuclear
gene trees (Figs 2, S2–S3, Supporting information). This
is similar to the previous ﬁnding by Morrison et al.
(2005). The rampant nonmonophyly of morphological
species in gene trees may indicate that the genus has
experienced recent radiations, and there is insufﬁcient
time for lineage sorting or speciation to be completed.
Alternatively, this could be attributed to frequent
hybridization between closely related species. The
above inference is corroborated by the results of the
incomplete lineage sorting vs. hybridization test (Fig.
S5, Supporting information).
Species diversiﬁcation in Paphiopedilum and the
Paphiopedilum is phylogenetically sister to the New
World conduplicate-leaved clade comprising Mexipedi-
um and Phragmipedium (Cox et al. 1997; Guo et al. 2012),
but species number of this Old World genus is almost
three times of that of its sister clade. Our preliminary
analysis did not detect a signiﬁcant difference in molec-
ular evolutionary rate between the two sister clades,
when the average branch lengths were compared based
on the data from Guo et al. (2012). Therefore, it would
be interesting to know why Paphiopedilum had a higher
The subgenus Paphiopedilum includes about 80 species
(hybrid species excluded), accounting for 83% of the
species richness in the genus. In particular, its three sec-
tions (sect. Barbata, sect. Cochlopetalum and sect. Paphio-
pedilum), which could have been involved in reticulate
evolution as discussed earlier, represent 62.5% of spe-
cies richness (60 species) in the genus. Hence, it appears
that reticulate evolution has played a vital role in speci-
ation of this genus. Furthermore, according to the phy-
logeny and distribution of the genus (Figs 1–3), the two
subgenera Parvisepalum and Brachypetalum that diverged
earlier are mainly conﬁned to mainland Asia, whereas
the later diverged subg. Paphiopedilum spreads across
South-East Asia, with a crown age dated back to the
Upper Miocene (7.62 Ma) (Fig. 5). This, together with
ancestral area reconstruction, suggests that the genus
Paphiopedilum initially diversiﬁed in the Indo-China
Peninsula and subsequently diversiﬁed and colonized
other regions of South-East Asia (Fig. 5). The net diver-
siﬁcation rate of subg. Paphiopedilum (Ɛ=0,c=0.4841;
Ɛ= 0.9, c=0.2801 sp/Myr) is much higher than that of
the genus Paphiopedilum (Ɛ=0, c=0.1894; Ɛ= 0.9,
c=0.1125 sp/Myr) (Table 2), and the LTT and BAMM
analyses also indicate a recent increase in the diversiﬁ-
cation rate of the genus (Figs 6 and 7). These ﬁndings
suggest that South-East Asia is a species pump for the
The relatively recent divergence of many species and
the diversiﬁcation rate shift in the genus appear to have
been closely linked to sea-level ﬂuctuations (Fig. 5). Pre-
vious island biogeography studies showed that the spe-
cies diversity of islands is strongly correlated with
geographical isolation and area size (Whittaker &
andez-Palacios 2007). Instead, the species diversity
of Paphiopedilum could be mainly attributed to the com-
plex and dynamic geography of South-East Asia caused
by climatic oscillations. In particular, contrary to the
general pattern of forest expansion and contraction in
the Northern Hemisphere, the geographical range of
species in South-East Asia very likely expanded in the
ice ages and contracted in the interglacial periods, as
reported in the study of rainforests by Cannon et al.
(2009). Climatic changes in the late Cenozoic induced
ﬂuctuations of sea levels (Fig. 5). For instance, during
the Last Glacial Maximum (LGM), the sea level
dropped approximately to 120 m below the present
level, the Malay Peninsula, Java, Sumartra and Borneo
were connected by exposed seabeds, and a land area
similar to Europe formed in Sundaland (Bird et al. 2005;
Woodruff 2010). The sea-level ﬂuctuations not only
gave rise to open niches, but also provided corridors
for species dispersal or geographical barriers that could
Fig. 7 A phylorate plot showing the heterogeneity of diversiﬁ-
cation rates in the genus Paphiopedilum estimated from BAMM.
©2015 John Wiley & Sons Ltd
2850 Y.-Y. GUO ET AL.
stimulate speciation. That is, when the sea level
dropped, some species could disperse among the
islands, and previously isolated populations or species
came into contact, leading to interspeciﬁc hybridization;
when the sea level rose, the isolation of the islands
facilitated allopatric speciation. For example, Sect. Coch-
lopetalum is restricted to Sumatra and Java, both located
at the periphery of South-East Asia; the relatively iso-
lated distribution might have promoted hybrid specia-
tion in this section. Mallet et al. (2014) also reported the
role of habitat isolation in the evolution of an orchid
from a small oceanic island. New niches can inhibit
backcross and competition with parental species, as
documented in many plant groups such as Helianthus
(Rieseberg 1991), Iris (Arnold 1997) and Pinus densata
(Song et al. 2002, 2003; Mao & Wang 2011).
Our inference about the effects of sea-level ﬂuctua-
tions on the evolution of species diversity in the genus
Paphiopedilum, especially in subg. Paphiopedilum, is also
supported by the ancestral area reconstruction and
diversiﬁcation rate estimation. Among the ﬁve sections
of subg. Paphiopedilum, two (sect. Cochlopetalum and
sect. Coryopedilum) comprising 21 species are endemic
to islands, and most species of the species-rich section
Barbata occur in islands (Fig. 1), although the genus
originated in the mainland of SE Asia (Fig. 5). Notably,
for subg. Paphiopedilum, the net diversiﬁcation rate of
the island species (Ɛ=0,c=0.4224; Ɛ= 0.9, c=0.2261
sp/Myr) is obviously higher than that of the mainland
species (Ɛ=0, c=0.3554; Ɛ= 0.9, c=0.1718 sp/Myr)
(Table 2). At the genus level, the net diversiﬁcation rate
of the island species is only a little higher than that of
the mainland species, which may be explained by the
fact that the two subgenera Parvisepalum and Paphiopedi-
lum only occur in the mainland, with the exception of
P.niveum. It may be postulated that the species diversi-
ﬁcation on islands could be mainly attributed to niche
diversiﬁcation. However, the available information indi-
cates that most species from islands grow in similar
altitudes and show similar habitats, that is in leaf litter
and crevices under the tree canopy or on limestone
rocks with leaf litter (Table S4, Supporting information),
although the possibility that fragmented habitats due to
topographical heterogeneity also led to the higher speci-
ation rate in the islands cannot be ruled out. One may
still argue that sect. Paphiopedilum is distributed in
mainland of South-East Asia, where the effects of land-
mass area ﬂuctuations may have been relatively weak,
but it is the second largest section of the genus and
comprises 20 species. In fact, however, most species of
this section originated in the Pliocene and Quaternary
(Fig. 5), and their evolution, especially for the species
Fig. 8 Dot plot comparison of divergence
times of the mainland and island species
in the subgenera and sections of Paphio-
pedilum (a) and in the genus (b). Barb,
sect. Barbata; Brac, subg. Brachypetalum;
Coch, sect. Cochlopetalum; Cory, sect.
Coryopedilum; Pard, sect. Pardalopetalum;
Paph, sect. Paphiopedilum; Parv, subg.
©2015 John Wiley & Sons Ltd
EVOLUTION OF SLIPPER ORCHIDS IN SE ASIA 2851
distributed in lowlands, may also have been affected
by landmass area ﬂuctuations caused by sea-level ﬂuc-
tuations. The network analysis also indicates that sect.
Paphiopedilum was involved in reticulate evolution
(Fig. 4). Actually, the important role of sea-level ﬂuctua-
tions in species diversiﬁcation has also been reported
in some studies of animals, such as tree squirrels
from South-East Asia (Mercer & Roth 2003) and shrews
from the Philippine Archipelago (Esselstyn & Brown
Finally, we should consider the biological features of
Paphiopedilum. Previous studies showed that all
observed species in this genus are characterized by
deceptive pollination (Atwood 1985; B€
1996, 2002; Shi et al. 2009). Deceptive species have
higher degrees of gene ﬂow than rewarding species
(Scopece et al. 2010), although the pollen dispersal by
insects is limited to short distances and many species
are pollinator-limited. In addition, given the weak
interspeciﬁc reproductive isolation in Paphiopedilum,
the sympatric distribution, secondary contact and ﬂow-
ering time overlap of some species could provide
opportunity for hybridization. The interspeciﬁc hybrid-
ization may also lead to the development of new ﬂoral
characters and specialized pollination systems, which
can drive the isolation between hybrids and their par-
ent species (Vereecken et al. 2010). Moreover, the tiny
seeds of Paphiopedilum with a large air space could be
advantageous for long-distance dispersal. Orchid seeds
are produced in great number (Arditti & Ghani 2000),
but the germination rate is low in the ﬁeld because
the seed germination needs symbiotic fungi. Neverthe-
less, occasional successful long-distance dispersals
could have occurred and led to interspeciﬁc hybridiza-
It is also interesting that aneuploids occur in the sub-
genus Paphiopedilum, such as 2n=28–42 in sect. Barbara,
2n=30–37 in sect. Cochlopetalum and 2n=26–30 in sect.
Paphiopedilum (Cox et al. 1998). In particular, the Paphio-
pedilum species show many chromosomal rearrange-
ments, and the double-strand break repair processes are
dynamic and ongoing (Lan & Albert 2011). Although
these phenomena might be imprints from reticulate
evolution, the perennial habit and vegetative reproduc-
tion, like in Paphiopedilum, are helpful to the survival of
hybrids (Ellstrand et al. 1996).
How species diversiﬁcation and adaptation have been
driven is a fundamental question in the evolutionary
study of Orchidaceae. In this study, we reconstructed
the phylogeny of Paphiopedilum with cpDNA and four
unlinked low-copy nuclear genes and detected reticu-
late evolution in the genus. Besides, we found that the
net diversiﬁcation rate of subg. Paphiopedilum was much
higher than that of the genus Paphiopedilum, which was
associated with climate ﬂuctuations in the late Ceno-
zoic. The results add valuable insights into the evolu-
tion of Paphiopedilum. For instance, sea-level ﬂuctuations
disrupted the species boundaries and led to interspe-
ciﬁc hybridization, which greatly promoted species
diversiﬁcation of this genus. Overall, ecological factors
played a vital role in this process. Although the LTT
analysis and diversiﬁcation rate estimation revealed the
tempo and mode of evolution in the genus, future stud-
ies should integrate evolutionary, ecological and popu-
lation genetic approaches. Given the high species
diversity and endemism in South-East Asia, the mecha-
nisms underlying the species richness of this region
deserve more studies. In addition, the investigation on
diversiﬁcation patterns of some other endemic genera
will undoubtedly shed more light on biodiversity evolu-
tion in this region.
The authors thank Dr. Peter Bernhardt of Saint Louis Univer-
sity and Ms. Blanche Wagner of Missouri Botanical Garden for
their kind help in sample collection; Mr. Zhen Ma and Mr.
Wei-Tao Li for plasmid extraction; Dr. Xing-Hua Sui for base-
map processing; and Ms. Wan-Qing Jin, Qing Cai and Rong-
Hua Liang for their assistance in DNA sequencing. We also
thank the Subject Editor and the anonymous reviewers for
their insightful comments and suggestions on the manuscript.
This study was supported by the National Natural Science
Foundation of China (Grant Nos. 31330008, 30730010,
31300179) and the Key Research Programme of the Chinese
Academy of Sciences (KJZD-EW-L07).
Aagaard SMD, S
astad SM, Greilhuber J, Moen A (2005) A sec-
ondary hybrid zone between diploid Dactylorhiza incarnata
ssp cruenta and allotetraploid D.lapponica (Orchidaceae).
Arditti J, Ghani AKA (2000) Numerical and physical properties
of orchid seeds and their biological implications. New Phytol-
Arnold ML (1997) Natural Hybridization and Evolution. Oxford
University Press, New York.
Atwood JT (1985) Pollination of Paphiopedilum rothschildianum:
brood-site deception. National Geographic Research,1, 247–254.
Baldwin BG, Wagner WL (2010) Hawaiian angiosperm radia-
tions of North American origin. Annals of Botany,105, 849–
anfer G, Moog U, Fiala B et al. (2006) A chloroplast genealogy
of myrmecophytic Macaranga species (Euphorbiaceae) in
Southeast Asia reveals hybridization, vicariance and long-
distance dispersals. Molecular Ecology,15, 4409–4424.
anziger H (1994) Studies on the natural pollination of three
species of wild lady-slipper orchids (Paphiopedilum) in South-
©2015 John Wiley & Sons Ltd
2852 Y.-Y. GUO ET AL.
east Asia. Proceedings of the 14th World Orchid Conference. pp.
anziger H (1996) The mesmerizing wart: the pollination strat-
egy of epiphytic lady slipper orchid Paphiopedilum villosum
(Lindl) Stein (Orchidaceae). Botanical Journal of the Linnean
anziger H (2002) Smart alecks and dumb ﬂies: natural pollina-
tion of some wild lady slipper orchids (Paphiopedilum spp.,
Orchidaceae). Proceedings of the 16th World Orchid Conference.
pp. 165–169. Vancouver: Vancouver Orchid Society.
Bateman RM, Smith RJ, Fay MF (2008) Morphometric and pop-
ulation genetic analyses elucidate the origin, evolutionary
signiﬁcance and conservation implications of Orchis 9angust-
icruris (O.purpurea 9O.simia), a hybrid orchid new to Brit-
ain. Botanical Journal of the Linnean Society,157, 687–711.
Bird MI, Taylor D, Hunt C (2005) Palaeoenvironments of insu-
lar Southeast Asia during the Last Glacial Period: a savanna
corridor in Sundaland? Quaternary Science Reviews,24, 2228–
Blanco-Pastor JL, Vargas P, Pfeil BE (2012) Coalescent simula-
tions reveal hybridization and incomplete lineage sorting in
mediterranean Linaria.PLoS ONE,7, e39089.
Brown RM, Siler CD, Oliveros CH et al. (2013) Evolutionary
processes of diversiﬁcation in a model island archipelago.
Annual Review of Ecology, Evolution, and Systematics,44, 411–
Cannon CH, Manos PS (2003) Phylogeography of the Southeast
Asian stone oaks (Lithocarpus). Journal of Biogeography,30,
Cannon CH, Morley RJ, Bush ABG (2009) The current refugial
rainforests of Sundaland are unrepresentative of their bioge-
ographic past and highly vulnerable to disturbance. Proceed-
ings of the National Academy of Sciences of the United States of
Carlquist SJ (1974) Island Biology. Columbia University Press,
Chochai A, Leitch IJ, Ingrouille MJ, Fay MF (2012) Molecular
phylogenetics of Paphiopedilum (Cypripedioideae; Orchida-
ceae) based on nuclear ribosomal ITS and plastid sequences.
Botanical Journal of the Linnean Society,170, 176–196.
Cortis P, Vereecken NJ, Schiestl FP et al. (2009) Pollinator con-
vergence and the nature of species’ boundaries in sympatric
Sardinian Ophrys (Orchidaceae). Annals of Botany,104, 497–
Cox AV, Pridgeon AM, Albert VA, Chase MW (1997) Phyloge-
netics of the slipper orchids (Cypripedioideae, Orchidaceae):
nuclear rDNA ITS sequences. Plant Systematics and Evolution,
Cox AV, Abdelnour GJ, Bennett MD, Leitch IJ (1998) Genome
size and karyotype evolution in the slipper orchids (Cypripe-
dioideae: Orchidaceae). American Journal of Botany,85, 681–
Cribb PJ (1998) The Genus Paphiopedilum, 2nd edn. Natural His-
tory Publications, Kota Kinabalu and Kew.
Darwin C (1859) The Origin of Species. John Murray, London.
Darwin C, Wallace A (1858) On the tendency of species to
form varieties; and on the perpetuation of varieties and spe-
cies by natural means of selection. Journal of the Proceedings of
the Linnean Society Zoology,3,45–62.
Doolittle WF (1999) Phylogenetic classiﬁcation and the univer-
sal tree. Science,284, 2124–2129.
Drummond AJ, Rambaut A (2007) BEAST: Bayesian evolution-
ary analysis by sampling trees. BMC Evolutionary Biology,7,
Ellstrand NC, Whitkus R, Rieseberg LH (1996) Distribution
of spontaneous plant hybrids. Proceedings of the National
Academy of Sciences of the United States of America,93, 5090–
Esselstyn JA, Brown RM (2009) The role of repeated sea-level
ﬂuctuations in the generation of shrew (Soricidae: Crocidura)
diversity in the Philippine Archipelago. Molecular Phylogenet-
ics and Evolution,53, 171–181.
Farris JS, Kallersjo M, Kluge AG, Bult C (1994) Testing signiﬁ-
cance of incongruence. Cladistics,10, 315–319.
Felsenstein J (1985) Conﬁdence limits on phylogenies: an
approach using the bootstrap. Evolution,39, 783–791.
Frajman B, Eggens F, Oxelman B (2009) Hybrid origins and ho-
moploid reticulate evolution within Heliosperma (Sileneae,
Caryophyllaceae)—a multigene phylogenetic approach with
relative dating. Systematic Biology,58, 328–345.
orniak M, Szlachetko DL, Kowalkowska AK, Bohdanowicz J,
Canh CX (2014) Taxonomic placement of Paphiopedilum canhii
(Cypripedioideae; Orchidaceae) based on cytological, molec-
ular and micromorphological evidence. Molecular Phylogenet-
ics and Evolution,70, 429–441.
Grant PR, Grant BR (2002) Unpredictable evolution in a 30-year
study of Darwin’s ﬁnches. Science,296, 707–711.
Grant PR, Grant BR (2008) How and Why Species Multiply: The
Radiation of Darwin’s Finches. Princeton University Press,
Princeton, New Jersey.
Grudinski M, Pannell CM, Chase MW, Ahmad JA, Muellner-
Riehl AN (2014) An evaluation of taxonomic concepts of the
widespread plant genus Aglaia and its allies across Wallace’s
Line (tribe Aglaieae, Meliaceae). Molecular Phylogenetics and
Guo YY, Luo YB, Liu ZJ, Wang XQ (2012) Evolution and bioge-
ography of the slipper orchids: Eocene vicariance of the con-
duplicate genera in the Old and New World tropics. PLoS
Hall TA (1999) BioEdit: a user-friendly biological sequence
alignment editor and analysis program for Windows 95/98/
NT. Nucleic Acids Symposium Series,41,95–98.
Harmon L, Weir J, Brock C, Glor R, Challenger W (2008) GEI-
GER: investigating evolutionary radiations. Bioinformatics,24,
Ho SYW, Phillips MJ (2009) Accounting for calibration uncer-
tainty in phylogenetic estimation of evolutionary divergence
times. Systematic Biology,58, 367–380.
Holloway JD, Hall R (1998) SE Asian geology and biogeogra-
phy: an introduction. In: Biogeography and Geological Evolution
of SE Asia (eds Hall R, Holloway JD), pp. 1–23. Backhuys,
Leiden, The Netherlands.
Huson DH, Bryant D (2006) Application of phylogenetic net-
works in evolutionary studies. Molecular Biology and Evolu-
Kelly LJ, Leitch AR, Clarkson JJ et al. (2010) Intragenic recombi-
nation events and evidence for hybrid speciation in Nicotiana
(Solanaceae). Molecular Biology and Evolution,27, 781–799.
Kelly LJ, Leitch AR, Clarkson JJ, Knapp S, Chase MW (2013)
Reconstructing the complex evolutionary origin of wild allo-
polyploid tobaccos (Nicotiana section suaveolentes). Evolution,
©2015 John Wiley & Sons Ltd
EVOLUTION OF SLIPPER ORCHIDS IN SE ASIA 2853
Lan TY, Albert VA (2011) Dynamic distribution patterns of
ribosomal DNA and chromosomal evolution in Paphiopedi-
lum, a lady’s slipper orchid. BMC Plant Biology,11, 126.
Librado P, Rozas J (2009) DnaSP v5: a software for comprehen-
sive analysis of DNA polymorphism data. Bioinformatics,25,
Linder CR, Rieseberg LH (2004) Reconstructing patterns of
reticulate evolution in plants. American Journal of Botany,91,
Liu ZJ, Chen SC, Chen LJ, Lei SP (2009) The Genus Paphiopedi-
lum in China. Science Press, Beijing.
Lohman DJ, de Bruyn M, Page T et al. (2011) Biogeography of
the Indo-Australian archipelago. Annual Review of Ecology,
Evolution, and Systematics,42, 205–226.
Losos JB, Ricklefs RE (2009) Adaptation and diversiﬁcation on
islands. Nature,457, 830–836.
Maddison WP, Maddison DR (2014) Mesquite: a modular sys-
tem for evolutionary analysis. Version 3.01. Available at
on S, Sanderson MJ (2001) Absolute diversiﬁcation rates
in angiosperm clades. Evolution,55, 1762–1780.
Mallet B, Martos F, Blambert L, Pailler T, Humeau L (2014)
Evidence for isolation-by-habitat among populations of an
epiphytic orchid species on a small oceanic island. PLoS
Mao JF, Wang XR (2011) Distinct niche divergence character-
izes the homoploid hybrid speciation of Pinus densata on the
Tibetan Plateau. The American Naturalist,177, 424–439.
Marques I, Draper D, Riofrio L, Naranjo C (2014) Multiple
hybridization events, polyploidy and low postmating isola-
tion entangle the evolution of neotropical species of Epiden-
drum (Orchidaceae). BMC Evolutionary Biology,14, 20.
Martin DP, Lemey P, Lott M et al. (2010) RDP3: a ﬂexible and
fast computer program for analyzing recombination. Bioinfor-
Maureira-Butler IJ, Pfeil BE, Muangprom A et al. (2008) The
reticulate history of Medicago (Fabaceae). Systematic Biology,
McBreen K, Lockhart PJ (2006) Reconstructing reticulate evolu-
tionary histories of plants. Trends in Plant Science,11, 398–
Mercer JM, Roth VL (2003) The effects of Cenozoic global
change on squirrel phylogeny. Science,299, 1568–1572.
Miller KG, Kominz MA, Browning JV et al. (2005) The phan-
erozoic record of global sea-level change. Science,310, 1293–
Miller MA, Pfeiffer W, Schwartz T (2010) Creating the CIPRES
Science Gateway for inference of large phylogenetic trees.
Proceedings of the Gateway Computing Environments Workshop
(GCE), 14 Nov. 2010, New Orleans, Louisiana.
Mittermeier RA, Myers N, Gil PR, Mittermeier CG (1999) Hot-
spots: Earth’s Biologically Richest and Most Endangered
Terrestrial Ecoregions. Conservation International and Agru-
ıon Sierra Madre, Monterrey, Mexico.
Mittermeier RA, Gil PR, Hoffman M et al. (2005) Hotspots Revis-
ited: Earth’s Biologically Richest and Most Endangered Terrestrial
Ecoregions. Conservation International and Agrupaci
Madre, Monterrey, Mexico.
Morley RJ (2000) Origin and Evolution of Tropical Rain Forests.
John Wiley & Sons Ltd, Chichester, UK.
Morrison CL, Hovatter K, Eackles M, Spidle AP, King TL
(2005) Molecular identiﬁcation of Cypripedioid orchids in
international trade. Selbyana,26, 196–216.
Muellner AN, Pannell CM, Coleman A, Chase MW (2008) The
origin and evolution of Indomalesian, Australasian and Paci-
ﬁc island biotas: insights from Aglaieae (Meliaceae, Sapin-
dales). Journal of Biogeography,35, 1769–1789.
Myers N, Mittermeier RA, Mittermeier CG, da Fonseca GAB,
Kent J (2000) Biodiversity hotspots for conservation priori-
ties. Nature,403, 853–858.
Nauheimer L, Boyce PC, Renner SS (2012) Giant taro and its
relatives: a phylogeny of the large genus Alocasia (Araceae)
sheds light on Miocene ﬂoristic exchange in the Malesian
region. Molecular Phylogenetics and Evolution,63,43–51.
Nylander JAA (2004) MrModeltest v2: program distributed by
the author. Evolutionary Biology Centre, Uppsala University.
Ohtani M, Kondo T, Tani N et al. (2013) Nuclear and chloroplast
DNA phylogeography reveals Pleistocene divergence and
subsequent secondary contact of two genetic lineages of the
tropical rainforest tree species Shorea leprosula (Dipterocarpa-
ceae) in South-East Asia. Molecular Ecology,22, 2264–2279.
Paradis E, Claude J, Strimmer K (2004) APE: analyses of phy-
logenetics and evolution in R language. Bioinformatics,20,
Peng D, Wang XQ (2008) Reticulate evolution in Thuja inferred
from multiple gene sequences: implications for the study of
biogeographical disjunction between eastern Asia and North
America. Molecular Phylogenetics and Evolution,47, 1190–1202.
Pinheiro F, de Barros F, Palma-Silva C et al. (2010) Hybridiza-
tion and introgression across different ploidy levels in the
Neotropical orchids Epidendrum fulgens and E.puniceoluteum
(Orchidaceae). Molecular Ecology,19, 3981–3994.
Plummer M, Best N, Cowles K, Vines K (2006) CODA: Conver-
gence diagnosis and output analysis for MCMC. R News,
Rabosky DL (2006a) LASER: a maximum likelihood toolkit for
detecting temporal shifts in diversiﬁcation rates from molec-
ular phylogenies. Evolutionary Bioinformatics Online,2, 247–
Rabosky DL (2006b) Likelihood methods for detecting temporal
shifts in diversiﬁcation rates. Evolution,60, 1152–1164.
Rabosky DL (2014) Automatic detection of key innovations,
rate shifts, and diversity-dependence on phylogenetic trees.
PLoS ONE,9, e89543.
Rabosky DL, Grundler M, Anderson C et al. (2014) BAMM-
tools: an R package for the analysis of evolutionary dynam-
ics on phylogenetic trees. Methods in Ecology and Evolution,5,
Ramadugu C, Pfeil BE, Keremane ML et al. (2013) A six
nuclear gene phylogeny of Citrus (Rutaceae) taking into
account hybridization and lineage sorting. PLoS ONE,8,
Ricklefs RE (2007) Estimating diversiﬁcation rates from phylo-
genetic information. Trends in Ecology & Evolution,22, 601–
Rieseberg LH (1991) Homoploid reticulate evolution in Helian-
thus (Asteraceae): evidence from ribosomal genes. American
Journal of Botany,78, 1218–1237.
Rogers SO, Bendich AJ (1988) Extraction of DNA from plant
tissues. Plant Molecular Biology (Manual),A6,1–10.
©2015 John Wiley & Sons Ltd
2854 Y.-Y. GUO ET AL.
Ronquist F, Huelsenbeck JP (2003) MrBayes 3: Bayesian phylo-
genetic inference under mixed models. Bioinformatics,19,
Russell A, Samuel R, Klejna V et al. (2010) Reticulate evolution
in diploid and tetraploid species of Polystachya (Orchidaceae)
as shown by plastid DNA sequences and low-copy nuclear
genes. Annals of Botany,106,37–56.
Scopece G, Cozzolino S, Johnson SD, Schiestl FP (2010) Pollina-
tion efﬁciency and the evolution of specialized deceptive pol-
lination systems. The American Naturalist,175,98–105.
Sessa EB, Zimmer EA, Givnish TJ (2012a) Reticulate evolution
on a global scale: a nuclear phylogeny for New World Dry-
opteris (Dryopteridaceae). Molecular Phylogenetics and Evolu-
Sessa EB, Zimmer EA, Givnish TJ (2012b) Unraveling reticulate
evolution in North American Dryopteris (dryopteridaceae).
BMC Evolutionary Biology,12, 104.
Shi J, Luo YB, Bernhardt P et al. (2009) Pollination by deceit in
Paphiopedilum barbigerum (Orchidaceae): a staminode exploits
the innate colour preferences of hoverﬂies (Syrphidae). Plant
Sodhi NS, Brook BW (2006) Southeast Asian Biodiversity in Crisis.
Cambridge University Press, New York.
Song BH, Wang XQ, Wang XR et al. (2002) Maternal lineages
of Pinus densata, a diploid hybrid. Molecular Ecology,11,
Song BH, Wang XQ, Wang XR, Ding KY, Hong DY (2003)
Cytoplasmic composition in Pinus densata and population
establishment of the diploid hybrid pine. Molecular Ecology,
okl J, Schl€
uter PM, Stuessy TF et al. (2008) Scent variation and
hybridization cause the displacement of a sexually deceptive
orchid species. American Journal of Botany,95, 472–481.
Su YCF, Saunders RMK (2009) Evolutionary divergence times
in the Annonaceae: evidence of a late Miocene origin of
Pseuduvaria in Sundaland with subsequent diversiﬁcation in
New Guinea. BMC Evolutionary Biology,9, 153.
Swofford DL (2002) PAUP*. Phylogenetic Analysis Using Parsi-
mony (* and other methods), Version 4. Sinauer Associates, Sun-
Thomas DC, Hughes M, Phutthai T et al. (2012) West to east
dispersal and subsequent rapid diversiﬁcation of the mega-
diverse genus Begonia (Begoniaceae) in the Malesian archi-
pelago. Journal of Biogeography,39,98–113.
Rambaut A, Drummond AJ (2009) TRACER version 1.5. Avail-
able at http://beast.bio.ed.ac.uk/Tracer.
Vereecken NJ, Cozzolino S, Schiestl FP (2010) Hybrid ﬂoral
scent novelty drives pollinator shift in sexually deceptive
orchids. BMC Evolutionary Biology,10, 103.
Vriesendorp B, Bakker FT (2005) Reconstructing patterns of
reticulate evolution in angiosperms: what can we do? Taxon,
Wallace AR (1860) On the zoological geography of the Malay
Archipelago. Journal of the Proceedings of the Linnean Society
Whittaker RJ, Fern
andez-Palacios JM (2007) Island Biogeography:
Ecology, Evolution, and Conservation. Oxford University Press,
Woodruff DS (2010) Biogeography and conservation in South-
east Asia: how 2.7 million years of repeated environmental
ﬂuctuations affect today’s patterns and the future of the
remaining refugial-phase biodiversity. Biodiversity Conserva-
Yang ZY, Ran JH, Wang XQ (2012) Three genome-based phylog-
eny of Cupressaceae s.l.: further evidence for the evolution of
gymnosperms and Southern Hemisphere biogeography.
Molecular Phylogenetics and Evolution,64, 452–470.
Yu Y, Harris AJ, He XJ (2010a) S-DIVA (Statistical Dispersal-
Vicariance Analysis): a tool for inferring biogeographic histo-
ries. Molecular Phylogenetics and Evolution,56, 848–850.
Yu Y, Harris AJ, He XJ (2010b) S-DIVA (Statistical Dispersal-
Vicariance Analysis) version 1.9 beta. Available at http://
X.Q.W. designed the study. Y.Y.G. performed the labo-
ratory work. X.Q.W., Z.J.L. and Y.B.L. contributed
reagents, materials and analysis tools. Y.Y.G. and
X.Q.W. analysed the data and wrote the article.
DNA sequences: GenBank accession nos KP311695–
Sequence alignments and tree ﬁles: Dryad
Additional supporting information may be found in the online ver-
sion of this article.
Fig. S1 Structure of the nuclear genes and the locations of the
primers used in this study.
Fig. S2 Majority-rule consensus tree obtained from the maxi-
mum parsimony analysis based on cpDNA.
Fig. S3 Majority-rule consensus tree obtained from the maxi-
mum parsimony analysis of each nuclear gene.
Fig. S4 Estimated divergence times of Paphiopedilum based on
cpDNA of 104 samples.
Fig. S5 Baseline and observed distributions of tree-to-tree dis-
Table S1 Sources of materials.
Table S2 PCR (P) and sequencing (S) primers used in this
Table S3 Results of MRMODELTEST.
Table S4 Habitat information of the species of Paphiopedilum.
©2015 John Wiley & Sons Ltd
EVOLUTION OF SLIPPER ORCHIDS IN SE ASIA 2855