CLADES, CLOCKS, AND CONTINENTS: HISTORICAL AND BIOGEOGRAPHICAL ANALYSIS OF MYRTACEAE, VOCHYSIACEAE, AND RELATIVES IN THE SOUTHERN HEMISPHERE Kenneth J. Sytsma, 1, * Amy Litt, y Michelle L. Zjhra, z J. Chris Pires,§ Molly Nepokroeff, k Elena Conti,# Jay Walker,* and Peter G. Wilson** *Department of Botany, University of Wisconsin, Madison, Wisconsin 53706, U.S.A.; yDepartment of Molecular, Cellular, and Developmental Biology, Yale University, New Haven, Connecticut 06520, U.S.A.; zDepartment of Biology, Georgia Southern University, Statesboro, Georgia, U.S.A.; §Department of Agronomy, University of Wisconsin, Madison, Wisconsin 53706, U.S.A.; k Department of Biology, University of South Dakota, Vermillion, South Dakota 57069, U.S.A.; #Institute of Systematic Botany, University of Zurich, Zurich, Switzerland; and **National Herbarium of New South Wales, Royal Botanic Gardens, Sydney, New South Wales 2000, Australia Some of the most interesting but still most contentious disjunct biogeographical distributions involve Southern Hemisphere tropical and warm temperate families. The PHMV clade of Myrtales includes four families (Psiloxylaceae, Heteropyxidaceae, Myrtaceae, and Vochysiaceae) that exhibit a number of these biogeographical patterns. The related Psiloxylaceae and Heteropyxidaceae are small families restricted in distribution to the recent volcanic Mascarene Islands to the east of Madagascar and to southeast Africa, respectively. Myrtaceae are found on three major Gondwanan regions (South America, Australasia, and Africa). Because the New World taxa are almost exclusively fleshy fruited, it is unclear whether the family distribution is a classic Gondwanan vicariance pattern or results from one or more long-distance dispersal events over ocean barriers. The Vochysiaceae represent one of a handful of families with amphi-Atlantic distributions vigorously argued to support both long-distance dispersal over the Atlantic and vicariance of western Gondwanan biota by Atlantic seafloor spreading. Molecular phylogenetic relationships, fossil dating of nodes, and penalized likelihood rate smoothing of maximum likelihood trees were employed for a Myrtales- wide analysis using rbcL and ndhF and an analysis of the PHMV analysis using ndhF and matK. The results indicate that the PHMV differentiated during the late Cretaceous. The African lineage of Vochysiaceae is nested within a South American clade and probably arose via long-distance dispersal in the Oligocene at a time when the Atlantic had already rifted 80 m.yr. at the equatorial region. The African/Mascarene Island families, most closely related to Myrtaceae, differentiated during the late Eocene, with subsequent but recent long- distance dispersal from Africa to the Mascarenes. Myrtaceae show a rapid differentiation of a basal, paraphyletic subf. Leptospermoideae in Australasia. Fleshy-fruited taxa (subf. Myrtoideae) are not monophyletic. Vicariance of a widespread warm temperate Southern Hemisphere distribution is likely in explaining the South American–Australasian disjunction, with subsequent dispersal events between the two and to Africa and the Mediterranean basin. Keywords: biogeography, dispersal, Heteropyxidaceae, matK, molecular clock, Myrtaceae, ndhF , phyloge- netics, Psiloxylaceae, rbcL, vicariance, Vochysiaceae. Introduction The distribution patterns and implied biogeographical events of southern hemisphere organisms have been vigor- ously studied and debated from the late 1960s since the in- corporation of continental drift ideas into biogeography. The breakup of Gondwana starting in the early Cretaceous has provided a strong vicariance focal point (Cracraft 1988) for explaining present-day disjunct patterns in early mam- mals (Murphy et al. 2001), passerine birds (Edwards and Boles 2002), flightless ratites (Haddrath and Baker 2001), Ar- aucaria beetles (Sequeira and Farrell 2001), and a number of angiosperm clades (Raven and Axelrod 1974; Manos 1997; Renner et al. 2000; Swenson et al. 2001; Vinnersten and Bremer 2001; Bremer 2002). In many of these examples, how- ever, evidence for more recent intercontinental dispersal in at least some subclades of these southern hemisphere groups has been presented (Haddrath and Baker 2001; Vinnersten and Bremer 2001; Bremer 2002). Further complicating the study of southern continent biogeography are the distinct separation times for southern temperate compared with tropical biota because of such factors as the counterclockwise rotation of Africa while separating from South America (Pittman et al. 1993), bridging of temperate flora of South America and Aus- tralia via Antarctica until ca. 50 Ma (Berry et al. 1990, 2004; Sytsma et al. 1991; Hallam 1994), the early separation (ca. 80 Ma) of New Zealand with apparently repeated disper- sal events from Australia (Winkworth et al. 2002), and the long isolation of India (Conti et al. 2002). 1 Author for correspondence; e-mail [email protected]. Manuscript received August 2003; revised manuscript received March 2004. S85 Int. J. Plant Sci. 165(4 Suppl.):S85–S105. 2004. Ó 2004 by The University of Chicago. All rights reserved. 1058-5893/2004/1650S4-0007$15.00
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
CLADES, CLOCKS, AND CONTINENTS: HISTORICAL AND BIOGEOGRAPHICALANALYSIS OF MYRTACEAE, VOCHYSIACEAE, AND RELATIVES
IN THE SOUTHERN HEMISPHERE
Kenneth J. Sytsma,1,* Amy Litt,y Michelle L. Zjhra,z J. Chris Pires,§ Molly Nepokroeff,k Elena Conti,#Jay Walker,* and Peter G. Wilson**
*Department of Botany, University of Wisconsin, Madison, Wisconsin 53706, U.S.A.; yDepartment of Molecular, Cellular, and DevelopmentalBiology, Yale University, New Haven, Connecticut 06520, U.S.A.; zDepartment of Biology, Georgia Southern University, Statesboro,
Georgia, U.S.A.; §Department of Agronomy, University of Wisconsin, Madison, Wisconsin 53706, U.S.A.; kDepartment ofBiology, University of South Dakota, Vermillion, South Dakota 57069, U.S.A.; #Institute of Systematic Botany,
University of Zurich, Zurich, Switzerland; and **National Herbarium of New South Wales,Royal Botanic Gardens, Sydney, New South Wales 2000, Australia
Some of the most interesting but still most contentious disjunct biogeographical distributions involveSouthern Hemisphere tropical and warm temperate families. The PHMV clade of Myrtales includes fourfamilies (Psiloxylaceae, Heteropyxidaceae, Myrtaceae, and Vochysiaceae) that exhibit a number of thesebiogeographical patterns. The related Psiloxylaceae and Heteropyxidaceae are small families restricted indistribution to the recent volcanic Mascarene Islands to the east of Madagascar and to southeast Africa,respectively. Myrtaceae are found on three major Gondwanan regions (South America, Australasia, andAfrica). Because the New World taxa are almost exclusively fleshy fruited, it is unclear whether the familydistribution is a classic Gondwanan vicariance pattern or results from one or more long-distance dispersalevents over ocean barriers. The Vochysiaceae represent one of a handful of families with amphi-Atlanticdistributions vigorously argued to support both long-distance dispersal over the Atlantic and vicariance ofwestern Gondwanan biota by Atlantic seafloor spreading. Molecular phylogenetic relationships, fossil datingof nodes, and penalized likelihood rate smoothing of maximum likelihood trees were employed for a Myrtales-wide analysis using rbcL and ndhF and an analysis of the PHMV analysis using ndhF and matK. The resultsindicate that the PHMV differentiated during the late Cretaceous. The African lineage of Vochysiaceae isnested within a South American clade and probably arose via long-distance dispersal in the Oligocene at a timewhen the Atlantic had already rifted 80 m.yr. at the equatorial region. The African/Mascarene Island families,most closely related to Myrtaceae, differentiated during the late Eocene, with subsequent but recent long-distance dispersal from Africa to the Mascarenes. Myrtaceae show a rapid differentiation of a basal,paraphyletic subf. Leptospermoideae in Australasia. Fleshy-fruited taxa (subf. Myrtoideae) are notmonophyletic. Vicariance of a widespread warm temperate Southern Hemisphere distribution is likely inexplaining the South American–Australasian disjunction, with subsequent dispersal events between the twoand to Africa and the Mediterranean basin.
The distribution patterns and implied biogeographicalevents of southern hemisphere organisms have been vigor-ously studied and debated from the late 1960s since the in-corporation of continental drift ideas into biogeography.The breakup of Gondwana starting in the early Cretaceoushas provided a strong vicariance focal point (Cracraft 1988)for explaining present-day disjunct patterns in early mam-mals (Murphy et al. 2001), passerine birds (Edwards andBoles 2002), flightless ratites (Haddrath and Baker 2001), Ar-aucaria beetles (Sequeira and Farrell 2001), and a number ofangiosperm clades (Raven and Axelrod 1974; Manos 1997;
Renner et al. 2000; Swenson et al. 2001; Vinnersten andBremer 2001; Bremer 2002). In many of these examples, how-ever, evidence for more recent intercontinental dispersal in atleast some subclades of these southern hemisphere groups hasbeen presented (Haddrath and Baker 2001; Vinnersten andBremer 2001; Bremer 2002). Further complicating the studyof southern continent biogeography are the distinct separationtimes for southern temperate compared with tropical biotabecause of such factors as the counterclockwise rotation ofAfrica while separating from South America (Pittman et al.1993), bridging of temperate flora of South America and Aus-tralia via Antarctica until ca. 50 Ma (Berry et al. 1990, 2004;Sytsma et al. 1991; Hallam 1994), the early separation (ca.80 Ma) of New Zealand with apparently repeated disper-sal events from Australia (Winkworth et al. 2002), and thelong isolation of India (Conti et al. 2002).
Manuscript received August 2003; revised manuscript received March 2004.
S85
Int. J. Plant Sci. 165(4 Suppl.):S85–S105. 2004.
� 2004 by The University of Chicago. All rights reserved.
1058-5893/2004/1650S4-0007$15.00
Vicariance as the sole explanation for disjunct patterns insouthern hemisphere groups not only has been tempered, asevident in some of the examples already cited, but has beenruled out in others. These include Tertiary migration andlong-distance dispersal for the pantropical Melastomataceae(Renner et al. 2001), the Laurasian origin and subsequentsouthern hemisphere dispersal of Lauraceae (Chanderbaliet al. 2001), and the recent long-distance dispersal in theamphi-Atlantic distributed Bromeliaceae and Rapateaceae(Givnish et al. 2000, 2004). The latter families are only twoexamples of the distinctive disjunction between South Amer-ica and Africa involving ca. 12 families (including Cactaceae,Humiriaceae, Mayacaceae, and Vochysiaceae). The relation-ships among these families of tropical South America andwest Africa have been the subject of much systematic andbiogeographic discussion before any recent sequence-basedbiogeographical analyses (Thorne 1972, 1973; Brenan 1978;Goldblatt 1978, 1993; Renner 2004). The vicariance versusdispersalist stances in explaining these amphi-Atlantic dis-junctions arose decades ago and have been largely untestedsince. Axelrod (1970, p. 293) suggested that the shared, dis-junct African–South American families indicate ‘‘the splittingapart of a more homogeneous flora by fragmentation ofGondwana.’’ In contrast, most others (Thorne 1972, 1973;Smith 1973; Barthlott 1983) strongly argued for more recentover-water dispersal to account for many, if not all, of thesedisjunctions.The issues involving vicariance of widespread taxa with
subsequent erection of barriers or dispersal of taxa acrosspreexisting barriers, however, cannot be satisfactorily ad-dressed unless three items are known: robust estimates ofphylogenetic relationships, ages of relevant clade formation,and a geological time sequence of barrier formation. Here wehave attempted with DNA sequence-based phylogenetic anal-yses, fossil dating of nodes, and molecular clock calibrationsto unravel the biogeographic history of a clade of rosid an-giosperms in Myrtales that exhibits many of these southernhemisphere disjunct patterns (fig. 1). This clade (hereafter re-ferred to as the PHMV clade) comprises two small families,Psiloxylaceae and Heteropyxidaceae, restricted to the Mas-carene Islands and southeast Africa, respectively, the diverseand largely South American and Australian Myrtaceae, andthe moderate-sized, amphi-Atlantic disjunct Vochysiaceae.The Myrtaceae represent one of the most striking ecologi-
cal and taxic radiations of the Gondwanan flora. Myrtaceaecomprise ca. 130–140 genera and nearly 4000 species and,with their not-too-distant relative the Melastomataceae, rep-resent the majority of species in the order Myrtales. Thefamily is widespread in the southern hemisphere, with con-centrations in the Neotropics and Australia, is somewhatrarer in the Australasian tropics, and is very rare in the Afri-can tropics. A single genus (Myrtus) occurs in the Mediterra-nean region. Table 1 shows the latest division of the familybased on Johnson and Briggs (1984). The traditional two-subfamily classification, the capsular-fruited Leptospermoi-deae and the indehiscent-fruited Myrtoideae, is maintainedhere, although recent cladistic analyses of morphological andmolecular data have demonstrated paraphyly of the Lepto-spermoideae (Wilson et al. 2001). Informally recognized ‘‘al-liances’’ and ‘‘suballiances’’ are used, although these also
have been reordered as new information has accumulated(Wilson et al. 2001). Of the two roughly equal subfamilies,the indehiscent-fruited Myrtoideae show a number of geo-graphical connections between the Old and New Worlds butare almost entirely unrepresented on very nutrient-deficientsites in Australia and southern Africa. Two genera, Eugeniaand Syzygium, occur in tropical Africa and vicinity. Withinthe Myrtoideae, all genera are endemic to their respectivegeographical areas, with the exception of Eugenia and Syzyg-ium. The dry-fruited Leptospermoideae are represented inthe New World (Tepaulia) and Africa (‘‘Metrosideros’’ angus-tifolia) by only one endemic genus each, their diversificationbeing largely Australasian.The remainder of the PHMV clade consists of Hetero-
pyxidaceae (three dioecious species) from southeast Africapossessing dry fruits; the monotypic, dioecious, but fleshy-fruited Psiloxylaceae from the Mascarenes, a volcanic archi-pelago 800 km east of Madagascar that is up to 8 m.yr. oldand is linked floristically to Africa and Madagascar; and thelargely Neotropical Vochysiaceae but with two genera en-demic to tropical west Africa (fig. 1). Vochysiaceae is of con-siderable interest not only because of its biogeographicaldistributional pattern that it shares with 11 other tropicalfamilies restricted to the Neotropics and the African tropicsbut also because the flowers are strikingly different from itsclosely related families. The flowers are zygomorphic witha single fertile stamen and with reduction seen in petal num-ber from five down to three or even one. The tribe Vochy-sieae have trilocular ovaries forming loculicidal capsules, and
Fig. 1 Current distribution of Myrtaceae, Vochysiaceae, Hetero-
pyxidaceae, and Psiloxylaceae (modified after Heywood 1993). The
occurrences of Myrtaceae and Vochysiaceae in Africa are due to only
three and two genera, respectively.
S86 INTERNATIONAL JOURNAL OF PLANT SCIENCES
the tribe Erismeae, including one Neotropical genus and bothAfrican genera, have unilocular ovaries forming indehiscent,samara-like fruits (Litt 1999; Litt and Cheek 2002).Earlier molecular evidence based solely on rbcL placed
these four families in a well-supported group (Conti et al.1996, 1997), although relationships of the four, except forthe strong support for the pair of African/Mascarene Islandfamilies (Heteropyxidaceae and Psiloxylaceae), were weak.This Myrtaceae group is sister to a clade of Melastomataceaein the broad sense and a biogeographically disjunct lineageof five small families referred to as the CAROP (Crypteronia-ceae, Alzateaceae, Rhynchocalycaeae, Oliniaceae, Penaea-ceae) lineage (Conti et al. 2002).The PHMV clade thus poses a number of interesting bio-
geographic, evolutionary, and ecological issues. For the pur-pose of this article, the issues are distilled to threeinterrelated problems: (1) What are the phylogenetic relation-ships within the PHMV clade (relationships of the four fami-lies found on three southern continents, relationships of theNew vs. Old World Myrtaceae, relationships of African vs.American members of Vochysiaceae)? (2) What are the agesof the various splits in the group (separation of African Het-eropyxidaceae and Mascarene endemic Psiloxylaceae, separa-tion of American vs. Old World Myrtaceae, timing of theAmerican-African disjunction within Vochysiaceae)? and (3)Are these major separations the result of vicariance of south-ern continents and/or more recent over-water dispersal?We employ here a nested, two-stage survey of phylogenetic
relationships and biogeographical history within the PHMVclade using maximum likelihood (ML) and penalized likeli-hood (PL) to examine these questions. The first is a broadsurvey of Myrtales using rbcL and ndhF sequences to exam-ine the placement of the PHMV clade in the Myrtales; thesecond survey focuses on the PHMV clade using ndhF andmatK sequences, both considerably faster-evolving than rbcL.Likelihood ratio tests of sequence evolution, penalized likeli-
hood analyses, and clock calibration with fossils of Myrta-ceae and other Myrtales were subsequently done to addressthe issue of date calibration of the resulting tree nodes andinference of biogeographical events, whether vicariance ordispersal. Presented elsewhere will be a full phylogeneticanalysis of the Myrtaceae using additionally maximum (MP)and Bayesian analyses and resulting conclusions on systemat-ics, evolution of morphological characters, and ecologicalspecializations (K. J. Sytsma, M. L. Zjhra, J. C. Pires, M. Nepo-kroeff, J. Walker, C. J. Quinn, and P. G. Wilson, unpublishedmanuscript).
Material and Methods
Sampling Design
The survey of all Myrtales using rbcLþ ndhF included 79ingroup plus four outgroup species (appendix). As this analy-sis was intended to place the PHMV clade within the orderMyrtales and to provide a ML framework under clock as-sumptions in order to date fossils outside the PHMV clade,sampling was more intensive in other families: 13 from thePHMV clade, 34 from Melastomataceae s.l., five from theCAROP clade, 17 from Onagraceae, six from Lythraceae,and four from Combretaceae. Most larger-scale phylogeneticstudies of rosids place Myrtales either isolated in Rosids (Sa-volainen et al. 2000; Soltis et al. 2000) or most closely re-lated to Malvales, Sapindales, and Brassicales (Conti et al.1996, 1997). The four outgroups selected came from Mal-vales and Brassicales. The majority of the rbcL sequencescame from previous studies (Conti et al. 1996; Clausing andRenner 2001). The ndhF sequences are new except for themajority of Melastomataceae (Clausing and Renner 2001;Renner and Meyer 2001; Renner et al. 2001).The ndhF þmatK analysis of the PHMV clade included
65 species of Myrtaceae representing all alliances and
Table 1
Biogeography of Myrtaceae Using an Informal Two-Subfamily Classification
Total Number of Genera : Number of Endemic Genera
Central and
South America Australasia Africa Mediterranean Total
Sources. Modified after Briggs and Johnson (1979) and Johnson and Briggs (1984).
Note. These ‘‘alliances’’ are now superseded by informal ‘‘groups’’ based on molecular results given in Wilson et al.
(2001), used in figure 5.
S87SYTSMA ET AL.—PHYLOGENETICS AND BIOGEOGRAPHY OF MYRTACEAE AND VOCHYSIACEAE
suballiances of Johnson and Briggs (1984), one species eachof Heteropyxidaceae and Psiloxylaceae, and seven species ofVochysiaceae, representing all American and African generaexcept for the recently discovered Korupodendron from westAfrica (Litt and Cheek 2002). The sister group to the PHMVis clearly Melastomataceae s.l. plus the CAROP clade (Contiet al. 1996). Because of long branches (and resulting prob-lems in clock tests) seen generally in Melastomataceae s.l. rel-ative to either the CAROP or PHMV clades based on thefirst survey and previous studies (Renner et al. 2001), Rhyn-chocalyx of the CAROP group was used as outgroup. All butthree ndhF sequences are new; 26 Myrtaceae and all Vochy-siaceae matK sequences are new. Different species were se-quenced for rbcL and ndhF in two genera, Mouriri(Memecylaceae) and Tococa (Melastomataceae), and forndhF and matK in four genera, Psidium and Xanthomyrtus(Myrtaceae) and Callisthene and Qualea (Vochysiaceae), buteach pair was treated as one taxon when combining datasets. The lack of evidence for nonmonophyly of these generaand the congruence of placements of each genera with phylo-genetic analysis of either gene indicate that this approach isnot providing misleading evidence.
DNA Extraction, Amplification, and Sequencing
DNA was extracted using a modified 6% CTAB extraction(Doyle and Doyle 1987; protocol D of Smith et al. 1991) orwith DNeasy Plant Mini Kit (QIAGEN, Valencia, Calif.). Inthe former method, DNA was precipitated with ethanol andeither sodium chloride or ammonium acetate because thesecause fewer polysaccharides to coprecipitate with the DNAthan is the case with isopropanol or with ethanol in the pres-ence of sodium acetate (Bult et al. 1992). Particularly in thecase of some Myrtaceae and Melastomataceae, subsequentsalt washes (Sytsma 1994) were required in order to consis-tently amplify the DNA. PCR amplification and cycle-sequencing followed the methods described elsewhere (Contiet al. 1993, 1996; Givnish et al. 2000; Hall et al. 2002). AllPCR products were cleaned using QIAquick PCR purificationkits (QIAGEN). Sequences were generated on an ABI Prism377 DNA sequencer. Contiguous alignments were edited us-ing Sequencher version 3.0 (Gene Codes, Ann Arbor, Mich.).Overlapping sequence fragments of rbcL were obtained
from both strands using up to 10 primers (Conti et al. 1997for details). Overlapping sequence fragments of the 39 end ofndhF were obtained from both strands using primers 5–14(Olmstead et al. 1993; Sytsma et al. 2002). Amplificationand sequencing of both strands for matK (Vochysiaceae only)used primers described in Johnson and Soltis (1994). Myrta-ceae matK primers and protocols have been published else-where (Wilson et al. 2001). Sequences of ndhF and matK, forwhich indels existed, were easily aligned visually in Se-Al ver-sion 2.0a6 (Rambaut 2001).
Phylogenetic Analyses
As the goal of this study was to explore the utility of se-quence data to provide a framework for clock calibration,only ML methodologies were employed here. However, the
main phylogenetic relationships important for the biogeo-graphical analyses are robust with either MP or Bayesiananalyses (K. J. Sytsma, M. L. Zjhra, J. C. Pires, M. Nepo-kroeff, J. Walker, C. J. Quinn, and P. G. Wilson, unpublishedmanuscript). ML analyses of the data sets of rbcL/ndhF forMyrtales and ndhF/matK for the PHMV clade were per-formed using PAUP* version 4.0b10 (Swofford 2002) ona Macintosh G4. Model test version 3.06 (Posada and Cran-dall 1998) was used to explore 56 ML models of DNA sub-stitution in a hierarchical testing framework for each data setwith a starting MP tree (obtained in PAUP*) instead of thedefault neighbor-joining tree. A heuristic ML search using de-fault settings with TBR branch-swapping was then conductedusing parameters determined for the best model of sequenceevolution and using a MP topology as a starting tree. MLbootstrap values (Felsenstein 1985) were obtained using theprocedure outlined in Nepokroeff et al. (2003) for large datasets. This method used a constraint tree of clades supported100% in MP bootstrap analyses, ML model of DNA substi-tution estimated previously, Addition Sequence ¼ random,and SEARCH ¼ FASTSTEP.
Fossils and Clock Calibration
As sequences from multiple species have accumulated overthe last decades, it has become apparent that a molecularclock is not fully supported in general (Britten 1986; Gilles-pie 1986; Li 1997; Wray 2001; Smith and Peterson 2002) orspecifically in angiosperms (Wilson et al. 1990; Bosquet et al.1992; Gaut et al. 1992; Clegg et al. 1995; Sytsma and Baum1996; Sanderson 1998; Sanderson and Doyle 2001). To eval-uate whether the chloroplast DNA sequences in each data setevolved in a clocklike fashion, a likelihood ratio test (Felsen-stein 1988) was performed by comparing the scores of MLtrees with and without the clock. Following rejection ofthe clock assumption, the ML trees (with branch lengths) foreach of the two data sets were rate smoothed with the pe-nalized likelihood (PL) method of Sanderson (2002) as im-plemented in r8s version 1.60 (Sanderson 2003) usingthe truncated Newton algorithm. PL averages local differen-ces in the rate of DNA evolution on different branches andassigns a penalty for rate changes among branches that aretoo rapid or frequent based on a smoothness parameter.Cross-verification in r8s was used to find the optimal valueof the smoothness parameter by varying the latter from 100
to 103 in steps of 0.25 in the exponent.Five fossil dates were used in PL to constrain specific nodes
to minimum, maximum, or fixed ages. Issues with these ap-proaches are known (Doyle and Donoghue 1993; Smith andPeterson 2002), but the use of multiple independent fossilcalibration points can provide increased confidence in result-ing dates. Two fossils of the Melastomataceae, utilized byRenner et al. (2001), were used with the rbcLþ ndhF dataset. The first leaf fossil (Wehr and Hopkins 1994; fossil 1,fig. 3) fixes the node representing the most recent commonancestor (MRCA) of Pternandra and all other Melastomata-ceae at 53 Ma. The second seed fossil (fossil 2, fig. 3) fixesthe node representing the MRCA of Monochaetum andRhexia at 23 Ma. We did not use the putative combretaceousfossils described as Esgueiria (Campanian-Maastrichtian of
S88 INTERNATIONAL JOURNAL OF PLANT SCIENCES
Portugal [Friis et al. 1991] and early Santonian of Japan[Takahashi et al. 1999]). Although Magallon et al. (1999)and Wikstrom et al. (2001) used Esgueiria to constrain theCombretaceae node at 84 Ma in their angiosperm-wide anal-yses, its flowers differ significantly in stylar branches from ex-tant Combretaceae, and placement in the family is not at allcertain or represents a member of an early stem lineage(Friis et al. 1991; E. M. Friis, personal communication; fig.3). We also did not use Diporites (Berry et al. 1990), repre-senting Old World Fuchsia (Onagraceae) pollen (fig. 3) be-cause our sampling with only one Fuchsia species precludesproper placement within the genus (Berry et al. 2004).The fossil status for the PHMV group is sketchy. No fossils
have been reported that unambiguously belong to Hetero-pyxidaceae, Psiloxylaceae, or the Vochysiaceae. The earliestfossil record for the PHMV group is Myrtaceidites (=Syncol-porites) pollen from Gabon during the Santonian or 86 Ma(Herngreen 1975; Muller 1981). This pollen is also seen laterin the Cretaceous from Colombia and from Borneo (van derHammen 1954; Muller 1968). Although Myrtaceidites is at-tributed to the Myrtaceae, the pollen of Myrtaceae, Hetero-pyxidaceae, and Psiloxylaceae are difficult to distinguish(Schmid 1980; Patel et al. 1984). Thus we gave a maximumage of 86 Ma to node C (figs. 3, 5), representing the MRCAof the three families in both analyses. However, we also ex-plored fixing this node at 86 Ma, or placing a maximum ageof 86 Ma at the crown node of Myrtaceae (node D; fig. 3),or simply removing the fossil date from the PL analyses. Thefourth fossil dates from the middle Eocene, is the earliest ofeucalyptoid fruits and leaves from central Australia (Green-wood 1991), and provides a fixed age of ca. 48 Ma for theMRCA of Eucalyptus and Angophora in the ndhF þmatKanalysis (fossil 4, fig. 5). Sine this node is not present in therbcLþ ndhF analysis, we constrained a minimum age of 48Ma (i.e., it will likely be older) on the node representing theMRCA of Angophora and Acmena (fossil 4, fig. 3). The fifthset of fossils is Paleomyrtinaea from the latest Paleocene ofNorth Dakota and early Eocene of British Columbia (Craneet al. 1990; Pigg et al. 1993; Manchester 1999). Paleomyrti-naea material comprises distinctive fruits and seeds seen onlyin the largely American subtribe Myrtinae (Pigg et al. 1993).Seed coat features of Paleomyrtinaea are similar to extantgenera (Lantern and Sharp 1989). In addition, the unorna-mented C-shaped embryo in Paleomyrtinaea is similar to thatof extant Myrtinae (McVaugh 1968; Lantern and Stevenson1986). Thus, Paleomyrtinaea can be used to date the node(56 Ma; Paleocene/Eocene border) from which Myrtinae firstdiverge within the subf. Myrtoideae. As the Myrtinae arebroadly paraphyletic in our analyses (fig. 4), this node is es-sentially the crown radiation of the tribe Myrteae (the Myr-toideae in the strict sense). We did not use the still unplacedfossil Syzygioides (Manchester et al. 1998; Manchester 1999.In addition, we constrained with a maximum age of 120 Mathe root of Myrtales, Sapindales, Brassicales, and Malvalesin the rbcLþ ndhF analysis (fig. 3) because this age is consis-tent with eudicot fossil records (125 Ma in Barremian-Aptianfor early eudicots; Magallon et al. 1999). The node repre-senting the crown group of PHMV (node B, fig. 5) was fixedat 93 Ma in the ndhF þmatK analysis based on the resultsfrom the larger-scale analysis.
The ML trees with PL-smoothed branch lengths were thenvisualized and saved in TreeEdit version 1.0a10 (Rambautand Charleston 2001) as chronograms for subsequent analy-sis. Geographical area was mapped onto resulting trees usingMacClade version 4.05 (Maddison and Maddison 2002). Forthe purposes of the broad biogeographical questions posedhere, areas of endemism were restricted to continental re-gions (South America, Africa, Australasia, Mediterranean). Amore detailed examination of Myrtaceae (K. J. Sytsma, M. L.Zjhra, J. C. Pires, M. Nepokroeff, J. Walker, C. J. Quinn,and P. G. Wilson, unpublished manuscript) defines more ex-clusive areas of endemism in Australia, Asia, and the Neo-tropics. For dates of geological periods, epochs, and ages, weused the Obradovich (1993) Cretaceous time scale as modi-fied by Everhart (2003) and the Harland et al. (1990) Ter-tiary time scale as modified by Poling (1997).
Results
Placement of PHMV Clade in Myrtales
The ML tree (fig. 2) for rbcLþ ndhF sequences usingGTRþ IþG likelihood settings obtained from Modeltesthighlights four important phylogenetic results in the Myrtales-wide survey: (1) Combretaceae is identified as the sister cladeto the rest of the order, although support is not strong. (2) ThePHMV clade is fairly well supported and is sister to the cladecomprising Melastomataceaeþ CAROP. (3) The African fam-ilies Heteropyxidaceae and Psiloxylaceae are strongly mono-phyletic and are sister to Myrtaceae, although not with strongML bootstrap support. and (4) Vochysiaceae is sister to allother members of the PHMV clade.
Clock Calibration and Lineage Dating inMyrtales-Wide Analysis
As is evident from a cursory glance of the ML tree (fig. 2),rates of sequence evolution are not clocklike in the Myrtalesas a whole, based on the likelihood ratio test. The woodyMyrtaceae lineage has a slow rate of sequence change,whereas the Melastomataceae have an accelerated rate—approaching that of Arabidopsis in the outgroup. Removal ofMelastomataceae and/or Arabidopsis still did not permitpassing of the likelihood ratio test. Cross verification of therbcLþ ndhF data set with PL indicated that the optimumsmoothing rate was 100.00. Figure 3 represents the PL chro-nogram using information from all five fossils and placinga maximum age of 120 Ma on the root. This chronogram isbased on the assumption that Myrtaceidites is a representativeof the stem lineage leading to Myrtaceae, Heteropyxidaceae,and Psiloxylaceae (i.e., node C, fig. 3, has a maximum age of86 Ma). This placement of Myrtaceidites is supported by al-most an identical chronogram obtained using only the fourother fossils and with Myrtaceidites removed from the analy-sis (table 2). The chronogram obtained by fixing node C at86 Ma (instead of maximum 86 Ma constraint) is similar tothat in figure 3 but gives slightly older dates within the Myr-taceae/Heteropyxidaceae/Psiloxylaceae clade (table 2). As-suming that Myrtaceidites belongs to the Myrtaceae lineageonly (i.e., node D, fig. 3, has a maximum age of 86 Ma) pro-vides considerably older dates within the Myrtaceae clade
S89SYTSMA ET AL.—PHYLOGENETICS AND BIOGEOGRAPHY OF MYRTACEAE AND VOCHYSIACEAE
that are not consistent with the dates obtained from the fourother fossils alone (table 2). The PL chronogram (fig. 3) thusplaces the 86 Ma Myrtaceidites nearly at the crown groupradiation of Myrtaceae/Heteropyxidaceae/Psiloxylaceae thatoccurred at 84 Ma. Esgueiria occurs farther back in timefrom the crown group radiation of Combretaceae sampledhere (fig. 3). As expected, Diporites is placed on the Fuchsialineage after the 41 Ma split of Fuchsia and Circaea in theOnagraceae (fig. 3).
Based on the preferred PL chronogram, the crown groupof Myrtales (node A, fig. 3) diverged 111 Ma in the earliestAlbian. By the end of the Albian, the four major lineageswithin Myrtales are already evident: Combretaceae,Onagraceaeþ Lythraceae, Melastomataceaeþ CAROP, andPHMV. The first divergence within the PHMV clade (nodeB, fig. 3) involving the separation into the stem lineages forVochysiaceae and for the other three families occurred at 93Ma. The Cretaceous dates for both the origin of Myrtales
Fig. 2 Maximum likelihood tree depicting phylogenetic relationships within the order Myrtales based on rbcL and ndhF sequences. ML
bootstrap values 70% and greater are above branches (those constrained in ML bootstrap runs are asterisked).
S90 INTERNATIONAL JOURNAL OF PLANT SCIENCES
Fig. 3 Penalized likelihood chronogram of Myrtales based on rbcLþ ndhF sequences. Information from five fossils (indicated by numbers)
was used; see text for details of implementation. Two fossils from the Melastomataceae (1, 2) were used to fix dates at two nodes. The date of
Myrtaceidites (86 Ma) was used as a maximum age constraint of node C (alternative analyses fixed the age of node C at 86 Ma or constrained
node D at 86 Ma; chronograms not shown). The date of the earliest eucalyptoid fossil (4) was used as a minimum age constraint on the mostrecent node leading to the sole eucalyptoid genus (Angophora). The date of Paleomyrtinaea (fossil 5; 56 Ma) was used to fix the age of node Eleading to the Myrtoid lineage. The rosid root of the tree was maximally constrained at 120 Ma. Probable placements of two fossils not used in the
analyses (Diporites, Esgueiria) are shown. Nodes identified by letter are discussed in the text.
and of the PHMV clade are clearly in the range indicatingGondwanan vicariance as probably important in theirbiogeographical history. However, the crown group of the ex-tant amphi-Atlantic Vochysiaceae (node G, fig. 3) dates backonly to 33 Ma from the Oligocene of the Tertiary, a time bywhich ca. two-thirds of the Atlantic seafloor rifting had al-ready occurred. The separation of southeast African Hetero-pyxidaceae and Psiloxylaceae from the Mascarenes (node F,fig. 3) is dated at ca. 40 Ma in the Eocene, well before theage of present-day Mascarenes (ca. 8 Ma).
Phylogenetic Relationships within the PHMV Clade
The ML analysis of Myrtaceae, Vochysiaceae, and relativesin the PHMV clade based on ndhF þmatK sequences usingGTRþG settings shows clear relationships among the fourfamilies in the PHMV clade but does not strongly resolve thebasal lineages within Myrtaceae (fig. 4). The inclusion ofMelastomataceae as outgroups (in addition to Rhynchocaly-caceae of the CAROP clade) does not affect relationshipswithin the PHMV clade (tree not shown). As in the previousMyrtales data set, the molecular clock is rejected in this fo-cused survey of the PHMV clade using ndhF and matK. TheML tree reveals five important points about relationshipswithin the PHMV clade: (1) The relationships of the fourfamilies are exactly the same as indicated by the broad Myr-tales survey using rbcL and ndhF. Heteropyxidaceae and Psi-loxylaceae are monophyletic and are sister to Myrtaceae and,with Vochysiaceae, are sister to these three families. (2)Within Vochysiaceae, the tribe Erismeae, represented herewith the South American Erisma and the tropical west Afri-can Erismadelphus, is clearly monophyletic. This clade inturn is sister to all other Vochysiaceae. (3) Although thebackbone of the Myrtaceae is not strongly supported, thefirst diverging lineages are consistently obtained with ndhFand matK singly or in combination: the ‘‘Lophostemon’’group diverges first, followed by the mangrove Osborniaand/or ‘‘Calothamnus-Melaleuca’’ group (fig. 5). (4) The
fleshy-fruited Myrtoid group is strongly supported as mono-phyletic and well derived within the family, making thesubfamily Leptospermoideae paraphyletic. and (5) Fleshy,indehiscent fruits are derived at least three times, once in theMyrtoid group, once in the ‘‘Acmena’’ group, and once forOsbornia, making the subfamily Myrtoideae polyphyletic.When geographical area is mapped onto the ML tree (fig.
5), three important results are seen: (1) The base of thePHMV clade is ambiguous for area with three major Gond-wanan regions (South America, Africa, and Australasia) oc-curring in the basal lineages. This basal ambiguity ingeographical area is mirrored in the outgroup CAROP clade(fig. 3), a result obtained whether the Melastomataceae areincluded or not. (2) The biogeographical shift within the Vo-chysiaceae between South America and Africa, whether dueto dispersal or vicariance, occurs after the crown group nodein Vochysiaceae has already diversified in South America.and (3) Despite the apparent rapid radiation of the majorlineages within Myrtaceae, which are often strongly sup-ported, there is a clear signal for the origin of the extantMyrtaceae within Australasia. Biogeographical shifts toSouth America (and to the Mediterranean and Africa) aremore recent.
Clock Calibration and Lineage Dating in PHMV Clade
The PL chronogram for ndhF þmatK (fig. 5) used a fixed48 Ma date for the eucalyptoid fossil node; a fixed 56 Madate for the base of Myrteae; a constrained maximum 86 Madate for the MRCA of Myrtaceae, Heteropyxidaceae, andPsiloxylaceae (node C, fig. 5); and a fixed date of 93 Ma forthe PHMV clade node based on the first rbcLþ ndhF PLanalysis. Dates of most nodes in the ndhF þmatK chrono-gram are quite similar to those seen in the rbcLþ ndhF chro-nogram except for the age of Myrtaceae, 80 versus 70 Ma,respectively (table 2). However, the ndhF þmatK analysishas considerably greater sampling of Myrtaceae, and thusnodes D (defining the Myrtaceae crown group) in figures 3 and
Table 2
Divergence Times for Clades within Myrtales and the PHMV Lineage Basedon Penalized Likelihood Analyses of Two Data Sets
Note. In the Myrtales analysis, four fossil dates were used with or without Myrtaceidites. The latter’s placement was
allowed to vary (maximum constraint for Myrtaceae or as either a maximum constraint or a fixed aged for the PHMclade). Two fossils plus Myrtaceidites were used as a maximum age constraint in the PHMV analysis.
a Clade age based on differential placement of 86 Ma Myrtaceidites.b Because Myrtaceae was sampled more extensively in the PHMV analysis, the Myrtaceae crown nodes for the two
analyses are not exactly the same. Values in parentheses indicate ages for the Myrtaceae node seen in the less-sampled
Myrtales analysis and for the comparable node in the PHMV analysis.
S92 INTERNATIONAL JOURNAL OF PLANT SCIENCES
5 are not exactly the same. Both data sets provide similar datesfor the separation of Heteropyxidaceae and Psiloxylaceae(node F; 40–38 Ma) in the late Eocene (table 2). TheVochysiaceae crown group divergence dates back to theEocene-Oligocene boundary (node G; 36–33 Ma). The Afri-
can Erismadelphus and its American sister Erisma (both oftribe Erismeae) show a separation at 28 Ma (node H, fig. 5)near the end of the Oligocene, long after complete separationof tropical South America and west Africa (ca. 80 Ma; seebelow). The Myrtaceae crown group diverged at 70–80 Ma
Fig. 4 Maximum likelihood tree depicting phylogenetic relationships within the PHMV clade based on ndhF and matK sequences. Subfamily
designation for Myrtaceae based on Briggs and Johnson (1979) and Johnson and Briggs (1984). ML bootstrap values 70% or greater are above
branches (those constrained in ML bootstrap runs are asterisked).
S93SYTSMA ET AL.—PHYLOGENETICS AND BIOGEOGRAPHY OF MYRTACEAE AND VOCHYSIACEAE
Fig. 5 Penalized likelihood chronogram of the PHMV clade based on ndhF and matK sequences. Information from three fossils (indicated bynumbers) also used in the Myrtales-wide survey (fig. 3) was used; see text for details of implementation. The date of Myrtaceidites (86 Ma) was
used as a maximum age constraint of node C. The date of the earliest eucalyptoid fossil (4) was used to fix the age of the node representing the
most recent common ancestor of all eucalyptoid genera. The date of Paleomyrtinaea (56 Ma) was used to fix the age of node E representing the
Myrtoids. The node representing the crown group of the PHMV clade was fixed at 93 Ma based on the Myrtales-wide analysis (fig. 3). Nodesidentified by letter are discussed in the text. Areas of endemism at the continental scale for each clade are provided. The branches leading to
Eugenia and Syzygium are indicated as polymorphic since both additionally contain African species (these not sampled). Informal ‘‘groups’’ within
Myrtaceae follow the designations of Wilson et al. (2001).
(node D, figs. 3, 5; table 2). The Myrtoid stem lineage is old(77 Ma), but the crown group diversified at 56 Ma (table 2)at the Paleocene-Eocene boundary. Separation of the Neotrop-ical elements of Myrtaceae (excluding capsular-fruited Tepau-lia, not sampled here) from the basal grade of AustralasianMyrtoid elements is less clear but occurred shortly thereafter(fig. 5).
Discussion
This two-tiered approach at estimating phylogenetic place-ment and relationships within the biogeographically disjunctPHMV clade comprising Myrtaceae, Vochysiaceae, Hetero-pyxidaceae, and Psiloxylaceae provides four important con-clusions on the phylogenetic relationships and biogeographichistory of the clade: (1) The PHMV clade separated from itssister clade of Melastomataceae and relatives in the Albianof the Lower Cretaceous and the PHMV crown group di-versified 12 m.yr. later around the Cenomanian-Turonianboundary of the Upper Cretaceous. (2) Vochysiaceae aremoderately supported as the sister clade to the other threefamilies in the PHMV clade, but extant genera of Vochysia-ceae did not diversify until ca. 60 m.yr. later in the early Oli-gocene of the Tertiary, and thus the South American–Africandisjunction (28 Ma) must be from long-distance dispersalrather than vicariance. (3) The African/Mascarene endemicfamilies, Heteropyxidaceae and Psiloxylaceae, are closely re-lated, and they diverged from each other in the late Eocene.(4) The widespread Myrtaceae date back to the Campanianof the late Cretaceous, extant members clearly are Austral-asian in origin (basal lineages of the paraphyletic subfamilyLeptospermoideae), and a more recent shift to America (in-cluding most members of the polyphyletic subfamily Myrtoi-deae) occurred in the early Eocene, perhaps via the Antarctictemperate land bridge.
Biogeographical History of Myrtales and Originof the PHMV Clade
Phylogenetic relationships presented here using rbcL andndhF expand on and clarify the emerging picture of familialrelationships within and biogeography of Myrtales based on-ly on rbcL (Conti et al. 1996, 1997, 2002). Probably an out-come of its ancient age, Myrtales have not been stronglyplaced within the Rosids and typically are placed outsideboth Eurosids I and II (APG 1998, 2003; Magallon et al.1999; Savolainen et al. 2000; Soltis et al. 2000). Based onthese analyses, the crown group of Myrtales (node A, fig. 3)began differentiation at 111 Ma during the early Albian ofthe Lower Cretaceous. This date is similar to the 107 Ma inthe Albian estimated by Wikstrom et al. (2001) based onrbcL alone.Combretaceae are shown now to represent the sister family
to all other Myrtales with the ML analysis (figs. 2, 3, al-though not strongly supported), whereas in previous MPrbcL analyses the family was weakly linked as sister to theLythraceae-Onagraceae clade. The combretaceous fossil Es-gueiria is placed considerably earlier than the node definingthe crown radiation of extant Combretaceae (fig. 3). Thisearlier occurrence is consistent with the issues (see above) of
placing Esgueiria in extant Combretaceae. Additionally,Combretaceae is the one family in Myrtales that has not beenadequately sampled yet (e.g., Strephonema of subf. Strepho-nematoideae; Tan et al. 2002; Stace, in press), so the crowngroup radiation might actually be older and thus more con-sistent with Esgueiria dates. Of the remaining Myrtales sisterto Combretaceae, two large clades diverged shortly thereafter(109 Ma): (1) PHMVþMelastomataceae s:l:þ CAROP and(2) LythraceaeþOnagraceae. The separation of the PHMVclade from Melastomataceae and the CAROP clade occurredin the late Albian at 105 Ma. Onagraceae and Lythraceae arestrongly supported as sister families; they diverged at 93 Manear the Cenomanian-Turonian border.Although the breakup of Gondwana into a western portion
(South America, Africa) and an eastern portion (Antarctica,Australia, India, Madagascar) occurred prior to these earlysplits within Myrtales (180–150 Ma; Scotese et al. 1988; Sco-tese and McKerrow 1990; Scotese 1997; McLoughlin 2001;Briggs 2003), subsequent continental movements broughtSouth America and Antarctica into contact and permitted di-rect west-east floral exchange well into the Tertiary (Hallam1994). Evidence was previously presented for the role of thebreakup of Gondwana in the diversification of the CAROPclade and subsequent rafting and dispersal out of the Indiansubcontinent for Crypteroniaceae (Conti et al. 2002; Rutsch-mann et al. 2004). Renner et al. (2001) suggested with ndhFdata that the diversification of Melastomataceae s.s. probablyoccurred during the Tertiary (Paleocene/Eocene) and in thenorth of the Tethys Sea (not Gondwanan), and their resultsimply a Melastomataceae s.s.–Memecylaceae split of 53 Maand an earlier separation with the CAROP clade at ca. 73Ma. These latter dates are consistent with our rbcLþ ndhFresults (56 and 82 Ma, respectively) and point toward Gond-wanan separation events in the early formation of the PHMVclade and its sister lineage (fig. 6A).
Vochysiaceae: Vicariance or Dispersal?
The molecular phylogenetic relationships and clock cali-bration presented here (fig. 5) provide unambiguous resolu-tion of the controversy involving the vicariance (Axelrod1970) versus dispersalist (Thorne 1972, 1973; Smith 1973;Barthlott 1983) interpretations of the amphi-Atlantic disjunc-tion for Vochysiaceae (fig. 1). Although Vochysiaceae repre-sent the sister lineage to Myrtaceae, Heteropyxidaceae, andPsiloxylaceae, and this separation dates back to ca. 93 Ma inthe Turonian of the Upper Cretaceous, the crown group ofextant American and African genera did not diversify untilca. 60 m.yr. later in the early Oligocene of the Tertiary (36–33 Ma, based on PHMV or Myrtales data set calibrations,respectively). Inclusion of the African Erismadelphus in thesecond PHMV analysis (figs. 4, 5) provides a separation timeof 28 Ma (late Oligocene) of this African genus from its sis-ter Erisma in South America. The topology of the ML treesargues that South America is the plesiomorphic geographicarea for Vochysiaceae, a finding maintained in matK analysesof many more taxa (Litt 1999; A. Litt, unpublished data,2001). Although the second African genus Korupodendronwas not available for DNA analysis, it is clearly more closelyrelated to Erismadelphus in Africa than to other members of
S95SYTSMA ET AL.—PHYLOGENETICS AND BIOGEOGRAPHY OF MYRTACEAE AND VOCHYSIACEAE
Fig. 6 Sequence of lineage diversification in the PHMV clade. A, Late Cretaceous; B, mid-Eocene; C, mid-Oligocene. Chronograms derived
from penalized likelihood analyses (fig. 5). Stem lineages for extant families indicated by proto. Informal ‘‘groups’’ within Myrtaceae follow the
designations of Wilson et al. (2001). Paleomaps showing relative placements of continents are based on Scotese (1997). Note that African
Vochysiaceae have not differentiated yet by the mid-Oligocene (C) and that the lineage leading to the Mascarene endemic Psiloxylaceae is shownin paleomap C although these volcanic islands in their present form are considerably younger. Additional intercontinental movements not shown
involve Eugenia (South America to Africa) and capsular Tepaulia (Australasia to South America).
Erismeae in South America (Litt and Cheek 2002). Synapo-morphies the African genera share with Erisma includepetal number greater than one, five-winged fruit (vs. three-winged), and simple (vs. stellate) hairs. Thus, the nesting ofAfrican taxa within the South American Vochysiaceae ratherthan sister to all South American taxa is strong with bothmolecular and morphological data. Additionally, as theAmerican-African lineage separation is on the order of ca. 28Ma, the disjunct distribution must be attributed to dispersalover a considerable Atlantic water barrier rather thandue to vicariance caused by disruption of a once more continu-ous American-African area of endemism (fig. 6C).Raven and Axelrod (1974) argued that it appears likely
that upwards of 80 large, fairly primitive to moderately spe-cialized angiosperm families were widely distributed in tropi-cal regions before the Gondwanan continents separated.According to Pittman et al. (1993), the beginning of the sepa-ration of Africa from South America occurred just before135 Ma in the early Cretaceous (Valanginian). The earlyphase of rifting lasted from 135 Ma to ca. 106 Ma (middleAlbian) but, due to the rotation of Africa away from SouthAmerica, tropical west Africa was still in contact with north-eastern South America. Complete separation of the two con-tinents was achieved sometime between 106 and 84 Ma.Pittman et al. (1993) suggested that by between 100 and 95Ma (Cenomanian), Africa and South America were effec-tively isolated from each other except to a few easily dis-persed terrestrial organisms (Goldblatt 1993). The distancefrom the eastern bulge of Brazil to the nearest point in westAfrica is 2900 km. Assuming an initial continental separationof tropical South America and Africa at 106 Ma and a steadyrate of separation (27.4 km/m.yr.), the movement of Vochy-siaceae to Africa would have entailed a long-distance disper-sal event over 2100 km of water during the late Oligocene(fig. 6C).Of the 12 or so families of flowering plants exhibiting the
striking amphi-Atlantic disjunct pattern involving only theAmericas and Africa, eight are essentially American families(e.g., Bromeliaceae, Cactaceae, Caricaceae, Humiriaceae,Loasaceae, Mayacaceae, Rapateaceae, Vochysiaceae) with ca.3500 species in the Americas compared with just over 10 inAfrica (Thorne 1973). Previously, Givnish et al. (2000, 2004)conclusively demonstrated with a molecular phylogeneticanalysis of Rapateaceae that the single African genus Mas-chocephalus is of recent origin (late Miocene) via long-distance dispersal. Arguments have been made for similaryoung ages of African taxa in Bromeliaceae and Mayacaceae(Givnish et al. 1999, 2000, 2004) and perhaps the remainderof these eight families. Many of the families possess seed orfruit dispersal mechanisms likely to account for long-distancedispersal over water: fleshy fruits in Rhipsalis (Cactaceae) forinternal bird dispersal; sticky seeds in Caricaceae for externalbird dispersal; small seeds in muddy or wet areas in Brome-liaceae, Mayacaceae, Rapateaceae for external bird dispersal;and drift fruits in Humiriaceae and Vochysiaceae (one speciesonly) for water dispersal. The fruits of Vochysiaceae areeither loculicidal capsules with few winged seeds (tribeVochysieae) or samaroid with a single seed encased by en-larged sepals (tribe Erismeae) (Kawasaki 1992). Of the latter,the fruit of one species of inundated Amazonian forests, Eris-
meae calcaratum, is further modified into a wingless, nutlike,and corky structure that is dispersed by water. Despite thelack of information on specific relationships of the Africanspecies to those in Erisma and on fruit type in the lineageleading to the African clade, the potential for long-distancedispersal exists in the tribe Erismeae in which occurred theamphi-Atlantic dispersal event.
Historical Biogeography of Heteropyxidaceae andPsiloxylaceae
The strong phylogenetic affinity of Heteropyxidaceae andPsiloxylaceae seen in these analyses is consistent with previ-ous MP results based solely on rbcL (Conti et al. 1996,1997) or matK (Gadek et al. 1996; Wilson et al. 2001). Thepresent ML results using combinations of rbcL, ndhF, andmatK (figs. 2–5) strongly support these two families as sisterto Myrtaceae, a relationship strongly supported (95% boot-strap) by matK alone (Wilson et al. 2001) but not rbcL alone(Conti et al. 1996, 1997); the latter could not resolve withhigh bootstrap support the polytomy of VochysiaceaeþðHeteropyxidaceaeþ PsiloxylaceaeÞ þMyrtaceae. Cladisticanalysis of morphological characters (Wilson et al. 2001)links Heteropyxidaceae and Psiloxylaceae to Myrtaceae bythe presence of glands in the anther connective and of secre-tory cavities in leaves (followed by loss of essential oils inPsiloxylon). The phylogenetic results support the recognitionof Heteropyxis and Psiloxylon as a clade within a morebroadly defined Myrtaceae and sister to the Myrtaceae s.s.(Gadek et al. 1996). The APG (2003) continues to recognizeall three families, although a single family is logically possi-ble. The argument, however, could be made that if these twosmall families are merged into Myrtaceae, then a similarmerging of the five small CAROP families should also occur,and perhaps also with Melastomataceae s.l.The separation of Heteropyxis and Psiloxylon from the
Myrtaceae lineage is old (86–84 Ma; fig. 6A). The restrictionof Heteropyxis to southeast Africa (West Gondwana) versusthe Australian or East Gondwanan nature of basal lineagesin Myrtaceae (see below) indicates that this early separationmight be Gondwanan. However, these dates are youngerthan typically are indicated for Gondwanan vicariance andmight indicate dispersal over at least moderate ocean bar-riers. Psiloxylon is restricted to Mauritius and Reunion, partof the hot spot basalt volcanic Mascarene Island chain in thewestern Indian Ocean formed after the final breakup of Indiafrom Madagascar and Seychelles at ca. 65 Ma (Storey et al.1995; Briggs 2003). The Mascarene Islands are situated ina line along a submerged ridge, the Seychelles-Mauritius Pla-teau, located 640–800 km east of Madagascar in the westernIndian Ocean, with the oldest, Mauritius, probably no olderthan 8 m.yr. (Bonneville et al. 1988; Hantke and Scheidegger1998). Since the divergence between Heteropyxis and Psilo-xylon occurred between 40 and 38 Ma in the late Eocene(figs. 3, 5, 6C), the lineage leading to Psiloxylon must haveoccupied other areas for most of the Tertiary. There is thepossibility that this lineage was situated on earlier islandsthat are now submersed, as has been argued for the ancestorof Fuchsia in Tahiti (Sytsma et al. 1991; Berry et al. 2004)and Hawaiian lobelioids (Givnish et al. 1995). The flora ofthe Mascarenes is highly endemic but exhibits ties mainly to
S97SYTSMA ET AL.—PHYLOGENETICS AND BIOGEOGRAPHY OF MYRTACEAE AND VOCHYSIACEAE
Africa and Madagascar (Renvoize 1979; D. Lorence, per-sonal communication), although connections also are seenwith Australasia and Polynesia (Carlquist 1974) and even theNeotropics (Graham 2002). Thus, the precise biogeographi-cal history of these two monogeneric families remains ob-scure. The most likely scenario involves an early Gondwananorigin in Africa for the small clade, a dispersal event of un-known age (but no earlier than 40 Ma) of the Psiloxylonlineage to the volcanic islands forming along the Seychelles-Mauritius Plateau after the India-Madagascar-Seychelles sep-aration of the early Tertiary and subsequent island hoppingand older island extinction events (fig. 6C).
Biogeographical History of the Amphi-TropicalMyrtaceae
The molecular phylogenetic tree (figs. 4, 5) depicts a rapidradiation at the base of the family Myrtaceae with only themore recent lineages well supported and the paraphyly ofsubf. Leptospermoideae. The detailed phylogenetic results ofthe combined ndhF þmatK data sets using MP and Bayesiananalyses, along with morphological data, will be presentedelsewhere, but they largely agree with that seen here withML and with the MP analysis of only matK (Wilson et al.2001). Despite the lack of strong support for basal relation-ships, the historical biogeographical results seen here (fig. 5)are strongly compelling for an origin of extant members ofMyrtaceae in Australasia and more recent shifts to the Amer-icas, Africa, the Mediterranean, with perhaps subsequent dis-persals back to Australasia (fig. 6). The ‘‘Lophostemon’’group (Lophostemon, Xanthostemon, Kjellbergiodendron),centered in northeast Australia, New Caledonia, NewGuinea, and out to Sulawesi, is sister to the remainder of thefamily. The Australasian-distributed mangrove, Osbornia,and the largely Australian ‘‘Calothamnus-Melaleuca’’ group(e.g., Calothamnus, Eremaea, Beaufortia, Melaleuca, andCallistemon) are also early diverging lineages in Myrtaceae.Thus the largely capsular-fruited subf. Leptospermoideae isa paraphyletic grade at the base of the family (figs. 4, 5). Inaddition, as suggested by Johnson and Briggs (1984) basedon a cladistic morphological analysis and Wilson et al.(2001) using only matK, the fleshy-fruited subf. Myrtoideaeis polyphyletic with the Acmena group (Acmena, Syzygium,and related genera) unrelated to the Myrtoid group (figs.4, 5).Whether or not the base of the stem lineage for Myrtaceae
is also Australasian is unclear. The closest relatives of Myrta-ceae in the PHMV clade are African (or African derived),and thus an origin in either west or east Gondwana is pos-sible (Wilson et al. 2001). This ambiguity is mirrored in thefossil record since the first fossil of myrtaceous type (includ-ing Heteropyxidaceae and Psiloxylaceae) is the Upper Creta-ceous Myrtaceidites from Africa (Gabon) but shortlythereafter also from South America (Colombia) and Asia(Borneo). This early pantropical distribution of the Myrta-ceae/Heteropyxidaceae/Psiloxylaceae clade must have beenfollowed by extinction in South America. Much of the diver-sification of the early diverging Australian lineages (includingNew Caledonia and New Guinea) of the Myrtaceae tookplace between 80 and 30 Ma (fig. 6A, 6B). This was a time
period of temperate connection with Antarctica (and thus toSouth America), followed by separation of the Australianplate from Antarctica, increasing aridification, and the rise ofthe distinctive scleromorphic/xeromorphic vegetation charac-terizing the late Miocene to the Recent of Australia (Christo-phel and Greenwood 1989; Hill 1994; White 1994).Myrtaceae now occur in three areas outside the plesiomor-
phic area of Australasia: the Americas, Africa, and the Medi-terranean (table 1; figs. 1, 6C). The Mediterranean endemicMyrtus appears to be a Tertiary dispersal event out of theMyrtoid group (fig. 5), discussed below. The African taxacomprise three unrelated groups: Syzygium of the Acmenagroup; Eugenia of the Myrtoid group; and ‘‘Metrosideros’’angustifolia of uncertain placement. Thus, the occurrenceof all extant Myrtaceae in Africa (except perhaps for thecapsular-fruited ‘‘Metrosideros’’) may all represent more re-cent dispersal events by fleshy-fruited taxa followed by spe-cies radiations. Recent long-distance dispersal via air currentsof the small seeds of Metrosideros from Australasia to thePacific islands has been argued with molecular data (Wrightet al. 2000, 2001).All Neotropical genera belong to the Myrtoid group (ex-
cept for the capsular-fruited Tepaulia not sampled here), thelargest fleshy-fruited lineage of the subf. Myrtoideae (Myrcia,Myrtus, Cryptorhiza, and Eugenia alliances; table 1). The stemlineage for the Myrtoid group spans ca. 20 m.yr. (77–56Ma; fig. 5), a time when the temperate Australia/Antarctica/South American connection was still in contact (fig. 6B,6C; see below). This portion of the Tertiary also experiencedthe height of the Cenozoic warm period and saw the incur-sion of a vast, climate-moderating epicontinental sea inNorth America and thus the extension of tropical and sub-tropical climates to high northern latitudes (50�N) (Graham1993). The Myrtoid group radiated quickly ca. 56 Ma (figs.4, 5) in both South America and Australasia, and relation-ships among the Neotropical and Australasian genera are notstrongly supported (only Eugenia occurs in both areas). Withthe ndhF þmatK data there is not enough resolution to ad-dress biogeographical relationships. Additional molecularmarkers are needed to verify the geographical pattern sug-gested by ndhF þmatK data (figs. 5, 6C)—a paraphyleticAustralasian Myrtoid group from which the American taxadiverged.The Australasian-American disjunct pattern is consistent
with a vicariance model based on geologically and fossil cali-brated phylogenetic trees with some subsequent dispersalevents occurring between Australasia and South America andAfrica (Tepaulia, ‘‘Metrosideros’’ angustifolia, Eugenia, andSyzygium) (fig. 6). The stem lineage of the Myrtoid group di-verged from Australasian lineages in the Campanian of theUpper Cretaceous (77 Ma) followed 20 m.yr. later by therapid crown group radiation in both Australasia and theAmericas (fig. 6B, 6C). Although tropical areas of Gondwanahad already been separated for at least 20 m.yr. when theMyrtoid stem lineage is seen, warm temperate connectionsexisted between South America and west Antarctica and be-tween east Antarctica and Australia well into the Tertiary(fig. 6C; Hallam 1994; Scotese 1997; McLoughlin 2001). Itis clear that neither the Drake Passage between Antarcticaand South America nor the passage between Australia and
S98 INTERNATIONAL JOURNAL OF PLANT SCIENCES
Antarctica was open at 40 Ma, but both passageways wereopen to middle and deep water circulation by the earlyOligocene (32 Ma) (Lawver and Gahagan 1998; McLoughlin2001). A widespread warm temperate Myrtoid stem lineageacross these southern continents would have experienced vi-cariance at about the time the crown group quickly radiatedin the earliest Eocene (56 Ma) or when the first American lin-eages are seen (48 Ma) (figs. 5, 6C). A similar warm temper-ate Antarctic link between South America and Australia/NewZealand explains the phylogenetic pattern, fossil (Diporites)dates, and distribution for Fuchsia (Onagraceae) (Berry et al.1990, 2004; Sytsma et al. 1991).
Phylogenetic-Biogeographic Summary for the PHMV Clade
To summarize these phylogenetic and biogeographic eventswithin the PHMV clade of four families, paleogeographic re-constructions of the continents are examined at three time in-tervals: late Cretaceous, 70 Ma (fig. 6A); mid-Eocene, 50 Ma(fig. 6B); and Oligocene, 30 Ma (fig. 6C). The relevant sub-clades (stem lineages and extant lineages) that have evolvedat each time interval are presented using a consensus clock-calibrated phylogenetic tree for the PHMV clade. Prior to theCretaceous/Tertiary boundary, 70 Ma (fig. 6A), Myrtales hadalready been diverging into at least four main clades for morethan 40 m.yr. in concert with the separation of Gondwana.Of these, the PHMV clade had also been diverging for almost20 m.yr. into the three lineages leading to Myrtaceae, to theAfrican/Mascarene Island families, and to Vochysiaceae. Atthe Cretaceous/Tertiary boundary (fig. 6A), stem lineages arestill present only for Vochysiaceae and the African/MascareneIsland families. However, the Myrtaceae are now diversifyinginto several Australasian groups, perhaps partly in responseto the movement, beginning isolation, and aridification ofAustralia. The Myrtoid stem lineage has just separated fromother early basal lineages.During the mid-Eocene of the Tertiary at 50 Ma (fig. 6B),
tropical regions are now quite in isolation from each other,
but a temperate connection between South America and Aus-tralia via the Antarctic temperate land bridge will be presentfor at least another 10 m.yr. The crown groups of extant Vo-chysiaceae and of Heteropyxidaceae and Psiloxylaceae stillare not evident. Most of the major Australasian lineages ofLeptospermoideae are now present and are actively diversify-ing. Two fleshy-fruited stem lineages in Myrtaceae are nowseen—the Acmena and the Myrtoid groups. The Myrtoidcrown group is differentiating in Australia, and the Americanand European lineages now appear. This may have involveda single overland migration of fleshy fruits to the Americas.Additional dispersal events to the Mediterranean (Myrtoidgroup) and Africa (Acmena group) occur from the Eoceneon.Finally, a majority of the genera sampled (at least their
stem lineages) are present by the mid-Oligocene at 30 Ma(fig. 6C). The Myrtoid lineages of both the Neotropics andAustralasia are almost fully differentiated. Heteropyxidaceaeand Psiloxylaceae have just differentiated, presumably inAfrica since the Mascarenes will not be available for coloni-zation by Psiloxylaceae for some time. Only American line-ages of Vochysiaceae are now evident, with the Africanlineage not to diverge for another 2 m.yr., well after the for-mation of a considerable Atlantic water barrier.
Acknowledgments
We are grateful to Chris Quinn, Marcelle O’Brien, andMargaret Heslewood for making available unpublished Myr-taceae matK sequences and to the Molecular SystematicsLaboratory, Royal Botanic Gardens, Kew, for initial help inextraction of Vochysiaceae DNAs. Graphical support wasprovided by Kandis Elliot. This research was supported byNSF grants DEB 9407270 and 9981587 to K. J. Sytsma andby Australian Research Council grant A19602649 to P. G.Wilson and C. J. Quinn.
Appendix
Table A1
Species Names, Sources, and Genbank Accession Numbers of the DNA Sequences Used in the Analyses
Sytsma 2004 Ancient vicariance or recent long-distance dispersal?inferences about phylogeny and South American–African disjunc-
tions in Rapateaceae and Bromeliaceae based on ndhF sequence
data. Int J Plant Sci 164(suppl):S35–S54.
Givnish TJ, KJ Sytsma, WJ Hahn, JF Smith 1995 Molecular
evolution, adaptive radiation, and geographic speciation in Cyanea(Campanulaceae, Lobelioideae). Pages 288–337 in WLWagner, VAFunk, eds. Hawaiian biogeography: evolution on a hot spot
diversity, and diversification of eudicots. Ann Mo Bot Gard 86:
297–372.
Manchester ST 1999 Biogeographical relationships of North Amer-
ican Tertiary floras. Ann Mo Bot Gard 86:472–522.
Manchester ST, DL Dilcher, SLWing 1998 Attached leaves and fruitsof myrtaceous affinity from the middle Eocene of Colorado, USA.
Rev Palaeobot Palynol 102:153–163.
Manos P 1997 Systematics of Nothofagus (Nothofagaceae) based on
rDNA spacer sequences (ITS): taxonomic congruence with mor-phology and plastid sequences. Am J Bot 84:1137–1155.
McLoughlin S 2001 The breakup history of Gondwana and its
impact on pre-Cenozoic, floristic provincialism. Aust J Bot 49:271–300.
McVaugh R 1968 The genera of American Myrtaceae: an interim
report. Taxon 17:354–418.
Muller J 1968 Palynology of the Pedawan and Plateau Sandstone
Formations (Cretaceous-Eocene) in Sarawak, Malaysia. Micropa-
leontology 14:1–37.
——— 1981 Fossil pollen records of extant angiosperms. Bot Rev47:1–142.
Murphy WJ, E Eizirik, SJ O’Brien, O Madsen, M Scally, CJ Douady, ETeeling, et al 2001 Resolution of the early placental mammal
radiation using Bayesian phylogenetics. Science 294:2348–2351.
Nepokroeff M, KJ Sytsma, WL Wagner, EA Zimmer 2003 Recon-
structing ancestral patterns of colonization and dispersal in theHawaiian understory tree genus Psychotria (Rubiaceae): a compar-
ison of parsimony and likelihood approaches. Syst Biol 52:
820–838.
Obradovich JD 1993 A Cretaceous time scale. InWGE Caldwell, EG
Kauffman, eds. Evolution of the Western Interior Basin. Geol AssocCan Spec Pap 39:379–396.
O’Brien MM, CJ Quinn, PG Wilson 2000 Molecular systematics of
the Leptospermum suballiance (Myrtaceae). Aust J Bot 48:
621–628.
Olmstead RG, JA Sweere, KH Wolfe 1993 Ninety extra nucleotidesin the ndhF gene of tobacco chloroplast DNA: a summary of
revisions to the 1986 genome sequence. Plant Mol Biol 22:
1191–1193.
Patel VC, JJ Skvarla, PH Raven 1984 Pollen characters in relation to
the delimitation of Myrtales. Ann Mo Bot Gard 71:858–969.
Pigg KB, RA Stockey, SL Maxwell 1993 Paleomyrtinaea, a newgenus of permineralized myrtaceous fruits and seeds from the
Eocene of British Columbia and Paleocene of North Dakota. Can J
Bot 71:1–9.
Pittman WC III, S Cande, J LaBrecque, J Pindell 1993 Fragmentationof Gondwana: the separation of Africa from South America. Pages
15–34 in P Goldblatt, ed. Biological relationships between Africa
and South America. Yale University Press, New Haven, Conn.
Poling J 1997 Geological ages of earth history. http://www.dinosauria.
com/dml/history.htm
Posada D, KA Crandall 1998 Modeltest: testing the model of DNAsubstitution. Bioinformatics 14:817–818.
Rambaut A 2001 Se-Al (sequence alignment editor), version 2.0a6.
http://evolve.zoo.ox.ac.uk/
Rambaut A, M Charleston 2001 TreeEdit: phylogenetic tree editor,
version 1.0a10. http://evolve.zoo.ox.ac.uk/
Raven PH, DI Axelrod 1974 Angiosperm biogeography and pastcontinental movements. Ann Mo Bot Gard 61:539–673.
Renner SS 2004 Plant dispersal across the tropical Atlantic by wind
and sea currents. Int J Plant Sci 165(suppl):S23–S33.
Renner SS, G Clausing, K Meyer 2001 Historical biogeography of
Melastomataceae: the roles of tertiary migration and long-distance
dispersal. Am J Bot 88:1290–1300.
Renner SS, DB Foreman, D Murray 2000 Timing transantarcticdisjunctions in the Atherospermataceae (Laurales): evidence
from coding and noncoding chloroplast sequences. Syst Biol 49:
579–591.
Renner SS, K Meyer 2001 Melastomeae come full circle: biogeo-graphic reconstruction and molecular clock dating. Evolution 55:
1315–1324.
Renvoize SA 1979 The origins of Indian Ocean island floras. Pages
107–129 in D Bramwell, ed. Plants and islands. Academic Press,
London.
Rutschmann F, T Eriksson, E Conti 2004 Did Crypteroniaceae reallydisperse out of India? molecular dating evidence from rbcL, ndhF,and rpl16 intron sequences. Int J Plant Sci 165(suppl):S69–S83.
Sanderson MJ 1998 Estimating rate and time in molecular phylog-
enies: beyond the molecular clock? Pages 242–264 in DE Soltis, PSSoltis, JJ Doyle, eds. Molecular systematics of plants. Vol 2. Kluwer,
Boston.
——— 2002 Estimating absolute rates of molecular evolution and
divergence times: a penalized likelihood approach. Mol Biol Evol
19:101–109.
——— 2003 r8s version 1.60: analysis of rates (‘‘r8s’’) of evolution(and other stuff). http://ginger.ucdavis.edu/r8s/
S104 INTERNATIONAL JOURNAL OF PLANT SCIENCES
Sanderson MJ, JA Doyle 2001 Sources of error and confidenceintervals in estimating the age of angiosperms from rbcL and 18S
rDNA data. Am J Bot 88:1499–1516.
Savolainen V, MW Chase, SB Hoot, CM Morton, DE Soltis, C Bayer,MF Fay, AY De Bruijn, S Sullivan, Y-L Qiu 2000 Phylogenetics of
flowering plants based on combined analysis of plastid atpB and
rbcL gene sequences. Syst Biol 49:306–362.
Schmid R 1980 Comparative anatomy and morphology of Psilo-xylon and Heteropyxis, and the subfamilial and tribal classification
90–0497. Department of Geology, University of Texas, Arlington.Scotese CR, LM Gahagan, RL Larson 1988 Plate tectonic of the
Cretaceous and Cenozoic ocean basins. Tectonophysics 155:27–48.
Scotese CR, WS McKerrow 1990 Revised world maps and intro-duction. Pages 1–21 in WS McKerrow, CR Scotese, eds. Paleozoic
paleogeography and biogeography. Geol Soc Mem 12:1–21.
Sequeira AS, BD Farrell 2001 Evolutionary origins of Gondwanan
interactions: how old are Araucaria beetle herbivores? Biol J LinnSoc 74:459–474.
Smith AB, KJ Peterson 2002 Dating the time of origin of major
clades: molecular clocks and the fossil record. Annu Rev Earth
Planet Sci 30:65–88.Smith AC 1973 Angiosperm evolution and the relationship of the
floras of Africa and America. Pages 49–61 in BJ Meggers, ES
Ayensu, WD Duckworth, eds. Tropical forest ecosystems in Africaand South America: a comparative review. Smithsonian Institution,
Washington, D.C.
Smith JF, KS Sytsma, JS Shoemaker, RL Smith 1991 A qualitative
comparison of total cellular DNA extraction protocols. PhytochemBull 23:2–9.
Soltis DE, PS Soltis, MW Chase, ME Mort, DC Albach, M Zanis, V
Savolainen, et al 2000 Angiosperm phylogeny inferred from 18S
rDNA, rbcL, and atpB sequences. Bot J Linn Soc 133:381–461.Stace CA In press Combretaceae. In K Kubitzki, C Bayer, eds. The
families and genera of vascular plants. Springer, Berlin.
Storey M, JJ Mahoney, AD Saunders, RA Duncan, SP Kelley, MF
Coffin in 1995 Timing of hot spot–related volcanism and thebreakup of Madagascar and India. Science 267:852–855.
Swenson U, RS Hill, S McLoughlin 2001 Biogeography of Notho-fagus supports the sequence of Gondwana break-up. Taxon 50:1025–1041.
Swofford DL 2002 PAUP*: phylogenetic analysis using parsimony
(*and other methods), version 4. Sinauer, Sunderland, Mass.
Sytsma KJ 1994 DNA extraction from recalcitrant plants: long, pure,and simple? Pages 69–81 in RPAdams, JS Miller, EM Golenberg, JE
Adams, eds. Conservation of plant genes. II. Intellectual property
rights and DNA utilization. Missouri Botanical Garden, St. Louis.
Sytsma KJ, DA Baum 1996 Molecular phylogenies and the di-versification of angiosperms. Pages 314–340 in DW Taylor, LJ
Hickey, eds. Flowering plant origin, evolution and phylogeny.
Chapman and Hall, New York.Sytsma KJ, J Morawetz, JC Pires, M Nepokroeff, E Conti, M Zjhra,
JC Hall, MW Chase 2002 Urticalean rosids: circumscription, rosidancestry, and phylogenetics based on rbcL, trnLF, and ndhFsequences. Am J Bot 89:1531–1546.
Sytsma KJ, JF Smith, PE Berry 1991 The use of chloroplast DNA toassess biogeography and evolution of morphology, breeding
systems, and flavonoids in Fuchsia sect. Skinnera (Onagraceae).
Syst Bot 16:257–269.
Takahashi M, PR Crane, H Ando 1999 Esgueiria futabensissp. nov., a new angiosperm flower from the Upper Cretaceous
(lower Coniacian) of northeastern Honshu, Japan. Paleontol Res
3:81–87.
Tan FX, SH Shi, Y Zhong, X Gong, YG Wang 2002 Phylogeneticrelationships of Combretoideae (Combretaceae) inferred from
plastid, nuclear gene and spacer sequences. J Plant Res 115:
475–481.Thorne RF 1972 Major disjunctions in the geographic ranges of seed
plants. Q Rev Biol 47:365–411.
——— 1973 Floristic relationships between tropical Africa and
tropical America. Pages 27–47 in BJ Meggers, ES Ayensu, WDDuckworth, eds. Tropical forest ecosystems in Africa and South
America: a comparative review. Smithsonian Institution, Washing-
ton, D.C.
van der Hammen T 1954 El desarrollo de la flora colombiana en losperiodos geologicos. I. Maestrichtiano hasta Terciario mas inferior.
Bol Geol 2:49–106.
Vinnersten A, K Bremer 2001 Age and biogeography of major cladesin Liliales. Am J Bot 88:1695–1703.
Wehr WC, DQ Hopkins 1994 The Eocene orchards and gardens of
Republic, Washington. Wash Geol 22:27–34.
White ME 1994 After the greening: the browning of Australia.Kangaroo, Kenthurst, N.S.W.
Wikstrom N, V Savolainen, MW Chase 2001 Evolution of the
angiosperms: calibrating the family tree. Proc R Soc Lond B 268:
2211–2220.Wilson MA, B Gaut, MT Clegg 1990 Chloroplast DNA evolves
slowly in the palm family (Arecaceae). Mol Biol Evol 7521:
303–314.
Wilson PG, MM O’Brien, PA Gadek, CJ Quinn 2001 Myrtaceaerevisited: a reassessment of infrafamilial groups. Am J Bot 88:
2013–2025.
Winkworth RC, SJ Wagstaff, D Glenny, PJ Lockhart 2002 Plantdispersal N.E.W.S. from New Zealand. TREE 17:514–520.
Wray GA 2001 Dating branches on the tree of life using DNA.
Genome Biol 3:reviews0001.1–0001.7.
Wright SD, CG Yong, JW Dawson, DJ Whittaker, RC Gardner2000 Riding the ice age El Nino: Pacifica biogeography and
evolution of Metrosideros subg. Metrosideros (Myrtaceae) inferred
from nuclear ribosomal DNA. Proc Natl Acad Sci USA 97: