Molecular Phylogenetics and Evolution 55 (2010) 621–630

Contents lists available at ScienceDirect

Molecular Phylogenetics and Evolution

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

Evidence of transoceanic dispersion of the genus Vanilla based on plastid DNAphylogenetic analysis

Anthony Bouetard a,1, Pierre Lefeuvre a,b, Rodolphe Gigant a,c, Séverine Bory a,c, Marc Pignal d, Pascale Besse c,Michel Grisoni a,*

a Cirad, UMR 53 Cirad-Université de la Réunion PVBMT, Pole de Protection des Plantes, 7 chemin de l’IRAT, 97410 Saint Pierre, La Réunion, Franceb Institute of Infectious Diseases and Molecular Medicine, University of Cape Town, Observatory 7925, South Africac University of La Réunion, UMR 53 Cirad-Université de la Réunion PVBMT, 15 avenue René Cassin, BP 7151, 97715 Saint Denis Messag Cedex 9, La Réunion, Franced Muséum national d’histoire naturelle, Département Systématique et Evolution, UMR 7205 Origine Structure et Evolution de la Biodiversité, 16 Rue Buffon, C.P. 39, 57 rueCuvier, 75231 Paris Cedex 05, France

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

Article history:Received 12 August 2009Revised 17 December 2009Accepted 20 January 2010Available online 28 January 2010

Keywords:OrchidaceaeVanillapsaBpsbBpsbCrbcLBayesian analysisPhylogeographyMolecular datingGondwanan vicarianceTransoceanic dispersion

1055-7903/$ - see front matter � 2010 Elsevier Inc. Adoi:10.1016/j.ympev.2010.01.021

* Corresponding author. Fax: +33 0262499293.E-mail address: michel.grisoni@cirad.fr (M. Grison

1 Present address: IRD, Research Unit #016, Polechemin de l’IRAT, 97410 Saint Pierre, La Réunion, Fran

The phylogeny and the biogeographical history of the genus Vanilla was investigated using four chloro-plastic genes (psbB, psbC; psaB and rbcL), on 47 accessions of Vanilla chosen from the ex situ CIRAD col-lection maintained in Reunion Island and additional sequences from GenBank. Bayesian methodsprovided a fairly well supported reconstruction of the phylogeny of the Vanilloideae sub-family and moreparticularly of the genus Vanilla. Three major phylogenetic groups in the genus Vanilla were differenti-ated, which is in disagreement with the actual classification in two sections (Foliosae and Aphyllae) basedon morphological traits. Recent Bayesian relaxed molecular clock methods allowed to test the two mainhypotheses of the phylogeography of the genus Vanilla. Early radiation of the Vanilla genus and diversi-fication by vicariance consecutive to the break-up of Gondwana, 95 million years ago (Mya), was incom-patible with the admitted age of origin of Angiosperm. Based on the Vanilloideae age recently estimatedto 71 million years ago (Mya), we conclude that the genus Vanilla would have appeared �34 Mya in SouthAmerica, when continents were already separated. Nevertheless, whatever the two extreme scenariostested, at least three long distance migration events are needed to explain the present distribution ofVanilla species in tropical areas. These transoceanic dispersions could have occurred via transoceanic pas-sageway such as the Rio Grande Ridge and the involvement of floating vegetation mats and migratorybirds.

� 2010 Elsevier Inc. All rights reserved.

1. Introduction

The genus Vanilla Plumier ex Miller is one of the most familiarof all orchids due notably to V. planifolia G. Jackson which is theprincipal source of the famous natural vanilla flavoring. The genuscontains as many as 110 different species distributed throughoutthe New and Old World tropics, but absent from Australia. Most(52) of the species are found in tropical America, 31 species arefound in south-east Asia and New Guinea, 14 in Africa, 10 in the In-dian Ocean islands, and 3 in the Pacific islands (Bory et al., 2008b;Portères, 1954).

The phylogenetic position of Vanilla among Orchidaceae is nowclearer thanks mostly to molecular phylogenetic studies based on

ll rights reserved.

i).de Protection des Plantes, 7ce.

combinations or individual matrices of plastid rbcL (Cameronet al., 1999; Soto Arenas, 2003), psaB (Cameron, 2004), or psbBand psbC (Cameron and Molina, 2006) gene sequences with goodsampling of Vanilloideae. Plastid genes present a real interest inphylogenetics due to unilateral inheritance, numerous copies percell, ease of amplification and sequencing and minimized risk ofcontamination by fungi (Clegg and Zurawski, 1992; Olmsteadand Palmer, 1994; Palmer et al., 1988). Other studies have alsoused nad1b-c mitochondrial gene sequences (Freudenstein et al.,2004) or nuclear 18S gene sequences (Cameron and Chase, 2000),to discriminate the Orchidaceae family. Thus, taxonomists nowagree that Vanilla belongs to the sub-tribe Vanillinae, in the tribeVanilleae, which forms with the tribe Pogonieae, a primitive mono-phyletic orchid lineage which is now recognized as a distinct sub-family Vanilloideae and comprises approximately 14 other genera(Cameron, 2004, 2009; Cameron et al., 1999; Cameron and Molina,2006; Soto Arenas, 2003).

However, the taxonomy and the phylogeography of the genusVanilla still remain uncertain. Rolfe (1896) was the first author to

622 A. Bouetard et al. / Molecular Phylogenetics and Evolution 55 (2010) 621–630

propose a classification of the genus from morphological observa-tions of flowers, leaves and stems. He recognized two sections:Foliosae and Aphyllae, respectively, leafy and leafless. Portères(1954) further divided the Foliosae section into three subsections:Papillosae (thick leaves and labellum with more or less fleshyhairs), Lamellosae (thick leaves and labellum with scaly lamellae)and Membranaceae (thin membraneous to sub-membraneousleaves). More recently, Soto Arenas (2003) invalidated this classifi-cation by stressing that subsections are heterogeneous and incom-plete. However, as no revision of the taxonomy has been proposedso far, Portères’ classification stays the most complete and is stillused.

On the basis of the discriminant floral characters, Portères(1954) suggested that the primary diversification center of genusVanilla could be Indo-Malaysian. This Indo-Malaysian stock wouldhave subsequently diversified and evolved in two groups: on onehand in Madagascar, Mascarene islands and Africa, and on theother hand in Oriental Asia and Occidental Pacific islands, withsubsequent migration towards America from the Pacific duringthe Tertiary (65.5–2.5 million years ago (Mya)) (Portères, 1954).Molecular phylogenetic studies invalidated this hypothesis and itis now established that the sub-tribe Vanillinae originated fromSouth America (Cameron, 2000; Cameron and Chase, 1999). Never-theless, previous molecular studies were based on a limited num-ber of Vanilla samples therefore the fine phylogeny of the genusVanilla still remains uncertain.

Following the 1960s general acceptance of the theory of con-tinental drift, vicariance became the predominant hypothesis inhistorical biogeography (Cuenca et al., 2008). However, anincreasing number of phylogenetic studies of tropical angio-sperms evolving in both the Old and New World already sug-gested the spread of taxa either from the Old World to theNew World for Meliaceae (Muellner et al., 2005), Lauraceae(Von Balthazar et al., 2007) and Annonaceae (Doyle and Thomas,1997), or in the opposite direction for Melastomataceae (Morleyand Dick, 2003) and Malpighiaceae (Davis et al., 2001) by trans-oceanic passageway.

Those two main hypotheses are relevant with regards to thedispersion history of the genus Vanilla. The first proposed a migra-tion of the genus before the break-up of Gondwana from SouthAmerica to Africa (�160–120 Mya), then, from Africa to Asia (Cam-eron, 2005). However, the recent discovery in the DominicanRepublic of well preserved orchid pollinarium (of Meliorchis cari-bea) attached to an extinct stingless bee (Problebeia dominica),recovered in Miocene amber has allowed to estimate that the com-mon ancestor of present day orchids lived in the late Cretaceous(76–84 Mya), at least 20 Mya after the break-up of Gondwana (Ra-mirez et al., 2007). This finding is therefore inconsistent with thehypothesis of transcontinental dispersion. Thus, transoceanicmigration would seem to better explain the present dispersion pat-tern of the genus Vanilla.

Here, we present a generic-level phylogeny of vanilloids includ-ing an extended set of Vanilla species using four chloroplasticgenes: psbB, psbC; psaB and rbcL, coding, respectively, for subunitsof proteins P680 (i.e. photosystem II), P700 (i.e. photosystem I) andRubisco (i.e. Calvin cycle). We used this phylogeny to address twoobjectives. Firstly, the obtained phylogenetic data are comparedwith previous hypotheses generated with morphological (Portères,1954) and molecular data (Cameron, 2004; Cameron et al., 1999;Cameron and Molina, 2006; Soto Arenas, 2003) in order to clarifythe phylogeny of the genus. Secondly, relative divergence timesof the two main phylogeographical hypotheses are estimated usingrecent Bayesian relaxed molecular clock methods, in order to ob-tain a better understanding of the temporal pattern of diversifica-tion of the genus between the New World (Neotropical America)and the Old World (Africa and Asia).

2. Materials and methods

2.1. Plant material

Forty-seven accessions of Vanilla, representing 23 specimenidentified on the basis of floral characters, three hybrids and 21unidentified species, were chosen from the ex situ collection main-tained at the CRB VATEL (CIRAD) in Reunion Island (Grisoni et al.,2007), to represent maximum variability within the genus on thebasis of previous AFLP and microsatellite characterization (Boryet al., 2008a,c). Twenty-eight accessions originate from America,18 from Africa or Asia and one is of unknown origin. Thirty-eightaccessions belong to the Foliosae section, of which sixteen are fromof the Lamellosae subsection, one from the Membranaceae subsec-tion and two from the Papillosae subsection. The remaining nineVanilla accessions belong to the Aphyllae section (Portères, 1954)(Table 1).

2.2. PCR and sequencing reactions

DNA from leaf or stem material was extracted as described pre-viously (Bory et al., 2008c). No further purification of the DNA be-fore PCR amplification was necessary. Two primer pairs were usedto amplify each of four chloroplastic genes, psaB, psbB, psbC andrbcL, in two overlapping DNA fragments, each with a length infe-rior to 900 bp in order to permit sequencing reactions (Table 2).Primers were designed using primer 3 software (Rozen and Skalet-sky, 2000) based on the sequences of four chloroplastic genes, psaB,psbB, psbC and rbcL from seven Vanilla species (Table 3) depositedin GenBank (Cameron, 2004; Cameron et al., 1999; Cameron andMolina, 2006). PCR reactions were conducted on a GeneAmp PCRSystem 9700 thermocycler (Applied Biosystems, Foster City, CA,USA) in a final volume of 50 lL containing 50 ng DNA, 200 lMdNTPs (New England Biolabs, Ipswich, MA, USA), 1.5 mM MgCl2

(Eurogentec, Liège, Belgium), 1� PCR buffer, 0.2 lM of each primerand 2 U Taq DAP GoldStar™ (Eurogentec, Liège, Belgium). Cyclingparameters were 94 �C for 4 min, then 35 cycles of 95 �C for 45 s,55 �C for 1 min, 72 �C for 2 min and a final elongation at 72 �C for7 min. The integrity of target loci was checked using electrophore-sis on a 1.5% agarose gel. Then, PCR products were purified and se-quenced reverse and forward using amplification primers, byMacrogen Inc. (South Korea).

2.3. Sequence analysis

Sequences were checked and cleaned by editing reverse andforward electrophoregrams using BioEdit Sequence Alignment Edi-tor v.5.0.6 (Hall, 1999) and MEGA 3.1 (Kumar et al., 2004) software.They have been deposited in GenBank under the accession num-bers FN545387 to FN545433 (psaB), FN545434 to FN545479(psbB), FN545480 to FN545527 (psbC) and FN545528 toFN545574 (rbcL). After alignment with ClustalW (Thompsonet al., 1994) implemented in Bioedit, the Reading Frame (RF) wasdetermined and a preliminary analysis of the polymorphism ofthe DNA matrices was carried out using DNaSP 4.0 software (Rozaset al., 2003). The four independent gene matrices showed compa-rable features, notably concerning the level of polymorphism andthe number of haplotypes differentiated (data not shown) suggest-ing they were submitted to similar evolution forces.

The DNA sequences previously published (Cameron, 2004;Cameron and Chase, 1999; Cameron and Molina, 2006) for psaB,psbB, psbC and rbcL genes for six Vanilla species, 17 others Vanilloi-deae (eight other Vanilleae and nine Pogoniae) and one superiorEpidendroïdeae have been added to the analysis to complete thesequence dataset. To root the phylogenetic trees psaB, psbB, psbC

Table 1Vanilla samples analyzed, listed by accession numbers of the CIRAD collection(Grisoni et al., 2007), species, geographical origin (inferred from place of collectionand/or molecular analysis), sections and subsections.

Accessionnumber

Species Geographicalorigina

Sectionb Subsectionb

CR0058 V. albida Blume Asia/Africa Foliosae LamellosaeCR0065 V. crenulata Rolfe Asia/Africa Foliosae LamellosaeCR0067 sp. Asia/Africa Foliosae LamellosaeCR0091 V. crenulata Rolfe Asia/Africa Foliosae NDCR0102 V. crenulata Rolfe Asia/Africa Foliosae NDCR0103 V. africana Lindl. Asia/Africa Foliosae LamellosaeCR0104 V imperialis Kraenzl Asia/Africa Foliosae PapillosaeCR0153 sp. Asia/Africa Foliosae NDCR0175 sp. Asia/Africa Foliosae NDCR0698 sp. Asia/Africa Foliosae NDCR0705 V. cf. polylepis Summerh Asia/Africa Foliosae LamellosaeCR0089 sp. Asia/Africa Aphyllae NDCR0108 V. humblotii Rchb. f. Asia/Africa Aphyllae NDCR0141 V. madagascariensis Rolfe Asia/Africa Aphyllae NDCR0146 V. phalaenopsis Rchb. f. ex

Van HoutteAsia/Africa Aphyllae ND

CR0697 sp. Asia/Africa Aphyllae NDCR0699 sp. Asia/Africa Aphyllae NDCR0810 V. roscheri Rchb. f. Asia/Africa Aphyllae NDCR0003 Hyb. V. planifolia � V.

tahitensisAmerica Foliosae ND

CR0060 sp. America Foliosae LamellosaeCR0079 sp. America Foliosae NDCR0083 V. palmarum (Salm. ex.

Lindl.) Lindl.America Foliosae Membranaceae

CR0087 sp. America Foliosae NDCR0095 sp. America Foliosae NDCR0098 V. bahiana Hoehne America Foliosae LamellosaeCR0100 sp. America Foliosae NDCR0109 V. leprieuri R. Porteres America Foliosae LamellosaeCR0115 sp. America Foliosae NDCR0117 sp. America Foliosae NDCR0131 Hyb. V. planifolia � (V.

planifolia � V. pompona)America Foliosae ND

CR0164 V. tahitensis J. W. Moore America Foliosae LamellosaeCR0166 Hyb. V. planifolia � V.

phaeanthaAmerica Foliosae ND

CR0169 V. pompona Schiede America Foliosae LamellosaeCR0171 sp. America Foliosae NDCR0174 V. ensifolia Rolfe America Foliosae LamellosaeCR0177 sp. America Foliosae NDCR0178 sp. America Foliosae NDCR0211 V. planifolia Jacks. ex.

AndrewsAmerica Foliosae Lamellosae

CR0665 sp. America Foliosae NDCR0666 V. chamissonis Klotzsch America Foliosae LamellosaeCR0682 V. lindmaniana Kraenzl America Foliosae PapillosaeCR0683 sp. America Foliosae NDCR0686 V. odorata C. Presl. America Foliosae LamellosaeCR0693 V. cf. grandiflora Lindl. America Foliosae LamellosaeCR0081 sp. America Aphyllae NDCR0794 V. dilloniana Correll America Aphyllae NDCR0068 sp. ND Foliosae Lamellosae

a According to Bory et al. (2008a,c).b According to Portères’ classification (1954).

Table 2Primer map used to amplify and sequence chloroplast genes psaB, psbB, psbC and rbcL.Primers designed on the basis of Vanilla sequences published by Cameron et al. (1999,2004) and Cameron and Molina (2006).

Primers Sequence 50 > 30 Size(bp)

Tm

(�C)Size ofamplifiedfragments(bp)

psaBPsaB49L/

PsaB848RCCGTCGCAAGGAAAACTATAA 21 59.26 800TTCGGGATTGGTCACAGTAT 20 57.86

PsaB766L/PsaB1526R

AGACCCTTATGYCCACGYC 19 59.94 761GCTTGGCAAGGAAATTTTGA 20 60.19

psbBPsbB119L/

PsbB508RGGAGACAARGCATGTTCGTT 20 60.12 390CCCACGCTGGATTTACAGAT 20 59.96

PsbB434L/PsbB1212R

TGGTCCTGGAATATGGGTGT 20 60.05 795CCTAATTGGGCACGTCTAGC 20 59.73

psbCPsbC25L/

PsbC786RGGTCTGGCTCTGAACCTACG 20 59.87 762GGGCTAAGGGTCAARTTGGT 20 60.36

PsbC596L/PsbC1379R

TCCTTTCCATTCTTCGGTTATG 22 60.3 784AAGAACCTAAAGGAGCATGAGTC 23 58.11

rbcLRcbL33L/

RcbL730RCTCCTGACTACGAAACCAAAGA 22 58.51 698TCTCTGGCAAATACCGCTCT 20 59.98

RcbL453L/RcbL1231R

TCGTCCCCTATTGGGATGTA 20 60.15 779CCTCATTACGAGCTTGCACA 20 60.01

A. Bouetard et al. / Molecular Phylogenetics and Evolution 55 (2010) 621–630 623

and rbcL genes of four dicots, Medicago truncatula, Populus alba,Guizotia abyssinica, Oenothera argillicola, have been added as out-group of the Orchidaceae in addition to Phalaenopsis aphroditewhich was used as an outgroup for the Vanilloideae (Table 3).

2.4. Molecular phylogenetic methods

The individual psaB (1420 bp), psbB (1086 bp), psbC (1320 bp)and rbcL (1191 bp) matrices, as well as the concatenated four-genematrices (5037 bp) for 76 taxa, were analyzed using four phyloge-netic methods: Neighbor-Joining (NJ) and Maximum Parsimony(MP), using Mega3.1, Maximum Likelihood (ML) using PhyML

(Guindon and Gascuel, 2003) and Bayesian (MB) analysis usingMrBayes v3.1.1 (Huelsenbeck and Ronquist, 2001).

For ML and MB methods, the General Time Reversible model + -gamma distribution + Invariant sites (GTR + C + I) was selectedaccording to the Akaike Information Criterion (AIC) using Model-Test v3.8. For MB phylogeny reconstruction, two simultaneousindependent runs of the Monte Carlo Markov Chain (MCMC) for5 millions generations were conducted in MrBayes, sampling every100 generations with an initial 10% burn-in.

Resulting trees obtained with the different matrices and the dif-ferent methods were very similar in overall topology (data notshown) although the concatenated-matrix offered a better differ-entiation between haplotypes and stronger bootstrap supports.The MB tree obtained the highest maximum likelihood in SH test(Shimodaira and Hasegawa, 1999) and was subsequently consid-ered as the best phylogeny for our dataset. Consequently, only re-sults obtained with the concatenated genes matrix (5037 bp) andlikelihood analyses conducted in a Bayesian framework arepresented.

2.5. Molecular clock estimation

We estimated divergence times of splits using the Bayesian re-laxed clock approach implemented in BEAST v1.4.7. (Drummondet al., 2006). The three molecular clock models proposed in BEAST(e.g. strict clock, uncorrelated exponential relaxed clock and uncor-related lognormal relaxed clock model) were tested on our se-quence dataset. Using the Yule process of speciation (Steel andMcKenzie, 2001), with the ingroup assumed to be monophyleticwith respect to the outgroup and with default prior. After run con-vergence, the three analyses were then compared with a Bayes Fac-tor analysis (Tracer v1.4; (Rambaut and Drummond, 2004)) on thetree likelihood. Consequently, the hypothesis of a strict clock wassignificantly rejected and the two relaxed molecular clock modelswere retained as the most appropriate. Results obtained with thesetwo molecular clocks were globally concordant and the Bayes Fac-tor presented a non-significant preference for the uncorrelatedexponential relaxed clock. This model will therefore be the one

Table 3Taxa added in the analysis. GenBank accession numbers are provided. Ranks of species in the classification were based on the NCBI’s Taxonomy Guide.

Classification Accession GenBank

psaB psbB psbC rbcL

OrchidaceaeEpidendroideae

VandeaeAeridinae

Phalaenopsis aphrodite Rchb. f. AY916449d AY916449d AY916449d AY916449d

VanilloideaeVanilleae

VanillineaeVanilla aphylla Bl. AY381088a AY705150c AY705180c AF074238a

Vanilla africana Lindl. AY381089a AY705151c AY705181c AF074239a

Vanilla cf. barbellata Rchb. f. AY381090a AY705153c AY705183c AF074240a

Vanilla imperialis Kraenzlin AY381091a AY705159c AY705179c AF074241a

Vanilla roscheri Rchb. f. AY381095a AY705154c AY705184c AF074243a

Vanilla inodora Schiede AY381092b AY705155c AY705186c AY381136b

Pseudovanilla ponapensis (Kaneh. & Yamam.) Garay AY381069b AY705156c AY705188c AY381131b

Clematepistephium smilacifolium Rchb. f. AY380964b AY705158c AY705190c AF074131a

Eriaxis rigida Rchb. f. AY381005b AY705157c AY705189c AF074165a

Epistephium parviflorum Lindl. AY381002b AY705162c AY705194c AF074162a

Epistephium cf. lucidum Chase AY381001b AY705161c AY705193c AF074161a

Epistephium subrepens Hoehne AY381003b AY705163c AY705195c AF074163a

GaleolinaeErythrorchis altissima Blume AY381008b AY705148c AY705178c AF074168a

Erythrorchis cassythoides (A.Cunn. ex Lindl.) Garay AY381009b AY705147c AY705177c AF074169a

PogonieaePogoniinae

Cleistes cipoana Hoehne AY380962b AY705167c AY705199c AY381112b

Cleistes divaricata (L.) Ames AY380957b AY705166c AY705198c AF074127a

Cleistes rosea Lindl. AY380958b AY705171c AY705203c AF074128a

Duckeella adolphii Porto & Brade AY380992b AY705175c AY705207c AF074154a

Isotria medeoloides (Pursh) Raf. AY381022b AY705164c AY705196c AY381123b

Isotria verticillata (Muehl. ex Willd.) Raf. AY381023b AY705165c AY705197c AF074180a

Pogonia japonica Rchb. f. AY381061b AY705173c AY705205c AF074219a

Pogonia minor Makino AY381062b AY705172c AY705204c AF074220a

Pogonia ophioglossoides (L.) Ker Gawl. AY381063b AY705174c AY705206c AF074221a

FabaceaePapilionoideae

TrifolieaeMedicago truncatula Gaertn. AC093544e AC093544e AC093544e AC093544e

SalicaceaeSaliceae

Populus alba L. AP008956f AP008956f AP008956f AP008956f

AsteraceaeAsteroideae

Guizotia abyssinica (L. f.) Cass. EU549769g EU549769g EU549769g EU549769g

OnagraceaeOenothera

Oenothera argillicola Mackenzie EU262887h EU262887h EU262887h EU262887h

a Cameron et al. (1999).b Cameron (2004).c Cameron and Molina (2006).d Chang et al. (2006).e Lin et al. (unpublished).f Okumura et al. (2006).g Kane et al. (unpublished).h Greiner et al. (2008).

624 A. Bouetard et al. / Molecular Phylogenetics and Evolution 55 (2010) 621–630

presented to estimate substitution rates for all nodes in the tree. Asingle calibration node was applied as priors to test for each migra-tion hypothesis. For the continental radiation hypothesis, the nodeseparating the Old World and the New World Vanilla clades wasfixed at 95 Mya (with normal prior and a standard deviation of5 Mya) according to a maximum age for the break up betweenSouth America and Africa (Cuenca et al., 2008; Morley, 2003).

According to Ramirez et al.’s results (2007) and to test for thetransoceanic migration hypothesis, 71 Mya (with normal prior dis-tribution and a standard deviation of 5 Mya) was applied as priors

as minimum Vanilloideae apparition age. Two independent MCMCruns of 2,000,000 generations were performed for each calibrationanalysis, with parameters logged every 1000 generations. Theseruns were combined for each analysis to estimate the posterior dis-tribution of the substitution model and tree model parameters, aswell as node ages. Analyses of these parameters in Tracer 1.4 (A.Rambaut, http://tree.bio.ed.ac.uk/software/tracer) suggested thatthe number of MCMC steps was more than adequate, with effectivesample sizes of all parameters superior to 200 (often exceeding1000) and Tracer plots showing strong equilibrium after discarding

A. Bouetard et al. / Molecular Phylogenetics and Evolution 55 (2010) 621–630 625

burn-in. Trees were then constructed and divergence times addedon the different nodes using the Figtree v.1.1 software (A. Rambaut,http://tree.bio.ed.ac.uk/software/figtree).

3. Results

3.1. Phylogenetic analysis

The Bayesian tree obtained for the concatenated genes withbest-fit model (GTR + I + G) (Fig. 1) is concordant with previousdata published for intergeneric relationships among the Vanilloi-deae sub-family (Cameron and Molina, 2006; Soto Arenas, 2003).The analysis highlights a clear differentiation (Bayesian posteriorprobability (BpP) of nodes = 1) between the monophyletic sistertribes Pogonieae and Vanilleae as well as among monophyleticsub-tribes Pogoniinae, Galeolinae and Vanillinae.

From the 53 Vanilla accessions analyzed in this study (Tables 1and 3), 37 haplotypes have been differentiated. They are located inthree major clades (BpP = 1), identified as groups a, b and c (Fig. 1).A first node separates the American Foliosae species V. mexicana(group a) to the remainder of the genus. Separation betweengroups b (i.e., the rest of the American Foliosae species) and c(i.e. Old World Foliosae species and Aphyllae) appears to be morerecent. Within these groups, a high supported differentiation be-tween species was obtained, permitting to segregate each groupinto eight and six main subgroups, respectively (Fig. 1).

Within group b, subgroup bA represented by V. palmarum, V.lindmaniana and V. sp. CR0100, is sister to the rest of the group.Then, subgroups bB (represented by V. ensifolia) in pair with bC(represented by V. leprieuri), bE (represented by V. chamissonis) inpair with bF (represented by V. odorata), bG (represented by V.planifolia, V. tahitensis and intra-subgroups hybrids), and bH (repre-sented by V. bahiana) are gradually discriminated from each other.

Within group c, the African Foliosae constitute subgroup cA,represented by V. africana and V. crenulata, which pairs with amoderate support (BpP = 0.87) with the Asian subgroup cB com-posed of V. aphylla and V. albida, respectively, Aphyllae and Folio-sae species. The rest of group c is composed by subgroup cC,represented by the African Foliosae V. imperialis, which is sisterto the Indian-Ocean Aphyllae species (subgroup cD) with a moder-ate support (BpP = 0.89). Finally, there is a moderately supportedclade grouping the African Foliosae unidentified species V. sp.CR0067 (subgroup cE) with the Caribbean Aphyllae species (sub-group cF) (Fig. 1).

3.2. Molecular dating

According to the Gondwanan dispersion hypothesis, the calibra-tion of the node separating the Old World and the New World Va-nilla clades was fixed at 95 ± 5 Mya as age for the break up betweenSouth America and Africa. Under this hypothesis and according tothe uncorrelated exponential relaxed clock analysis, the age of thecrown of the Orchidaceae (excluding Apostasioideae) family andthe genus Vanilla were estimated to 310.25 Mya (HPD95% range:188–455 Mya) and 129.2 Mya (HPD95% range: 98–176 Mya),respectively.

In the transoceanic dispersion hypothesis a calibration at71 ± 5 Mya was applied as age for most recent ancestor of theVanilloideae sub-tribe. The age of the crown of the Orchidaceaefamily (excluding Apostasioideae) was estimated at 83.47 Mya(HPD95% range: 68–104 Mya), the genus Vanilla would have ap-peared 34.65 Mya (HPD95% range: 21–49 Mya) and migration ofthe genus from the New World to the Old World would have oc-curred around 25.5 Mya (HPD95% range: 15–38 Mya).

All datings of nodes obtained with the two relaxed molecularclock for these two dispersion hypotheses are detailed in Table 4.

4. Discussion

4.1. Phylogenetic relationships and classification of the genus Vanilla

The present study provides a comprehensive representation ofthe genetic relationships between members of the genus Vanilla,based on molecular data. The plastid DNA sequences analyzedindicated the presence of an external clade represented by V. mex-icana of the Foliosae section and Membranaceae subsection. This iscongruent with previous analysis of the plastid gene rbcL se-quences of 21 Vanilla species realized by Soto Arenas (2003), aswell as with a recent cladogram based on plastid, mitochondrialand nuclear DNA sequences of 22 Vanilla species (Cameron, inpress). These molecular phylogenies are in agreement with mor-phological analyses (Portères, 1954), with the presence in the V.mexicana clade (a) of species from the Foliosae section and Mem-branaceae subsection such as V. edwallii, V. angustipetala and V.martinezii, which are confined to the Neotropics and more diversein South America than in northern neotropical areas. The flowers ofVanilla species with membranaceous leaves are close to those ofEpistephium, although overall flower morphology may be corre-lated to pollination by large bees (Soto Arenas, 2003).

The remaining species are separated in two main clades b and c,one including the rest of American Foliosae species and the otherincluding the Old World and Caribbean species.

The earliest diverging species in group b (V. palmarum, V. lind-maniana and V. sp. CR0100) share some traits with the species ofgroup a, such as membranaceous to sub-membranaceous leavesor self-fertilization ability. However these species bear fragrantfruits like most of the remaining species of the group b gatheringAmerican species belonging to the Foliosae section and Lamellosaesubsection (Soto Arenas, 1999, 2003; Cameron, in press). Thereforethe aromatic properties of the fruit might be a synapomorphiccharacter specific to the monophyletic group b. Aromatic Vanillafruits are suggested to be an adaptation to fruit dispersal by bats(Soto Arenas, 1999, 2003) or to sticky seed dispersion by orchidsbees as observed recently for V. grandiflora (Lubinsky et al., 2006).

Sister to clade b is clade c, which includes all the remaining spe-cies from the Paleotropics and the leafless Caribbean species. Asianspecies from the Aphyllae and Foliosae section (cA) seem to be clo-ser to subgroup (cB) regrouping African Foliosae species of theLamellosae subsection such as V. africana and V. cf. crenulata. Thesetwo subgroups are sister to the last four subgroups, which differen-tiate each other gradually, with species of the Foliosae section andPapillosae subsection such as V. imperialis on one hand, and speciesof the Aphyllae section from Indian Ocean islands (V. roscheri, V.madagascariensis, V. phalaenopsis and V. humblotii), the Africanunidentified Foliosae species V. sp. CR0067 and the Caribbean spe-cies of the Aphyllae section such as V. dilloniana and V. barbellata onthe other hand.

Soto Arenas (2003) previously mentioned that the classificationof the section Foliosae into subsections was heterogeneous andincomplete. Moreover, leaflessness (a characteristic of the Aphyllaesection) has appeared at least three times in the evolutionary pat-tern of Vanilla species, in the Caribbean (cF), in the Indian Ocean(cD) and in Asia (cA). Their speciation originated from differentancestors belonging to the Foliosae section of the Old World clade(c). It could have resulted from an adaptation of ancestral Vanillaspecies to xeric conditions occurring in coastal and insular areasin the tropics. Although anatomical data suggested that the Aphyl-lae group was monophyletic (Stern and Judd, 1999), the well sup-ported tree obtained here with plastid DNA sequences confirmed

Epidendroideae V

anilleae

0.02 substitutions / site

V. crenulata CR0065

V. aphylla*

V. sp. CR0175

V. sp. CR0153

V. africana* V. crenulata CR0102

V. albida CR0058

V. africana CR0103

V. crenulata CR0091

Medicago truncatula*

Oenothera argillicola* Guizotia abyssinica*

Populus alba*

Cleistes divaricata* Pogonia japonica*

Pogonia minor *

Cleistes rosea*

Pogonia ophioglossoides*

Isotria verticillata* Isotria medeoloides*

Cleistes cipoana* Phalaenopsis aphrodite*

Duckeella adolphii*

Epistephium cf.*

Eriaxis rigida*

Pseudovanilla ponapensis*

Epistephium subrepens*

Erythrorchis altissima* Erythrorchis cassythoides*

Epistephium parviflorum*

Clematepistephium smilacifolium *

V. chamissonis CR0666

V. sp. CR0117

V. sp. CR0683

V. sp. CR0100V. lindmaniana CR0682

V. sp. CR0178

V. palmarum CR0083

V. leprieuri CR0109

V. ensifolia CR0174

V. sp. CR0665

V. mexicana *

V. odorata CR0686

V. sp. CR0079

V. cf. grandiflora CR0693

V. sp. CR0171

Hyb. CR0166

V. bahiana CR0098

V. sp. CR0087

V. pompona CR0169

Hyb. CR0003

V. sp. CR0068

V. sp. CR0060

V. planifolia CR0211

V. tahitensis CR0164

Hyb. CR0131

V. sp. CR0115

V. sp. CR0095

1

1

0,7

0.9

1

1

0,9

1

1

0,9

1

0,8

1

0,9

1

1

1

0,9

1

1

1

1

1

1

1

1

1

1

1

1

1

0,9

V. imperialis*

V. imperialis CR0104

V. cf. polylepis CR0705

V. sp. CR06981

10,9

0,8

1

1

1

1

1

V. sp. CR00810,9

V. sp. CR0067

V. sp CR0089

V. cf. barbellata*

V. sp. CR0699

V. madagascariensis CR0141

V. dilloniana CR0794

V. roscheri*

V. phalaenopsis CR0146

V. sp. CR0697

V. humblotii CR0108

V. roscheri CR0810

10,8

0.9

0,9

1

1

1

1

0,9

1

1 1

1

1

1

1

1

1

1

group α Membranaceous spp.

group β American

fragrant spp.

group γ Old World +

Caribbean spp.

A

B

C

D

E

F

G

H

A

B

C

D

E

F

Pogonieae

V a n i l l a

Orchidaceae

V. sp. CR0177

Vanilloideae

Eudicotyledons

Fig. 1. Bayesian analysis of the chloroplastic psaB, psbB, psbC and rbcL data set for the Vanilloideae. Consensus of 45,000 trees (burn-in of 5000 trees) generated after5,000,000 generations based on the GTR + I + C model of DNA substitution. Bayesian node support probabilities are indicated above the branches. Circle and triangle indicatesleafy and leafless Vanilla species, respectively. The geographic origin is specified by shading code: white for American taxa, black for African taxa and grey for Asian taxa.Asterisk (�) indicates sequences obtained from GenBank (see Table 3).

626 A. Bouetard et al. / Molecular Phylogenetics and Evolution 55 (2010) 621–630

Table 4Estimated divergence times derived from Bayesian relaxed clock analyses based on a combined analysis of four chloroplastic gene psaB, psbB, psbC and rbcL, testing the two mainhypothesis of radiation of the genus Vanilla. Mean and 95% HPD of the posterior probability distribution are in Mya.

Transoceanic dispersion hypothesis Gondwanan dispersion hypothesis71 Mya as prior on Vanilloidae node* 95 Mya as prior on OW/NW node*

Molecular clock model UEa ULb UEa ULb

Node Mean [95% HPD] Mean [95% HPD] Mean [95% HPD] Mean [95% HPD]

Orchidaceae 83.47 [68.4–104] 83.42 [69.9–99.5] 310.25 [187.5–455] 526.6 [283.9–786.8]Vanilloidae 70.75* 70.77* 264.65 [167.4–381] 445.78 [240.6–658.3]Pogonieae 48.96 [31–64.5] 50.09 [36–62.9] 189.22 [104.1–286] 310.09 [156.2–465.7]Galeolinae 9.87 [2.4–20.5] 9.75 [4–16.8] 36.05 [10.4–71.3] 60.38 [19.6–107.1]Vanilleae 54.26 [40.6–66.4] 48.87 [36.2–62.3] 201.16 [131.9–294.3] 300.3 [175.4–425.1]Vanilla 34.65 [20.9–48.6] 26.92 [16.3–38.4] 129.2 [97.6–176.2] 165.65 [107.6–235.2]Separation between New and Old World V. spp. 25.71 [14.7–37.8] 15.37 [9.2–22.7] 94.8* 94.79*

Separation between African and Asian V. spp. 10.29 [3.1–18.7] 7.25 [3.4–11.6] 37.43 [11.8–65.4] 44.23 [20.4–67.8]American Fragant V. spp. 9.84 [3.3–17.4] 5.3 [2.7–8.3] 36.65 [11.5–63.8] 32.46 [17.1–49.8]Aphyllae spp. from the Indian Ocean area 4.4 [0.3–10.6] 1.62 [0.4–3.3] 15.68 [1.6–37.8] 9.93 [2.4–20.3]Aphyllae spp. from the Caribbean area 3.53 [0.2–9.9] 1.72 [0.2–3.7] 12.32 [0.6–33.5] 10.61 [1.7–23.1]

An asterisk marks nodes constrained by fixed age according to the two hypothesis tested. HPD, high probability divergence.a Result obtained with the uncorrelated exponential relaxed molecular clock model.b Result obtained with the uncorrelated lognormal relaxed molecular clock model.

A. Bouetard et al. / Molecular Phylogenetics and Evolution 55 (2010) 621–630 627

that the Vanilla section Aphyllae is paraphyletic, in agreement withprevious molecular analyses (Soto Arenas, 2003). Soto Arenas(2003) even suggested that a fourth leafless habit could haveevolved independently within group b, with notably V. penicillata,discovered in the Amazonian forest in Venezuela although thisspecies was not included in the study presented. Consequently,both the Aphyllae and the Folosiae sections are not monophyletic,confirming that Portères’ classification needs a major revision.Morphological traits as well as biochemical characteristics (for fra-grant species) should be good tools to assist molecular data for anew nomenclature, but they cannot explain alone relationshipsamong species. Floral features are hypothesized to be especiallyprone to selective pressure from pollinators and, hence, are likelyto display high levels of convergence or parallelism (Bory et al.,2008b; Cameron, 1999). According to this study and results ob-tained by Soto Arenas (2003), three sections corresponding to theclades denoted a, b and c seem to be more appropriate to classifythe genus Vanilla. To be nomenclaturally valid, these three subdivi-sions of the genus Vanilla should include the generic name and pro-vide a typified Latin diagnosis (Soto Arenas, 2003).

4.2. Molecular dating and biogeography of the genus Vanilla

The understanding of the early evolutionary history of flower-ing plant families centered in the tropics may be aided by the inte-gration of evidence from biogeography, paleobotany, climatology,and molecular phylogeny (Zerega et al., 2005). In this study, wehave tested the two main hypotheses of dispersal of the genus Va-nilla. Since the dating method relies on a single orchid fossil andevolution rates can show heterogeneity through time or across treebranches (Magallon and Castillo, 2009), it is difficult to ascertainaccuracy of the nodes’ age estimates. However, given the largetime span separating the two hypotheses and the confidence inter-val of the node provided by BEAST software analysis, we couldchoose between the two scenarios.

According to the dating based on the hypothesis of a transcon-tinental migration of the genus Vanilla before the break up of theGondwana, the radiation of the Orchidaceae family would have be-gun at least 310 Mya during the carboniferous after the secondmassive extinction. Subsequently, the sub-family Vanilloideaeand the genus Vanilla would have appeared around 265 Mya (dur-ing Permian) and 129 Mya (during Cretaceous), respectively. Thesedates are to consider as minima because we calibrated the nodeseparating the New World species and the Old World species at

95 Mya which is a minimum assumption. Indeed, the age of theseparation between the South American and the African plates isestimated between 165 and 96 Mya (Cuenca et al., 2008; Morley,2003). Nevertheless, according to paleobotany and phylogeneticstudies, the origin of the Angiosperm has been estimated aroundthe Middle Jurassic-Early Cretaceous (Crane et al., 1995; Sunet al., 2002; Zerega et al., 2005) that is around 180–120 Mya.Therefore the hypothesis of an early radiation of the Vanilla genusby vicariance consecutive to the Gondwanan separation appearsincompatible with the currently admitted time span of Angiospermevolution.

Calibrating the analysis from the dating of Vanilloideae sub-family emergence 71 Mya as proposed, as a minimum, by Ramirezet al. (2007), we estimated the origin of the Orchidaceae family83 Mya (excluding Apostasioideae) in the late Cretaceous (Fig. 2).This estimation is close to that obtained by Ramirez et al. (2007),about 78 Mya, based on the age of fossilized Meliorchis caribea poli-narium. Under this scenario, the genus Vanilla appeared �34 Myain South America, and migrated to Africa �25 Mya during Oligo-cene. At that time landmasses were already well separated and atransoceanic migration is therefore necessary to explain migrationfrom tropical America to Africa. Under this scenario it is howeverquite surprising that the Vanilla species, so prone to transoceanicdispersion, remained absent from tropical Australia.

However it is important to keep in mind the following twopoints: (i) fossils calibration inevitably underestimate the trueage of divergence and (ii) there is growing evidences that thebreak-up of Gondwana might have happened more recently thantraditionally believed (around 70 Mya, during the late Cretaceous)(Van Bocxlaer et al., 2006). Some authors have also suggested theprolonged existence of a South America–Africa corridor in the earlyLate Cretaceous (Van Bocxlaer et al., 2006). The Gondwananhypothesis for the America to Africa migration of the Vanilla genusmight therefore prove to be correct but only under suchassumptions.

On the other hand, several long-distance transoceanic routes formega thermal angiosperms dispersion have been proposed thatmay have played a role in shaping the distribution of many taxaduring the early tertiary (Paleocene-Eocene), facilitated by thecombination of warm climates and relative proximity of the majorcontinent (Cuenca et al., 2008; Morley, 2000; Muellner et al., 2006;Pennington and Dick, 2004). The mid-oceanic Rio Grande Ridge,which would have been above sea level via series of islands untilthe Oligocene (Morley, 2000), could have allowed the passage of

120 110 100 90 80 70 60 50 40 30 20 10 0

Africa & South America

separate

Onset of abrupt global cooling

yraitreT suoecaterC

Early Late Pa Eo Ol Mi Pl

K/T boundary

-34,49

-23,93

-49,00

-59,05

-25,50

-46,79

-38,89

-83,43

-70,75

-106,75

-54,20 -9,67

Phalaenopsis aphrodite

Pogoniinae

Pseudovanilla ponapensis

Outgroup(eudicotyledons)

-19,08

-12,87

-4,60

-10,24 A

B

-3,59

-4,41

-14,41

-10,86

C

D

F

Galeolinae

-6,25

-3,54

-18,12

-13,64

-0,89

-0,75 -6,30

-6,58

-4,79

-9,80

A

B

C

D

E

F

G

H

group β

American Fragant spp.

group α Membranaceous spp.

group

Old World + Caribbean spp.

Vanilloideae*

Orchidaceae

Vanilleae

Vanilla

Closing of the Rio Grande Ridge and the North Antlantic Land

Bridge

My

Epistephium spp Eriaxis rigida Clemepistephium smilacifolium

Fig. 2. Chronogram of Vanilla divergence based on Bayesian analysis of the chloroplast gene psaB, psbB, psbC and rbcL. Posterior estimates of mean (in millions of year) and95% HPD of divergence times (presented in Table 4) were estimated using 71 Myr as the minimum ages of Vanilloideae published by Ramirez et al. (2007) for the subfamillyVanilloideae, the GTR + I + C model of DNA substitution and the uncorrelated exponential relaxed clock. Means of divergence times are indicated above branches. Division ofthe genus Vanilla in groups and subgroups is specified as in Fig. 1. Geological time scale is shown at bottom.

628 A. Bouetard et al. / Molecular Phylogenetics and Evolution 55 (2010) 621–630

species ancestral to groups b and c between South America andWestern Africa, at least 25 Mya according to the chronogram(Fig. 2). Another scenario could involve the Northern Atlantic LandBridge (NALB) that has also played a role in the formation of theAfrican flora from New World taxa, as suggested by phylogenetic

analyses (Davis et al., 2002; Lavin et al., 2000). After 20 Mya, theNALB was closed (Sanmartìn and Ronquist, 2001).

Whatever the calibration of our tree, and therefore whateverthe scenario retained for the early migration from tropical Americato Africa (Gondwanan vicariance or long distance dispersal), our

A. Bouetard et al. / Molecular Phylogenetics and Evolution 55 (2010) 621–630 629

results nevertheless demonstrate the occurrence of three long dis-tance dispersal events in the evolutionary history of the Vanillagenus.

Only transoceanic dispersion can explain the passage from Afri-ca to Asia. Indeed, the node segregating the Asian clade (cA) fromthe African clade (cB) was dated �37–10 Mya, according to thetwo hypotheses tested, well after the rifting of Madagascar–In-dia–Seychelles from Africa estimated by mid-Jurassic (�167 Mya)to early-Cretaceous (Aptian: 112–125 Mya) (McLoughlin, 2001;Morley, 2003; Zerega et al., 2005). The Asian clade seems to resultfrom the migration of a common ancestor of the African clade (cB)regrouping species of the Vanilla Foliosae section such as V. afri-cana. Although many African taxa failed to cross to Asia as the dis-persal path from Africa to India was likely to be a filter (Morley,2000), a largest sampling set of Asian taxa could reveal furthermore migration events from other African or Indian Ocean islandsclades. Two others transoceanic migrations giving rise to the Carib-bean and Indian Ocean islands leafless species from African ances-tors were dated about 12–3.5 Mya and 15–4.4 Mya, respectively.Results obtained for the speciation of the Caribbean leafless speciesindicate a second transatlantic dispersion event from Africa toCaribbean islands during the Pliocene. At that time the continentswere widely disconnected and no transoceanic ridge is suspected.Speciation of leafless species in the Indian Ocean islands from anAfrican ancestor happened also well after the rifting of Madagas-car–India–Seychelles from Africa, and is also too recent to fit withthe alternative hypothesis of the existence of a ‘‘central corridor”between Africa and Indo-Madagascar until around 90 Mya (Chater-jee and Scotese, 1999). Other means of transoceanic migrationshould be involved such as the dispersion of common ancestors(apparently close to the Vanilla sp. 0067) via floating vegetationmats following ocean currents and predominant wind directions(Renner, 2004; Zhang et al., 2007). Another explanation could in-volve migratory birds (Bory et al., 2008b). It was proposed that V.planifolia seeds could be endoornithochors with bird digestive flu-ids helping germination (Arditti and Ghani, 2000; Bouriquet,1954). These means of dispersion could have also been implicatedin the first transatlantic migration for the passage from islands toislands.

5. Conclusion

Combined analyses of the chloroplastic genes psbB, psbC; psaBand rbcL, have provided a fairly well supported reconstruction ofthe phylogeny of the Vanilloideae sub-family and more particu-larly of the genus Vanilla. This phylogeny segregates the genusinto three major clades and invalidates the current sub-genericclassification (Portères, 1954). Using recent Bayesian relaxedmolecular clock methods with two extreme calibrating dates,we conclude that at least three transoceanic migration eventsexplain the biogeographic pattern of the genus Vanilla; one fromAfrica to Asia, one from Africa to Indian Ocean islands and onefrom Africa to Caribbean islands. Furthermore, according to themost likely dating, based on the Vanilloideae age estimated byRamirez et al. (2007) and admitted age of Gondwana break-up,transoceanic migration from America to Africa remains the mostplausible explanation for the early radiation of the genus, sepa-rating groups b and c after Cretaceous (�25 Mya) using theexisting evidence that is available to date. However, with moreorchid fossil evidences, such as the Miocene orchids from NewZealand (Conran et al., 2009), greater sampling, new methodsof molecular age estimations and a better understanding of con-tinental drift, resolving the age and origin of Vanilla and orchidsin general will, with no doubt, continue to be an exciting area offurther investigations.

Acknowledgments

This research was funded by the following institutions: ConseilRégional of Reunion Island (program Regenda), the CIRAD and theUMR PVBMT of the University of Reunion Island, the European Un-ion (FEDER) and GIS Centre de recherche et de veille sanitaire surles maladies émergentes dans l’océan Indien (N�PRAO/AIRD/CRVOI/08/03). Special thanks to Delphine Ramalingom for guidingus in using the calculation platform TITAN of the University of Re-union Island, to Claire Micheneau for her advices on parsimonyanalysis, to Jean Bernard Dijoux and Katia Jade for their technicalsupport.

The authors would like to thank Dr. Kenneth Cameron for hisconstructive review and helpful comments for improving themanuscript.

References

Arditti, J., Ghani, A.K.A., 2000. Numerical and physical properties of orchid seeds andtheir biological implications. New Phytologist 145, 367–421.

Bory, S., Da Silva, D., Risterucci, A.M., Grisoni, M., Besse, P., Duval, M.F., 2008a.Development of microsatellite markers in cultivated vanilla: polymorphism andtransferability to other Vanilla species. Scientia Horticulturae 115, 420–425.

Bory, S., Grisoni, M., Duval, M.F., Besse, P., 2008b. Biodiversity and preservation ofvanilla: present state of knowledge. Genetic Resources and Crop Evolution 55,551–571.

Bory, S., Lubinsky, P., Risterucci, A.M., Noyer, J.L., Grisoni, M., Duval, M.-F., Besse, P.,2008c. Patterns of introduction and diversification of Vanilla planifolia(Orchidaceae) in Reunion Island (Indian Ocean). American Journal of Botany95, 805–815.

Bouriquet, G., 1954. Le vanillier et la vanille dans le monde. Encyclopédie BiologiqueXLVI. Lechevalier, Paris VI.

Cameron, K.M., 1999. Biogeography of Vanilloideae (Orchidaceae). In: 16thInternational Botanical Congress, 1–7 August 1999, St. Louis, Missouri(abstract).

Cameron, K.M., 2000. Gondwanan biogeography of Vanilloideae (Orchidaceae). 3rdSouthern Connection Congress, 17–22 January 2000, Canterbury, New Zealand(abstract).

Cameron, K.M., 2004. Utility of plastid psaB gene sequences for investigatingintrafamilial relationships within Orchidaceae. Molecular Phylogenetic andEvolution 31, 1157–1180.

Cameron, K.M., 2005. Recent advances in the systematic biology of Vanilla andrelated orchids (Orchidaceae: subfamily Vanilloideae). In: 1st InternationalVanilla Congress, 11–12 November 2003, Princeton, NJ, USA (Poster).

Cameron, K.M., 2009. On the value of nuclear and mitochondrial gene sequences forreconstructing the phylogeny of vanilloid orchids (Vanilloideae, Orchidaceae).Annals of Botany 104, 377–385.

Cameron, K.M., in press. Vanilloid orchids systematics and evolution. In: Odoux, E.,Grisoni, M. (Eds.), Vanilla. Medicinal and Aromatic Plants – Industrial profiles.Taylor and Francis, NY.

Cameron, K.M., Chase, M.W., 1999. Phylogenetic relationships of Pogoniiae(Vanilloideae, Orchidaceae): an herbaceous example of the eastern northAmerica–eastern Asia phylogeographic disjunction. Journal of Plant Research112, 317–329.

Cameron, K.M., Chase, M.W., 2000. Nuclear 18S rDNA sequences of Orchidaceaeconfirm the subfamilial status and circumscription of Vanilloideae. In: Wilson,K.L., Morrison, D.A. (Eds.), Monocots, Systematics and Evolution. CSIRO,Collingwood, Victoria, Australia, pp. 457–464.

Cameron, K.M., Chase, M.W., Whitten, W.M., Kores, P.J., Jarrell, D.C., Albert, V.A.,Yukawa, T., Hills, H.G., Goldman, D.H., 1999. A phylogenetic analysis of theOrchidaceae: evidence from rbcL nucleotide sequences. American Journal ofBotany 86, 208–224.

Cameron, K.M., Molina, M.C., 2006. Photosystem II gene sequences of psbB and psbCclarify the phylogenetic position of Vanilla (Vanilloideae, Orchidaceae).Cladistics 22, 239–248.

Chang, C.C., Lin, H.C., Lin, I.P., Chow, T.Y., Chen, H.H., Chen, W.H., Cheng, C.H., Lin,C.Y., Liu, S.M., Chang, C.C., Chaw, S.M., 2006. The chloroplast genome ofPhalaenopsis aphrodite (Orchidaceae): comparative analysis of evolutionaryrate with that of grasses and its phylogenetic implications. Molecular Biologyand Evolution 23, 279–291.

Chaterjee, S., Scotese, C.R., 1999. The break-up of Gondwana and the evolution andbiogeography of the Indian plate. Proceedings of the Indian National ScienceAcademy A 65, 397–425.

Clegg, M.T., Zurawski, G., 1992. Chloroplast DNA and the study of plant phylogeny:present status and future prospects. In: Soltis, P., Soltis, D., Doyle, J. (Eds.),Molecular Systematics of Plants. Chapman and Hall, New York, NY.

Conran, J.G., Bannister, J.M., Lee, D.E., 2009. Earliest orchid macrofossils: earlyMiocene Dendrobium and Earina (Orchidaceae: Epidendroideae) from NewZealand. American Journal of Botany 96, 466–474.

630 A. Bouetard et al. / Molecular Phylogenetics and Evolution 55 (2010) 621–630

Crane, P.R., Friis, E.M., Pedersen, K.R., 1995. The origin and early diversification ofangiosperms. Nature 374, 27–33.

Cuenca, A., Asmussen-Lange, C.B., Borchsenius, F., 2008. A dated phylogeny of thepalm tribe Chamaedoreeae supports Eocene dispersal between Africa, Northand South America. Molecular Phylogenetics and Evolution 46, 760–775.

Davis, C.C., Anderson, W.R., Donoghue, M.J., 2001. Phylogeny of Malpighiaceae:evidence from chloroplast ndhF and trnl-F nucleotide sequences. AmericanJournal of Botany 88, 1830–1846.

Davis, C.C., Bell, C.D., Mathews, S., Donoghue, M.J., 2002. Laurasian migrationexplains Gondwanan disjunctions: evidence from Malpighiaceae. Proceedingsof the National Academy of Sciences USA 99, 6833–6837.

Doyle, J.A., Thomas, A.L., 1997. Phylogeny and geographic history of Annonaceae.Géographie Physique et Quaternaire 51, 353–361.

Drummond, A.J., Ho, S.Y.W., Phillips, M.J., Rambaut, A., 2006. Relaxed phylogeneticsand dating with confidence. PLoS Biology 4, e88.

Freudenstein, J.V., van den Berg, C., Goldman, D.H., Kores, P.J., Molvray, M., Chase,M.W., 2004. An expanded plastid DNA phylogeny of Orchidaceae and analysis ofjackknife branch support strategy. American Journal of Botany 91, 149–157.

Greiner, S., Wang, X., Herrmann, R.G., Rauwolf, U., Mayer, K., Haberer, G., Meurer, J.,2008. The complete nucleotide sequences of the 5 genetically distinct plastidgenomes of Oenothera, subsection Oenothera: II. A microevolutionary viewusing bioinformatics and formal genetic data. Molecular Biology and Evolution25 (9), 2019–2030.

Grisoni, M., Moles, M., Besse, P., Bory, S., Duval, M.-F., Kahane, R., 2007. Towards aninternational plant collection to maintain and characterize the endangeredgenetic resources of vanilla. Acta Horticulturae (ISHS) 760, 83–91.

Guindon, S., Gascuel, O., 2003. A simple, fast, and accurate algorithm to estimatelarge phylogenies by maximum likelihood. Systems biology 52, 696–704.

Hall, T.A., 1999. BIOEDIT: a user-friendly biological sequence alignment editor andanalysis program for Windows 95/98/NT. Nucleic Acids Symp. Ser. 41, 95–98.

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

Kumar, S., Tamura, K., Nei, M., 2004. MEGA3: integrated software for MolecularEvolutionary Genetics Analysis and sequence alignment. Briefings inBioinformatics 5, 150–163.

Lavin, M., Thulin, M., Labat, J.-N., Pennington, R.T., 2000. Africa, the odd man out:molecular biogeography of Dalbergioid Legumes (Fabaceae) suggests otherwise.Systematic Botany 25, 449–467.

Lubinsky, P., Van Dam, M., Van Dam, A., 2006. Pollination of Vanilla and evolution inOrchidaceae. Orchids 75, 926–929.

Magallon, S., Castillo, A., 2009. Angiosperm diversification through time. AmericanJournal of Botany 96. doi:10.3732/ajb.0800060.

McLoughlin, S., 2001. The break up history of Gondwana and its impact on pre-Cenozoic floristic provincialism. Australian Journal of Botany 49, 271–300.

Morley, R.J., 2000. Origin and evolution of tropical rain forests. John Wiley and Sons,Chichester, UK, 378p.

Morley, R.J., 2003. Interplate dispersal paths for megathermal angiosperms.Perspectives in Plant Ecology, Evolution and Systematics 6, 5–20.

Morley, R.J., Dick, C.W., 2003. Missing fossils, molecular clocks, and the origin of theMelastomataceae. American Journal of Botany 90, 1638–1644.

Muellner, A.N., Samuel, R., Chase, M.W., Pannell, C.M., Greger, H., 2005. Aglaia(Meliaceae): an evaluation of taxonomic concepts based on DNA data andsecondary metabolites. American Journal of Botany 92, 534–543.

Muellner, A.N., Savolainen, V., Samuel, R., Chase, M.W., 2006. The mahogany family‘‘out-of-Africa”: divergence time estimation, global biogeographic patternsinferred from plastid rbcL DNA sequences, extant, and fossil distribution ofdiversity. Molecular Phylogenetics and Evolution 40, 236–250.

Okumura, S., Sawada, M., Park, Y.W., Hayashi, T., Shimamura, M., Takase, H.,Tomizawa, K., 2006. Transformation of poplar (Populus alba) plastids andexpression of foreign proteins in tree chloroplasts. Transgenic Research 15,637–646.

Olmstead, R., Palmer, J., 1994. Chloroplast DNA systematics: a review of methodsand data analysis. American Journal of Botany 81, 1205–1224.

Palmer, J., Jansen, R., Michaels, H., Chase, M., Manhart, J., 1988. Chloroplast DNAvariation and plant phylogeny. Annals of the Missouri Botanical Garden 75.

Pennington, R.T., Dick, C.W., 2004. The role of immigrants in the assembly of theSouth American rainforest tree flora. Philosophical Transactions of the RoyalSociety B: Biological Sciences 359, 1611–1622.

Portères, R., 1954. Le genre Vanilla et ses espèces. In: Bouriquet, G. (Ed.), Le vanillieret la vanille dans le monde, Paul Lechevailer, Paris, pp. 94–290.

Rambaut, R., Drummond, A.J., 2004. Tracer v1.3. Available from: http://evolve.zoo.ox.ac.uk/software.html.

Ramirez, S.R., Gravendeel, B., Singer, R.B., Marshall, C.R., Pierce, N.E., 2007. Datingthe origin of the Orchidaceae from a fossil orchid with its pollinator. Nature 448,1042–1045.

Renner, S.S., 2004. Multiple Miocene Melastomataceae dispersal betweenMadagascar, Africa and India. Philosophical Transactions of the Royal SocietyB: Biological Sciences 359, 1485–1494.

Rolfe, R.A., 1896. A revision of the genus Vanilla. Journal of the Linnean Society 32,439–478.

Rozas, J., Sánchez-DelBarrio, J.C., Messeguer, X., Rozas, R., 2003. DnaSP, DNApolymorphism analyses by the coalescent and other methods. Bioinformatics19, 2496–2497.

Rozen, S., Skaletsky, H.J., 2000. PRIMER3 on the WWW for general users and forbiologist programmers. In: Krawetz, S., Misener, S. (Eds.), BioinformaticsMethods and Protocols: Methods in Molecular Biology. Humana Press,Totowa, NJ, pp. 365–386. Source code available from: http://fokker.wi.mit.edu/primer3/.

Sanmartìn, I., Ronquist, H.E.F., 2001. Patterns of animal dispersal, vicariance anddiversification in the Holarctic. Biological Journal of the Linnean Society 73,345–390.

Shimodaira, H., Hasegawa, M., 1999. Multiple comparisons of log-likelihoods withapplications to phylogenetic inference. Molecular Biology and Evolution 16,1114–1116.

Soto Arenas, M.A., 1999. Filogeografia y recursos genéticos de las vainillas deMéxico, Herbario de la Asociación Mexicana de Orquideología, México.

Soto Arenas, M.A., 2003. Vanilla. In: Pridgeon, A.M., Cribb, P.J., Chase, M.W.,Rasmussen, F.N. (Eds.), Genera Orchidacearum: Orchidoideae. Oxford UniversityPress, USA, pp. 321–334.

Steel, M., McKenzie, A., 2001. Properties of phylogenetic trees generated by Yule-type speciation models. Mathematical Biosciences 170, 91–112.

Stern, W.L., Judd, W.S., 1999. Comparative vegetative anatomy and systematics ofVanilla (Orchidaceae). Botanical journal of the Linnean Society 131, 353–382.

Sun, G., Ji, Q., Dilcher, D.L., Zheng, S., Nixon, K.C., Wang, X., 2002. Archaefructaceae, anew basal Angiosperm family. Science 296, 899–904.

Thompson, J.D., Higgins, D.G., Gibson, T.J., 1994. CLUSTAL W: improving thesensitivity of progressive multiple sequence alignment through sequenceweighting, positions-specific gap penalties and weight matrix choice. NucleicAcids Research 22, 4673–4680.

Van Bocxlaer, I., Roelants, K., Biju, S.D., Nagaraju, J., Bossuyt, F., 2006. LateCretaceous vicariance in Gondwanan amphibians. PLoS One 1, e74.

Von Balthazar, M., Pedersen, K.R., Crane, P.R., Stampanoni, M., Friis, E.M., 2007.Potomacanthus lobatus gen. et sp. nov., a new flower of probable Lauraceae fromthe Early Cretaceous (Early to Middle Albian) of eastern North America.American Journal of Botany 94, 2041–2053.

Zerega, N.J.C., Clement, W.L., Datwyler, S.L., Weiblen, G.D., 2005. Biogeography anddivergence times in the mulberry family (Moraceae). Molecular Phylogeneticsand Evolution 37, 402–416.

Zhang, L.-B., Simmons, M.P., Renner, S.S., 2007. A phylogeny of Anisophylleaceaebased on six nuclear and plastid loci: ancient disjunctions and recent dispersalbetween South America, Africa, and Asia. Molecular Phylogenetics andEvolution 44, 1057–1067.