Phylogeographical structure and temporal complexity in American sweetgum (Liquidambar styraciflua; Altingiaceae)

Abstract

Eastern North American plant biogeography has traditionally focused on two primary issues: (i) the location of temperate Pleistocene refugia and their proximity to the southern margin of the ice sheet during the last glacial maximum, and (ii) the origin of the temperate element of northern Latin America. While numerous population genetic and phylogeographical studies have focused on the first issue, few (if any) have considered the second. We addressed these issues by surveying 117 individuals from 24 populations of Liquidambar styraciflua (American sweetgum; Altingiaceae) across the southeastern USA, eastern Mexico, and Guatemala, using more than 2200 bp of chloroplast DNA sequence data. To specifically address the issue of timing, we estimated intraspecific divergence times on the basis of multiple fossil-based calibration points, using taxa from Altingiaceae (Liquidambar and Altingia) and Hammamelidaceae (Hamamelis) as outgroups. More than half of the sampled localities exhibited multiple haplotypes. Remarkably, the greatest variation was observed within the USA, with Mexico and Guatemala sharing widespread haplotypes with Texas, Mississippi, Kentucky, Ohio, and northern Virginia. This lack of differentiation suggests shared ancestral polymorphisms, and that the genetic signal we observed is older than the disjunction itself. Our data provide support for previously proposed hypotheses of Pleistocene refugia in peninsular Florida and along the eastern Atlantic, but also for deeper divergences (approximately 8 million years ago) within the USA. These patterns reflect a dynamic biogeographical history for eastern North American trees, and emphasize the importance of the inclusion of a temporal component in any phylogeographical study.

Full text

Molecular Ecology (2008) 17, 3889–3900 doi: 10.1111/j.1365-294X.2008.03875.x
© 2008 The Authors
Journal compilation © 2008 Blackwell Publishing Ltd
Blackwell Publishing Ltd
Phylogeographical structure and temporal complexity in
American sweetgum (Liquidambar styraciflua; Altingiaceae)
ASHLEY B. MORRIS,* STEFANIE M. ICKERT-BOND,† D. BURKE BRUNSON,* DOUGLAS E. SOLTIS‡
and PAMELA S. SOLTIS§
*Department of Biology, University of South Alabama, Mobile, AL 36688, USA, †Department of Biology and Wildlife, University of
Alaska, Fairbanks, AK 99775, USA, ‡Department of Botany, University of Florida, Gainesville, FL 32611, USA, §Florida Museum of
Natural History, University of Florida, Gainesville, FL 32611, USA
Abstract
Eastern North American plant biogeography has traditionally focused on two primary
issues: (i) the location of temperate Pleistocene refugia and their proximity to the southern
margin of the ice sheet during the last glacial maximum, and (ii) the origin of the temperate
element of northern Latin America. While numerous population genetic and phylogeo-
graphical studies have focused on the first issue, few (if any) have considered the second.
We addressed these issues by surveying 117 individuals from 24 populations of Liquidambar
styraciflua (American sweetgum; Altingiaceae) across the southeastern USA, eastern Mexico,
and Guatemala, using more than 2200 bp of chloroplast DNA sequence data. To specifically
address the issue of timing, we estimated intraspecific divergence times on the basis of
multiple fossil-based calibration points, using taxa from Altingiaceae (Liquidambar and
Altingia) and Hammamelidaceae (Hamamelis) as outgroups. More than half of the sampled
localities exhibited multiple haplotypes. Remarkably, the greatest variation was observed
within the USA, with Mexico and Guatemala sharing widespread haplotypes with Texas,
Mississippi, Kentucky, Ohio, and northern Virginia. This lack of differentiation suggests
shared ancestral polymorphisms, and that the genetic signal we observed is older than the
disjunction itself. Our data provide support for previously proposed hypotheses of Pleistocene
refugia in peninsular Florida and along the eastern Atlantic, but also for deeper divergences
(~8 million years ago) within the USA. These patterns reflect a dynamic biogeographical
history for eastern North American trees, and emphasize the importance of the inclusion of
a temporal component in any phylogeographical study.
Keywords: divergence time estimation, fossil calibration, Latin America, phylogeography, tem-
perate disjunction
Received 27 March 2008; revision received 13 June 2008; accepted 25 June 2008
A rapidly growing database of phylogeographical literature
for eastern North American plants and animals indicates
greater spatial and temporal complexity than previously
suggested (Soltis et al. 2006). In the absence of divergence
time estimates, most of these studies assume Pleistocene
glaciation as the primary causal factor associated with
observed phylogeographical breaks. Numerous researchers
have indicated the potential for interpretative pitfalls in the
absence of temporal data (Bermingham & Avise 1986;
Cunningham & Collins 1994; Avise 2000; Donoghue et al.
2001; Xiang & Soltis 2001; Donoghue & Moore 2003) yet
few phylogeographical studies have dealt with this issue
(but see Church et al. 2003; Sota & Hayashi 2007). The
possibility of convergent patterns being derived from
different evolutionary processes at different times, often
referred to as pseudocongruence (Cunningham & Collins
1994), has become an increasingly important consideration
in phylogeographical studies. In fact, several recent studies
suggest that observed phylogeographical breaks may
predate the last glacial maximum (LGM), or even the
Pleistocene (Klicka & Zink 1997; Austin et al. 2002; Church
et al. 2003; Zamudio & Savage 2003; Near & Keck 2005;
Howes et al. 2006). Divergence time estimation using fossil
Correspondence: Ashley B. Morris, Fax: 251-414-8220; Email:
amorris@jaguar1.usouthal.edu

3890 A. B. MORRIS ET AL.
© 2008 The Authors
Journal compilation © 2008 Blackwell Publishing Ltd
constraints of phylogenetic trees is becoming a standard
analytical tool for many biogeographical studies of
plants, and has provided a great deal of evidence for
pseudocongruent patterns among codistributed plant
genera in the Northern Hemisphere (Donoghue et al. 2001;
Xiang & Soltis 2001). The same theory and methods used
in higher-level taxa should be extended to intraspecific
phylogeographical data to formally assess the commonly
cited hypothesis of Pleistocene-driven divergence.
As many as 50 species of plants exhibit a disjunction
between the temperate floras of the eastern USA and the
mountains of eastern Mexico (Graham 1973, 1999). Oddly,
this pattern is only seen in a few animal taxa (Martin &
Harrell 1957), including the flying squirrel (Glaucomys
volans, Arbogast 2007), the red-bellied snake (Storeria
occipitomaculata), and certain Pselaphid beetle genera (Carlton
1990). Deevey (1949) suggested that this disjunction is the
result of Pleistocene cooling in the southeastern USA, forcing
temperate taxa into the southern reaches of the region,
including peninsular Florida and the mountains of eastern
Mexico. Braun (1950) countered that these elements arrived
in Mexico as early as the mid-Cenozoic, and were later
isolated by more arid conditions during the Pliocene. Fossil
pollen data support Braun’s hypothesis of early arrival,
indicating the occurrence of 10 temperate woody genera
(including Liquidambar) in the Mexican palynoflora as early
as the mid-Pliocene, approximately 5.2 to 1.6 million years
ago (Graham 1973, 1999). Graham (1999) further noted that
the appearance of these temperate woody taxa is consistent
with a major temperature decline in the mid-Miocene,
before which very few of these elements are found in the
northern Mexican palynofloras. He suggested that the
absence of temperate pollen from earlier Mexican records,
and the presence of temperate pollen in the Eocene record
of the southeastern USA, provide evidence for a southeastern
USA origin for the Mexican temperate element. However,
Axelrod (1975) argued that the Mexican temperate element
represents ‘specialized outliers’ of what was once a con-
tinentally extensive Paleogene temperate rainforest, and not
the result of a southward migration from the southeastern
USA.
For each of the taxa with a disjunct biogeographical
distribution documented by Graham (1999), the extent of
the disjunction varies (as does the degree of morphological
variation). This might be expected given differences in life
histories, habitat requirements, and responses to environ-
mental change. The taxonomic rank of taxa involved varies
as well, with some Mexican disjunct lineages considered
conspecific to their US counterparts (e.g. Carpinus caroliniana,
Fagus grandifolia, and Nyssa sylvatica), and others considered
sister species to their US counterparts (e.g. Illicium floridanum/
mexicanum, Taxodium distichum/mucronatum, and Liquidambar
styraciflua/macrophylla; Graham 1973). While these differences
are likely to result in some variability in phylogeographical
structure, the overall expectation is one of phylogeographical
convergence, given some common underlying biogeo-
graphical history. However, as mentioned above, shared
phylogeographical patterns can also be the result of
pseudocongruence, and it is essential that timing be taken
into consideration. Divergence time estimation based on
molecular sequences should provide some resolution to
this issue.
The objectives of this study were to use L. styraciflua
(American sweetgum; Altingiaceae) to (i) assess the degree
of genetic divergence between populations from eastern
North America and northern Latin America, and (ii) test
hypotheses related to the location of Pleistocene refugia for
temperate taxa. The integration of chloroplast DNA (cpDNA)
sequence data and fossil-based calibrations for the estimation
of intraspecific divergence times should shed new light on
this topic.
Materials and methods
Liquidambar styraciflua is considered a bottom-land species
and is most abundant in the Lower Mississippi Valley
(LMV). Its modern-day distribution is primarily in the
southeastern USA, extending from Connecticut to central
Florida and west to eastern Texas (Fig. 1). Liquidambar
styraciflua (including Liquidambar macrophylla as a synonym
of L. styraciflua) occurs frequently throughout eastern and
central Mexico, and extends as far south as Nicaragua
(Little 1971; Kormanik 1990). It is a relatively fast-growing
pioneer species, has an average lifespan of 200 years, and
reaches reproductive maturity at 20–30 years of age. It is a
monoecious species, with wind-pollinated flower production
in mid- to late spring and fruit production in late fall.
Fruits, known colloquially as gumballs, open to release
wind-dispersed seeds, which are also eaten (and ultimately
dispersed) by birds, squirrels, and chipmunks. The
relationship between North American Liquidambar and its
Asian and European relatives is complex, but molecular
data suggest that divergence between L. styraciflua and its
sister, Liquidambar orientalis, occurred between the Oligocene
and Miocene (Ickert-Bond & Wen 2006), during which time
the North Atlantic Land Bridge is hypothesized to have
been available for migration between these major areas
(Donoghue et al. 2001). Post-glacial fossil pollen recon-
structions for eastern North American Liquidambar support
a western Gulf Coast refugium, with subsequent northeastern
spread (Williams et al. 2004).
Data collection
arcview 3.2 (ESRI 1992–2000) was used to update the
boundaries of Little’s (1971) species distribution map for
L. styraciflua using occurrence data acquired from North
American herbarium records online (Fig. 1). Target sampling

PHYLOGEOGRAPHY OF LIQUIDAMBAR STYRACIFLUA 3891
© 2008 The Authors
Journal compilation © 2008 Blackwell Publishing Ltd
localities were evenly distributed throughout the species
range by superimposing a grid (cell size approximately
241 km2) on the distribution map and identifying localities
near the vertices of each cell. Actual sampled localities
depended on the ability to identify suitable sites at or near
those vertices, such that distances between sites varied
(Fig. 1; Table 1). Leaf material was obtained from a
maximum of 30 individuals at each locality and preserved
in silica gel desiccant.
DNA extraction, polymerase chain reaction amplification,
and DNA sequencing
Total genomic DNA was extracted from silica-dried material
using a modified cetyltrimethyl ammonium bromide
(CTAB) protocol (Doyle & Doyle 1987). Four plastid regions
were surveyed for genetic variation within and among
populations of L. styraciflua: atpB-rbcL (Hodges & Arnold
1994), psbA-trnH (Sang et al. 1997; Tate & Simpson 2003),
psbE-petL, and the trnL intron (Taberlet et al. 1991). For
psbE-petL, primers used for initial amplification were
taken from an unpublished source. Because those primers
resulted in differential amplification success, Liquidambar-
specific primers were designed for this study: PSBE1 5′-
ATGCCGAGCTCCACATATTC-3′; PSBE2 5′-CGTTGTTCTC
TTTCTTTCATCG-3′; PSBE3 5′-CGATGAAAGAAAGAGAA
CAACG-3′; and PSBE4 5′-AGGCTGAAGGAACTAAATGAAA-
3′. Final concentrations of polymerase chain reaction (PCR)
components were as follows: 1× PCR buffer, 3 mm MgCl2,
200 nm dNTPs, 200 nm forward primer, 200 nm reverse
primer, 1 m Betaine, and 1.25 U Ta q polymerase. The PCR
profile followed that of Taberlet et al. (1991). PCR products
were cleaned with Exo-Sap and were sequenced at a com-
mercial sequencing facility (High-Throughput Genomics Unit,
Department of Genome Sciences, University of Washington).
Sequence alignment and haplotype network construction
All sequences were aligned using clustal_x version 1.83
(Thompson et al. 1997) and were manually checked for
corrections using se-al Sequence Alignment Editor version
2.0a11, available for download from Andrew Rambaut
(http://evolve.zoo.ox.ac.uk/software.html). Haplotype
networks were constructed for L. styraciflua intraspecific
Table 1 Collection locality information and cpDNA haplotypes recovered for Liquidambar styraciflua
Locality State code Latitude Longitude NHaplotype(s)†
Hackneyville, AL AL 33.07 –85.88 3 K Q R
Pinnacle Mountain State Park, AR AR1 34.84 –92.48 12 K T V
Blaylock Mountain, AR AR2 36.09 –94.37 1 S
Apalachicola Bluffs and Ravines Preserve, FL FL1 30.49 –84.98 10 K L
San Felasco Hammock State Park, FL FL2 29.70 –82.46 4 E
Wekiva Springs State Park, FL FL3 28.71 –81.49 4 E G
George L. Smith State Park, GA GA1 32.54 –82.12 4 E F
Tallulah Gorge State Park, GA GA2 34.74 –83.39 2 A
Green River Lake State Park, KY KY 37.25 –85.34 4 M P
Chemin-a-Haute State Park, LA LA1 32.91 –91.85 4 H I T
Louisiana State Arboretum, LA LA2 30.80 –92.28 4 K N
Ragland Hills, MS MS1 31.20 –89.18 3 P
Starkeville, MS MS2 33.46 –88.79 3 J T W
Ev-Henwood Nature Preserve, NC NC 34.16 –78.12 9 A C D K
Hueston Woods, OH OH 39.51 –84.75 2 P
Columbia, SC SC 34.00 –81.04 4 T
Great Smoky Mountains National Park, TN TN1 35.65 –83.51 8 T U
Paris Landing State Park, TN TN2 36.43 –88.08 4 T
Sewanee, TN TN3 35.20 –85.92 4 T
Big Thicket National Preserve, TX TX 30.43 –94.10 4 P
Chesapeake, VA VA1 36.58 –76.15 3 A B
Pace Estate, VA VA2 37.92 –78.53 3 P
Mesa de la Yerba, Veracruz MEX‡ 19.56 –97.01 11 O P
San Pedro Carcha, Alta Verapaz GUA§ 15.54 –90.24 7 P
†Haplotype codes are from haplotype network in Fig. 1 and maximum likelihood topology in Fig. 2;
‡single collecting locality from Mexico; §single collecting locality from Guatemala.

3892 A. B. MORRIS ET AL.
© 2008 The Authors
Journal compilation © 2008 Blackwell Publishing Ltd
data only. A haplotype network was constructed using
statistical parsimony as implemented in tcs 1.21 (Clement
et al. 2000) with gaps treated as a fifth character state. While
it is more desirable to treat indels as additional coded
characters than as a fifth character state, tcs does not
currently allow for the inclusion of symbols in the data
matrix. All gaps were the result of length variation in
mononucleotide repeat regions. While we acknowledge
the potential for homoplasy in such regions (reviewed in
Kelchner 2000), the observed patterns among individuals
of L. styraciflua were largely congruent with those exhibited
by substitution data, warranting the consideration of these
repeats as potentially informative characters.
Phylogenetic analyses
Maximum parsimony (MP), maximum-likelihood (ML),
and Bayesian analyses were conducted using paup * (Swofford
2002) and beast version 1.43 (Drummond & Rambaut
2007). Analyses were performed on the combined data,
Fig. 1 Chloroplast haplotype distribution for Liquidambar styraciflua sampled in this study. The haplotype network constructed using
statistical parsimony is given on the left side of the figure. Each circle represents a unique haplotype, with circle size reflecting frequency
of that haplotype. Lines between circles indicate single mutational steps, while hash marks indicate unsampled or extinct haplotypes.
Dotted boxes indicate the two major intraspecific clades recovered in phylogenetic analyses. Haplotype codes (A–W) correspond to those
given in all other Tables and Figures. Colours correspond to major clades of interest as identified by maximum-likelihood analysis (Fig. 3),
and are consistent throughout Figs 2 and 3. For the distribution map given on the right of the figure, the current species range is indicated
b
y black outline. Pie charts provide haplotype frequency data for sampled localities, with colours being consistent with those given in the
haplotype network. MR, Mississippi River; AR, Apalachicola/Chattahoochee River.

PHYLOGEOGRAPHY OF LIQUIDAMBAR STYRACIFLUA 3893
© 2008 The Authors
Journal compilation © 2008 Blackwell Publishing Ltd
collapsing all individuals with the same haplotype to a
single representative (see Supplementary material). To
allow for fossil-based calibrations of our intraspecific data,
we included accessions of Liquidambar acalycina (EU595851,
EU595860, EU595869, DQ352216), Liquidambar formosana
(EU595852, EU595861, EU595870, DQ352221), L. orientalis
(EU595846, EU595855, EU595864, DQ352222), Altingia
chinensis (EU595847, EU595856, EU595865, DQ352203),
Altingia excelsa (EU595850, EU595859, EU595868, DQ352226),
Altingia obovata (EU595853, EU595862, EU595871, DQ352208),
Altingia poilanei (EU595849, EU595858, EU595867, DQ352210),
and Altingia yunnanensis (EU595848, EU595857, EU595866,
DQ352211), as well as Hamamelis virginiana (Hamamelidaceae;
EU595854, EU595863, EU595872, DQ352196), the latter
serving as an outgroup to Altingiaceae.
MP analyses were performed both with gaps treated as
missing and with gaps coded as additional characters
following the simple indel-coding method of Simmons &
Ochoterena (2000). For each analysis, we performed a heu-
ristic search with 1000 random addition replicates (holding
one tree each step), tree-bisection–reconnection (TBR) branch
swapping, and multrees in effect. All characters were
treated as unordered and equally weighted. Support for
recovered nodes was inferred by bootstrap analysis (1000
replicates, search parameters same as above).
Recent work by Posada & Buckley (2004) suggests that
while the hierarchical likelihood-ratio tests (HLRT) are the
most commonly employed model selection strategy, the
Akaike information criterion (AIC) is a superior approach.
We therefore used the AIC as implemented in modeltest
version 3.6 (Posada & Crandall 1998) to determine an
appropriate model of evolution for use in ML analyses. The
best-fit model for the combined cpDNA data was TVM + G,
with base frequencies of A = 0.3096, C = 0.1533, G = 0.1512,
and T = 0.3858, and a gamma distribution shape parameter
equal to 0.1785. A heuristic search was performed using
1000 random addition sequence replicates, holding one
tree at each step, with multrees in effect, and saving all
trees, and TBR branch swapping. Internal support was
inferred from 100 bootstrap replicates, using the full heuristic
search option and the same search strategy as above.
Estimating divergence times using fossil calibrations
The ML topology recovered above was used to test for rate
constancy. Log-likelihood scores were generated with (+cl)
and without (–cl) a molecular clock enforced. A likelihood-
ratio test (LRT) was performed to determine if there was a
significant difference in evolutionary rates among lineages
(Felsenstein 1988). The hypothesis of a molecular clock was
rejected (P value = 0.015). Divergence time estimation was
performed under a Bayesian approach as implemented in
beast version 1.43 (Drummond & Rambaut 2007). For
readers unfamiliar with beast version 1.43, two substitution
models [HKY (Hasegawa–Kishino–Yano) or GTR (general
time reversible)], and three site heterogeneity models (gamma,
invariant sites, or gamma + invariant sites) are available.
Based on the results of AIC model selection above, we
used a GTR + G model of sequence evolution, under an
uncorrelated lognormal relaxed clock model. We con-
strained several groups to be monophyletic: L. styraciflua;
L. styraciflua +L. orientalis; and A. obovata +L. acalycina +
L. formosana (see below for details). Because the tree priors
available in the current version of beast are not designed
to model mixed inter- and intraspecific data (A. Drummond,
personal communication.), we performed a sensitivity
analysis comparing two approaches to modelling the tree
prior: (i) the Yule speciation process, and (ii) a coalescent
model assuming logistic population growth. Fossil calibration
points were used to determine specific node priors (see
discussion in Ickert-Bond & Wen 2006), and lognormal
distributions were used for all priors to approximate
minimum ages while allowing nodes to be slightly younger
or considerably older. Microaltingia Zhou, Crepet, and
Nixon from the Late Cretaceous of New Jersey was used to
calibrate the root node, approximating a median age of 90
million years [lognormal mean 4.5, SD 0.2, zero offset 0;
range of 60–133 million years (Myr)]. In a morphological
study comparing Liquidambar changii Pigg, Ickert-Bond,
and Wen from the Middle Miocene of eastern Washington
to all other recognized species of Liquidambar, L. changii
was found to most closely resemble L. acalycina (Pigg et al.
2004; Ickert-Bond & Wen 2006). The original survey of L.
changii recovered the fossil as sister to all other Liquidambar
species, which have since been shown by molecular data to
be nested within Altingia (Shi et al. 2001; Ickert-Bond & Wen
2006). However, Ickert-Bond et al. (2007) recently published
a more extensive morphological survey of Altingiaceae
and found strong support for Altingia and Liquidambar as
mutually exclusive sister clades, citing a complex relationship
between morphological convergence and evolutionary
diversification rates.
Previous molecular work recovered a clade containing
L. formosana, A. obovata, and L. acalycina, the latter of which
(as described above) is most similar to L. changii, leading
Ickert-Bond & Wen (2006) to use L. changii to constrain the
age of this clade, which we do here as well (approximate
median age of 15.6 Myr; lognormal mean 2.71, SD 0.5, zero
offset 0; range of 15–40 Myr). For the L. styraciflua crown
group, we set a median age of 3 Myr (lognormal mean 1.1,
SD 1.0, zero offset 0; range of 3–21 Myr) on the basis of fossil
material from the Citronelle formation of southern Alabama,
which appears to be identical to modern L. styraciflua (B.
Axsmith, personal communication, unpublished data).
We compared the estimated mean ages and 95% confidence
intervals for individual fossil calibrations (i.e. just Microaltingia;
Microaltingia and L. changii; Microaltingia and L. styraciflua)
as well as for the three calibration points combined.

3894 A. B. MORRIS ET AL.
© 2008 The Authors
Journal compilation © 2008 Blackwell Publishing Ltd
Results
Haplotype network
The aligned haplotype data set for Liquidambar styraciflua
(excluding all other taxa) was 2165 bp and included two
variable mononucleotide repeat regions. We initially
sampled a maximum of four individuals from each of 24
populations for a total of 96 individuals (Fig. 1). Twenty-
three haplotypes were recovered with these samples, with
13 of 25 localities exhibiting more than one haplotype
(including five populations with three haplotypes each).
Samples from Guatemala and Mexico shared one of the
most frequently recovered haplotypes (P, shown in dark
red, Figs 1–3), present in samples from Texas, southern
Mississippi, Kentucky, Ohio, and northern Virginia. Given
the higher than expected haplotype diversity recovered
from our initial sampling, we increased the sample sizes for
six populations: Pinnacle Mountain State Park (AR2);
Apalachicola Bluffs and Ravines Preserve (FL1); Ev-
Henwood Nature Preserve (NC); Great Smoky Mountains
National Park (TN1); Mesa de la Yerba, Mexico (MEX); and
San Pedro Carcha, Guatemala (GUA). For each of these
populations, we sequenced an additional six to eight
individuals, bringing the total number sampled to 117. No
additional unique haplotypes were recovered (within or
among populations), but one haplotype (K, shown in
dark red, Figs 1–3) was recovered in a new locality (North
Fig. 2 Maximum-likelihood topology for all inter- and intraspecific cpDNA sequences collected for Altingiaceae. Thicker branches
correspond to bootstrap values of 93% or greater from all maximum-likelihood and maximum parsimony analyses. Additional limited
support (55–74%) was recovered within each of the major clades (data not shown). State codes within the ENA/CA clade (Liquidambar
styraciflua) correspond to those given in Table 1; haplotype codes (A–W) correspond to those in Table 1 and Fig. 1. Colour-coding indicates
major clusters of interest and is used throughout all figures. ENA/CA, Eastern North America/Central America; EAS, Eastern Asia;
WAS, Weste rn Asia .

PHYLOGEOGRAPHY OF LIQUIDAMBAR STYRACIFLUA 3895
© 2008 The Authors
Journal compilation © 2008 Blackwell Publishing Ltd
Carolina). Haplotype distribution and GenBank Accession
nos are provided in Appendix S1, Supplementary
material. Descriptions of unique haplotypes defined by
polymorphic sites are provided in Appendix S2, Supple-
mentary material.
Phylogenetic analyses
The aligned Liquidambar/Altingia data matrix included 33
taxa and 2297 bp of data. Several indels, ranging in length
from 5 to 20 bp, were required to align the data. Eight
Fig. 3 Chronogram for Altingiaceae based on a coalescent-based Bayesian approach assuming logistic population growth. Calibrated nodes
are indicated by numbers (1–3): (1) Microaltingia was used to calibrate the root; (2) Liquidambar changii was used to calibrate the clade
including Liquidambar formosana, Liquidambar acalycina, and Altingia obovata. All calibration points were calibrated using lognormal priors
to approximate minimal ages (see methods for details). Gray bars indicate 95% HPD intervals for nodes of particular interest, with ages and
95% HPD given (in millions of years) above the bars. These nodes all have posterior probabilities of 0.99–1.00 (Table 2). Haplotype letter
codes (A–W) and colour codes correspond to those in all other Tables and Figures. Divergence time estimates are further summarized in
Table 2. ENA/CA, Eastern North America/Central America; EAS, Eastern Asia; WAS, Western Asia.

3896 A. B. MORRIS ET AL.
© 2008 The Authors
Journal compilation © 2008 Blackwell Publishing Ltd
synapomorphic indels were coded for parsimony analyses.
MP analyses with gaps treated as missing recovered 26
most parsimonious trees (MPT) with a shortest length of
187 (48 parsimony-informative characters; CI = 0.7761;
HI = 0.2239; RC = 0.8515), while analyses with gaps coded
as additional characters recovered 52 MPTs with a shortest
length of 197 (56 parsimony-informative characters;
CI = 0.7792; HI = 0.2208; RC = 0.8476). While all gaps were
phylogenetically informative, their inclusion decreased
bootstrap support for some nodes (data not shown).
Relationships among the species of Liquidambar and Altingia
are largely congruent with previously published data
(Ickert-Bond & Wen 2006), with the exception of the position
of Liquidambar orientalis. This species has been shown to be
sister to L. styraciflua in almost every other study of the
group (Hoey & Parks 1994; Li et al. 1997; Li & Donoghue
1999; Shi et al. 2001; Ickert-Bond & Wen 2006), but its
position is somewhat unresolved here. The strict consensus
constructed from MPTs with gaps treated as missing
recovered L. orientalis as sister to all other members of the
family, while the strict consensus constructed from MPTs
with gaps coded as additional characters recovered the L.
styraciflua sister relationship, as previously published.
However, neither position is strongly supported (< 50%
and 67% bootstrap support, respectively). Short branch
lengths across the tree indicate limited divergence within
L. styraciflua and among members of otherwise well-
supported clades (i.e. the Indochina and Eastern Asian
clades). ML analyses recovered one tree with limited
intraspecific resolution; major clades were congruent with
those recovered using parsimony (Fig. 3).
Divergence time estimation
Results from divergence time estimation using the two tree
prior alternatives (Yule speciation process and coalescent
assuming logistic population growth) are summarized in
Table 2. Implementation of individual fossil calibration
points relative to an approach with three calibration points
result in considerably older nodes with much wider 95%
higher posterior density intervals (HPD; data not shown).
For the combined calibration approach, the coalescent model
assuming logistic population growth yielded considerably
younger mean ages and narrower 95% HPD intervals than
did the Yule speciation prior. Of particular interest are
those nodes within L. styraciflua. Under the coalescent model,
the two major intraspecific clades diverged approximately
8.34 million years ago (95% HPD = 3.14–15.97) relative to
19.21 million years ago (95% HPD = 8.52–39.31) under
the Yule speciation model. Additionally, the Carolina
coastal clade (haplotypes A–D, shown in greens and blues,
Figs 1–3) diverged during the Pleistocene (mean 1.89, 95%
HPD = 0.27–2.66) under the coalescent model, relative to
the Miocene age estimated under the Yule speciation model
(mean 5.48, 95% HPD = 0.85–9.34). Given the differences in
the two approaches, and the predominantly intraspecific
nature of the data, the coalescent tree prior appears to be a
better fit to the data.
Discussion
The patterns revealed in Liquidambar styraciflua reflect
the dynamic phylogeographical history of eastern North
American taxa, and emphasize the importance of the
integration of a temporal component into any phylo-
geographical study. The greatest divergence observed
within the species was not between the disjunct US and
eastern Mexican populations, as may have been expected
due to the great geographical distance between extant
populations, but was instead within the US populations
(Figs 1–3). Our estimates attribute this break to a pre-
Pleistocene event, roughly 8 million years ago, with
Node
Yule speciation prior Coalescent prior
PP Age (95% CI) PP Age (95% CI)
Root 1.00 101.78 (50.35, 107.50) 1.00 93.11 (64.81, 149.61)
EAS/ENA* 0.90 43.55 (19.45, 72.45) 0.99 22.16 (9.74, 44.59)
IND/EAS† 0.98 28.29 (10.41, 51.36) 0.99 19.20 (6.39, 30.75)
LORIE/LS‡ 1.00 35.85 (13.18, 58.80) 1.00 14.44 (6.51, 31.64)
LS major clades§ 1.00 19.21 (8.52, 39.31) 1.00 8.34 (3.14, 15.97)
LS Carolina clade¶ 0.99 5.48 (0.85, 9.34) 0.99 1.89 (0.27, 2.66)
*Split between the Eastern Asian + Indochina and Eastern North American clades of
Altingiaceae.
†Split between the Eastern Asian and Indochina clades of Altingiaceae.
‡Split between Liquidambar orientalis and Liquidambar styraciflua.
§Split between the two major clades within L. styraciflua.
¶Split between L. styraciflua clade containing haplotypes A–D and all other clades.
Table 2 Summary of divergence time
estimation results using two alternative
Bayesian tree priors using multiple fossil
calibration points. Estimates are given
as mean ages (in millions of years) with
95% confidence intervals in parentheses.
See methods for explanation of fossil
calibrations and a discussion of models.
Posterior probabilities (PP) are given for
each model

PHYLOGEOGRAPHY OF LIQUIDAMBAR STYRACIFLUA 3897
© 2008 The Authors
Journal compilation © 2008 Blackwell Publishing Ltd
admittedly broad confidence intervals (Table 2; Fig. 3). The
family Altingiaceae has a long and diverse fossil record in
the Northern Hemisphere that spans the Late Cretaceous
(Microaltingia) into the Recent in North America, Europe,
and Asia. There are North American Miocene reports of
Liquidambar from Alaska (Seldovian point), Idaho (Clarkia),
Georgia, Kansas, and eastern Mexico, but the relationships
of these fossil taxa to extant L. styraciflua are poorly known.
Given this information, the two clades of L. styraciflua
recovered here may represent ancient relicts of a once
widespread Liquidambar species complex. A lack of differen-
tiation among Latin American and North American
populations may actually reflect the retention of ancestral
haplotypes, as these widespread types are found in the
interior of the haplotype network (Fig. 1).
Phylogeographical structure in L. styraciflua
Previous allozyme work by Hoey & Parks (1994) uncovered
lower than average genetic diversity in L. styraciflua when
compared to other wind-pollinated taxa. From their data,
they inferred two closely allied population centres (i.e. the
southeastern US and eastern Mexico) representing one
species. Hoey & Parks (1994) further suggested that the
Mexican populations must be relatively old, consistent with
fossil data supporting a Miocene arrival of the temperate
element into Mexico, while the disjunction itself is relatively
recent. They found greater structure among Mexican
populations than among US populations, and suggested
that any observable structure in the US would have been
erased by persistent glacial expansion and contraction
during the Pleistocene, whereas Mexican populations
would have remained relatively unaffected. The data
presented here are consistent with this hypothesis, in that
only ancestral, widespread haplotypes were recovered in
Latin American populations sampled for this study
(Fig. 1). Unlike Hoey & Parks (1994), we recovered limited
genetic differentiation within and among Latin American
populations. Differences in our results and those of the
previous study could be attributed to variation in pollen-
and seed-mediated gene flow. Nuclear markers, such as
allozymes, are bi-parentally inherited, reflecting a combination
of both pollen and seed movement. Chloroplast markers,
which are maternally inherited in most plants, reflect only
seed movement. Liquidambar styraciflua is wind-pollinated
and its seeds are largely wind-dispersed, although they are
also known to be eaten by birds, squirrels, and chipmunks
(Kormanik 1990). In a study of seed dispersal in bottomland
hardwood forests, Nuttle & Haefner (2005) found that the
majority of L. styraciflua seeds fell within 50 to 100 m of the
parent tree, with many gumballs containing as many as 50
seeds falling directly below the tree. Such limited dispersal
would likely result in greater localized structure for chloro-
plast markers, while nuclear markers should exhibit a
more homogeneous signal, due to wind pollination, which
is the inverse of the patterns seen here. It may be more likely
that the cpDNA regions used here evolve more slowly than
the allozyme markers used by Hoey & Parks (1994), such
that the resulting patterns reflect different timescales.
In the present study, we recovered only two haplotypes
from Mexico and Guatemala (haplotypes O and P, shown
in dark red on Figs 1–3). Furthermore, haplotype O, which
was shared by both Latin American localities, was one of
the two most common types recovered within and among
US populations. Graham’s (1999) palynological data indicate
the presence of Liquidambar in eastern Mexico as early as
the Pliocene, and there are additional macrofossil reports
of Liquidambar from this region as early as the late Miocene
(Berry 1923). Graham (1999) suggested that the arrival of
temperate taxa such as Liquidambar into eastern Mexico
was the result of range expansion from the southeastern
USA in response to climatic cooling, such that the observed
modern disjunction would be a consequence of repeated
range expansion and contraction in response to dynamic
climate change over the last 3–6 Myr. From an evolutionary
perspective, it is possible that frequent population extinction
and recolonization of ephemeral sinks from a stable source
(within Mexico) could have prevented population subdivision
by drift, and resulted in relatively short coalescence times
among populations (Avise 2000). Alternatively, the genetic
signal we observed may be older than the disjunction itself,
reflecting a time of admixture among continuously distributed
stands of Liquidambar. Additional genetic markers (e.g.
nuclear DNA sequences and microsatellites) will be needed
to provide additional resolution to this question.
Comparisons with other ENA taxa
Some phylogeographical structuring in L. styraciflua is
consistent with previously published biogeographical
patterns for eastern North America (reviewed in Soltis et al.
2006). While phylogenetic support was limited, there
appears to be a trend towards an ‘out of Florida’ track, with
haplotypes from peninsular Florida, eastern and northern
Georgia, and coastal North Carolina and Virginia clustering
together in the haplotype network (haplotypes A–G, shown
in blues and greens, Figs 1–3). Based on our estimates, the
age of this group corresponds with the Pleistocene (Table 2;
Fig. 3), which is consistent with hypotheses that peninsular
Florida served as an ice-age refugium to many temperate
plants and animals. Alternatively, some authors have
suggested that the coastal areas of the Carolinas may have
also played a role as temperate refugia, based in part on
modern species diversity and endemism (Estill & Cruzan
2001; Sorrie & Weakley 2001), and in part on the presence
of early Pleistocene fossils for temperate species (including
Liquidambar; Whitehead 1983). Most recently, in a phylo-
geographical survey of the eastern tiger salamander

3898 A. B. MORRIS ET AL.
© 2008 The Authors
Journal compilation © 2008 Blackwell Publishing Ltd
(Ambystoma tigrinum), Church et al. (2003) found evidence
for two independent refugia along the mid-Atlantic Coastal
Plain, one of which is centred on the Carolina coast, which
is consistent with what we have found here. Given this
information, it is possible that this primarily coastal clade
recovered for L. styraciflua may represent multiple refugia
(Florida refugium in blue, Carolina refugium in green),
although additional data are needed to assess this
hypothesis.
Caveats to divergence time estimation
While our study represents one of only a few examples of
intraspecific dating in plants, there are numerous caveats
to the analyses presented here. First, limited phylogenetic
resolution within L. styraciflua likely dramatically increased
confidence intervals associated with estimated mean ages.
Future work including more rapidly evolving sequence
data may provide a solution to this issue. Second, generally
speaking, fossil position is somewhat subjective when
not resolved phylogenetically through inclusion in a
morphological matrix. In this case, previously published
morphological studies on Liquidambar (Pigg et al. 2004;
Ickert-Bond et al. 2005), Altingia (Ickert-Bond et al. 2007),
and Altingioid relatives (Zhou et al. 2001; Ickert-Bond &
Wen 2006) provided strong rationale for our choices for
fossil placement. Third, theoretical models for divergence
time estimation currently do not explicitly allow for the
analysis of mixed inter- and intraspecific data. Recent studies
have indicated the potential for decreasing substitution
rates with increasing calibration depth, which may ultimately
overestimate intraspecific divergence times (Ho et al. 2005,
2007, 2008). Approaches to rectify this issue are in the early
stages of development, and will ultimately serve as powerful
tools for estimating divergence times in such cases. Finally,
divergence time estimates are based on gene trees, and
may not reflect species trees. Particularly in the case of
more recently diverged taxa (both intra- and interspecific),
gene divergence times are likely to precede species
divergence, such that the latter may be overestimated in
attempts to estimate divergences within phylogeographical
studies (Jennings & Edwards 2005). Comparison of multiple
gene trees from independent loci coupled with coalescent
approaches my aid in the resolution of this issue (Jennings
& Edwards 2005).
Given these caveats, the work presented here still pro-
vides an important lesson. In the absence of any divergence
time estimates, a standard phylogeographical approach to
data interpretation for L. styraciflua would be to assume
two primary Pleistocene refugia on the basis of two
intraspecific clades, one potentially originating along the
Gulf Coast (shown in red, blue, and green, Figs 1–3), and
one from the highlands of the Cumberland Plateau and the
Southern Appalachians (shown in orange and yellow,
Figs 1–3). Our divergence time estimates indicate that such
inferences would be misleading, given that those two
primary clades appear to have diverged at least 8 million
years ago. Regardless of the potential challenges associated
with intraspecific divergence time estimation, the importance
of such attempts is obvious, as has been seen in other tree
species (Dick et al. 2003; Magri et al. 2007).
Importance of recognizing pseudocongruence
Pleistocene glaciation is often presumed to be the primary
factor resulting in observed phylogeographical breaks, but
studies rarely include data that provide support for temporal
associations. It is obvious from comparing molecular
topologies of codistributed organisms that while major clades
may be congruent, there tends to be great heterogeneity
among mutation rates of these codistributed species (reviewed
in Avise 2000). Such heterogeneity does not preclude the
assumption of a shared vicariant event, but it does suggest
the potential for other possibilities (Avise 2000). In particular,
relatively frequent climatic oscillations during the course
of the last three million years likely resulted in numerous
cycles of species range expansions and contractions,
overlaying multiple evolutionary signals on the landscape
of the genome. To better evaluate the extent to which this
occurred, it will be necessary to approach phylogeographical
studies much the same way phylogeneticists are approaching
the reconstruction of relationships among species and
genera, which is with the integration of fossil data (e.g.
Manos et al. 2007). While this will not be a viable option for
many population-level studies due to lack of appropriate
fossils, a ‘hybrid phylogeographical’ approach aimed at
both inter- and intraspecific resolution (e.g. Liston et al. 2007)
will allow for the integration of a temporal element in groups
with good fossil histories. Additionally, such an approach
would better allow for the inclusion of fossil DNA, providing
potentially more accurate divergence time estimates as the
relative position of the fossil in the topology becomes more
certain. Another approach that is gaining attention for testing
phylogeographical hypotheses is that of niche modelling of
paleodistributions (Hugall et al. 2002; Carstens & Richards
2007). Because these models are entirely dependent on modern
collection records and historical climate reconstructions,
they may provide an additional line of evidence towards
the resolution of biogeographical histories.
Acknowledgements
This work was funded by the Canon National Parks Science Scholars
Program through a fellowship awarded to AB Morris. Additional
support was provided by the Botanical Society of America Karling
Award, Society of Systematic Biologists Graduate Student Award,
and Deep Time NSF RCN DEB-0090283. All analyses were completed
on the University of Florida Phyloinformatics Cluster for High
Performance Computing in the Life Sciences, funded by NSF grant

PHYLOGEOGRAPHY OF LIQUIDAMBAR STYRACIFLUA 3899
© 2008 The Authors
Journal compilation © 2008 Blackwell Publishing Ltd
DEB-0083659/0196412. The authors thank the many individuals
who aided in the collection of leaf material, particularly S. Church,
T. Cross, J. Evans, M. Fishbein, E. Hardcastle, B. McMillan, P.
Manos, B. Mattingly, M. Mendez, J. Nelson, S. Rosso, L. Thien,
K. Tobiason, and R. Zipp. Thanks to A. Bigger and Z. Damji for
laboratory assistance, and J. Clayton, A. Drummond, S. Stutsman,
and M. Gitzendanner for discussions regarding analyses. Finally,
thanks to R. Petit and three anonymous reviewers for comments on
previous versions of this manuscript.
References
Arbogast BS (2007) A brief history of the new world flying squir-
rels: phylogeny, biogeography, and conservation genetics.
Journal of Mammalogy, 88, 840–849.
Austin JD, Lougheed SC, Neidrauer L, Chek AA, Boag PT (2002)
Cryptic lineages in a small frog: the post-glacial history of the
spring peeper, Pseudacris crucifer (Anura: Hylidae). Molecular
Phylogenetics and Evolution, 25, 316–329.
Avise JC (2000) Phylogeography: The History and Formation of Species.
Harvard University Press, Cambridge, Massachusetts.
Axelrod D (1975) Evolution and biogeography of Madrean-
Tethyan sclerophyll vegetation. Annals of the Missouri Botanical
Garden, 62, 280–334.
Bermingham E, Avise JC (1986) Molecular zoogeography of
fresh-water fishes in the southeastern United States. Genetics,
113, 939–965.
Berry EW (1923) Miocene fossil plants from southeastern Mexico.
Proceedings of the US National Museum, 62, 1–27.
Braun EL (1950) Deciduous Forests of Eastern North America.
Blakiston, Philadelphia.
Carlton CE (1990) Biogeographic affinities of Pselaphid beetles of
the Eastern United States. Florida Entomologist, 73, 570–579.
Carstens BC, Richards CL (2007) Integrating coalescent and
ecological niche modeling in comparative phylogeography.
Evolution, 61, 1439–1454.
Church SA, Kraus JM, Mitchell JC, Church DR, Taylor DR (2003)
Evidence for multiple pleistocene refugia in the postglacial
expansion of the eastern tiger salamander, Ambystoma tigrinum
tigrinum. Evolution, 57, 372–383.
Clement M, Posada D, Crandall KA (2000) tcs: a computer pro-
gram to estimate gene genealogies. Molecular Ecology, 9, 1657–
1659.
Cunningham C, Collins T (1994) Developing model systems for
molecular biogeography: vicariance and interchange in marine
invertebrates. In: Molecular Ecology and Evolution: Approaches and
Applications (eds Schierwater B, Streit B, Wagner P, DeSalle R),
pp. 405–433. Birkhauser Verlag, Basel, Switzerland.
Deevey ES (1949) Biogeography of the Pleistocene. Bulletin of the
Geological Society of America, 60, 1315–1416.
Dick CW, Abdul-Salim K, Bermingham E (2003) Molecular sys-
tematic analysis reveals cryptic tertiary diversification of a
widespread tropical rain forest tree. American Naturalist, 162,
691–703.
Donoghue MJ, Bell CD, Li J (2001) Phylogenetic patterns in north-
ern hemisphere plant geography. International Journal of Plant
Sciences, 162, S41–S52.
Donoghue MJ, Moore BR (2003) Toward an integrative historical
biogeography. Integrative and Comparative Biology, 43, 261–270.
Doyle JJ, Doyle JL (1987) A rapid DNA isolation procedure from
small quantities of fresh leaf tissues. Phytochemical Bulletin, 19,
11–15.
Drummond A, Rambaut A (2007) beast: Bayesian evolutionary
analysis by sampling trees. BMC Evolutionary Biology, 7, 214.
Estill JC, Cruzan MB (2001) Phytogeography of rare plant species
endemic to the southeastern United States. Castanea, 66, 3–23.
Felsenstein J (1988) Phylogenies from molecular sequences: infer-
ence and reliability. Annual Review of Genetics, 22, 521–565.
Graham A (1973) History of the arborescent temperate element in
the northern Latin American Biota. In: Vegetation and Vegetational
History of Northern Latin America (ed. Graham A), pp. 301–314.
Elsevier Scientific Publishing Co., New York.
Graham A (1999) The tertiary history of the northern temperate
element in the northern Latin American biota. American Journal
of Botany, 86, 32–38.
Ho SYW, Phillips MJ, Cooper A, Drummond AJ (2005) Time
dependency of molecular rate estimates and systematic overesti-
mation of recent divergence times. Molecular Biology and Evolution,
22, 1561–1568.
Ho SYW, Shapiro B, Phillips MJ, Cooper A, Drummond AJ (2007)
Evidence for time dependency of molecular rate estimates.
Systematic Biology, 56, 515–522.
Ho SYW, Saarma U, Barnett R, Haile J, Shapiro B (2008) The effect
of inappropriate calibration: three case studies in molecular
ecology. PLoS ONE, 3, e1615.
Hodges SA, Arnold ML (1994) Columbines: a geographically
widespread species flock. Proceedings of the National Academy of
Sciences, USA, 91, 5129–5132.
Hoey MT, Parks CR (1994) Genetic divergence in Liquidambar
styraciflua, L. formosana, and L. acalycina (Hamamelidaceae).
Systematic Botany, 19, 308–316.
Howes BJ, Lindsay B, Lougheed SC (2006) Range-wide phylog-
eography of a temperate lizard, the five-lined skink (Eumeces
fasciatus). Molecular Phylogenetics and Evolution, 40, 183–194.
Hugall A, Moritz C, Moussalli A, Stanisic J (2002) Reconciling
paleodistribution models and comparative phylogeography in
the wet tropics rainforest land snail Gnarosophia bellendenkerensis
(Brazier 1875). Proceedings of the National Academy of Sciences,
USA, 99, 6112–6117.
Ickert-Bond SM, Wen J (2006) Phylogeny and biogeography of
Altingiaceae: evidence from combined analysis of five non-
coding chloroplast regions. Molecular Phylogenetics and Evolution,
39, 512–528.
Ickert-Bond SM, Pigg KB, Wen J (2005) Comparative infructescence
morphology in Liquidambar (Altingiaceae) and its evolutionary
significance. American Journal of Botany, 92, 1234–1255.
Ickert-Bond SM, Pigg KB, Wen J (2007) Comparative infructescence
morphology in Altingia (Altingiaceae) and discordance between
morphological and molecular phylogenies. American Journal of
Botany, 94, 1094–1115.
Jennings BW, Edwards SV (2005) Speciational history of Australian
grass finches (Poephila) inferred from thirty gene trees. Evolution,
59, 2033–2047.
Kelchner S (2000) The evolution of non-coding chloroplast DNA
and its application in plant systematics. Annals of the Missouri
Botanical Garden, 87, 482–498.
Klicka J, Zink RM (1997) The importance of recent ice ages in
speciation: a failed paradigm. Science, 277, 1666–1669.
Kormanik PP (1990) Liquidambar styraciflua L. In: Silvics of North
America: 2. Hardwoods (eds Burns RM, Honkala BH). US Depart-
ment of Agriculture, Washington, DC.
Li JH, Donoghue MJ (1999) More molecular evidence for inter-
specific relationships in Liquidambar (Hamamelidaceae). Rhodora,
101, 87–91.

3900 A. B. MORRIS ET AL.
© 2008 The Authors
Journal compilation © 2008 Blackwell Publishing Ltd
Li JH, Bogle AL, Klein AS (1997) Interspecific relationships and
genetic divergence of the disjunct genus Liquidambar (Hamamel-
idaceae) inferred from DNA sequences of plastid gene matK.
Rhodora, 99, 229–240.
Liston A, Parker-Defeniks M, Syring JV, Willyard A, Cronn R
(2007) Interspecific phylogenetic analysis enhances intraspecific
phylogeographic inference: a case study in Pinus lambertiana.
Molecular Ecology, 16, 3926–3938.
Little E (1971) Atlas of United States Trees, Vol 1: Conifers and Impor-
tant Hardwoods. US Department of Agriculture, Forest Service,
Was hing ton, D C.
Magri D, Fineschi S, Bellarosa R et al. (2007) The distribution of
Quercus suber chloroplast haplotypes matches the palaeogeo-
graphical history of the western Mediterranean. Molecular
Ecology, 16, 5259–5266.
Martin PS, Harrell BE (1957) The Pleistocene history of temperate
biotas in Mexico and eastern United States. Ecology, 38, 468–
480.
Manos PS, Soltis PS, Soltis DE et al. (2007) Phylogeny of extant and
fossil Juglandaceae inferred from the integration of molecular
and morphological data sets. Systematic Biology, 56, 412–430.
Near TJ, Keck BP (2005) Dispersal, vicariance, and timing of
diversification in Nothonotus darters. Molecular Ecology, 14,
3485–3496.
Nuttle T, Haefner JW (2005) Seed dispersal in heterogeneous
environments: bridging the gap between mechanistic dispersal
and forest dynamics models. American Naturalist, 165, 336–349.
Pigg KB, Ickert-Bond SM, Wen J (2004) Anatomically preserved
Liquidambar (Altingiaceae) from the Middle Miocene of Yakima
Canyon, Washington State, USA and its biogeographic implica-
tions. American Journal of Botany, 91, 499–509.
Posada D, Buckley TR (2004) Model selection and model averaging
in phylogenetics: advantages of akaike information criterion and
Bayesian approaches over likelihood ratio tests. Systematic
Biology, 53, 793–808.
Posada D, Crandall KA (1998) modeltest: testing the model of
DNA substitution. Bioinformatics, 14, 817–818.
Sang T, Crawford DJ, Stuessy TF (1997) Chloroplast DNA phylogeny,
reticulate evolution, and biogeography of Paeonia (Paeoniaceae).
American Journal of Botany, 84, 1120–1136.
Shi S, Huang Y, Zhong Y et al. (2001) Phylogeny of the Altingiaceae
based on cpDNA matK, PY-IGS and nrDNA ITS sequences.
Plant Systematics and Evolution, 230, 13–24.
Simmons MP, Ochoterena H (2000) Gaps as characters in sequence-
based phylogenetic analyses. Systematic Biology, 49, 369–381.
Soltis DE, Morris AB, McLachlan JS, Manos PS, Soltis PS (2006)
Comparative phylogeography of unglaciated eastern North
America. Molecular Ecology, 15, 4261–4293.
Sorrie BA, Weakley AS (2001) Coastal Plain vascular plant endemics:
phytogeographic patterns. Castanea, 66, 50–82.
Sota T, Hayashi M (2007) Comparative historical biogeography of
Plateumaris leaf beetles (Coleoptera: Chrysomelidae) in Japan:
interplay between fossil and molecular data. Journal of Bio-
geography, 34, 977–993.
Swofford DL (2002) PAUP*. Phylogenetic Analysis Using Parsimony
(* and Other Methods). Sinauer & Associates, Sunderland,
Massachusetts.
Taberlet P, Gielly L, Pautou G, Bouvet J (1991) Universal primers
for amplification of three non-coding regions of chloroplast
DNA. Plant Molecular Biology, 17, 1105–1109.
Tate JA, Simpson BB (2003) Paraphyly of Ta r a s a (Malvaceae) and
diverse origins of the polyploid species. Systematic Botany, 28,
723–737.
Thompson J, Gibson T, Plewniak F, Jeanmougin F, Higgins D
(1997) The clustal_x windows interface: flexible strategies for
multiple sequence alignments aided by quality analysis tools.
Nucleic Acids Research, 24, 4876–4882.
Whitehead DR (1983) Pollen analysis of the peat member from the
Lee Creek mine. Smithson Contributions to Paleobiology, 53, 265–285.
Williams JW, Shuman BN, Webb T, Bartlein PJ, Leduc PL (2004)
Late-quaternary vegetation dynamics in North America: scaling
from taxa to biomes. Ecological Monographs, 74, 309–334.
Xiang Q-Y, Soltis DE (2001) Dispersal-vicariance analyses of inter-
continental disjunctions: historical biogeographical implications
for angiosperms in the northern hemisphere. International Journal
of Plant Science, 162, S29–S39.
Zamudio KR, Savage WK (2003) Historical isolation, range
expansion, and secondary contact of two highly divergent
mitochondrial lineages in spotted salamanders (Ambystoma
maculatum). Evolution, 57, 1631–1652.
Zhou ZK, Crepet WL, Nixon KC (2001) The earliest fossil evidence
of the Hamamelidaceae: late Cretaceous (Turonian) inflores-
cences and fruits of Altingioideae. American Journal of Botany, 88,
753–766.
Ashley Morris is interested in phylogeography of disjunct
populations. This work was part of her PhD at the University of
Florida. Steffi Ickert-Bond is an expert on biography and character
evolution of Altingiaceae. Burke Brunson is a Master’s student
in the Morris lab. Pam and Doug Soltis are interested in
phylogeography, systematics, and polyploidy in angiosperms.
Supplementary material
The following supplementary material is available for this article:
Appendix S1 Haplotype distribution and GenBank Accession
nos for Liquidambar styraciflua. Haplotype codes correspond to
those in Figs 2–4 and Table 1. Sample ID, GenBank isolate name;
localities, all sampling localities in which a given haplotype was
recovered
Appendix S2 Polymorphic sites among cpDNA haplotypes
recovered within Liquidambar styraciflua. Numbers indicate base
position; ‘.’, same as previous sample; and ‘–’, an indel
This material is available as part of the online article from:
http://www.blackwell-synergy.com/doi/abs/
10.1111/j.1365-294X.2008.03875.x
(This link will take you to the article abstract).
Please note: Blackwell Publishing are not responsible for the con-
tent or functionality of any supplementary materials supplied by
the authors. Any queries (other than missing material) should be
directed to the corresponding author for the article.