Content uploaded by Peter Del Tredici
Author content
All content in this area was uploaded by Peter Del Tredici on Mar 01, 2019
Content may be subject to copyright.
Research Article
Phylogenetics and biogeography of Theaceae based on
sequences of plastid genes
1,2
Mi‐Mi LI
3,4
Jian‐Hua LI
*4
Peter DEL TREDICI
3
Jeffrey CORAJOD
1
Cheng‐Xin FU
1
(College of Life Sciences,Zhejiang University, Hangzhou 310058, China)
2
(Jiangsu Provincial Key Laboratory for Plant Ex Situ Conservation,Institute of Botany, Jiangsu Province and Chinese Academy of Sciences,
Nanjing 210014, China)
3
(Department of Biology,Hope College, MI 49423, USA)
4
(Arnold Arboretum,Harvard University, Jamaica Plain, MA 02130, USA)
Abstract Despite several morphological and molecular analyses of Theaceae, several outstanding issues remain in
the phylogenetics and biogeography of the family including the disputed relationships among the tribes Gordonieae,
Stewartieae, and Theeae, the controversial taxonomic status of Hartia and Stewartia, and the unclear biogeographic
history of Gordonieae and Stewartieae. In this study we gathered DNA sequences of multiple plastid genes from 27
species of Theaceae representing all genera except Laplacea, conducted phylogenetic analyses using parsimony,
likelihood, and Bayesian methods, and estimated divergence times within a Bayesian framework with fossil
calibrations and molecular data. Our data provided further support for the three tribes in the family and for the sister‐
group relationship of Theeae to Stewartieae plus Gordonieae. Within Gordonieae, our study for the first time offered
strong molecular support for the sister relationship of Franklinia and Schima. Within Stewartieae, our data supported
the paraphyly of Stewartia including Hartia. Within Stewartia, our data for the first time suggested that North
American (NA) species Stewartia ovata was more closely related to eastern Asian (EA) species than to the other NA
species Stewartia malacodendron. Biogeographic analyses indicated that disjunct endemic species of Gordonieae
might have originated from NA and those of Stewartieae from EA. Divergence times of the EA‐NA disjunct pairs
identified in this study (Franklinia and Schima in Gordonieae and S. ovata (NA) and Asian species of Stewartia) were
estimated to be in the Mid‐Miocene. Population exchanges in Gordonieae and Stewartieae may have occurred over the
Bering land bridge prior to the Mid‐Miocene.
Key words Gordonieae, plastid, small single copy, Stewartieae, Theaceae, Theeae.
Ericales as defined by the angiosperm phylogeny
group (APG, 2009) consists of 23 families that have
been traditionally placed in Rosidae, Dilleniidae, or
Asteridae (Cronquist, 1981). Theaceae is one of the
families and has been recognized narrowly excluding
the genera of Pentaphylacaceae (Schonenberger et al.,
2005; APG, 2009). A possible synapomorphy of
Theaceae is the presence of specialized pseudopollen
(Tsou, 1997).
Several phylogenetic studies of Theaceae have
been published based on morphological (Luna &
Ochoterena, 2004) and molecular data (Prince &
Parks, 2001; Li et al., 2002; Prince, 2002; Yang
et al., 2004; Wang et al., 2006). Three major lineages or
tribes have been recognized including Gordonieae
(Gordonia J. Ellis, Franklinia W. Bartram ex H.
Marshall and Schima Reinwardt ex Blume), Stewar-
tieae (Stewartia L. and Hartia Dunn), and Theeae
(Camellia L., Laplacea Kunth, Apterosperma Chang,
Polyspora Sweet ex Don, and Pyrenaria Blume)
(Prince & Parks, 2001; Yang et al., 2004; Wang
et al., 2006). However, relationships among the three
tribes remained controversial. In Prince & Parks (2001)
study using plastid gene matK and flanking region
(2118 sites), Gordonieae and Stewartieae were sister to
each other with 62% bootstrap support, while in Yang
et al.’s (2004) study, sequences from nrDNA ITS region
(756 sites) provided strong support (100% bootstrap)
for the sister relationship of Gordonieae and Theeae.
This sister relationship was also recognized by a
phylogenetic analysis based on a non‐molecular data set
of 47 characters (Wang et al., 2006); however, the
bootstrap support was <50%. Relationships within
tribes also remained either unresolved or debated in
previous studies (Prince & Parks, 2001; Yang
et al., 2004; Wang et al., 2006). Within Gordonieae,
Received: 4 November 2012 Accepted: 25 March 2013
* Author for correspondence. E‐mail: li@hope.edu. Tel.: 1‐616‐395‐
7460. Fax: 1‐616‐395‐7125.
Journal of Systematics and Evolution 51 (4): 396–404 (2013) doi: 10.1111/jse.12017
© 2013 Institute of Botany, Chinese Academy of Sciences
in Prince & Parks (2001) study, unweighted phyloge-
netic analyses of plastid genes matK and rbcL (3440
sites) did not resolve the relationships among the three
genera, while equally weighted analysis of the same
sequence data provided support for sister relationship of
Schima and Franklinia. In Yang et al.’s (2004) study,
none of the sequence data (nrDNA ITS: 756 sites,
plastid trnL‐F: 982 sites, or mitochondrial matR: 1727
sites) resolved the intergeneric relationships of Gordo-
nieae. However, in Wang et al.’s (2006) study, non‐
molecular data recognized the sister relationship of
Franklinia and Schima with 86–87% bootstrap support.
Within Stewartieae, based on non‐molecular characters,
some authors (Chun, 1934; Merrill, 1938; Wu, 1940;
Yan, 1981; Ye, 1982, 1990; Chang & Ren, 1998; Heo
et al., 2011) recognized the deciduous Stewartia and
evergreen Hartia, while others preferred to merge the
two genera (Airy‐Shaw, 1936; Spongberg, 1974;
Li, 1996). Phylogenetic studies based on molecular
data have also drawn different conclusions for the
generic status of Stewartia and Hartia. In Li et al.’s
(2002) study using nrDNA ITS data and Franklinia as
the outgroup to root parsimonious trees, Hartia and
Stewartia formed reciprocal monophyletic groups.
However, in Prince’s (2002) study using both nrDNA
ITS and plastid sequence data, and Apterosperma and
Gordonia as the outgroups, Hartia appeared to be
derived from within the paraphyletic Stewartia (boot-
strap ¼80%). Paraphyly of Stewartia was also shown
in the nrDNA ITS trees reconstructed using the maxi-
mum likelihood (ML) method (Xiang et al., 2004). Both
parsimony and ML analyses of nrDNA ITS data
suggested that the two North American (NA) Stewartia
malacodendron and Stewartia ovata formed a weakly‐
supported clade (Li et al., 2002; Xiang et al., 2004).
However, the NA clade was not recovered from either
nrDNA ITS or plastid data in Prince (2002) study.
All three tribes in Theaceae show a disjunct
distribution between eastern Asia (EA) and NA. In
Gordonieae, Franklinia, and Gordonia are monotypic,
NA taxa (Prince, 2007). The former is extinct in the wild
but has flourished in cultivation (Fry, 2000). Schima has
a wide distribution in subtropical and tropical areas of
southern Asia. Species of Hartia in the Stewartieae
occur in central and southern China, whereas Stewartia
has two species in southeastern United States and the
remaining species in EA. Laplacea of Theeae occurs in
the West Indies, and Central and South Americas
(WCSA), while Polyspora,Pyrenaria, and Aptero-
sperma are found in Southeast Asia.
Although the biogeographic patterns of the
disjunction between EA and NA are complex and
vary across lineages (Wen, 1999; Donoghue et al.,
2001), there is a predominant direction of migration of
ancestral lineages from Asia to North America—an
“out of Asia”hypothesis (Donoghue & Smith, 2004).
Few disjunct lineages between EA and eastern NA
(ENA) showed the opposite direction of migration from
ENA to EA (Donoghue & Smith, 2004). A recent
biogeographic analysis indicated that most eastern and
western NA endemic lineages of inter‐continental
disjunct genera originated from widespread ancestral
areas (Harris et al., 2013). In Stewartieae, fossils of
Stewartia have been described in the Miocene of EA
and the Oligocene of Europe (Caspary, 1872; Mai &
Walther, 1985; Suzuki & Hirata, 1989), but not in NA.
This implies an Old World origin of the disjunct
lineage. The Asian origin hypothesis is also supported
by the paraphyly of Asian Stewartia including the two
NA species of Stewartia and Hartia (Prince, 2002).
Nevertheless, phylogenetic relationships of Stewartia
remain debated and the Asian origin hypothesis needs
further testing. In Gordonieae, however, the earliest
reliable fossils have been found in NA (Grote &
Dilcher, 1992). Previous phylogenetic analyses sug-
gested that Franklinia and Schima may be sister to each
other (Prince & Parks, 2001; Wang et al., 2006).
Therefore, Gordonieae may represent one of few
lineages with migration of ancestral populations from
ENA to EA.
Some fossil leaf remains in the Upper Cretaceous
and Eocene of North America have been assigned to
Theaceae (Berry, 1916, 1924, 1925; Taylor, 1990).
However, their identities have been questioned (Grote
& Dilcher, 1989). Nevertheless, Theaceae have been
suggested to have radiated at least since the early
Eocene (Grote & Dilcher, 1992). Detailed comparative
analyses of fossil fruits and seeds from the middle
Eocene Claiborne Formation in western Kentucky and
Tennessee (Grote & Dilcher, 1989) have shown that the
fossils are closely related to Gordonia lansianthus and
Franklinia of North America or to Polyspora of
Southeast Asia (Grote & Dilcher, 1989, 1992). Fossil
capsules similar to Asian Polyspora were found in the
upper Eocene of eastern Germany (Mai & Walther,
1985). These reliable fossils provide calibration points
in molecular phylogenetic trees to estimate the
divergence times of EA‐NA disjunct lineages.
Because previous phylogenetic studies have
produced different phylogenetic trees likely due to
limited data points and/or the use of different analytical
methods along with other unknown factors, we
gathered the largest DNA sequence data set to date
for Theaceae (>9000 sites) from the small single copy
region (SSC) of plastid genome and reconstruct
phylogenetic relationships of Theaceae utilizing
© 2013 Institute of Botany, Chinese Academy of Sciences
LI et al.: Phylogeny and biogeography of Theaceae 397
maximum parsimony (MP), ML, and Bayesian meth-
ods. Specific objectives of the study were (i) to further
evaluate relationships among the three major tribes of
Theaceae (Gordonieae, Stewartieae, and Theeae); (ii) to
clarify intergeneric relationships of Franklinia,Gordo-
nia, and Schima; (iii) to test relationships of Stewartia
and Hartia; and (iv) to infer the biogeographical history
of EA‐NA disjunct lineages in Gordonieae and
Stewartieae.
1 Material and methods
1.1 Materials
Twenty‐seven samples were included in this study
(see Appendix A) representing all seven genera of
Theaceae (Prince, 2007) except for Laplacea (unavail-
able for the study). One species from Symplocaceae and
one from Styracaceae were also sampled as outgroups
to root the phylogenetic trees because the families are
most closely related to Theaceae (Geuten et al., 2004).
Within Theaceae (Prince, 2007), our taxon sampling
represents morphological and geographic diversity of
the tribes and genera, with a focus on Gordonieae and
Stewartieae. In Gordonieae we sampled Gordonia
lasianthus (L.) Ellis from Gordonia (2 species in total),
Franklinia alatamaha Bartr. ex Marshall from Frank-
linia (monotypic), and three species of Schima, which
has been treated as a monotypic genus with a complex‐
polymorphous species (Bloembergen, 1952) or as a
genus with 20 species (Keng, 1994). In Stewartieae, we
included seven species of Stewartia (8 species)
(Ye, 1982) and four species of Hartia (15 species),
which is a monophyletic group based on nrDNA ITS
data of 11 species (Li et al., 2011, unpublished data).
1.2 Molecular techniques
Genomic DNAs were extracted from silica gel
dried leaves using a Plant DNeasy Mini Kit following
the manufacturer’s instructions (Qiagen, Valencia,
CA). Polymerase chain reactions (PCR) were per-
formed to amplify SSC regions of the plastid genome
using an MJ‐Research thermocycler or Eppendorf
Mastercycler and eight primer pairs designed by
Dhingra & Folta (2005), which amply a 9000‐bp
fragment containing 10 genes: rps12,ndhF,rpl32,
ccsA,ndhD,psaC,ndhE,ndhG,ndhI, and ndhA. Each
10‐mL PCR reaction contained 2 mL buffer (5),
20 mmol/L primers, Phusion polymerase (New Eng-
land Biolabs, Cat# F540L, Ipswich, MA, USA), and
50–100 ng genomic DNA. PCR products were gel
purified in 1% agarose gel with a Qiagen Gel Extraction
Kit and sequenced using the BigDye Terminator Ready
Reaction Kit and automated Genetic Analyzers (GA
3130 at Hope College or GA 3730 at Harvard
University). Sequences were edited in Sequencher
(version 4, Gene Coding, Ann Arbor, MI).
1.3 Sequence alignment and phylogenetic analysis
The sequences were aligned using the computer
program MUSCLE (Edgar, 2004). Gaps were treated as
missing data. Phylogenetic analyses were conducted
using MP, ML, and Bayesian inference (BI) methods.
MP analyses were carried out using a heuristic tree
search with the following options in MEGA 5.05
(Tamura et al., 2011): close‐neighbor interchange
(CNI) on random trees, 10 initial trees, and search
level set to 1. Characters were equally weighted and
character states were unordered. MODELTEST (Posa-
da & Crandall, 1998) was used to select the optimal
model of sequence evolution for the ML and BI
analyses. ML analyses were conducted using the
selected model (GTR þG) and implemented in
MEGA with number of discrete gamma categories set
to 4 and heuristic tree inference with NNI swapping and
automatic making of the initial tree (Tamura
et al., 2011). Bayesian analyses were performed with
GTR þGmodel using MrBayes (Ronquist et al., 2012)
with two runs, each with four chains and 10 million
generations. The split frequency was <0.01. Trees were
sampled every 1000th generation. The first 250 000
generations were discarded as Burnin generations when
ML scores have stabilized. Majority consensus indices
of clades in the remaining trees were taken as the
posterior probability for individual clades. To assess
relative support for individual clades in both MP and
ML analyses we used nonparametric bootstrap analyses
(Felsenstein, 1985) with 10 000 replicates for MP and
500 replicates for ML, and the options were as in the MP
and ML analyses.
1.4 Biogeographic analysis
To gain insights into the geographic origin of
EA‐NA disjunct clades in Gordonieae and Stewartieae,
we used dispersal‐vicariance analysis (DIVA on RASP
by Y. Yu, A. J. Harris, and X. J. He available at http://
mnh.scu.edu.cn/soft/blog/RASP) to reconstruct the
ancestral distribution of the disjunct pairs. The 50%
consensus tree from BI analysis was used for the DIVA
analysis because the disjunct nodes and all nodes below
the disjunct nodes were well supported in MP, ML, and
BI analyses. The number of samples for species with
two or more individuals included in the reconstructed
phylogeny was reduced to one due to the requirement of
DIVA for a bifurcating tree. Because Laplacea from the
West Indies, and Central and South Americas (WCSA)
© 2013 Institute of Botany, Chinese Academy of Sciences
398 Journal of Systematics and Evolution Vol. 51 No. 4 2013
was not included in the phylogeny, the DIVA analyses
were then conducted in two ways: without this genus
and with this genus manually connected to the tree as
the sister of the Camellia‐Pyrenaria clade (double
arrow in Fig. 2) based on the phylogenetic study of
Prince & Parks (2001) using plastid sequence data. The
maxarea in DIVA was set to 3 (EA, NA, and WCSA).
Outgroups were removed from DIVA analysis because
it is not entirely clear about where their ancestral areas
might be and whether they are sister taxa to Theaceae.
1.5 Molecular dating
Fossils with known ages and the inferred
phylogeny were used to estimate the divergence times
of sister clades between EA and NA. Fossil capsules of
the early Eocene (ca. 49 mya) represent the minimal age
of the tribe Theeae because they are similar to Asian
Polyspora (Mai & Walther, 1985). Fossil fruits and
seeds of the middle Eocene (ca. 40.4 mya) assignable to
either Gordonia or Franklinia suggest the minimal age
of the tribe Gordonieae (Grote & Dilcher, 1992).
Placing the fossil ages to the crown clades generated
unreasonably old age for the Theaceae (>125 mya).
Therefore, we used the fossils as the ages of the stem
clades for both tribes Theeae and Gordonieae. The
molecular dating exercise was conducted using BEAST
program (Drummond et al., 2012) with the following
parameters: GTR þGmodel, estimated base frequen-
cies, gamma categories 4, uncorrelated lognormal
relaxed clock model, Yule speciation process with a
UPGMA starting tree, and lognormal prior of calibra-
tion points with the offsets of 40.4 and 49 mya for
Gordonieae and Theeae, respectively, and the log
standard deviations of 2.365 and 2.2474, respectively.
For both Gordonieae and Theeae, the standard
deviations put the 95% quantiles to 89.3 mya, which
is the estimated age of the entire Ericales (Martínez‐
Millán, 2010). The MCMC analyses in BEAST were
run for 50 million generations with 2 million
generations as burnin, as determined by Tracer 1.5
(Drummond et al., 2012), and trees were sampled every
1000th generation. Effective sample size (ESS) was
ensured to be over 100 as shown in Tracer 1.5. Means
and 95% range of the estimated divergence times were
calculated in TreeAnnotator of the BEAST software
package (Drummond et al., 2012).
2 Results
2.1 Phylogenetic analyses
Individual analyses of each plastid gene did not
produce phylogenetic trees with strongly supported
but conflicting clades, we therefore combined them in
a single data set, which had 9787 bp, 572 of which
were parsimony informative (5.8%). MP analysis
generated 65 parsimonious trees of 1685 steps (CI ¼
0.89, RI ¼0.92). Trees from ML (Fig. 1) and BI
showed congruent relationships as in the MP trees.
Within Theaceae all three tribes (Gordonieae, Stew-
artieae, and Theeae) received strong support. Theeae
was sister to the clade containing Stewartieae and
Theeae. This relationship was more strongly supported
in the MP tree (93%) than in both ML (56%)
and BI (0.68) trees. In Gordonieae, Gordonia was
sister to the clade containing Franklinia and Schima
with strong support (MP/ML/BI: 98%/97%/1), while
within Stewartieae, Stewartia malacodendron was
sister to the clade consisting of Hartia and remaining
species of Stewartia (99%/100%/1). Hartia formed
a robust clade (99%/99%/1), and so did Stewartia
(minus S. malacodendron) (99%/99%/1). EA‐NA
disjunct clades were Franklinia (NA)—Schima (EA)
(98%/97%/1) (C1 in Fig. 2) and Stewartia ovata
Fig. 1. Maximum likelihood tree (ln likelihood ¼22960.55) of
Theaceae based on 10 plastid genes. Numbers at branches are bootstrap
percentages of MP and ML and posterior probability of BI.
© 2013 Institute of Botany, Chinese Academy of Sciences
LI et al.: Phylogeny and biogeography of Theaceae 399
(NA)—Asian Stewartia species (87%/91%/1) (C2 in
Fig. 2).
2.2 Biogeographic analyses and divergence time
estimation
Biogeographic analyses with or without Laplacea
generated similar inferred ancestral areas for the EA‐
NA disjunct lineages. NA was inferred as the ancestral
area of Gordonieae (C3 in Fig. 2). Dispersal from NA to
EA and subsequent vicariance produced the disjunct
distribution in Gordonieae. In Stewartieae, however,
DIVA suggested that either NA or EA þNA is the
likely ancestral area of the disjunction between S. ovata
and Asian species of Stewartia (C4 in Fig. 2).
Results of divergence time estimation using
BEAST showed that Franklinia and Schima diverged
at 9.9 mya (95% CI: 5–17 mya) (C1 in Fig. 2),
Stewartia malacodendron and the remaining species
of Stewartia (including Hartia) at 18.7 mya (95% CI:
11.3–29.1 mya, C5 in Fig. 2), and S. ovata and the
Asian Stewartia species at 9.3 mya (95% CI: 5.2–
15.1 mya) (C2 in Fig. 2).
Fig. 2. Divergence times of lineages in Theaceae estimated based on the plastid phylogeny and two fossil calibration points as shown by the shaded
squares. The bars represent 95% range of the estimated times of divergence. Letters A (EA, eastern Asia), B (NA, North America), and C (West indies,
Central and South America) indicate geographic distribution of extant species and inferred area of ancestral populations. C1 to C5 are nodes discussed in
the text.
© 2013 Institute of Botany, Chinese Academy of Sciences
400 Journal of Systematics and Evolution Vol. 51 No. 4 2013
3 Discussion
3.1 Phylogenetic relationships and systematic
implications
Several classification systems have been proposed
in the past century for Theaceae (Melchior, 1925; Airy‐
Shaw, 1936; Sealy, 1958; Keng, 1962; Prince &
Parks, 2001). Recent phylogenetic analyses using
molecular data have provided strong support for three
clades or tribes (Gordonieae, Stewartieae, and Theeae)
in Theaceae (Prince & Parks, 2001; Yang et al., 2004;
Prince, 2007). Morphologically, Theeae differs from
the other two tribes in having multiple bracteoles on
pedicels, sepal to petal gradation, three capsule valves,
and lack of endosperm (Keng, 1962). Gordonieae is
unique in seeds with thin layers of endosperm, while
Stewartieae shows a few unique features including
seeds with copious endosperm and ovaries without a
central column (Keng, 1962). Our sequence data from
the plastid genome of all genera except for Laplacea
further support the three clades (Fig. 1).
Phylogenetic relationships among the three tribes
have been in debate over decades (Airy‐Shaw, 1936;
Keng, 1962). Airy‐Shaw (1936) included Stewartieae
in Gordonieae indicating their close affinity. Keng
(1962), however, considered Gordonieae to be more
closely related to Theeae than Stewartieae. Cladistic
analysis of morphological characters did not provide
resolution for the generic relationships (Luna &
Ochoterena, 2004; Wang et al., 2006). Molecular
studies have also offered conflicting insights (Prince &
Parks, 2001; Yang et al., 2004). Prince & Parks (2001)
provided weak support (bs ¼62% and decay index
¼1) for the sister relationship of Gordonieae and
Stewartieae, whereas Yang et al.’s (2004) nrDNA ITS
data suggested a closer relationship of Gordonieae with
Theeae than to Stewartieae with strong support
(bs ¼100%). However, it is unclear whether their
ML analysis of the ITS data also provided support for
the closer relationship of Gordonieae with Theeae. In
addition, the combined analyses of ITS, trnL‐F, and
matR data sets did not show >50% bootstrap support
for the sister relationship of Gorodineae and Theeae
(Yang et al., 2004). In this study, results from all
analyses are consistent with Airy‐Shaw (1936) and
Prince & Parks (2001). However, the support is weak
from both ML (56%) and BI (0.68) analyses. Therefore,
tribal relationships of Theaceae remain unclear, and
genome‐wide data are needed to provide further
resolution to the problem.
Generic relationships of Gordonieae have not been
fully resolved by previous molecular studies (Prince &
Parks, 2001; Yang et al., 2004; Wang et al., 2006). Our
study with more plastid DNA sequence data provides
strong support for the sister‐group relationship of
Gordonia to the remaining Gordonieae (Wang
et al., 2006), and for the sister relationship of Franklinia
and Schima (Fig. 1). Although it is not clear what
morphological characters support the close relationship
of Franklinia and Schima, successful artificial crossing
of the two genera may lend further support for their
close affinity (Ranney et al., 2003).
One of the long‐lasting debates in the systematics
of Theaceae concerns the taxonomy of Hartia and
Stewartia (Chun, 1934; Airy‐Shaw, 1936; Merrill,
1938; Wu, 1940; Spongberg, 1974; Yan, 1981; Ye,
1982, 1990; Li, 1996; Chang & Ren, 1998). Many
authors (Chun, 1934; Merrill, 1938; Wu, 1940; Yan,
1981; Ye, 1982, 1990; Chang & Ren, 1998) recognized
the deciduous Stewartia and evergreen Hartia, while
several authors (Airy‐Shaw, 1936; Spongberg, 1974;
Li, 1996) suggested the combination of the two genera.
Parsimony analyses of the nuclear sequence data from
nrDNA ITS region supported the separation of the two
genera (Li et al., 2002), while ML analyses of the same
data set placed Hartia within Stewartia (Xiang et al.,
2004). In the tree (Fig. 1), species of Hartia form a
robust clade nested within Stewartia with strong
support. Thus, the plastid data support the taxonomic
treatment combining the two genera (Li, 1996). Another
taxonomic option is to recognize a new genus for
Stewartia malacodendron so that each genus represents
a separate lineage in the phylogenetic tree. However,
there are no distinct morphological or ecological charac-
ters to distinguish S. malacodendron from the remaining
species of Stewartia (Li, 1996). Therefore, we do not
advocate the further split of Stewartia into three genera.
Stewartia ovata has been taxonomically separated
as a subgenus or section (Gray, 1849; Li, 1996) from the
remaining species of Stewartia because of its unique-
ness in having a single bract (vs. two in the rest of
species) subtending a flower and free styles (vs. fused
styles). Parsimony analysis of the DNA sequence data
from the nrDNA ITS region provided weak support for
the sister relationship of S. ovata and S. malacodendron
(Li et al., 2002). In Prince’s (2002) study using plastid
gene data sets, S. ovata and S. malacodendron did not
form a clade; however, together they were in a clade
with Hartia species. Our plastid sequence data in this
study provide strong support for the distant relationship
of S. ovata and S. malacodendron (Fig. 1), and the latter
species appears to have diverged much earlier at 11.3–
29.1 mya from the remaining species of Stewartia
including Hartia (Fig. 2). Therefore, out data do not
support the recognition of S. ovata as a separate
subgenus or section.
© 2013 Institute of Botany, Chinese Academy of Sciences
LI et al.: Phylogeny and biogeography of Theaceae 401
3.2 Biogeographic implications
Our phylogenetic study has identified two EA‐NA
disjunctions: Franklinia (NA)‐Schima (EA) in Gordo-
nieae and S. ovata (NA)‐Asian Stewartia species in
Stewartieae. DIVA analyses based on the well‐resolved
phylogeny and molecular dating exercises using
BEAST suggest that the ancestral populations of
Gordonieae may have migrated from NA to EA in
the Mid‐Miocene producing a widespread distribution
(C1 in Fig. 2), which was subsequently interrupted by a
vicariant forming the EA‐NA disjunction. In the
Stewartieae, however, the migratory direction is
unresolved; the ancestral populations of the EA‐NA
disjunct lineage may be in EA or EA þNA (C4 in
Fig. 2). Nevertheless, this suggests that EA was
probably involved in the formation of the lineage,
which is dated at the mid‐Miocene. The fossil record
has shown that EA and NA exchanged flora before the
Mid‐Miocene, that is, 15 mya (Manchester, 1999).
Declining temperatures since then may have resulted in
differential survival and extinction of lineages in the
two continents, thus forming the current disjunct
distribution (Milne & Abbott, 2002). Our estimated
times of divergence between EA‐NA fall within the
range of divergent times estimated for many other EA‐
NA disjunct angiosperm clades (Wen, 1999; Xiang
et al., 2000; Milne & Abbott, 2002; Donoghue &
Smith, 2004). Two migratory pathways have been
suggested for the population exchanges between EA
and NA: the Bering and North Atlantic land bridges
(Tiffney, 1985a, 1985b). However, the latter bridge,
albeit as stepping stones available in the early Miocene,
might not be readily available in the Mid‐Miocene
(Tiffney, 1985b). Therefore, the widespread distribu-
tion of the stem lineage of the EA‐NA disjunction in
both Gordonieae and Stewartieae was more likely due
to the exchange via the Bering land bridge. Nonethe-
less, because species of both Gordonieae and Stew-
artieae occur in the temperate and subtropical regions, if
the ancestral populations of the disjunct lineages had
migrated via the Bering land bridge, they would have to
survive the cold and dark winters of the high latitude
area (Tiffney, 2000).
Based on the biogeographic reconstruction of
100 disjunct lineages in the Northern Hemisphere,
Donoghue & Smith (2004) concluded that most EA‐NA
disjuncts had an Asian origin. Recently, Harris et al.
(2013) used a Bayes‐DIVA approach to infer the
biogeographic origins of 185 endemic lineages within
23 inter‐continental disjunct genera, and their results
suggested that lineages in NA and Europe more
often originated from widespread ancestors than
EA. This study of Theaceae provides examples of
NA and EA origins of disjunct lineages in the two
areas.
In conclusion, our data support the three tribes of
Theaceae sensu Prince & Parks (2001), the sister
relationship of Theeae with the remaining Theaceae,
and the closer relationship of Franklinia and Schima
than either is to Gordonia, and the merger of Stewartia
and Hartia. Our molecular dating and DIVA analyses
suggest that Schima and Franklinia may have originat-
ed from the ancestral populations in NA, while the
ancestral area of S. ovata and Asian Stewartia species
may be EA, and that the EA‐NA disjunctions in both
Gordonieae and Stewartieae may have formed via
Bering land bridge in the Mid‐Miocene.
Acknowledgements The authors thank Richard
OLSEN of the United States Agricultural Research
Service in Washington DC, Arnold Arboretum,
Kunming Botanical Garden, Polly Hill Arboretum,
Quarryhill Botanical Garden, University of British
Columbia Botanical Garden, and Wenbo LIAO of
Zhongshan University for providing leaf material for
the study, and Arnold Arboretum, China Scholarship
Council, and Janine Luke and Mel Seiden for providing
financial support for the project.
References
Airy‐Shaw HK. 1936. Notes on the genus Schima and on the
classification of the Theaceae‐Camellioideae. Kew Bulletin
1936: 496–500.
APG (Angiosperm Phylogeny Group). 2009. An update of the
Angiosperm Phylogeny Group classification for the orders
and families of flowering plants: APG III. Botanical Journal
of the Linnean Society 161: 105–121.
Berry EW. 1916. The lower Eocene floras of southeastern North
America. U.S. Geological Survey Professional Paper 91:
1–481.
Berry EW. 1924. The middle and upper Eocene floras of south‐
eastern North America. U.S. Geological Survey Profes-
sional Paper 92: 1–206.
Berry EW. 1925. The flora of the Ripley formation. U.S.
Geological Survey Professional Paper 136: 1–94.
Bloembergen S. 1952. A critical study in the complex‐
polymorphous genus Schima (Theaceae). Reinwardtia 2:
133–183.
Caspary R. 1872. Einige Pflanzenreste aus der Bernsteinzeit.
Schriften der Koniglichen Physikalisch‐Okonomischen
Gesellschaft zu Königsberg 13: 17–18.
Chang HT, Ren SX. 1998. Theaceae. In: Flora Reipublicae
Popularis Sinicae. Beijing: Science Press. 1–251.
Chun WY. 1934. Additions to the flora of Kwantung and south‐
eastern China (2). Sunyatsenia 2: 49–87.
Cronquist A. 1981. An integreated system of classification
of flowering plants. New York: Columbia University
Press.
© 2013 Institute of Botany, Chinese Academy of Sciences
402 Journal of Systematics and Evolution Vol. 51 No. 4 2013
Dhingra A, Folta KM. 2005. ASAP: Amplification, sequencing
& annotation of plastomes. BMC Genomics 6: 176.
Donoghue MJ, Bell CD, Li JH. 2001. Phylogenetic patterns in
Northern Hemisphere plant geography. International Jour-
nal of Plant Sciences 162: S41–S52.
Donoghue MJ, Smith SA. 2004. Patterns in the assembly of
termperate forests around the Northern Hemisphere
Philosophical Transactions of the Royal Society of London
B: Biological Sciences 359: 1633–1644.
Drummond AJ, Suchard MA, Xie Dong, Rambaut A. 2012.
Bayesian phylogenetics with BEAUti and the BEAST 1.7.
Molecular Biology and Evolution 29: 1969–1973.
Edgar RC. 2004. MUSCLE: Multiple sequence alignment with
high accuracy and high throughput. Nucleic Acids Research
32: 1792–1797.
Felsenstein J. 1985. Confidence limits on phylogenies: An
approach using the bootstrap. Evolution 39: 783–791.
Fry JT. 2000. Franklinia alatamaha, a history of that ‘very
curious’shrub, Part 1: Discovery and naming of the
Franklinia.Bartram Broadside pp. 1–24.
Geuten K, Smets E, Schols P, Yuan YM, Janssens S, Küpfer P,
Pyck N. 2004. Conflicting phylogenies of balsaminoid
families and the polytomy in Ericales: Combining data in a
Bayesian framework. Molecular Phylogenetics and Evolu-
tion 31: 711–729.
Gray A. 1849. The genera of the plants of the United States. New
York: George P. Putnam.
Grote PJ, Dilcher DL. 1989. Investigations of angiosperms from
the Eocene of North America: A new genus of the Theaceae
based on fruit and seed remains. Botanical Gazette 152:
190–206.
Grote PJ, Dilcher DL. 1992. Fruits and seeds of tribe Gordonieae
(Theaceae) from the Eocene of North America. American
Journal of Botany 79: 744–753.
Harris AJ, Wen J, Xiang QY. 2013. Inferring the biogeographic
origins of inter‐continental disjunct endemics using a
Bayes‐DIVA approach. Journal of Systematics and Evolu-
tion 51: 117–133.
Heo KI, Lee S, Lee C, Kim SC. 2011. Generic delimitation
and infrageneric classification of Stewartia and Hartia
(Theaceae; Stewartieae): Insight from pollen morphology.
Plant Systematics and Evolution 297: 33–50.
Keng H. 1962. Comparative morphological studies in Theaceae.
University of California Publications in Botany 33: 269–
384.
Keng H. 1994. Flora Malesianae prevursors LVIII, part 4: The
genus Schima (Theaceae) in Malesia. Gardens’Bulletin
(Singapore) 46: 77–87.
Li J. 1996. A systematic study on the genera Stewartia and
Hartia (Theaceae). Acta Phytotaxonomica Sinica 34: 48–67.
Li JH, Del Tredici P, Yang SX, Donoghue MJ. 2002.
Phylogenetic relationships and biogeography of Stewartia
(Camellioideae, Theaceae) inferred from nuclear ribosomal
DNA ITS sequences. Rhodora 104: 117–133.
Luna I, Ochoterena H. 2004. Phylogenetic relationships of the
genera of Theaceae based on morphology. Cladistics 20:
223–270.
Mai DH, Walther H. 1985. Die obereozänen Floren des
Weisselster‐Beckens und seiner Randgebiete. Leipzig:
Deutscher Verlag fur Grundstoffindustrie.
Manchester SR. 1999. Biogeographical relationships of North
American Tertiary floras. Annals of the Missouri Botanical
Garden 86: 472–522.
Martínez‐Millán M. 2010. Fossil record and age of the Asteridae.
Botanical Review 76: 83–135.
Melchior H. 1925. Theaceae. In: Engler A, Prantl E eds.
Die naturlichen pflanzenfamilien. 2nd ed. Leipzig: Wilhelm
Engelmann. 21: 109–154.
Merrill ED. 1938. New or noteworthy Indo‐Chinese plants.
Journal of Arnold Arboretum 19: 21–70.
Milne RI, Abbott RJ. 2002. The origin and evolution of Tertiary
relict floras. Advances in Botanical Research 38: 282–314.
Posada D, Crandall KA. 1998. MODELTEST: Testing the model
of DNA substitution. Bioinformatics Application Note 14:
817–818.
Prince LM. 2002. Circumscription and biogeographic patterns in
the eastern North American‐East Asian genus Stewartia
(Theaceae: Stewartieae): Insight from chloroplast and
nuclear DNA sequences data. Castanea 67: 290–301.
Prince LM. 2007. A brief nomenclatural review of genera and
tribes in Theaceae. Aliso 24: 105–121.
Prince LM, Parks CR. 2001. Phylogenetic relationships of
Theaceae inferred from chloroplast DNA sequence data.
American Journal of Botany 88: 2309–2320.
Ranney TG, Eaker TA, Fantz PR, Parks CR. 2003. xSchimlinia
floribunda (Theaceae): A new intergeneric hybrid between
Franklinia alatamaha and Schima argentea. HortScience
38: 1198–1200.
Ronquist F, Teslenko M, van der Mark P, Ayres DL, Darling A,
Höhna S, Larget B, Liu L, Suchard MA, Huelsenbeck JP.
2012. MrBayes 3.2: Efficient Bayesian phylogenetic
inference and model choice across a large model space.
Systematic Biology 61: 539–542.
Schonenberger J, Anderberg AA, Sytsma KJ. 2005. Molecular
phylogenetics and patterns of floral evolution in the
Ericales. International Journal of Plant Sciences 166:
265–288.
Sealy JR. 1958. A revision of the genus Camellia. London: Royal
Horticlutural Society.
Spongberg SA. 1974. A review of deciduous‐leaved species of
Stewartia (Theaceae). Journal of Arnold Arboretum 55:
182–214.
Suzuki M, Hirata C. 1989. Fossil wood flora from the pumice tuff
of Yanagida Formation (Lower Miocene) at Mawaki, Noto
Peninsula (Japan). Annals of Science. Kanazawa University
26: 47–75.
Tamura K, Peterson D, Peterson N, Stecher G, Nei M, Kumar S.
2011. MEGA5: Molecular evolutionary genetics analusis
using maximum likelihood, evolutionary distance, and
maximum parsimony methods. Molecular Biology and
Evolution 28: 2731–2739.
Taylor DW. 1990. Paleobiogeographic relationships of
angiosperms from the Cretaceous and early Tertiary of
the North America area. The Botanical Review 56: 279–
320.
Tiffney BH. 1985a. Perspectives on the origin of the floristic
similarity between Eastern Asia and eastern North America.
Journal of the Arnold Arboretum 66: 73–94.
Tiffney BH. 1985b. The Eocene North Atlantic land bridge: Its
importance in tertiary and modern phytogeography of the
© 2013 Institute of Botany, Chinese Academy of Sciences
LI et al.: Phylogeny and biogeography of Theaceae 403
northern hemisphere. Journal of the Arnold Arboretum 66:
243–273.
Tiffney BH. 2000. Geographic and climatic influence on the
Cretaceous and Tertiary history of Euramerican floristic
similarity. Acta Universitatis Carolinae Geologica (Prague)
44: 5–16.
Tsou CH. 1997. Embryology of the Theaceae—Anther and
ovule development of Camellia,Franklinia, and Schima.
American Journal of Botany 84: 369–381.
Wang YH, He H, Min TL, Zhou LH, Fritsch PW. 2006. The
phylogenetic position of Apterosperma (Theaceae) based
on morphological and karyotype characters. Plant System-
atics and Evolution 260: 39–52.
Wen J. 1999. Evolution of eastern Asian and eastern North
American disjunct distributions in flowering plants. Annual
Review of Ecology and Systematics 30: 421–455.
Wu YC. 1940. Beiträge zur Kenntnis der Flora Süd‐Chinas.
Botanische Jahrbucher fur Systematik, Pflanzengeschichte
und Pflanzengeographie 71: 169–199.
Xiang QY, Soltis DE, Soltis PS, Manchester SR, Crawford DJ.
2000. Timing the eastern Asian‐eastern North American
floristic disjunction: Molecular clock corroborate paleonto-
logical estimates. Molecular Phylogenetics and Evolution
15: 462–472.
Xiang QY, Zhang WH, Ricklefs RE, Qian H, Chen ZD, Wen J,
Li J. 2004. Regional differences in rates of plant speciation
and molecular evolution: A comparison between eastern
asia and eastern north america. Evolution 58: 2175–2184.
Yan SZ. 1981. On the Chinese genera Stewartia Linn. and Hartia
Dunn. Acta Phytotaxonomica Sinica 19: 462–471.
Yang SX, Yang JB, Lei LG, Li DZ, Yoshino H, Ikeda T. 2004.
Reassessing the relationships between Gordonia and
Polyspora (Theaceae) based on the combined analyses of
molecular data from the nuclear, plastid and mitochondrial
genomes. Plant Systematics and Evolution 248: 45–55.
Ye CX. 1982. Systematic study of Stewartia and Hartia. Acta
Scientiarum Naturalium Universitatis Sunyatseni 21: 108–
116.
Ye CX. 1990. A discussion on relationship among genera
in Theoideae (Theaceae). Acta Scientiarum Naturalium
Universitatis Sunyatseni 29: 74–81.
Appendix A
Samples used in this study in the order of species
(in bold); family; voucher; source; SSC GenBank
accessions. A ¼Arnold Arboretum, KUN ¼Kunm-
ing Institute of Botany, Chinese Academy of Sciences,
NCSU ¼North Carolina State University, SYSU ¼
Zhongshan University, UBC ¼University of British
Columbia, ZJU ¼Zhejiang University.
Apterosperma oblata; Theaceae; 6423 (SYSU);
Guangdong, China; JX987783. Camellia obtusifolia;
Theaceae; Li, M. 8237 (ZJU); Hangzhou Botanical
Garden, Zhejiang, China; HM164083. Franklinia
alatamaha; Theaceae; Li, J. 4822 (A); Arnold Arbore-
tum, MA, USA; HM164089. Gordonia lasianthus;
Theaceae; Olsen, R. 4191 (NCSU); North Carolina,
USA; HM164088. Hartia crassifolia; Theaceae; Li, M.
8026 (ZJU); Mangshan, Hunan, China; HM164107.
Hartia sinensis; Theaceae; Li, M. 8000 (ZJU);
Kunming Botanical Garden, China; HM164105. Har-
tia sinii; Theaceae; Li, J. 4551 (ZJU); Laoshan,
Guangxi, China; HM164104. Hartia villosa; Theaceae;
Li, M. 8055 (ZJU); Heishiding, Guangdong, China;
HM164106. Polyspora chrysandra; Theaceae; Li, J.
3177 (A); Decanso Botanical Gardern, California,
USA; HM164084. Pyrenaria hirta; Theaceae; Li, M.
8247 (ZJU); Hangzhou Botanical Garden, Zhejiang,
China; HM164082. Schima argentea; Theaceae; Li, M.
8189 (ZJU); Lushan, Jiangxi, China; HM164086.
Schima khasiana; Theaceae; Li, J. 4809 (ZJU);
Yunnan, China; HM164087. Schima wallichii; The-
aceae; GLGS 18299; Li, J. 5891 (KUN); Gaoligong-
shan, Yunnan, China; HM164085. Stewartia
malacodendron; Theaceae; Li, J. 4827 (A); Polly
Hill Arboretum, MA, USA; HM164090. Stewartia
malacodendron; Theaceae; Del Tredici, P. and John-
son, J. 4e, 8277 (A); Alabama, USA; HM164091.
Stewartia malacodendron; Theaceae; Del Tredici, P.
and Johnson, J. 4c, 8278 (A); Alabama, USA;
HM164092. Stewartia monadelpha; Theaceae; Li, J.
6088; Quarryhill Botanical Garden, 1989.319B;
HM164096. Stewartia ovata; TheaceaeLi, J. 1601
(A); Arnold Arboretum 18847A; HM164101. Stew-
artia ovata; Theaceae; Li, J. 1602 (A); Arnold
Arboretum 18244B; HM164102. Stewartia ovata;
Theaceae; Del Tredici, P. and Johnson, J. 2–6, 8279
(A); South Carolina, USA; HM164103. Stewartia
ovata; Theaceae; Del Tredici, P. and Johnson, J. 4b,
8280 (A); Georgia, USA; HM164100. Stewartia
pseudocamellia; Theaceae; Li, J. 6090 (A); Quarryhill
Botanical Garden, 1989.071A, Japan; HM164094.
Stewartia rubiginosa; Theaceae; Li, M. 8025 (ZJU);
Ruyuan, Guangdong, China; HM164093. Stewartia
serrata; Theaceae; Li, J. 4821 (UBC); University of
British Columbia Botanical Garden, Canada;
HM164095. Stewartia sinensis; Theaceae; Li, M.
8078 (ZJU); Anhui, China; HM164098. Stewartia
sinensis; Theaceae; Li, M. 8190 (ZJU); Jiangxi, China;
HM164099. Stewartia sinensis; Theaceae; Li, M. 8267
(ZJU); Wuyishan, Jiangxi, China; HM164097. Styrax
obassia; Styracaceae; Li, J. 6156 (A); Arnold Arbore-
tum, MA, USA; HM164078. Symplocos paniculata;
Symplocaceae; Li, J. 6155 (ZJU); Arnold Arboretum,
MA, USA; HM164073.
© 2013 Institute of Botany, Chinese Academy of Sciences
404 Journal of Systematics and Evolution Vol. 51 No. 4 2013