Molecular Phylogenetics and Evolution xxx (2010) xxx–xxx

Contents lists available at ScienceDirect

Molecular Phylogenetics and Evolution

journal homepage: www.elsevier .com/ locate /ympev

Chloroplast phylogeny and phylogeography of Stellera chamaejasmeon the Qinghai-Tibet Plateau and in adjacent regions

Yong-Hong Zhang a, Sergei Volis b, Hang Sun a,⇑a Key Laboratory of Biodiversity and Biogeography, Kunming Institute of Botany, Chinese Academy of Sciences, Kunming, Yunnan 650204, People’s Republic of Chinab Life Sciences Department, Ben-Gurion University of the Negev, P.O. Box 653, Beer Sheva 84105, Israel

a r t i c l e i n f o a b s t r a c t

Article history:Received 1 April 2010Revised 27 August 2010Accepted 31 August 2010Available online xxxx

Keywords:Stellera chamaejasmePhylogenyPhylogeographyQinghai-Tibet PlateauRefugiumPost-glacial expansion

1055-7903/$ - see front matter � 2010 Elsevier Inc. Adoi:10.1016/j.ympev.2010.08.033

⇑ Corresponding author. Fax: +86 871 5215002.E-mail address: hsun@mail.kib.ac.cn (H. Sun).

Please cite this article in press as: Zhang, Y.-H., ein adjacent regions. Mol. Phylogenet. Evol. (201

Historic events such as the uplift of Qinghai-Tibet Plateau (Q-T Plateau) and climatic oscillations in theQuaternary period greatly affected the evolution and modern distribution of Sino-Tibetan flora. Stellerachamaejasme, a perennial herb with flower color polymorphism that is distributed from the mountainoussoutheastern Q-T Plateau (Hengduan Mountains, H-D Mountains) to the vast platform of the Q-T Plateauand the adjacent plain of northern China, provides an excellent model to explore the effects of historicevents on the origination and variation of species. In this study, we conducted a phylogenetic and phylo-geographical study using three chloroplast sequences (trnT-L, trnL-F and rpL16) in 26 populations of S. cha-maejasme and 12 outgroups from the Thymeleaceae. Phylogenetic analysis and molecular clock estimationrevealed that the monophyletic origin of S. chamaejasme occurred ca. 6.5892 Ma, which is consistent withthe radical environment changes caused by the rapid uplift of the Q-T Plateau ca. 7 Ma. Intra-specific dif-ferentiation of S. chamaejasme is estimated to have occurred after ca. 2.1 Ma. Twelve haplotypes wererevealed from combined trnL-F and rpL16 sequences. High genetic diversity (hT = 0.834) and populationdifferentiation (NST = 0.997 and GST = 0.982) imply restricted gene flow among populations and significantgeographical or environmental isolation. All populations from the vast plain of northern China were dom-inated by one haplotype (H1), and the same haplotype was fixed in most populations from the high ele-vation platform of the western and northern Q-T Plateau. In contrast, the majority of the haplotypeswere found in the relatively narrow area of the H-D Mountains, in the southeastern distribution of S. cha-maejasme. The contrasting haplotype distribution patterns suggested that the H-D Mountains were eithera refugium for S. chamaejasme during the Quaternary climatic oscillations or a diversification center of thisspecies. The present wide distribution of this species on the Q-T Plateau platform and in northern China islikely to have resulted from a rapid post-glacial population expansion from the southeastern refugiuminvolving founder effects, facilitated by the adjacent geographic range with a similar grassland habitat.

� 2010 Elsevier Inc. All rights reserved.

1. Introduction

The Qinghai-Tibet Plateau (Q-T Plateau) is the highest andlargest plateau in the world, with a mean elevation of 4500 mand an area of 2.5 � 106 km2 (Zheng, 1996). More than 12,000species from 1500 genera are found in this region, of which morethan 20% are endemic (Wang, 1993, 1994; Wu, 1987; Wu andWu, 1996). The uplift of the Q-T Plateau, which was followedby the formation of high mountains and deep valleys within theplateau (Li et al., 1995), was one of the most important geologicalevents to have occurred since the Cenozoic, and this event mod-ified the global climate and influenced monsoon intensity (Anet al., 2001; Raymo and Ruddiman, 1992; Spicer et al., 2003).The Hengduan Mountains (H-D Mountains) are located at thesoutheastern edge of the Q-T Plateau (Li, 1987, 1989; Wu,

ll rights reserved.

t al. Chloroplast phylogeny and0), doi:10.1016/j.ympev.2010.0

1988), and this area is referred to as the core area of the Himala-yan hotspot, one of the world’s biodiversity hotspots, due to itshigh level of species and generic richness (Boufford and Dijk,2000; Boufford et al., 2004; Myers et al., 2000; Raven and Axel-rod, 1978; Wilson, 1992; Ying et al., 1993). The high species rich-ness in the H-D Mountains (ca. 8000 angiosperm species) has leadto the hypothesis that this region is a distribution and diversitycenter for many alpine genera, for example Gentiana, Androsace,Pedicularis, Primula, Rhododendron and Saussurea (Sun, 2002; Sunand Li, 2003; Wu, 1988). Many taxa (genera or species) arethought to have originated recently in this region as a result ofthe severe alteration of topography and climate since the upliftof the Q-T Plateau between the Miocene and the Quaternary(Chen et al., 2005; Liu et al., 2006; Wang et al., 2005, 2007). Fur-thermore, the H-D Mountains have been hypothesized to havebeen a refugium for plants on the Q-T Plateau platform and inthe north temperate zone of eastern Asia during the QuaternaryIce Age (Wu, 1987).

phylogeography of Stellera chamaejasme on the Qinghai-Tibet Plateau and8.033

2 Y.-H. Zhang et al. / Molecular Phylogenetics and Evolution xxx (2010) xxx–xxx

The dominant vegetation types of the Q-T Plateau platform arealpine meadow and desert-steppe, with scattered and disjunct for-ests (Wu, 1980). It was suggested that the present vegetation in theregion developed during the late Pliocene with the major uplift ofthe plateau (Axelrod et al., 1996; Shi et al., 1998; Wu, 1980), whichwas followed by large-scale climatic changes (Ni, 2000; Shi et al.,1998; Thompson et al., 2000; Zhang et al., 1996). During the Qua-ternary, climatic oscillations caused repeated retreats and ad-vances of many species’ ranges, resulting in shifts in thedominant vegetation during glacial and interglacial cycles (Hewitt,1996, 1999; Tang and Shen, 1996). The distribution of many Q-TPlateau plant species is characterized by colonization due to a ra-pid expansion from the refugia located on the eastern edge area,especially the H-D Mountains (Chen et al., 2008; Meng et al.,2007; Yang et al., 2008), although some species are suspected tohave survived in ice-free areas on the plateau platform duringthe glaciations (Shi et al., 1998; Wang et al., 2009).

The natural vegetation of northern China is dominated bysteppes with a relatively homogeneous environment (Wu, 1980,1995). Numerous plant species have extended their distributionfrom Q-T Plateau to northern China (e.g., Stellera chamaejasme,Incarvillea sinensis, Ranunculus tanguticus, Hyecoum leptocarpum),and some of these species could have originated and expandednortheastwardly from the H-D Mountains(Wang, 1992a, 1992b).Reduction of genetic diversity is expected under a scenario of rapidpost-glacial expansion into homogeneous grassland, as has beenfound in northern Europe and America (Hewitt, 1996). Becausenorthern China is believed not to have been glaciated during theQuaternary (Hewitt, 2000; Liu, 1988; Zhao and Li, 1990), it is anopen question whether species with the range from southwesternChina (Q-T Plateau) to northern China experienced past northernexpansion and southern retreat during the Quaternary climaticoscillations. Northward expansion from southern refugia was dem-onstrated for Picea crassifolia and Pinus tabulaeformis, which aretree species of the fragmented forest (Chen et al., 2008; Meng

Fig. 1. Different flower color populations of S. chamaejasme. A. white-red flower morphoyellow flower morphological type (TypeWY). D. pure-red flower morphological type (Ty

Please cite this article in press as: Zhang, Y.-H., et al. Chloroplast phylogeny andin adjacent regions. Mol. Phylogenet. Evol. (2010), doi:10.1016/j.ympev.2010.0

et al., 2007). However, no similar phylogeographical study had everbeen conducted on herbaceous plants.

The availability of new phylogenetic and phylogeographicalmolecular tools, such as non-recombining and uniparentally inher-ited chloroplast DNA markers (cpDNA) (Dumolin et al., 1995; Rajo-ra and Dancik, 1992), has resulted in intensive investigation ofvegetation of the Q-T Plateau and the adjacent eastern mountainarea (Chen et al., 2005; Liu et al., 2006; Meng et al., 2007; Wanget al., 2009; Yang et al., 2008; Zhang et al., 2005). However, thesestudies did not consider the relationship of the studied vegetationwith the vast adjacent northeastern plain in northern China, whichis dominated by homogeneous grassland, despite the fact thatmany plant species are common to these two regions. In addition,most of the species studied have been trees (Meng et al., 2007;Zhang et al., 2005), while herbaceous vegetation has receivedmuch less attention (Chen et al., 2008; Yang et al., 2008). Therefore,it would be of great interest to study the phylogeography of an her-baceous species with a distribution that covers both desert steppeand grassland to understand the evolution and modern distribu-tion of the vegetation of the Q-T Plateau and adjacent regions.

We conducted a phylogenetic and phylogeographic study on aperennial herbaceous plant, S. chamaejasme L. This species has awide geographic range from the northwestern Yungui Plateauthrough the H-D Mountains northwestward to the Q-T Plateauplatform and northeastward to the plain of northern China(Grey-Wilson, 1995; Wang and Gilbert, 2007; Zhang, 1999). Thisplant exhibits flower color polymorphism (Fig. 1), which servedin the past as the basis for the delimitation of two separate species,S. chamaejasme L. and S. bodinieri H. Léveillé (Léveillé, 1912; Lin-naeus, 1753), or two forms of this species (Diels, 1912; Huang,1985). The white-red flower type (hereafter referred to asTypeWR), S. chamaejasme f. chamaejasme, with a red calyx tubeand white tube lobe flowers (Fig. 1A), has a vast distribution fromthe Q-T Plateau to northern China (Fig. 2). The pure-yellow flowertype (hereafter referred to as TypeY), S. chamaejasme f. angustifolia

logical type (TypeWR). B. pure-yellow flower morphological type (TypeY). C. white-peR).

phylogeography of Stellera chamaejasme on the Qinghai-Tibet Plateau and8.033

Fig. 2. The distribution range of different flower color morphs. Vertical line area: the distribution range of TypeWR morphs. Horizontal line area: the distribution range ofTypeWY morphs. Diagonal line area: the distribution range of TypeY morphs. Ringed star: the location of TypeR morphs.

Y.-H. Zhang et al. / Molecular Phylogenetics and Evolution xxx (2010) xxx–xxx 3

Diels (Diels, 1912), with a yellow calyx tube and lobes (Fig. 1B),grows south from the H-D mountains to the Yungui Plateau(Fig. 2) (Huang, 1985). During 4 years (from 2003 to 2006) fieldinvestigation and collections, two new flower color types were dis-covered. Of these, the white-yellow flower type (hereafter referredto as TypeWY), has yellow flower calyx tube and white tube lobes(Fig. 1C) and is mainly distributed in the southeastern H-D Moun-tains (Fig. 2). The pure-red flower type (hereafter referred to asTypeR; Fig. 1D), is only found in Yajiang County of the H-D Moun-tains (Fig 2). For any given population, flower color is a monomor-phic character. All four flower morphs can be found in the H-DMountains.

The special visual flower color polymorphism and wide geo-graphic range of S. chamaejasme make it ideally suited for examina-tion of topographic and climatic effects on plant evolution andspeciation and the glacial/post-glacial contraction and expansionpattern. Our objectives included: (1) reconstruction of the chloro-plast-based molecular phylogeny and flower color evolution in thisspecies; (2) assessment of the impact of the uplift of Q-T Plateau onthe origin and evolution history of the species; and (3) analysis ofthe population genetic consequences of the Quaternary climaticoscillations of the ice age.

2. Material and methods

2.1. Plant material

Sampling was conducted from 2003 to 2006 in 26 populationscovering most of the S. chamaejasme range (Table 1). In total, 130 S.chamaejasme individuals were collected, with sampled individualsbeing separated by at least 50 m to avoid collecting clones or closerelatives. The collected leaf material was dried using silica gel inthe field and kept under -80 �C in the lab. Voucher specimens weredeposited in the Herbarium of the Kunming Institute of Botany, Chi-

Please cite this article in press as: Zhang, Y.-H., et al. Chloroplast phylogeny andin adjacent regions. Mol. Phylogenet. Evol. (2010), doi:10.1016/j.ympev.2010.0

nese Academy of Sciences (KUN). Thirteen outgroups were selectedfrom the genus Aquilaria, Daphne, Edgeworthia, Stelleropsis, Thyme-laea and Wikstroemia of Thymelaeaceae (Table 2). Leaf samples ofthree species (Thymelaea passerina, T. villosa and T. hirsuta) were ta-ken from herbarium specimens. Two DNA samples (T. sanamundaand D. mezereum) were acquired from the DNA Bank of The Royal Bo-tanic Gardens at Kew (DNA bank No. 12002 and 6160).

2.2. DNA extraction, amplification, sequencing and alignment

Genomic DNA was extracted using the CTAB protocol of Doyleand Doyle (1987). Total DNA was additionally cleaned with aDNA cleanup kit (Bejing Sunbiotech co., Ltd, China). The primersa and b, c and f of Taberlet et al. (1991) were used to amplifynon-coding regions of the chloroplast trnT-trnL intergenic spacer(trnT-L), the trnL intron and the trnL-trnF intergenic spacer (trnL-F), respectively. The primers F71 and R1516 of Jordan et al.(1996) were used to amplify the rpL16 intron (rpL16). Polymerasechain reactions (PCR) were carried out in a total volume of 20 llcontaining 2 ll of 10 � PCR reaction buffer (Takara, Japan), 1.6 llof 25 mM MgCl2, 1 ll of each primer (Sangon, Shanghai, Chana)at 50 ng/ll, 1.6 ll of 2.5 mM dNTP solution in an equimolar ratio,0.1 ll of Taq DNA-polymerase (5 U/ll, Takara, Japan) and 2 ll ofgenomic DNA at 5 ng/ll. PCR was performed using a PTC-100 ther-mocycler (MJ Research, Waltham, Massachusetts, USA) with thefollowing program: an initial 4-min hotstart at 94 �C, 33 cycles of1 min of denaturing at 94 �C, 1 min of annealing at 55 �C, and2 min of extension at 72 �C and a final extension of 10 min at 72 �C.

The PCR products were purified with a purification kit (Prome-ga, Madison, Wisconsin, USA). The same primers that were used forthe PCR reactions were used in sequencing from both strands.Additional primers d and e of Taberlet et al. (Taberlet et al.,1991) were used when sequencing the trnL-F fragments. PurifiedPCR products were sequenced by the dideoxy chain termination

phylogeography of Stellera chamaejasme on the Qinghai-Tibet Plateau and8.033

Table 1Details of sample locations and cpDNA haplotype frequencies in 26 populations of Stellera chamaejamse. The abbreviations, IM, HB, QH, XZ, SC, YN and Alt. denote inner Mongolia,Hebei, Qinghai, Xizang, Sichuan, Yunnan and altitude, respectively.

cpDNA halotype

Population Code Latitude Longitude Alt. (m) H1 H2 H3 H4 H5 H6 H7 H8 H9 H10 H11 H12

Ergula, IM E 50�1800000 120�1405400 600 5Guyang, IM GY 40�4204700 110�1004700 1390 5Hohehot, IM HHHT 40�5500400 111�4501000 1140 5Zuolu, HB ZL 40�0604300 115�0600400 1180 5Gangcha, QH GC 37�2404600 100�0403000 3510 5Gangga, Dingri, XZ GG 28�3500200 86�3603900 4330 5Rendui, Nanmulin, XZ RD 29�5701300 89�0602600 4320 5Namucuo, XZ NMC 30�4703000 91�0303000 4740 5Changdu, XZ CD 31�2603500 97�1101600 3350 5Mangkang, XZ MKa 29�4105400 98�3501500 3900 5Mangkang, XZ MKb 29�4704700 98�5902600 3610 5Seda, SC SD 32�0804000 100�2504700 3780 5Luhuo, SC LH 31�0402900 100�5905700 3030 5Xiangcheng, SC XC 28�4500700 99�5105400 2960 5Derong, SC DR 29�0301900 99�2204500 3200 5Kangding, SC KD 30�0000600 101�5700200 2800 5Xinduqiao,Kangding,SC XDQ 30�0302800 101�2805400 3480 5Litang, SC LT 29�3900700 100�2102200 3760 5Daocheng, SC DC 29�0500200 100�1601400 3750 5Dimucaoba,Derong,SC DMCB 29�0704300 99�2005000 4020 5Yajiang, SC YJ 30�0403600 101�0801500 2870 5Zhongdian, YN ZD 27�4801600 99�4005400 3270 5Lijiang, YN LJ 26�5901300 100�1104900 2650 5Dali, YN DL 25�5900700 100�0500400 2010 5Kunming, YN KM 25�0901000 102�4300500 2230 5Qiaojia, YN QJ 26�5100900 102�4501200 2360 4 1

Table 2Information of the geographic location, coordinate and voucher of outgroups in this study. ‘‘ – ” denote the information of the location or coordinates is not collected.

Taxa Localities Coordinate Voucher

Aquilaria sinensis Zhangpu, Zhangzhou, Fujian, China 24�0603200 , 117�3600800 Zhang YH 07053Daphne mezereum Mildenhall, Suffolk, UK – M. Fay s.n.D. tangutica Ganzi, Sichuan, China 31�2905000 , 99�5903300 Boufford et al. 34156Edgeworthia chrysantha Kunming, Yunnan, China 25�0802700 , 102�4402500 Zhang YH 06120901Stelleropsis altaica Zhaoshu, Xingjiang, China 43�1003000 , 81�0605800 Zhang JW 20060804Thymelaea hirsuta Sierra de la Contraviesa, Granada, Spain – DGH95107T. passerina Perales de Tajuna, Madrid, Spain – DGH96169T. sanamunda – – Civegrel et al. 546T. villosa Los Barrios, Cádiz – DGH97204Wikstroemia delavayi Deqing, Yunnan, China 28�1503500 , 99�1602500 Zhang YH YN04-004W. dolichantha Jiulong, Sichuan, China 28�4403600 , 101�4005100 Boufford et al. 33012W. jiulongensis Jiulong, Sichuan, China 28�4403600 , 101�4005100 Boufford et al. 32996W. stenophylla Maerkang, Sichuan, China 31�4800300 , 101�3601300 Boufford et al. 34363

4 Y.-H. Zhang et al. / Molecular Phylogenetics and Evolution xxx (2010) xxx–xxx

method with an ABI PRISM BigDye Terminator Cycle SequencingReady Reaction kit (Applied Biosystem, Foster City, California,USA) with AmliTaq DNA polymerase FS (Perkin-Elmer). Reactionsand programs were chosen according to the recommendations ofthe manufacturer’s handbook (Perkin-Elmer, Applied Biosystem).Samples were electrophoresed in an ABI310 automated sequencer.

The DNA sequences were edited in Seqman (Lasergene, DNASTARInc., Madison, Wisconsin, USA). Consensus sequences were obtainedfor each individual. All sequences were submitted to GenBank. Mul-tiple sequence alignment was carried out in Clustal X 1.81 (Thomp-son et al., 1997) and refined manually. The aligned sequences andtrees are deposited in TreeBase with submission ID 10776.

2.3. Data analysis

2.3.1. Phylogeny and divergence estimatesOne sample per population of S. chamaejasme together with 13

outgroups were used. Three amplified and sequenced cpDNA frag-ments, trnT-L, trnL-F and rpL16 were combined before analysis. Phy-logenetic analyses were conducted in PAUP*4.0b10 (Swofford, 2000)using maximum parsimony (MP) (Fitch, 1971), with equal weighting

Please cite this article in press as: Zhang, Y.-H., et al. Chloroplast phylogeny andin adjacent regions. Mol. Phylogenet. Evol. (2010), doi:10.1016/j.ympev.2010.0

of all characters and with gaps treated as missing data. Heuristicsearches were replicated 1000 times with random taxon addition se-quences, tree bisection-reconnection (TBR) branch swapping, andwith the options Multrees and Steepest Descent in effect. Supportfor monophyletic groups was assessed by bootstrapping (resam-pling of data 100 times) using the heuristic search strategy, as indi-cated above. In addition, phylogenetic reconstructions were alsoperformed using maximum likelihood (ML) and Bayesian inference(Huelsenbeck and Ronquist, 2001). The appropriate model of DNAsubstitution for the maximum likelihood analysis was determinedusing the AIC as the selection criterion in Modeltest, version 3.6 (Po-sada and Buckley, 2004). Maximum likelihood searches were per-formed using a heuristic search method, each with 10 randomsequence additions. Bayesian inference was conducted using MrBa-yes, version 3.0 (Huelsenbeck and Ronquist, 2001). The MCMC algo-rithm was run for 1000,000 generations with four incrementallyheated chains, starting from random trees and sampling one of every100 generations. The first 2000 trees were discarded as burn in. Theremaining trees were used to construct a 50% majority rule tree.Internodes with posterior probabilities >95% were considered statis-tically significant.

phylogeography of Stellera chamaejasme on the Qinghai-Tibet Plateau and8.033

Y.-H. Zhang et al. / Molecular Phylogenetics and Evolution xxx (2010) xxx–xxx 5

BEAST v1.4.8 (Drummond and Rambaut, 2007) was used to esti-mate the divergence times of S. chamaejasme. The BEAUti interfacewas used to create an input file for BEAST, in which a general timereversible (GTR) nucleotide substitution model withgamma + invariant sites was applied; an uncorrelated lognormalmodel was used to describe the relaxed clock; and the birth–deathprocess was set for tree prior. Two fossil records were chosen asour primary calibration points. The root of the tree, which is the splitbetween Aquilaria and other species, was constrained between 33.9–23.03 million years ago (Ma) based on the known Aquilaria fossil re-cord (Mai and Walther, 1978). This time approaches the origin ofAquilaria (32–26 Ma) that has been estimated from molecular data(Wikström et al., 2001). The second calibration point used was thesplit between Thymelaea and Daphne, which was set between16.4–5.3 Ma based on the known fossils of Daphne (Palmarev,1989). The topology of Thymelaea and Daphne enforced the mono-phyly of the fossil placement. Ten million generations of the MCMCchains were run, sampling every 1000 generations. After discardingthe first 1000 trees as burn in, the samples were summarized in themaximum clade credibility tree using TreeAnnotator v1.4.8 (Drum-mond and Rambaut, 2007) with the posterior probability limit set to0.5 and summarizing mean node heights. The results were visualizedusing Figtree 1.0 (Rambaut, 2007).

2.3.2. Flower color character tracingWe traced the evolution of flower color, which has been re-

garded as crucial taxonomic character of S. chamaejasme (Diels,1912; Huang, 1985; Léveillé, 1912). The four color morphs (WR,R, WY and Y) were taken as four character states. Character tracingwas conducted onto the tree generated from the combined analysisof the cpDNA dataset using Mesquite software (Maddison andMaddison, 2009), and the optimality criterion was parsimony.

2.3.3. Genetic variation and geographical differentiation analysisThe trnL-F and rpL16 fragments were amplified and sequenced in

five individuals per S. chamaejasme population. All indels were trea-ted as single point mutations and evenly weighted with other muta-tions, except for those in polynucleotides. The cpDNA haplotypeswere determined from the concatenated alignment sequences of130 individuals of S. chamejasme. Diversity and differentiationparameters (within population diversity, hS; total diversity, hT; dif-ferentiation for unordered and ordered alleles, GST and NST, respec-tively) were calculated according to Pons and Petit (1996) usingPERMUT (http://www.pierroton.inra.fr/genetics/labo/Software/).This program was also used for a test of whether NST was larger thanGST, which would be indicative of phylogeographic structure (a situ-ation when closely related haplotypes are more often found in thesame area than less closely related haplotypes) (Pons and Petit,1995, 1996). The significance of the difference between NST and GST

was assessed with 1000 permutations. Analysis of molecular vari-ance (AMOVA) was performed using the Arlequin ver. 3.0 program(Excoffier et al., 2005) to partition the genetic variation into differentlevels, with the statistical significance determined by 1000 permuta-tions. Intra-specific relationships among the cpDNA haplotypeswere constructed using the program TCS 1.06 (Clement et al., 2000).

3. Results

3.1. Phylogenetic analysis and divergence time estimation

All sequences generated from this study were deposited in NCBIGenBank (accession numbers GU736293–GU736331 for rpL16,GU736332–GU736370 for trnL-F and GU736371–GU736409 fortrnT-L). The concatenated alignment data including the trnT-L,trnL-F and rpL16 sequences resulted in a length of 3229 positions,

Please cite this article in press as: Zhang, Y.-H., et al. Chloroplast phylogeny andin adjacent regions. Mol. Phylogenet. Evol. (2010), doi:10.1016/j.ympev.2010.0

among which 2431 were constant, 484 uninformative and 314informative. For the trnT-L sequence, one gap with a length of150–153 bp was detected in most individuals of S. chamaejasme,except for MKa1 and MKb1, and in all of the outgroups. With allgaps treated as missing data, a heuristic search of the combined se-quence dataset yielded only a single most parsimonious (MP) tree(CI = 0.8912, RI = 0.8991, RC = 0.8013), which is shown in Fig. 3. Allindividuals of S. chamaejasme formed a monophyletic group (cladeA, with 100% bootstrap value) split into two main clades (clade Band clade C), each with 100% bootstrap support (Fig. 3).

Clade B (Fig. 3) is represented by the very closely related haplo-types Mka and Mkb, which are possessed by all individuals fromMangkang county, SE Tibet (southwestern H-D Mountains), all ofwhich have TypeWR flowers. Clade C (Fig. 3) has a geographicallystructured tree topology consistent with flower color variation.Its sub-clade D comprises TypeWY individuals from the KD, LT,DC, DMCB, XDQ populations, grouped together with 74% bootstrapsupport (clade E), and TypeR individuals from the YJ population inthe base of the clade. All populations of clade D are distributedfrom the central to eastern H-D Mountains. The YJ population loca-tion was between the KD and LT population locations. Clade F, with86% bootstrap support, consists of Clade G and Clade H. In Clade G,which has 66% bootstrap support, all individuals are of TypeWR,and the populations cover a wide area from the central H-D Moun-tains west to the Q-T Plateau platform and northeast to northernChina. Though populations from both clades can be found in thecentral H-D Mountains and exhibit mosaic distribution, they arefound at different altitudes in different habitats. TypeWY popula-tions are predominantly located in the alpine meadow at a highaltitude, while the TypeWR populations are mostly found at thefoot of the mountain. For example, populations DR (TypeWR) andDMCB (TypeWY) are separated by several kilometers but differ inthe altitude, being found at 3200 and 4020 m, respectively.

Clade H comprises TypeY individuals of the KM, LJ, DL, QJ andZD populations from the southern H-D Mountains and adjacentnorthern Yungui Plateau (88% bootstrap support). Geographically,these populations extend southward through forestland, with a de-crease in altitude from 3270 to 2010 m. The ZD population of cladeH and the XC population of clade G, which are separated by severalkilometers, are distributed along the Big Snow Mountain, whichlies at the border of Sichuan and Yunnan in the southern H-Dmountains, and exhibited pure-yellow flower color on the south-ern side and white-red flower color on the northern side of theBig Snow Mountain.

Maximum likelihood and Bayesian analyses produced a similartopology; the Bayesian posterior values are marked on the phylo-genetic tree (Fig. 3). The divergence times for the major clades ofS. chamaejasme were estimated according to the two calibrationpoints from the fossil record (Fig. 4). The estimated divergencetime between S. chamaejasme and Wikstroemia was 6.5892 Ma,and the basal divergence time of S. chamaejasme was estimatedat 2.1053 Ma (both with a posterior probability = 1).

3.2. Flower color character tracing

The optimization of flower color onto the shortest tree for S.chamaejasme from the combined cpDNA analysis recovered oneoptimal reconstruction of three steps (Fig. 5). Our reconstructionindicated that the white-red flower color (TypeWR) is ancestralin this species, and the pure-yellow (TypeY), white-yellow (Typ-eWY) and pure-red (TypeR) flower colors are derived. The distribu-tion of color morphs (Fig. 2) shows that TypeWR underwentnorthern range expansion, while TypeY and TypeWY expanded tothe south and east, respectively..

phylogeography of Stellera chamaejasme on the Qinghai-Tibet Plateau and8.033

Fig. 3. Strict consensus tree of based on combined cpDNA sequences. Number above and below branches are bootstrap percentages and Bayesian posterior value,respectively. Rectangles with different color denote outgroups and different flower color individuals. Clade A–H denote intra-specific clades of S. chamaejasme discussed indetail in text.

6 Y.-H. Zhang et al. / Molecular Phylogenetics and Evolution xxx (2010) xxx–xxx

3.3. Population and phylogeographic analysis

The analyzed lengths of the trnL-F and rpL16 sequences were928–938 bp and 1006–1017 bp, respectively. Three poly-T se-quences (284–298,680–691,862–874) were found in the trnL-F se-quences, and one poly-T sequence (588–600) was found in rpL16,and these were omitted from the analysis. Based on concatenatedsequence data from the trnL-F and rpL16 sequences, a total of 31polymorphic sites and 12 haplotypes were detected (Table 3).The haplotype frequencies in the 26 investigated populations ofS. chamaejasme are listed in Table 1. The sequences of these haplo-types were deposited in NCBI GenBank (accession numbersGU736346 - GU736357 for trnL-F and GU736307 - GU736318 forrpL16).

Please cite this article in press as: Zhang, Y.-H., et al. Chloroplast phylogeny andin adjacent regions. Mol. Phylogenet. Evol. (2010), doi:10.1016/j.ympev.2010.0

Genetic diversity analysis of S. chamaejasme revealed high ge-netic diversity (hT = 0.834 ± 0.0610, and vT = 0.573 ± 0.1582) and ahigh degree of differentiation between populations(GST = 0.982 ± 0.0182 and NST = 0.997 ± 0.0033). When GST wascompared with the estimate of genetic differentiation that takesinto account haplotype similarities (NST), the difference betweenthe two estimates was not significant (P = 0.069). Similar resultswere produced by AMOVA, with the among population componentof variation being 99.8%.

The haplotype distribution is shown in Fig. 6A, and the phyloge-netic relationships among haplotypes are presented in Fig. 6B.Haplotype H1 took a central position in the network with the high-est proportion of individuals (19.2%). Haplotype H4 was private inpopulation Mka and joined haplotype H1 directly, with 15 muta-

phylogeography of Stellera chamaejasme on the Qinghai-Tibet Plateau and8.033

Fig. 4. The divergence times estimated by Beast software. The topology was obtained with Figtree software. The profile of time estimation results including all outgroups isshown at the top left corner. Arrows denote nodes where fossil calibration was applied. Shadow area is amplified and shown. Ringed characters denote the consistent androbust nodes in molecular time estimation: a to k represent the estimated time on origin and intra-specific divergence of S.chamaejasme with value 6.5892 Ma (a), 2.1053 Ma(b), 0.1132 Ma (c), 1.1491 Ma (d), 0.5957 Ma (e), 0.1475 Ma (f), 0.0643 Ma (g), 0.6086 Ma (h), 0.502 Ma (i), 0.5058 (j), 0.1181 Ma (k), respecitively.

Y.-H. Zhang et al. / Molecular Phylogenetics and Evolution xxx (2010) xxx–xxx 7

tions including 8 nucleotide substitutions and 7 indels (Fig. 6B). Se-ven haplotypes (H3, H4, H5, H6, H8, H10 and H12) were private.The geographic distribution of all 12 haplotypes is very asymmet-rical (Fig. 6A). Haplotype H1 was found to be the predominant hap-lotype distributed from the northwestern H-D Mountains outwardto the Q-T Plateau platform and northern China. In contrast, 16populations from the H-D Mountains occupying a small part ofthe species range harbored 9 haplotypes (75% of the total).

4. Discussion

4.1. Origin, intra-specific variation and taxonomy

Phylogenetic analysis of S. chamaejasme based on cpDNA se-quences revealed its monophyletic origin at approximately6.5892 Ma during the late Miocene. This estimation is consistentwith the radical environmental change that occurred in the Tibetanregion caused by the rapid uplift of the Q-T Plateau ca. 7 Ma (Anet al., 2001; Burbank et al., 1993; Harrison et al., 1992; Molnar,2005; Qiang et al., 2001; Wang et al., 2006). It has been suggestedthat the uplift of the Qinghai-Tibet plateau affected monsoonintensity (An et al., 2001; Burbank et al., 1993; Qiang et al.,2001) and modified the global climate (An et al., 2001; Kutzbachet al., 1989; Molnar et al., 1993; Raymo and Ruddiman, 1992). Dur-ing that period, the vegetation of the Q-T Plateau experienced atransition from forest to grassland due to aridization (Li, 1998;Quade et al., 1995; Wang et al., 2006). Thus, the origin of the mono-typic genus Stellera and its separation from Wikstroemia appear tohave been triggered by the uplift and to have resulted from adap-tation to the grassland habitat.

The majority of S. chamaejasme morphological and genetic var-iation is concentrated in the H-D Mountains. This variation pattern

Please cite this article in press as: Zhang, Y.-H., et al. Chloroplast phylogeny andin adjacent regions. Mol. Phylogenet. Evol. (2010), doi:10.1016/j.ympev.2010.0

is typical for many species having a similar distribution; it has beensuggested to be caused by geographic and ecological barriers with-in this region (Ge et al., 2005; Xia et al., 2005). Molecular datingtests revealed that intra-specific differentiation occurred in S. cha-maejasme after ca. 2.1053 Ma (Fig. 4), which is consistent with geo-graphical and ecological alterations that followed both the uplift ofthe Q-T Plateau and the glacial advance/retreat during the Quater-nary ice age.

Geographical and, probably, ecological isolation is likely to beresponsible for the population differentiation of S. chamaejasme.Populations of clade B (Fig. 3) are geographically isolated frompopulations of the other clades by the eastern Jinsha River andthe western Lancang River, which explains why this clade divergedearlier from other clades. Similarly, populations of clades G and H(Fig. 3) are geographically separated by the Big Snow Mountain(the border of SW Sichuan and NW Yunnan). This difference couldbe a result of genetic drift due to the reproductive isolation of twopopulation groups by the mountain.

Vegetation on the Q-T Plateau is considered to be highly sensi-tive and vulnerable to global climate change (Zheng, 1996). Be-cause plants have the potential to respond to climate changethrough either plastic adaptation or migration (Jump and Penuelas,2005; Körner, 1999; Pauli et al., 2007), the impacts of past climaticchanges on the evolutionary history of plant taxa are of particularinterest. Quaternary climatic oscillations have been hypothesizedto be the cause of the adaptive population differentiation of somespecies on the Q-T Plateau, especially in the H-D Mountains (Geet al., 2005; Wang et al., 2009; Xia et al., 2005; Yang et al., 2008).In this study, populations from two distinct flower color clades(clade G with TypeWR and clade E with TypeWY, Fig. 3) exhibitedhabitat-specific geographical distributions in the H-D Mountains.All of the TypeWR populations from clade G grew below the treeline at the mountain foot, while the TypeWY populations of clade

phylogeography of Stellera chamaejasme on the Qinghai-Tibet Plateau and8.033

Fig. 5. The single shortest tree obtained for S. chamaejasme from the combinedcpDNA analysis, onto which four flower color states have been mapped.

8 Y.-H. Zhang et al. / Molecular Phylogenetics and Evolution xxx (2010) xxx–xxx

E mostly occupied the high altitudinal alpine meadow on themountaintop. For example, the DR (TypeWR, 3200 m) populationwas located at the base of a mountain, while the DMCB (TypeWY,4020 m) population was located very close to DR, but at a higheraltitude, above the tree line separating the two populations. A pos-sible explanation is that a glacier caused differential plant adapta-tion to alpine meadows on mountain tops or to grasslands belowthe tree line during the Quaternary glacial period. Then, the exis-

Table 3Distribution of informative polymorphic sites in the combined chloroplast DNA sequences r– ”. #1, #2, #3, #4, #5, #6 and #7 represent AAAC, GAGAGAAA, TA, TA, TA, TG and AAA, r

Nucleotide position rpL16

191 227 292 3 353 355 4 5 6 6 719 7– – – 2 – – 5 7 1 5 – 1194 234 293 9 354 356 6 8 3 7 720 1

H1 #1 #2 #3 C #4 – C G T C #6 TH2 #1 #2 #3 C #4 – C G T C #6 TH3 #1 #2 #3 C – – C G T C #6 TH4 – #2 – C #4 – A G T T – CH5 – #2 – C #4 #5 A G T T – CH6 #1 #2 #3 C #4 – C G T C #6 TH7 #1 #2 #3 C #4 – C G T C #6 TH8 #1 #2 #3 C #4 – C G A C #6 TH9 #1 #2 #3 C #4 – C G T C #6 TH10 #1 #2 #3 G #4 – C G T C #6 TH11 #1 – #3 C #4 – C G T C #6 TH12 #1 – #3 C #4 – C A T C #6 T

Please cite this article in press as: Zhang, Y.-H., et al. Chloroplast phylogeny andin adjacent regions. Mol. Phylogenet. Evol. (2010), doi:10.1016/j.ympev.2010.0

tent intervening tree line served as a barrier preventing gene flowamong populations with different adaptations, thus maintainingtheir genetic and morphological distinctness. This hypothesismay be also used to interpret the distribution of the TypeR popu-lation (YJ, 2870 m), which is located in the lower valley betweentwo alpine TypeWY populations (XDQ at 3480 m and LT at3760 m) at higher altitudes.

An alternative explanation that cannot be neglected for the iso-lation between populations from clade E and clade G within thesame region is that populations below the tree line are relicts thatsurvived glacial advances. Populations above the tree line wouldhave then colonized or re-colonized from the relict populations be-low the tree line during the interglacial period. Thus, the repeatedadvance and retreat of the ice sheet could promote different adap-tations in populations above or below the tree line, finally causingisolation among them.

Because the monophyly for TypeY, TypeWY and TypeR withinclade C was supported by the phylogenetic tree (Fig. 3) and parsi-mony network (Fig. 6B) that we produced, and because popula-tions from each clade had separate distributions, one may arguethat the different flower color morphs of S. chamaejasme are differ-ent species or sub-species. In fact, several former studies suggestedtreating S. chamaejasme as two species (Léveillé, 1912) or twoforms (Diels, 1912) based on its flower color. However, the fact thatTypeWY and TypeR individuals were clustered into one clade (D)and that TypeWR individuals were scattered into two differentclades (C and G) with distinct positions in the phylogenetic tree(Fig. 3) opposes this separation. In addition, character tracing basedon flower color, which was the critical taxonomic trait investigatedin this study, revealed that white-red flower color is the ancestralcharacter, and the other flower colors are derived. Thus, we suggestfollowing the taxonomic treatment of Flora of China (Wang andGilbert, 2007) and consider all flower color morphs as one species.

4.2. Extent and structure of genetic diversity

Species genetic diversity and population structure can be af-fected by life history traits (e.g., life cycle, seed dispersal, breedingsystem, pollination mechanism) and environmental effects (e.g.,geographical range, climate, topography) (Hamrick et al., 1979;Loveless and Hamrick, 1984; Nybom, 2004). S. chamaejasme plantspossess a sexual reproductive mode, a long life cycle as perennialherbs, and seed dispersal through gravity, all of which are traitsassociated with low total genetic diversity (Hamrick and Godt,1989, 1996; Hamrick et al., 1979). However, high genetic diversity(hT = 0.834) was observed in our study, and this diversity appears toresult from a wide species distribution and high among-population

pL16 and trnL-F in the 12 haplotype. Gap was treat as one mutation and denote with ‘‘espectively.

trnL-F

8 8 8 9 9 119 1 1 1 1 2 3 3 3 4 4 7 80 5 7 0 0 6 – 2 2 2 3 6 2 6 9 1 5 9 96 3 7 5 7 1 121 2 7 8 7 2 5 7 0 4 1 4 6

A A A T A T #7 G T A A T – T C A C C CA A A T A T #7 G T A A T – C C A C C CA A A T A T #7 G T A A T – T C A C C CG G A – A T – – T A A T A T C T T C TG G A – A T – – T A A T A T C T T C TA A C T A A #7 G T C – C – T C A C C CA A C T A T #7 G T A – T – T C A C A CA A C T A T #7 G T C – T – T C A C C CA A A T T T #7 G T A A T – T G A C C CA A A T T T #7 G T A A T – T G A C C CA A A T T T #7 G T A A T – T G A C C CA A A T T T #7 G A A A T – T G A C C C

phylogeography of Stellera chamaejasme on the Qinghai-Tibet Plateau and8.033

Fig. 6. A(left): The geographical distribution of sampled S. chamaejasme populations and haplotype frequencies. Filled black circles indicate locations of sampled populations.Haplotype frequencies of each population are shown in the pie charts. The abbreviated population name is marked in the circle. The haplotype names accompanying thecolorful rectangle symbols correspond to those in Table 1. The geographic regions discussed in this paper, Qinghai-Tibet Plateau, Yungui Plateau and northern China, aremarked on the map. The rough location of H-D Mountains is highlighted with a rectangle. B(right): Network of 12 haplotype based on rpL16 and trnL-F sequences of S.chamaejasme. The relative sizes of the circles in the network are proportional to haplotype frequencies. Haplotype name is marked near the circle, and its distribution ismarked in the circle except for QJ1. Solid and open bars represent nucleotide substitutions and indels, respectively.

Y.-H. Zhang et al. / Molecular Phylogenetics and Evolution xxx (2010) xxx–xxx 9

genetic variation. Population differentiation is expected to be weakin long-lived, perennial, outcrossing species (Loveless and Hamrick,1984). However, very high genetic differentiation (GST = 0.982) wasfound in S. chamaejasme, greatly exceeding the mean value for out-crossing species (GST = 0.22) reported by Nybom (Nybom, 2004).Most populations of this plant are spatially isolated, especially inthe H-D Mountains. Isolation and reduction of gene flow in this spe-cies may be enhanced by the different pollination modes observedacross the species range. For instance, butterfly pollination (psy-chophily) of S. chamaejasme was observed in Nepal (Grey-Wilson,1995), while in Kunming, plants were pollinated by moths (pha-laenophily; personal observations).

4.3. Impacts of geological and climatic alteration on the distributionalpattern of haplotypes

The twelve haplotypes detected in S. chamaejasme exhibited asurprising distribution pattern in the H-D Mountains (Fig. 6A).The H-D Mountains have been reported to be a region of high hap-lotype richness for several species. For example in Pedicularis longi-folia, 73% of the total haplotypes that have been found weredistributed in the H-D Mountains (referred to as the southeasternQ-T Plateau) (Yang et al., 2008). Eight haplotypes out of 10 weredetected in this region for Metagentiana striata (Chen et al.,2008). The characteristic topographic features of this region areparallel mountain ridges incised by a deep river valley. S. chamaej-asme grows discontinuously in different habitats along the altitudi-nal gradient from 2010 to 4020 m. Restricted gene flow due to thecomplex topography and great altitudinal variability of the regionmay be responsible for the high degree of genetic population dif-ferentiation (Till-Bottraund and Gaudeul, 2002). Limited or nonex-istent interpopulation migration coupled with a heterogeneousenvironment could promote local adaptation and fixation of differ-ent alleles in S. chamaejasme. Additionally, the current distributionof haplotypes could be a result of past climatic changes related to

Please cite this article in press as: Zhang, Y.-H., et al. Chloroplast phylogeny andin adjacent regions. Mol. Phylogenet. Evol. (2010), doi:10.1016/j.ympev.2010.0

the advance/retreat of glaciations (Yang et al., 2008). During theQuaternary period, climatic oscillations resulted in repeated dras-tic environmental changes, which further caused massive rangeshifts of most plants and animals, leading to accumulated geneticdifferences and particular phylogeographic patterns (Hewitt,2000, 2004). The south-north parallel valleys and mountain chainsin the H-D Mountains could have provided an ecological corridoror refugium during the Pleistocene when northern populationsimmigrated to the southeastern area of the species range. The low-er altitudinal environment could also provide a suitable habitat forplants immigrating from the alpine flora during an ice age. Re-peated re-colonization of the H-D Mountains by S. chamaejasmeunder climatic oscillations could be another explanation for itshigh level of haplotype diversity. Similarly, a relatively highamount of genetic divergence has been observed for many speciesof temperate Europe and North America within their southernrefugia (Hewitt, 1996, 2004). During the glacial period, some pop-ulations of S. chamaejasme could have expanded southward to thecentral Yungui Plateau of the southern Hengduan Mountains,reaching Kunming (the most southern location). The parsimonyhaplotype network that we constructed revealed that haplotypeH9 from the Kunming populations was close to haplotype H1, withonly two nucleotide mutations, but two other haplotypes, H11(found in the ZD and LJ populations) and H12 (found in the DL pop-ulation), exhibited a greater number of nucleotide mutations(Fig. 6B). The ZD, LJ and DL populations were geographically closerto populations with the H1 haplotype. These results indicate thatsouthward migration might have occurred during the glacial peri-od, and subsequently accompanying the glacial retreat, the south-ern edge population KM could have survived in the mountainousarea of Kunming. Additional nucleotide mutations could haveaccumulated when populations ZD, LJ and DL retreated north intonew habitats.

Haplotype H1 was the most dominant haplotype and occurredin 10 populations from the Q-T Plateau platform and northern Chi-

phylogeography of Stellera chamaejasme on the Qinghai-Tibet Plateau and8.033

10 Y.-H. Zhang et al. / Molecular Phylogenetics and Evolution xxx (2010) xxx–xxx

na (Fig. 6A). The absence of cpDNA haplotype variation within andamong these populations may indicate recent population expan-sions. One or more population(s) carrying haplotype H1 rapidlycolonized the Q-T Plateau platform and extended to northern Chi-na facilitated by the adjacent geographic range with a similargrassland habitat after the glacial period. Founder effects duringsuch a rapid expansion would be responsible for genetic homoge-neity in these populations (Fig. 6A). Although our finding of a rapidexpansion from a southern refugium (H-D Mountains) to the Q-TPlateau platform is similar to previously reported results (Chenet al., 2008; Yang et al., 2008; Zhang et al., 2005), our results re-vealed a further finding that the rapid expansion process may haveextended to the vast northern China region from the southeasternrefugium.

5. Conclusions

The study presented here on S. chamaejasme is one of only a fewstudies to date that have been conducted on herbaceous plants onthe Q-T Plateau and in adjacent northern China. Therefore, the fol-lowing conclusions should provide important insight into the evo-lutionary history of plants with similar life history traits anddistribution pattern: 1. the origin of S. chamaejasme coincides withthe large-scale uplift of the Q-T Plateau during the Miocene; 2. thewhite-red color is the ancestral state for flower color in this spe-cies; 3. geographical and ecological heterogeneity in the H-DMountains and the southeastern edge of the Q-T Plateau, togetherwith climatic oscillations during the Quaternary may have pro-moted intra-specific differentiation and high haplotype richnessthrough blocking gene flow among populations; and 4. the H-DMountains served as an important glacial refugium for S. chamaej-asme. The current, wide distribution of this species on the Q-T Pla-teau platform and in northern China may have resulted from rapidpost-glacial population expansion from a southeastern refugium.

Acknowledgments

We are grateful to Jipei Yue, Jiahui Chen and Ticao Zhang ofKunming Institute of Botany for their help in collecting plant mate-rial of S. chamaejasme. We thank Dr. David Galicai-Herbada ofAutonomous University of Madrid for kindly presenting threespecimens of Thymelaea for our study. Two DNA samples, D. mezer-eum and T. sanamunda, were obtained from the DNA Bank of theRoyal Botanic Gardens, Kew, with the help of Proff. Mark W. Chaseand Edith Kapinos. This research was supported by the NationalNatural Science Foundation of China (30625004 40930209 to Dr.Sun Hang), the Key Project of the Yunnan Natural Science Founda-tion (2008CC013 to Sun Hang) and Knowledge Innovation Projectof the Chinese Academy of Sciences (KSCX2-YW-Z-1019 to Sun).

References

An, Z., Kutzbach, J.E., Prell, W.L., Porter, S.C., 2001. Evolution of Asian monsoons andphased uplift of the Himalayan-Tibetan plateau since Late Miocene times.Nature 411, 62–66.

Axelrod, D.I., Al-Shehbaz, I., Raven, P.H., 1996. History of the modern flora of China.In: Zhang, A., Wu, S.G. (Eds.), Floristic Characteristics and Diversity of East AsianPlants. China Higher Education Press, Beijing, pp. 43–55.

Boufford, D.E., Dijk, P.P.V., 2000. South-Central China. In: Mittermeier, R.A., Myers,N., Mittermeier, C.G. (Eds.), Hotspots: Earth’s Biologically Richest and MostEndangered Terrestrial Ecoregions. Cemex, Mexico, pp. 338–351.

Boufford, D.E., Dijk, P.P.V., Zhi, L., 2004. Mountains of southwest China. In:Mittermeier, R.A., Robles-Gil, P., Hoffmann, M., Pilgrim, J.D., Brooks, T.M.,Mittermeier, C.G., Lamoreux, J.L., Fonseca, G. (Eds.), Hotspots Revisited: Earth’sBiologically Richest and Most Endangered Ecoregions. Cemex, Mexico, pp. 159–164.

Burbank, D.W., Derry, L.A., France-Lanord, C., 1993. Reduced Himalayan sedimentproduction 8 Myr ago despite an intensified monsoon. Nature 364, 48–50.

Please cite this article in press as: Zhang, Y.-H., et al. Chloroplast phylogeny andin adjacent regions. Mol. Phylogenet. Evol. (2010), doi:10.1016/j.ympev.2010.0

Chen, S.T., Guan, K.Y., Zhou, Z.K., Olmstead, R., Cronk, Q., 2005. Molecular phylogenyof Incarvillea (Bignoniaceae) based on ITS and trnL-F sequences. Am. J. Bot. 92,625–633.

Chen, S.Y., Wu, G.L., Zhang, Z.J., Gao, Q.B., Duan, Y.Z., Zhang, F.Q., Chen, S.L., 2008.Potential refugium on the Qinghai-Tibet Plateau revealed by the chloroplastDNA phylogeography of the alpine species Metagentiana striata (Gentianaceae).Bot. J. Linn. Soc. 157, 125–140.

Clement, M., Posada, D., Crandall, K.A., 2000. TCS: a computer program to estimategene genealogies. Mol. Ecol. 9, 1657–1659.

Diels, L., 1912. Plantae Chinenses Forrestianae: Numerical Catalogue of All the PlantsCollected by G. Forrest During his First Exploration of Yunnan, Eastern Tibet inthe Years 1904, 1905, 1906. Notes Roy Bot. Gard., Edinburgh XXXII. 88.

Doyle, J.J., Doyle, J.L., 1987. A rapid DNA isolation procedure for small quantities offresh leaf tissue. Phytochem. Bull. 19, 11–15.

Drummond, A.J., Rambaut, A., 2007. BEAST: Bayesian evolutionary analysis bysampling trees. BMC Evol. Biol. 7, 214.

Dumolin, S., Demesure, B., Petit, R.J., 1995. Inheritance of chloroplast andmitochondrial genomes in pedunculate oak investigated with an efficient PCRmethod. Theor. Appl. Genet. 91, 1253–1256.

Excoffier, L., Laval, G., Schneider, S., 2005. Arlequin ver. 3.0: an integrated softwarepackage for population genetics data analysis. Evol. Bioinform. Online 1, 47–50.

Fitch, W.M., 1971. Toward defining the course of evolution: minimal change for aspecific tree topology. Syst. Zool. 20, 406–416.

Ge, X.J., Zhang, L.B., Yuan, Y.M., Hao, G., Chiang, T.Y., 2005. Strong geneticdifferentiation of the East-Himalayan Megacodon stylophorus (Gentianaceae)detected by Inter-Simple Sequence Repeats (ISSR). Biodivers. Conserv. 14, 849–861.

Grey-Wilson, C., 1995. Stellera chamaejasme: an overview. The New Plantsman 2,43–49.

Hamrick, J.L., Godt, M.J.W., 1989. Allozyme diversity in plant species. In: Brown,A.H.D., Clegg, M.T., Kahler, A.L., Weir, B.S. (Eds.), Plant Population Genetics.Breeding and Genetic Resources. Sinauer, Sunderland MA, pp. 43–63.

Hamrick, J.L., Godt, M.J.W., 1996. Effects of life history traits on genetic diversity inplant species. Philos. T. Roy. Soc. B. 351, 1291–1298.

Hamrick, J.L., Linhart, Y.B., Mitton, J.B., 1979. Relationships between life historycharacteristics and electrophoretically detectable genetic variation in plants.Annu. Rev. Ecol. Syst. 10, 173–200.

Harrison, T.M., Copeland, P., Kidd, W.S.F., Yin, A., 1992. Raising Tibet. Science 255,1663–1670.

Hewitt, G.M., 1996. Some genetic consequences of ice ages, and their role indivergence and speciation. Biol. J. Linn. Soc. 58, 247–276.

Hewitt, G.M., 1999. Post-glacial re-colonization of European biota. Biol. J. Linn. Soc.68, 87–112.

Hewitt, G.M., 2000. The genetic legacy of the Quaternary ice ages. Nature 405, 907–913.

Hewitt, G.M., 2004. Genetic consequences of climate oscillations in the Quaternary.Philos. T. Roy. Soc. B. 359, 183–195.

Huang, S.C., 1985. Taxa nova Thymelaeacearum Sinicarum. Acta Bot. Yunnan. 7, 277–291.

Huelsenbeck, J.P., Ronquist, R., 2001. MRBAYES: Bayesian inference of phylogenetictrees. Bioinformatics 17, 754–755.

Jordan, W.C., Courtney, M.W., Neigel, J.E., 1996. Low levels of intraspecific geneticvariation at a rapidly evolving chloroplast DNA locus in north Americanduckweeds (Lemnaceae). Am. J. Bot. 83, 430–439.

Jump, A.S., Penuelas, J., 2005. Running to stand still: adaptation and the response ofplants to rapid climate change. Ecol. Lett. 8, 1010–1020.

Körner, C., 1999. Alpine Plant Life: Functional Plant Ecology of High MountainEcosystems. Springer, Berlin.

Kutzbach, J.E., Guetter, P.J., Ruddiman, W.F., Prell, W.L., 1989. The sensitivity ofclimate to late Cenozoic uplift in Southern Asia and the American west:Numerical experiments. J. Geophys. Res. 94, 18393–18407.

Léveillé, H., 1912. Decades plantarum novarum. Repert. Spec. Nov. Regni Veg. 10, 581.Li, B.Y., 1987. On the boundaries of the Hengduan Mountains. Mount. Res. 5, 74–82.Li, B.Y., 1989. Geomorphologic Regionalization of the Hengduan Mountainous

Region. Mount. Res. 7, 13–20.Li, W.Y., 1998. Quaternary vegetation and environment of China. Science Press,

Beijing.Li, J.J., Shi, Y.F., Li, B.Y., 1995. Uplift of the Qinghai-Xizang (Tibet) Plateau and Global

Change. Lanzhou Univ. Press, Lanzhou.Linnaeus, C., 1753. Species Plantarum, Stockholm.Liu, K.B., 1988. Quaternary history of the temperate forests of China. Quatern. Sci.

Rev. 7, 1–20.Liu, J.Q., Wang, Y.J., Wang, A.L., Hideaki, O., Abbott, R.J., 2006. Radiation and

diversification within the Ligularia-Cremanthodium-Parasenecio complex(Asteraceae) triggered by uplift of the Qinghai-Tibetan Plateau. Mol. Phylogen.Evol. 38, 31–49.

Loveless, M.D., Hamrick, J.L., 1984. Ecological determinants of genetic structure inplant populations. Annu. Rev. Ecol. Syst. 15, 65–95.

Maddison, W.P., Maddison, D.R., 2009. Mesquite: a modular system for evolutionaryanalysis, Version 2.71, http://mesquiteproject.org.

Mai, D.H., Walther, H., 1978. Die Floren des Haselbacher Serie im Weisselster-Becken (Bezirk Leipzig, DDR). Abh. Staatl. Mus. Mineral. Geol. Dresden. 28, 1–200.

Meng, L.H., Yang, R., Abbott, R.J., Miehe, G., Hu, T.H., Liu, J.Q., 2007. Mitochondrialand chloroplast phylogeography of Picea Crassifolia Kom. (Pinaceae) in theQinghai-Tibetan Plateau and adjacent highlands. Mol. Ecol. 16, 4128–4137.

phylogeography of Stellera chamaejasme on the Qinghai-Tibet Plateau and8.033

Y.-H. Zhang et al. / Molecular Phylogenetics and Evolution xxx (2010) xxx–xxx 11

Molnar, P., 2005. Mio-Pliocene growth of the Tibetan Plateau and evolution of EastAsian climate. Palaeontol. Electron. 8, 1–23.

Molnar, P., England, P., Martinod, J., 1993. Mantle dynamics, uplift of the TibetanPlateau and the Indian monsoon. Rev. Geophys. 31, 357–396.

Myers, N., Mittermeier, R.A., Mittermeier, C.G., da Fonseca, G.A.B., Kent, J., 2000.Biodiversity hotspots for conservation priorities. Nature 403, 853–858.

Ni, J., 2000. A simulation of biomes on the Plateau and their response to globalclimate change. Mt. Res. Dev. 20, 80–89.

Nybom, H., 2004. Comparison of different nuclear DNA markers for estimatingintraspecific genetic diversity in plants. Mol. Ecol. 13, 1143–1155.

Palmarev, E., 1989. Paleobotanical evidences of the Tertiary history and origin of theMediterranean sclerophyll dendroflora. Plant Syst. Evol. 162, 93–107.

Pauli, H., Gottfried, M., Reier, K., Klettner, C., Grabherr, G., 2007. Signals of rangeexpansions and contractions of vascular plants in the high Alps: observations(1994–2004) at the GLORIA*master site Schrankogel, Tyrol, Austria. Global.Change Biol. 13, 147–156.

Pons, O., Petit, R.J., 1995. Estimation, variance and optimal sampling of genediversity. I. Haploid locus. Theor. Appl. Genet. 90, 462–470.

Pons, O., Petit, R.J., 1996. Measuring and testing genetic differentiation with orderedversus unordered alleles. Genetics 144, 1237–1245.

Posada, D., Buckley, T.R., 2004. Model selection and model averaging inphylogenetics: advantages of the AIC and Bayesian approaches over likelihoodratio tests. Syst. Biol. 53, 793–808.

Qiang, X.K., Li, Z.X., Powell, C.M., Zheng, H.B., 2001. Magnetostratigraphic record ofthe Late Miocene onset of the East Asian monsoon and Pliocene uplift ofnorthern Tibet. Earth Planet. Sci. Lett. 187, 83–93.

Quade, J., Cater, J.M.L., Ojha, T.P., Adam, J., Harrison, T.M., 1995. Late Mioceneenvironmental change and in Nepal and the northern Indian subcontinent;stable isotopic evidence from paleosols. Geol. Soc. Am. Bull. 107, 1381–1397.

Rajora, O.P., Dancik, B.P., 1992. Chloroplast DNA inheritance in Populus. Theor. Appl.Genet. 84, 280–285.

Rambaut, A., 2007. Figtree, version 1.0, http://www.beast.bio.ed.ac.uk/FigTree.Raven, P.H., Axelrod, D.I., 1978. Origin and relationships of the California flora. Univ.

Calif. Publ. Bot. 72, 1–134.Raymo, M.E., Ruddiman, W.F., 1992. Tectonic forcing of late Cenozoic climate

change. Nature 359, 117–122.Shi, Y.F., Li, J.J., Li, B.Y., 1998. Uplift and Environmental Changes of Qinghai-Tibetan

Plateau in the Late Cenozoic. Guangdong Science and Technology Press,Guangzhou.

Spicer, R.A., Harris, N.B.W., Widdowson, M., Herman, A.B., Guo, S., Valdes, P.J., Wolfe,J.A., Kelley, S., 2003. Constant elevation of southern Tibet over the past 15million years. Nature 421, 622–624.

Sun, H., 2002. Evolution of Arctic-Tertiary flora in Himalayan-Hengduan mountains.Acta Bot. Yunnan. 24, 671–688.

Sun, H., Li, Z.M., 2003. Qinghai-Tibet Plateau uplift and its impact on Tethys flora.Adv. Earth Sci. 18, 852–862.

Swofford, D.L., 2000. PAUP4.0b10. Phylogenetic Analysis Using Parsimony andOther Methods, Version 4. Sinauer Associates, Sunderland, Massachusetts.

Taberlet, P., Gielly, L., Pautou, G., Bouvet, J., 1991. Universal primers foramplification of three non-coding regions of chloroplast DNA. Plant Mol. Biol.17, 1105–1109.

Tang, L.Y., Shen, C.M., 1996. Late Cenozoic vegetational history and climaticcharacteristics of Qinghai-Xizang Plateau. Acta Micropalaeon. Sin. 13, 321–337.

Thompson, J.D., Gibson, T.J., Plewniak, F., Jeanmougin, F., Higgins, D.G., 1997. TheClustalX windows interface. flexible strategies for multiple sequence alignmentaided by quality analysis tools. Nucleic Acids Res. 25, 4876–4882.

Thompson, L.G., Yao, T., Mosley-Thompson, E., 2000. A high-resolution millennialrecord of the South Asian monsoon from Himalayan ice cores. Science 289,1916–1919.

Please cite this article in press as: Zhang, Y.-H., et al. Chloroplast phylogeny andin adjacent regions. Mol. Phylogenet. Evol. (2010), doi:10.1016/j.ympev.2010.0

Till-Bottraund, I., Gaudeul, M., 2002. Intraspecific genetic diversity in alpine plants.In: Körner, C., Spehn, E.M. (Eds.), Mountain Biodiversity: A Global Assessment.Parthenon Publishing, New York, pp. 23–34.

Wang, W.C., 1992a. On some distribution patterns and some migration routes foundin the Eastern Asiatic region. Acta Phytotaxon. Sin. 30, 1–24.

Wang, W.C., 1992b. On some distribution patterns and some migration routes foundin the Eastern Asiatic region (count.). Acta Phytotaxon. Sin 30, 97–117.

Wang, W.C., 1993. Vascular Plants of the Hengduan Mountains, vol. 1. Science Press,Beijing.

Wang, W.C., 1994. Vascular Plants of the Hengduan Mountains, vol. 2. Science Press,Beijing.

Wang, Y.Z., Gilbert, M.G., 2007. Stellera Linnaeus. In: Wu, C.Y., Raven, P.H. (Eds.),Flora of China, vol. 13. Science Press, Beijing, p. 250.

Wang, A.L., Yang, M.H., Liu, J.Q., 2005. Molecular phylogeny, recent radiation andevolution of gross morphology of the Rhubarb genus Rheum (Polygonaceae)inferred from chloroplast DNA trnL-F sequences. Ann. Bot. 96, 489–498.

Wang, Y., Deng, T., Biasatti, D., 2006. Ancient diets indicate significant uplift ofsouthern Tibet after ca. 7 Ma. Geology 34, 309–312.

Wang, Y.J., Liu, J.Q., Miehe, G., 2007. Phylogenetic origins of the Himalayan endemicDolomiaea, Diplazoptilon and Xanthopappus (Asteraceae: Cardueae) based onthree DNA regions. Ann. Bot. 99, 311–322.

Wang, L.Y., Abbott, R.J., Zheng, W., Chen, P., Wang, Y.J., Liu, J.Q., 2009. History andevolution of alpine plants endemic to the Qinghai-Tibetan Plateau: Aconitumgymnandrum (Ranunculaceae). Mol. Ecol. 18, 709–721.

Wikström, N., Savolainen, V., Chase, M.W., 2001. Evolution of the angiosperms:calibrating the family tree. Proc. R. Soc. Lond. B Biol. Sci. 268, 2211–2220.

Wilson, E.O., 1992. The Diversity of Life. Belknap Press of Harvard Univ. Press,Cambridge, MA.

Wu, C.Y., 1980. The Vegetation of China. Science Press, Beijing.Wu, C.Y., 1987. Origin and Evolution of Flora of Xizang. In: Wu, C.Y. (Ed.), Flora

Xizangica. Science Press, Beijing, pp. 874–902.Wu, C.Y., 1988. The Hengduan Mountain flora and her significance. J. Jpn. Bot. 63, 1–

14.Wu, C.Y., 1995. Vegetation of China. Science Press, Beijing.Wu, C.Y., Wu, S.G., 1996. A proposal for new floristic kingdom (realm). In: Zhang,

A.L., Wu, S.G. (Eds.), Floristic Characteristics and Diversity of East Asian Plants.Springer-Verlag, Hong Kong, pp. 3–42.

Xia, T., Chen, S.L., Chen, S.Y., Ge, X.J., 2005. Genetic variation within and amongpopulations of Rhodiola alsia (Crassulaceae) native to the Tibetan plateau asdetected by ISSR makers. Biochem. Genet., 87–101.

Yang, F.S., Li, Y.F., Ding, X., Wang, X.Q., 2008. Extensive population expansionof Pedicularis longiflora (Orobanchaceae) on the Qinghai-Tibetan Plateau andits correlation with the Quaternary climate change. Mol. Ecol. 17, 5135–5145.

Ying, T.S., Boufford, D.E., Zhang, Y.L., 1993. The Endemic Genera of Seed Plants ofChina. Science Press, Beijing.

Zhang, Z.R., 1999. Stellera Linn. In: Anonymous (Ed.), Flora Reipublicae PopularisSinicae, vol. 52(1). Science Press, Beijing, pp. 397–399.

Zhang, X.S., Yang, D.A., Zhou, G.S., Liu, C.Y., Zhang, J., 1996. Model expectation ofimpacts of global climate change on biomes of the Tibetan Plateau. In: Omasa,K., Kai, K., Taodao, H., Uchijima, Z., Yoshino, M. (Eds.), Climate Change and Plantsin East Asia. Springer-Verlag, Tokyo, pp. 25–38.

Zhang, Q., Chiang, T.Y., George, M., Liu, J.Q., Abbott, R.J., 2005. Phylogeography of theQinghai-Tibetan Plateau endemic Juniperus przewalskii (Cupressaceae) inferredfrom chloroplast DNA sequence variation. Mol. Ecol. 14, 3513–3524.

Zhao, S.L., Li, G.G., 1990. Desertization on the shelves adjacent China in the LaterPleistocene. Oceanol. Limnol. Sin. 8, 289–298.

Zheng, D., 1996. The system of physico-geographical regions of the Qinghai-Tibet(Xizang) Plateau. Sci. China, Ser. D Earth Sci. 39, 410–417.

phylogeography of Stellera chamaejasme on the Qinghai-Tibet Plateau and8.033