Investigating the evolution of Lantaneae (Verbenaceae...

Investigating the evolution of Lantaneae (Verbenaceae)using multiple loci

PATRICIA LU-IRVING* and RICHARD G. OLMSTEAD

Department of Biology, University of Washington, Seattle, Washington 98195, USA

Received 3 February 2012; revised 29 June 2012; accepted for publication 23 August 2012

Lantaneae are an example of a taxonomically problematic, widespread and recently radiated Neotropical lineage.Taxonomy in Lantaneae is difficult because of complex, overlapping patterns of shifts in morphological traits amongmembers; monophyly of the traditional genera cannot be assumed without additional information from moleculardata. We took a multi-locus approach to infer phylogenetic relationships in Lantaneae, resolving major cladesamong a broad representative sample that covers the morphological, taxonomic and geographical diversity of thisgroup. Data from multiple, independent loci reveal individual gene trees that are incongruent with one another,with varying degrees of support. Without reliable, applicable methods to determine the sources of such incongru-ence and to resolve it, we present the consensus between well-supported topologies among our data sets as the bestestimate of Lantaneae phylogeny to date. According to this consensus tree, fleshy fruits in Lantaneae have beenderived from dry fruits at least five times; taxonomic schemes separating genera based on fruit characteristics areartificial. Lantaneae have shifted into the Neotropics from the southern temperate subtropics and have colonizedAfrica in at least two separate long-distance dispersal events. This study provides a first pass at a broad Lantaneaephylogeny, but two important areas remain unresolved: the position of Acantholippia relative to Aloysia; andspecies-level relationships in the Lantana–Lippia clade. © 2012 The Linnean Society of London, BotanicalJournal of the Linnean Society, 2013, 171, 103–119.

ADDITIONAL KEYWORDS: Lantana –Lippia – long-distance dispersal – Neotropics – pentatricopeptiderepeat (PPR) loci – recent radiation.

INTRODUCTION

The Neotropics are globally renowned as a region ofremarkable floristic diversity. Much of this speciesrichness is concentrated in large, endemic (or nearlyendemic) lineages, such as Cactaceae, Bromeliaceaeand Bignoniaceae (Gentry, 1982). In these and othercharacteristic Neotropical lineages, ‘problematic’ taxaare common: plant groups in which traditional clas-sifications are at odds with newly obtained molecularevidence. Examples include Mammillaria Haw. inCactaceae (Butterworth & Wallace, 2004), subfamilyBromelioideae in Bromeliaceae (Schulte, Barfuss &Zizka, 2008; Sass & Specht, 2010), Tabebuia Gomesex DC. (Grose & Olmstead, 2007) and tribe Bignon-ieae (Lohmann, 2006) in Bignoniaceae.

With the increasing range of modern tools availableto systematists, great progress has been made in thelast several years in untangling the evolutionaryhistories of difficult taxa in important Neotropicalfamilies. Recent examples, in addition to those citedabove, can be found in cycads (González, Vovides &Bárcenas, 2008), palms (Eiserhardt et al., 2011;Ludeña et al., 2011), Fabaceae (Torke & Schaal, 2008),Annonaceae (Chatrou et al., 2012) and Podostemaceae(Tippery et al., 2011). Each of the lineages studied inthese examples has in common particular characteris-tics that make it problematic: it is species-rich andgeographically widespread, classifications within thegroup are historically difficult and previous broad,molecular phylogenetic studies fail to resolve relation-ships within it. Here we present an additional examplefrom our work in Lantaneae: a morphologically diversegroup of several hundred species forming the mostspecies-rich tribe of Verbenaceae.*Corresponding author. E-mail: [email protected]

bs_bs_banner

Botanical Journal of the Linnean Society, 2013, 171, 103–119. With 5 figures

© 2012 The Linnean Society of London, Botanical Journal of the Linnean Society, 2013, 171, 103–119 103

BACKGROUND INFORMATION

After recent recircumscription (Marx et al., 2010),tribe Lantaneae is monophyletic, containing twomajor genera (Lantana L. and Lippia L.) and sevensmaller genera. It is sister to tribe Verbeneae (Yuanet al., 2009b; Marx et al., 2010). The two principalgenera of Lantaneae comprise c. 75% of the species(Lippia with c. 200 species and Lantana with c. 150species; Atkins, 2004). However, some taxonomistsconsider that there are too many names in Lantana(López-Palacios, 1991; Verdcourt, 1992; Santos, 2002),preferring to recognize as few as 55 species (Sanders,2001). Other genera are smaller: Aloysia Palau (30species); Phyla Lour. (five species; O’Leary &Múlgura, 2012); Nashia Millsp. (seven species); Acan-tholippia Griseb. (six species); Coelocarpum Balf. f.(five species); Burroughsia Moldenke (two species;Moldenke, 1940); and the monotypic XeroaloysiaTronc. (numbers of species from Atkins, 2004, unlessotherwise attributed). Many members of Lantaneaeare of ecological and ethnobotanical significance intheir natural settings; for example, Acantholippia sal-soloides Griseb., a community dominant in the alti-plano used locally as a culinary herb. Others are ofglobal economic and/or ecological importance; forexample, Aloysia citriodora Palau (lemon verbena),commonly cultivated for its medicinal and culinaryuses, and Lantana camara (lantana), a popular orna-mental and weed of global significance.

The evolutionary history of Lantaneae presents adifficult problem. The large numbers of species inLantaneae encompass a great deal of morphologicalvariation, ranging from herbs to shrubs and smalltrees, with a diverse spectrum of leaf morphologiesand inflorescence architectures. Members of Lanta-neae are found in many different habitats, includingmoist lowland forests, the fire-prone cerrado and thedry altiplano, each with accompanying morphologicaladaptations. Attempts to partition this wide rangeof variation according to generic and infragenericboundaries traditionally rely heavily on fruit mor-phology (Chamisso, 1832; Schauer, 1847; Briquet,1895, 1904; Moldenke, 1959; Troncoso, 1974). Accord-ing to one scheme, species with schizocarpous fruitare assigned to Lippia and species with fleshy drupesare placed in Lantana (Schauer, 1847; Troncoso,1974). Alternatively, the number of mericarps orpyrenes per fruit has also been used to separateLantana from Lippia (Chamisso, 1832; Silva, 1999).However, generic boundaries in Lantaneae areblurred by species that are difficult to assign unam-biguously to genus, presumably attributable to con-vergence in these (and other) important diagnostictraits. These confounding morphological patterns areconsistent with recent radiation, as are the short

branch lengths in Lantaneae found in the molecularstudy of Marx et al. (2010).

Adding to the problems associated with describingthe wide range of morphologies in Lantaneae, thetribe is also geographically wide-ranging. The originof Lantaneae is in subtropical South America, andthe centre of diversity is in the Neotropics (Atkins,2004; Marx et al., 2010; Olmstead, 2012). The nativedistribution spans the southern states of the USA,Mexico and Central America, the Caribbean andSouth America; a few species also occur on theother side of a trans-Atlantic disjunction, in Africaand Madagascar. Some members, most notably theLantana camara L. species group, have been globallyintroduced as ornamentals and spread as weeds,apparently hybridizing with native species in someparts of the Neotropics (Sanders, 1987), further con-fusing taxonomic efforts. Native African species areassigned to both Lantana and Lippia, suggesting atleast two distinct colonization events.

There is a growing effort to address the troublesomeclassification schemes in Lantaneae and to producegeneric revisions (e.g. Silva, 1999; Salimena, 2002;Silva & Salimena, 2002; Santos, 2002; Sanders, 2001,2006; Siedo, 2008; O’Leary et al., 2012; O’Leary &Múlgura, 2012). However, because Lantaneae arespecies-rich, geographically widespread and recentlyradiated, these taxonomic efforts are hindered by thecommon problems that such a group presents. Theirfocus is often on specific geographical regions, usuallydefined by political boundaries, which may or may notbe of biogeographic significance. Additionally, manytaxonomic revisions focus on single genera, tradition-ally circumscribed, under the implicit assumption thatgeneric boundaries are of evolutionary significance.There is a clear need for a broad, well-resolved phy-logenetic hypothesis for Lantaneae, which has yet tobe addressed in detail in a molecular phylogeneticstudy.

PHYLOGENY RECONSTRUCTION USING MULTIPLE

INDEPENDENT LOCI

Phylogenetic systematic studies in plants over thelast three decades have made great use of sequencedata from plastid DNA, and recent studies thatsample very broadly across large Neotropical groupscontinue to rely on it (e.g. Lohmann, 2006; Olmsteadet al., 2008, 2009; Marx et al., 2010; Bárcenas, Yesson& Hawkins, 2011; Givnish et al., 2011). However, theplastid genome has a lower rate of molecular changethan the nuclear genome, and individual plastidloci are often insufficiently variable to provide reso-lution between species in recently diversified groups(Small, Cronn & Wendel, 2004). The nuclear genomeis an extensive source of variable DNA regions, and

104 P. LU-IRVING and R. OLMSTEAD

© 2012 The Linnean Society of London, Botanical Journal of the Linnean Society, 2013, 171, 103–119

variable nuclear loci are often much richer sources ofinformation for molecular phylogenetic studies insuch groups (Small et al., 2004; Whittall et al., 2006;Steele et al., 2008). Additionally, hybridization and/orincomplete lineage sorting may be common amongrecently diverged species; their effects can only beexposed by multi-locus approaches. For example,tribe Verbeneae have a complicated evolutionaryhistory of plastid transfer, incomplete lineage sortingand convergent character evolution, which was onlyrevealed by molecular phylogenetic studies usingmultiple loci (Yuan & Olmstead, 2008a, b; O’Learyet al., 2009; Yuan et al., 2009b). As genomic resourcesand sequencing technologies continue to be developed,the information content of the nuclear genome hasbecome increasingly accessible to and drawn upon byphylogenetic studies; the COSII genes in Solanaceaeare one example of this (Levin, Whelan & Miller,2009).

Yuan et al. (2009a) developed approaches toutilize the pentatricopeptide repeat (PPR) gene familyas a source of multiple nuclear loci suitable for usein phylogenetic studies and optimized primers toamplify and sequence several of these loci in Verben-aceae (Yuan et al., 2009b). PPR genes encode peptideswith unusually high substitution rates. There are alarge number of PPR loci, and these are highly diver-gent from one another. The shared presence of manyof these loci in such distantly related groups asBrassicaceae (Arabidopsis thaliana (L.) Heynh.) andPoaceae (rice, maize) suggests that the presentdiversity of PPR genes is attributable to ancientduplications (Yuan et al., 2009a, b). Yuan et al.(2009a) screened the genomes of A. thaliana and ricefor intron-less PPR genes with a single orthologue ineach, and published a list of > 100 of these. The loci onthis list are valuable as phylogenetic tools becausethey can be directly sequenced and easily and unam-biguously aligned, problems caused by doubtfulorthology are avoided, and they can potentially bedeveloped for use in any plant group.

We took a multi-locus approach to reconstruct aphylogeny for Lantaneae, in order to test monophylyof the genera, investigate the extent to which fruitcharacters are homoplasious, and seek evolutionarypatterns in geographical distribution in the tribe. Wecollected DNA sequences across a broad sample of thetribe, from three PPR genes along with the nuclearexternal transcribed spacer (ETS) region and threeplastid loci (trnT-L, rpl32-trnL and trnQ-rps16). Twoof the PPR loci used in this study were amplifiedusing primers designed by Yuan et al. (AT1G09680and AT5G39980; 2009b); a third (AT3G25970) wasselected from the original list of those with a singleorthologue in A. thaliana and rice (Yuan et al., 2009a)and new primers were designed to amplify it.

MATERIAL AND METHODSSAMPLING

Taxa were chosen to broadly represent the morphologi-cal and geographical variation found in Lantaneae. Allgenera belonging to the tribe were sampled (Acanthol-ippia, Aloysia, Burroughsia, Coelocarpum, Lantana,Lippia, Nashia, Phyla, Xeroaloysia). Forty-sevenspecies of Lantaneae were chosen as the ingroup, andseven species from related lineages were chosen asoutgroups. Voucher information and GenBank acces-sion numbers for all taxa sampled are listed in theSupporting Information (Table S1).

DNA EXTRACTION, AMPLIFICATION AND SEQUENCING

DNA was extracted from dried leaf tissue that wascollected in the field and preserved in silica gel orsampled from herbarium specimens. Extractions werecarried out following a standard 2 ¥ cetyl trimethyl-ammonium bromide (CTAB) method (modified fromDoyle & Doyle, 1987); DNA was purified by isopropa-nol precipitation and some extractions were furtherpurified using a DNA cleanup kit (Promega Corp.).

PCRs were performed in a Perkin-Elmer thermocy-cler, under the following general reaction conditions:94 °C for 2 min, followed by 35 cycles of 94 °C for 30 s,50 °C for 30 s, 72 °C for 1.5–2.5 min, followed by 72 °Cfor 10 min. Universal primers were used to amplify thetrnT-L (Taberlet et al., 1991), rpl32-trnL (Shaw et al.,2007) and trnQ-rps16 (Shaw et al., 2007) regions fromthe plastid genome. The ETS region of the nuclear18S/26S rDNA was amplified using the 18S-IGSprimer of Baldwin & Markos (1998) with a customprimer designed to amplify ETS in Lamiales (ETS-B:5′-ATAGAGCGCGTGAGTGGTG-3′). The AT1G09680and AT5G39980 PPR genes (hereafter referred to asPPR11 and PPR123, from the order in which they arelisted by Yuan et al., 2009a) were amplified usingprimers optimized for use in Verbenaceae by Yuanet al. (2009b). Primers specific to the AT3G25970region (hereafter referred to as PPR81; Yuan et al.,2009a) in Verbenaceae were designed following theprocedure outlined by Yuan et al. (2009a); the followingprimers were successfully used to amplify a fragmentof the coding sequence of approximately 1.2 kb inlength: PPR81-400f (5′-AGTGCRCTTTTWGATATGTAYGCAAAGTG-3′) and PPR81-1630r (5′-TCRACTGCACATGCRTAATKTTCCAT-3′). All PCR products werepurified by polyethylene glycol (PEG) precipitation.

Cycle sequencing reactions were carried out ina Perkin-Elmer thermocycler using BigDye ver. 3.1(Applied Biosystems Inc.), following a standardApplied Biosystems sequencing protocol. For all lociexcept ETS, internal sequencing primers were usedin addition to PCR primers to obtain overlapping

PHYLOGENY OF LANTANEAE 105


reads across fragments (see also Supporting Informa-tion, Table S2). Products of sequencing reactionswere purified by precipitation in sodium acetateand ethanol or by passing through Sephadex G-50columns. An Applied Biosystems genetic analyser wasused to generate raw sequence data; the reads werethen edited and assembled using Sequencher (GeneCodes Corp.).

PHYLOGENETIC ANALYSES

Sequences were aligned using MAFFT ver. 6 online(Katoh et al., 2002); alignments were then inspectedand manually adjusted where necessary. Sequencealignments for the three plastid loci (trnT-L, trnL-rpl32 and trnQ-rps16) were concatenated and ana-lysed as a single data set. Alignments for nuclearloci were treated as separate data sets. Phylogeneticreconstructions were performed individually for eachdata set, and for a supermatrix consisting of datafrom all loci (plastid and nuclear) in concatenation.The supermatrix was treated as consisting of a singlepartition.

The suitability of different models of evolution to thedata was assessed using jModeltest 0.1 (Posada, 2008).The GTR + I + G model was selected and applied to allanalyses. Phylogenetic reconstructions for individualdata sets and supermatrices were carried out usingmaximum likelihood and Bayesian approaches, asimplemented in GARLI (ver. 2.0; Zwickl, 2006) andMrBayes (ver. 3.1.2; Ronquist & Huelsenbeck, 2003).Shimodaira–Hasegawa (SH) tests (Shimodaira &Hasegawa, 1999) were carried out to gauge the com-patibility of the results of analyses of individual lociwith one another. Tree likelihood scores were calcu-lated and SH tests performed using PAUP* ver. 4b10(Swofford, 2000), with RELL optimization and 5000replicates under the GTR + I + G model.

Maximum likelihood analyses used two replicateruns, which were run with the generation thresholdfor termination at 20 000 generations, and termina-tion score threshold 0.05. Bootstrapping was carriedout with 100 replicates, with the generation thresholdfor termination lowered to 10 000 to facilitate fasteranalysis (as recommended in the GARLI manual, ver.0.96).

Bayesian analyses used two replicate runs, eachconsisting of four chains, which were run for at leastone million generations, and sampled every 1000 gen-erations. Convergence between runs was assessed byexamining standard deviations of split frequencies,and by using AWTY (Wilgenbusch, Warren & Swof-ford, 2004) to plot split frequencies over differentruns. Analyses that had not converged after onemillion generations were run until convergence diag-nostics indicated they had reached stationarity (up to

50 million generations). Longer MrBayes analyseswere carried out using the NSF TeraGrid via theCIPRES portal (Miller, Pfeiffer & Schwartz, 2010).When summarizing consensus trees over all runs, thefirst 25% of sampled trees were considered burn-inand were discarded.

FRUIT EVOLUTION AND BIOGEOGRAPHY

A semi-strict (combinable component) consensus treebetween trees inferred from different loci was con-structed using PAUP* (Swofford, 2000); relationshipsthat were not well supported in individual trees (boot-strap value > 80% and posterior probability > 0.9)were considered unresolved and collapsed beforecreating the consensus. We used Mesquite ver. 2.75(Maddison & Maddison, 2011) to score taxonomicallyimportant fruit characters and geographical distribu-tions, to map them onto the consensus tree and toinfer the most parsimonious character states anddistributions at ancestral nodes.

RESULTSDATA COLLECTION

Complete or nearly complete sequences of each targetlocus were obtained for the majority of taxa includedin this study. Only the sequences of the PPR81 locusfor three taxa (Lippia rehmannii H. Pearson, Lantanarugosa Thunb., Burroughsia fastigiata (Brandegee)Moldenke) were not available; sequences for thesetaxa were treated as missing data in the phylogeneticanalyses from all concatenated sequences, and notincluded in the individual analyses of the PPR81locus. A few other sequences were partial for sometaxa, or included short regions of missing data (DNAfrom herbarium specimens was occasionally of poorquality, making amplification difficult). The ETSregion for Lippia origanoides Kunth. was amplifiedand sequenced from a different DNA accession(individual) from that which provided sequences forother loci; ETS could not be sequenced directly fromthe original accession because of a length polymor-phism. The sequences from ETS and from the PPRloci contained some single nucleotide allelic differ-ences within individuals, which were scored as poly-morphisms in alignments.

The total aligned sequence data gathered were400 bp of ETS (all taxa), 1180 bp of PPR11 [exceptLippia lupulina Cham.: 761 bp; Lippia diamantinensisGlaz.: 854 bp; Lantana trifolia L.: 753 bp; Citharexy-lum montevidense (Spreng) Moldenke: regions ofmissing sequence totalling 253 bp], 1059 bp of PPR81[except Dipyrena glaberrima (Gillies & Hook.) Hook.:914 bp, Lippia dulcis Trevir.: 913 bp, Lippia javanica(Burm.f.) Spreng: 923 bp; sequences from Lippia



rehmannii, Lantana rugosa and Burroughsia fastig-iata were excluded], 1047 bp of PPR123 (except Lippialupulina: 773 bp). Plastid loci were completely ampli-fied and sequenced for all taxa [except trnQ-rps16 ofPhyla nodiflora (L.) Greene, for which approximately250 bp were missing from the 3′ end]. Plastid locivaried in length, from 626–698 bp for trnT-L frag-ments, 738–1010 bp for trnL-rpl32 fragments and1065–1652 bp for trnQ-rps16 fragments and, in com-bination, provided 4335 bp of aligned sequence data.After alignment and concatenation, the supermatrix ofall sequence data consisted of 8734 aligned positions.

PHYLOGENETIC ANALYSES

The results of phylogenetic reconstructions fromindividual data sets are depicted in Figures 1–3A;Figure 3B shows the results of phylogenetic analysisof the supermatrix consisting of all data in concate-nation. In SH tests, individual data sets all rejectedthe best likelihood trees from the other data sets withP < 0.000 (see also Supporting Information, Table S3).The combined tree was rejected with P < 0.05 by theplastid data, PPR81 and PPR123, but not by ETS(P = 0.118) and PPR11 (P = 0.09).

Well-supported clades are consistent between themaximum likelihood and Bayesian analyses for eachdata set; relationships that are resolved differently bymaximum likelihood and Bayesian analyses receivelow support. Three out of five gene trees place Coe-locarpum in a sister relationship with the rest ofLantaneae, with good support; conflicting topologiesreceive poor support in the other two gene trees. Twowell-supported clades of Aloysia spp. are present inall gene trees: the Aloysia citriodora clade, and theAloysia gratissima (Gillies & Hook.) L. D.Bensonclade, which includes Xeroaloysia ovatifolia (Mold-enke) Tronc. However, there is conflict between genetrees about whether these two clades together forma clade (ETS and plastid trees do not feature thisclade; all three PPR genes do). The tree inferred fromplastid data places Acantholippia salsoloides as sisterto the A. citriodora clade, with good support, but treesfrom the four nuclear loci place this species in variousother relationships, with varying levels of support.The tree inferred from all loci in concatenation isconsistent with the plastid gene tree with regard tothe placement of Coelocarpum, the two Aloysia cladesmentioned above, and A. salsoloides. Acantholippiaseriphioides (A.Gray) Moldenke is consistentlyreconstructed in a well-supported sister relationshipwith a large clade comprising all sampled species ofLantana and Lippia. This large Lantana–Lippiaclade also contains the sampled members of Nashia,Burroughsia and Phyla and one Aloysia sp. [Aloysiabarbata (Brandegee) Moldenke].

FRUIT EVOLUTION AND BIOGEOGRAPHY

The consensus between well-supported topologies ofindividual data sets is shown in Figures 4 and 5. Fruitcharacters important in separating Lantana fromLippia are mapped in Figure 4, with parsimony recon-structions of ancestral states. Geographical rangesof members of Lantaneae sampled in this study aremapped in Figure 5, with putative ancestral distribu-tions inferred by parsimony.

DISCUSSION

These results provide the first phylogenetic hypoth-eses for Lantaneae, which are broadly sampled andsufficiently resolved to reveal the major groups withinthe tribe. These major clades are consistent betweengene trees, despite some points of incongruence intheir relationships to one another and the relation-ships among taxa within them. The monophyly ofLantaneae sensu Marx et al. (2010) is confirmed. Theshort branch lengths in the tribe, particularly in theLantana–Lippia clade, are consistent with a recentradiation. We find strong evidence for the non-monophyly of the major genera of Lantaneae. Speciesof Lantana and Lippia are interspersed throughoutthe Lantana–Lippia clade, and Nashia, Burroughsiaand Phyla are nested in it, as is a lineage of Aloysiaspp. The remaining Aloysia spp. sampled here areallied with Acantholippia spp. and Xeroaloysia in agrade leading to the Lantana–Lippia clade. Majortaxonomic revisions are required in Lantaneae; inorder to achieve monophyletic genera, Lantana andLippia must either be fragmented into many smallergenera, or lumped into a single genus. Our phyloge-netic analysis reveals multiple independent shifts inthe fruit characteristics historically used to diagnosegenera (fleshiness and number of pyrenes); we alsoshow that the African members of Lantaneae repre-sent at least two independent colonization events.The finding that the Lantana camara species complexis not immediately related to most other Lantanaspp. is of note to tropical conservationists investigat-ing biological means to control invasive L. camarapopulations.

ANALYSES OF INDIVIDUAL DATA SETS

Areas of each individual tree that did not receive goodsupport were sometimes reconstructed differently bythe different methods of phylogenetic inference usedhere (indicated by dashed lines in Figs 1–3). This isprobably indicative of a lack of phylogenetic signal inthe data in these areas.

The contrast between the relatively slow rateof change of the plastid genome and the higher sub-stitution rates of the nuclear genome is evident in the



Figure 1. See caption on next page.



branch lengths and resolution of the trees shownin Figures 1–3A (note that the ETS tree is drawn tohalf the scale of the other trees). The concatenatedplastid matrix was several times the length (alignedpositions) of any other locus sequenced, but didnot provide enough information to resolve relation-ships in the Lantana–Lippia clade (although deepernodes in Lantaneae were resolved with confidence).This is consistent with our expectations and withfindings in the sister group to Lantaneae, Verbeneae(Yuan & Olmstead, 2008a, b). Plastid sequencewould be needed in great quantities compared withnuclear sequence in order to provide enough informa-tion to resolve relationships at the species level inLantaneae.

Plastid data could not resolve relationshipsbetween closely related species and the rapidly evolv-ing nuclear ETS region failed to resolve many of thedeeper nodes with confidence. In contrast, sequencesfrom PPR genes provided the greatest resolutionover the whole tree. All nuclear loci sequenced for thisstudy had polymorphic sites in some individuals,which did not affect direct sequencing (they werecoded as polymorphisms in alignments). Some allelicvariation is to be expected of nuclear loci, but wouldrequire the isolation of individual alleles via cloningin order to be studied in more detail.

INCONGRUENCE BETWEEN LOCI

Trees reconstructed from different individual datasets differ in their topologies, and are not compatiblewith one another according to SH topology tests (seealso Supporting Information, Table S3). However,most of the differences are in relationships thatare not well supported, and are thus probably bestexplained by insufficent information and/or noise(‘soft incongruence’; Seelanan, Schnabel & Wendel,1997). Our results also include a few instancesof well-supported incongruence between loci withrespect to the placement of (1) Dipyrena glaberrimaamong the outgroups, (2) Acantholippia salsoloides,(3) Lippia rhodocnemis Mart. & Schauer/Lippia her-mannioides Cham. and (4) Lippia aristata Schauer.

Conflict between different loci over the placement ofDipyrena glaberrima has been previously reported(Marx et al., 2010) and, whereas it lies outside the

scope of this study, the question of which topologybest reflects the evolutionary history of this speciesremains open. The position of Acantholippia salsolo-ides relative to the two Aloysia clades will affect howAcantholippia and Aloysia are recircumscribed andshould be resolved before revision can take place.Given the generally poor resolution of the backboneof the Lantana–Lippia clade, a future study usingdenser sampling and additional loci would berequired to study the evolution of this group in detail,and the placement of Lippia rhodocnemis and Lippiaaristata would be best addressed therein. The situa-tion in which the position of a few lineages are instrongly supported conflict between gene trees wasalso found in Verbeneae, the sister tribe of Lantaneae(Yuan & Olmstead, 2008a, b; O’Leary et al., 2009;Yuan et al., 2009b), and in the problematic Neotropi-cal palm tribe Bactridinae (Eiserhardt et al., 2011;Ludeña et al., 2011). In these examples, the questionof how the conflicting lineages are related to oneanother, and to other lineages within their respectivetribes, also has yet to be resolved.

When phylogenetic signals between gene trees arein conflict, the pattern of species divergence is some-times best represented by the combined phylogeneticsignals; i.e. the best estimate of the species tree isprovided by analysing the conflicting loci in concate-nation (the total evidence approach; Kluge, 1989).This approach provides a good approximation of thespecies tree under circumstances when stochasticerror in the finite data partitions is the cause ofincongruence (Olmstead & Sweere, 1994; Gadakgar,Rosenberg & Kumar, 2005) and is an attractive pros-pect when individual data sets do not provide enoughinformation to resolve a tree. Combined analyseshave been commonly performed in phylogeneticstudies over the last 10–20 years (reviewed briefly byEdwards, 2009; recent examples in Neotropical plantsinclude studies by Sass & Specht, 2010; Eiserhardtet al., 2011). However, analysis of combined datadoes not reliably reflect the species tree under othercircumstances, such as when conflicting evolutionaryhistories underlie individual genes as a result ofincomplete lineage sorting, hybridization or geneduplication and extinction (Maddison, 1997; Slowin-ski & Page, 1999; Kubatko & Degnan, 2007). Alter-native approaches, most commonly assuming that

Figure 1. A, maximum likelihood phylogeny inferred from DNA sequences from nuclear ETS (400 bp), for 47 species ofLantaneae and seven outgroup species. Branches in bold are supported by > 80% of bootstrap replicates and posteriorprobability values > 0.9 in Bayesian analyses of the same data. Dashed lines indicate branches not present in thephylogeny inferred by Bayesian analysis. B, maximum likelihood phylogeny inferred from DNA sequences from nuclearlocus PPR 11 (1180 bp), for 47 species of Lantaneae and seven outgroup species. Branches in bold are supported by > 80%of bootstrap replicates and posterior probability values > 0.9 in Bayesian analyses of the same data. Dashed lines indicatebranches not present in the phylogeny inferred by Bayesian analysis.�



incomplete lineage sorting is the cause of incongru-ence, rely on coalescent theory to infer the most likelyspecies tree from a number of individual gene trees(e.g. Liu, 2008; Kubatko, Carstens & Knowles, 2009;Heled & Drummond, 2010; for review, see Knowles,2009; Degnan & Rosenberg, 2009).

Unfortunately, no widely accessible method yetexists to tease apart the effects of incomplete lineagesorting from hybridization and gene duplication/extinction (but see Than & Nakhleh, 2009; Choi & Hey,2011). Any of these mechanisms could be the causeof the incongruence seen among our data sets forLantaneae. It might even be the case that there is nosingle bifurcating tree that adequately describes thepattern of descent of the species of Lantaneae fromtheir common ancestor; polytomy and reticulation maybe characteristic of evolutionary history in difficult,recently diversified groups such as Lantaneae.

Although the tree inferred from our combined datais fully resolved with reasonable support (Fig. 3B), wedo not assume that it necessarily corresponds withthe Lantaneae species tree. Relationships that are inconflict between loci are often resolved in favour ofthe larger data sets, or of the majority of data sets; i.e.minority conflicting signals from individual loci aremasked in the combined analysis, although they mayprovide equally valid alternative estimates of phylog-eny. We feel that it is more conservative and morerepresentative of our current understanding to leaveunresolved any nodes where well-supported conflictexists. We thus consider the semi-strict consensusbetween well-supported topologies of individual genetrees to be the best current estimate of Lantaneaephylogeny.

TAXONOMIC IMPLICATIONS

Marx et al. (2010) considered the assignment of Coe-locarpum to Lantaneae to be discordant, given themajor morphological differences between this genusand the other members of the tribe, but could notplace it with confidence as a lineage separate from therest of Lantaneae. Our results open the possibility ofexcluding Coelocarpum from Lantaneae and confirm-ing the monophyly of the tribe, whether Coelocarpumis included or not. However, none of the genera ofLantaneae represented here by more than one species

is monophyletic. Acantholippia contains two distinctlineages, Aloysia contains at least two (the relation-ship of the A. citriodora clade to the A. gratissimaclade should be considered equivocal, pending furtherinvestigation, and denser sampling, of these groupsand of Acantholippia). Lantana spp. form two distinctclades, a Lantana trifolia clade and a Lantanacamara clade. Lippia spp. are distributed throughoutthe Lantana–Lippia clade, and form the backgroundfrom which Nashia inaguensis Millsp., Burroughsiafastigiata, Phyla nodiflora, Aloysia barbata and thetwo Lantana clades are derived.

Our results show that assuming correspondencebetween traditional taxa and evolutionary lineages isnot valid in Lantaneae and should not be accepteduncritically in other, difficult Neotropical groups.Generic revisions in Lantaneae should proceed care-fully, contingent on thorough re-evaluation of themorphological characters that correspond with evolu-tionary lineages. Based on our results, Lantana andLippia will need to be fragmented or lumped togetherwith the smaller genera which nest in the Lantana–Lippia clade. In either scenario, genera will not beeasy to define morphologically. We can identify nomorphological characteristics that have not under-gone multiple, parallel shifts among the major cladesof Lantaneae. Taxonomic revisions in the tribe willprobably involve recircumscribing genera based oncombinations of traits, rather than on one to a fewdiagnostic characters. Densely sampled molecularphylogenetic studies are needed to investigate eachclade of Lantaneae, guided by the broad phylogeneticresults published here, before reliable revisions canbe made.

FRUIT EVOLUTION

Classifications in Lantaneae have relied largely onfruit characteristics to separate its principal genera,Lantana and Lippia. Schauer (1847), followed byTroncoso (1974), assigned species with fleshy drupesto Lantana and species with dry schizocarps toLippia. Under this scheme, Lippia brasiliensis (Link)T. R. S.Silva and Lippia macrophylla Cham. areplaced in Lantana section Sarcolippia. More recentrevisions (Silva, 1999) follow Chamisso (1832) bydefining Lippia as species with divided fruits, thus

Figure 2. A, maximum likelihood phylogeny inferred from DNA sequences from nuclear locus PPR 81 (1059 bp), for 44species of Lantaneae and seven outgroup species. Branches in bold are supported by > 80% of bootstrap replicates andposterior probability values > 0.9 in Bayesian analyses of the same data. Dashed lines indicate branches not present inthe phylogeny inferred by Bayesian analysis. B, maximum likelihood phylogeny inferred from DNA sequences fromnuclear locus PPR 123 (1047 bp), for 47 species of Lantaneae and seven outgroup species. Branches in bold are supportedby > 80% of bootstrap replicates and posterior probability values > 0.9 in Bayesian analyses of the same data. Dashed linesindicate branches not present in the phylogeny inferred by Bayesian analysis.�



Figure 3. A, maximum likelihood phylogeny inferred from DNA sequences from three plastid loci in combination (4335aligned positions) for 47 species of Lantaneae and seven outgroup species. Branches in bold are supported by > 80% ofbootstrap replicates and posterior probability values > 0.9 in Bayesian analyses of the same data. Dashed lines indicatebranches not present in the phylogeny inferred by Bayesian analysis. B, maximum likelihood phylogeny inferred from allDNA sequences in combination (8734 aligned positions), for 47 species of Lantaneae and seven outgroup species. Branchesin bold are supported by > 80% of bootstrap replicates and posterior probability values > 0.9 in Bayesian analyses of thesame data. Dashed lines indicate branches not present in the phylogeny inferred by Bayesian analysis.�

Figure 4. Semi-strict consensus between well-supported topologies of individual phylogenies for Lantaneae, with fruitcharacters mapped as indicated (left: fruit type; right: number of pyrenes/mericarps). Character states at ancestral nodesare parsimony reconstructions.



grouping dry schizocarps together with dipyrenousdrupes under Lippia, and limiting Lantana to includeonly species with monopyrenous drupes. This morerecent scheme reassigns dipyrenous fleshy-fruitedspecies such as L. brasiliensis and L. macrophylla toLippia. Our results show that both of these classifi-

cation schemes are artificial, confounded by charac-ters that have undergone multiple independent shiftsin different lineages.

There have been at least five origins of a fleshy orleathery outer layer on the fruit in Lantaneae, fourof them in the Lantana–Lippia clade (Fig. 4). Fleshy

Figure 5. Semi-strict consensus between well-supported topologies of individual phylogenies for Lantaneae, with geo-graphical distributions mapped as indicated; species occurring in more than one coded region are denoted with anadditional circle. Distributions at ancestral nodes are parsimony reconstructions. Inset A, distribution of occurrencerecords for the species of Lantaneae included in this study (data from GBIF; records of globally invasive species andspecies with no georeferenced records omitted).



fruited lineages identified in our results are: (1)the Lantana trifolia clade; (2) the Lantana camaraclade; (3) Nashia; (4) the clade corresponding to thetraditional Lantana section Sarcolippia (representedhere by L. brasiliensis and L. macrophylla); and (5)Xeroaloysia. Whether or not the common ancestorof Coelocarpum + Lantaneae had fleshy fruits isdifficult to infer, attributable to the difficulty inplacing fleshy-fruited Dipyrena relative to dry-fruitedVerbeneae and Lantaneae. If Dipyrena is sister toVerbeneae + Lantaneae, it is most parsimonious toreconstruct a dry-fruited ancestor for Lantaneae andhypothesize that Coelocarpum represents anotherindependent derivation of fleshy fruits (as shown inFig. 4). If, however, Dipyrena is sister to Verbeneae(rather than to Verbeneae + Lantaneae), a fleshy-fruited ancestor for Lantaneae is the more parsimo-nious hypothesis.

In the Lantana–Lippia clade, the independentderivation of fleshy drupes from dry schizocarpshas resulted in dipyrenous fruits in two lineages (theSarcolippia clade and Nashia) and monopyrenousfruits in two lineages (the L. camara clade, andthe L. trifolia clade). In the L. trifolia clade,Lippia aristata represents a subsequent shiftfrom monopyrenous fruits to dipyrenous fruits. Thepattern of shifts in fruit type (dry to fleshy) andsubdivision (two mericarps to two pyrenes or to onepyrene; one pyrene to two pyrenes) in the Lantana–Lippia clade reveals a complex history of fruit evolu-tion, which has had the consequence of misleadingtaxonomic efforts based on fruit characteristics.

BIOGEOGRAPHIC PATTERNS

Major clades in Lantaneae are geographically hetero-geneous, suggesting that migration has been animportant and common element in the evolutionof Lantaneae (Fig. 5). Old World representatives ofLantaneae can be accounted for by at least threeintercontinental colonization events. Coelocarpum,endemic to Madagascar and Socotra, is sister to therest of Lantaneae, and represents one lineage thathas dispersed to the Old World (Marx et al., 2010;Olmstead, 2012). Similar patterns of disjunctionbetween sister lineages (with distributions in theNew World and in Madagascar) are found in otherfamilies; for example, Tsoala Bosser & D’Arcy inSolanaceae (Olmstead et al., 2008) and groups ofFabaceae (Lavin et al., 2000; 2004). The legumes areparticularly well-studied examples, in which largeshifts in geographical range belie a high degree ofniche conservatism (Lavin et al., 2004). In addition toCoelocarpum, two long-distance dispersals from theNeotropics to Africa are inferred in the Lantana–Lippia clade: a lineage in a Lippia clade (represented

here by L. rehmannii and L. javanica); and a lineagein a Lantana clade [represented here by L. vibur-noides (Forssk.) Vahl and L. rugosa]. This frequencyof colonization of Africa seems high, given that Lan-taneae are a young lineage and that long-distancedispersal between Africa and South America has beenfound to be relatively infrequent in other lineages(Crisp et al., 2009).

In the Americas, a geographical shift fromtemperate/subtropical regions into the tropics can beseen in Lantaneae (Fig. 5). Aloysia and Acantholip-pia, which form a grade at the base of the tribe, aredistributed primarily in arid temperate regions ofSouth America, extending north into the Andes.Aloysia has an amphitropical distribution with a sec-ondary radiation in Mexico and the south-westernUSA, which may be the result of long-distance dis-persal (P. Lu-Irving & R. G. Olmstead, unpubl. data).Members of the Lantana–Lippia clade, derived fromthe grade of Aloysia and Acantholippia, are foundthroughout the tropics. This suggests a generalpattern of movement into the tropics from the aridtemperate or subtropical regions of South Americaduring the evolution of Lantaneae.

Members of Lantaneae mainly occur in dry to semi-arid habitats, and rarely in wet forest environments.For example, Acantholippia seriphioides, which issister to the rest of the Lantana–Lippia clade, inhab-its arid uplands in Argentina, whereas the nextlineage to diverge consists of low or creeping suf-frutescent herbs found in dry scrub and dry to mesicdisturbed habitats. Most of the rest of the clade arewoody shrubs of open and disturbed habitats, forestedges, dry hills and cerrado. Occurrence records forthe species of Lantaneae sampled here (Fig. 5, insetA) reveal geographical distributions that mostlyexclude the Amazon or wet coastal forests, andcorrespond with the distribution of seasonally drytropical forest and chaco biomes as outlined by Pen-nington, Lavin & Oliveira-Filho (2009). The lineagecorresponding to Lantana section Sarcolippia repre-sents a shift to wetter and more closed forest envi-ronments, but, this shift notwithstanding, the overallbiogeographical pattern in Lantaneae is one of nicheconservatism. Verbeneae, the sister clade to Lanta-neae, generally occur in dry to semi-arid habitats intemperate zones, and are not diverse in the tropics.Aloysia spp. echo this pattern and, in the colonizationof the tropics represented by the Lantana–Lippiaclade, the environmental preferences of most of thesespecies reflect those of their ancestors. This is con-sistent with findings that biome shifts are uncommonamong plant lineages (Crisp et al., 2009; but see alsoSimon et al., 2009).

There is no discernible correlation between fruittype (whether fleshy or dry) and biogeographic



patterns. A more densely sampled and fully resolvedphylogenetic hypothesis might reveal such a correla-tion, but, to date, if there is any consistent dispersaladvantage possessed by fleshy-fruited species in Lan-taneae, it is not apparent. In many dry-fruited species,segments of the hairy calyx persistently enclose themericarp, facilitating ectozoochory, just as the fleshyfruits are adapted to endozoochory. The different dis-persal strategies employed among members of Lanta-neae have not been broadly studied, and are likely tobe diverse in such a large and varied tribe.

FUTURE PROSPECTS

With a broadly representative sample of Lantaneae,we have identified major clades in the tribe andrevealed the extent to which they do or do not corre-spond with accepted genera. With the evolutionaryhistory of lineages of Lantaneae outlined here, futuresystematic studies can target specific groups for thedense sampling that will probably be necessary toelucidate relationships at the species level. Particu-larly important areas that have yet to be resolvedare: (1) the relationship of Acantholippia salsoloidesand its (unsampled) affiliates with Aloysia spp. (thiswill determine how these genera are redefined); and(2) species-level relationships in the Lantana–Lippiaclade (these will reveal the patterns of trait andbiogeographic evolution among these many species).

In Lantaneae, as in other problematic Neotropicalgroups (e.g. Bactridinae, Bromelioideae and otherexamples cited above), a phylogenetic estimate usingmolecular data is essential as a basis for reliabletaxonomic revisions and speculation on evolutionaryhistory. The difficult taxonomy of such groups hints atthe complex pattern of homoplasy that may exist inmorphological characters used to define taxa. Sharedancestry among lineages cannot be unambiguouslyinferred from morphology alone. Molecular phyloge-netic studies of difficult Neotropical groups shouldconsider evidence from multiple, independent loci. Ifmajor points of departure between gene historiesexist among the species under investigation, theycan be discovered by taking a multi-locus approach.It is important to evaluate possible incongruencebetween gene trees, to avoid providing an inappropri-ate interpretation of the species tree. Lineages thatare species-rich and recently radiated may be particu-larly prone to the incongruence among phylogeneticsignal from different loci that is attributable to incom-plete lineage sorting and hybridization.

In recently radiated lineages, nucleotide variabilitybetween taxa is an important criterion when selectingloci from which to infer phylogeny. Resolving mater-nal relationships at the level of species is a valuablecomponent of phylogenetic studies, but is likely to

require large quantities of sequence data from rapidlyevolving DNA regions in problematic, species-rich lin-eages. Individual plastid loci are unlikely to providesufficient phylogenetic information in such groups.If a molecular systematic study is to be undertakenin a difficult group, such as Lantaneae, a period ofextensive preliminary work should first be carried outin order to develop, evaluate and select the loci toprovide data for it. We expect that the potential of thenuclear genome as a resource for phylogenetic infor-mation will be largely realized over the next decade.Growing access to complete genome sequences acrossa range of plant species will enable a variety ofmulti-locus approaches to be developed and applied indivergent groups of flowering plants. With continuingadvances in sequencing technologies, we predict thatlarge-scale sequencing approaches such as restrictionsite-associated DNA (RAD) tagging (Miller et al.,2007; Baird et al., 2008) and large-scale alignment ofentire linkage groups will replace the use of sets ofwell-characterized loci for phylogenetic studies.

Most taxonomic and phylogenetic studies in large,geographically widespread plant groups are subjectto trade-offs between geographical and taxonomiccomprehensiveness and between breadth and depthin treating the taxa in question. Broad molecularsystematic studies across large groups guide the sam-pling of subsequent work focused on particular line-ages in those groups. Our phylogenetic estimate forLantaneae was guided by a previous, broader studyof Verbenaceae (Marx et al., 2010) and, in turn, willprovide a foundation for further efforts to revisegenera, elucidate patterns of trait evolution at thespecies level and understand patterns of migrationand colonization among the Neotropical flora better.As phylogenetic data become more easily obtainablein larger quantities, the trade-off between breadthand depth should become less limiting, and we expectthat large, data-rich studies that are broadly anddensely sampled will become more common. Movingforward, collaborative efforts will be needed to thor-oughly represent species-rich and geographicallywidespread groups in molecular phylogenetic studiesat a range of taxonomic levels. The development ofcollaborative networks across international bounda-ries will be important in coming years, as we pool ourefforts and expertise to advance our understanding ofevolution in problematic Neotropical plant groups.

ACKNOWLEDGEMENTS

This study was supported by the National ScienceFoundation (NSF) (grants DEB 0542493 and DEB1020369). The first author (P.L.I.) was able to partici-pate in this Symposium thanks to travel funds pro-vided by the NSF to students of the American Society



of Plant Taxonomists and the Botanical Society ofAmerica (DEB 1120802). Additional support, in theform of Graduate Student Research Awards, wasprovided by the Society of Systematic Biologists,the American Society of Plant Taxonomists and theBotanical Society of America. We are grateful toSegundo Leiva, Fátima Salimena, Lyderson Viccini,Loreta Freitas and Verônica Thode for assistance inthe field and for providing plant material for thisstudy. Additional plant material was obtained withthe help of the Plant Resources Center at the Uni-versity of Texas (TEX-LL), the Missouri BotanicalGarden (MO) and the US National Herbarium (US).Valuable discussion and comments on the manuscriptwere provided by Yuan Yaowu, Ryan Miller, theLeaché laboratory at the University of Washingtonand three anonymous reviewers.

REFERENCES

Atkins S. 2004. Verbenaceae. In: Kadereit JW, ed. The fami-lies and genera of flowering plants, Vol. 7. Berlin: Springer-Verlag, 449–468.

Baird NA, Etter PD, Atwood TS, Currey MC, Shiver AL,Lewis ZA, Selker EU, Cresko WA, Johnson EA. 2008.Rapid SNP discovery and genetic mapping using sequencedRAD markers. PLoS ONE 3: e3376.

Baldwin B, Markos S. 1998. Phylogenetic utility of theexternal transcribed spacer (ETS) of 18S-26S rDNA:congruence of ETS and ITS trees of Calycadenia (Composi-tae). Molecular Phylogenetics and Evolution 10: 449–463.

Bárcenas RT, Yesson C, Hawkins JA. 2011. Molecularsystematics of the Cactaceae. Cladistics 27: 470–489.

Briquet I. 1895. Verbenaceae. In: Engler A, Prantl K, eds.Die natürlichen Pflanzenfamilien, Vol. 4. Leipzig: W. Engel-mann, 132–182.

Briquet I. 1904. Verbenaceae. Plantae Hassleriane. Bulletinde l’Herbier Boissier ser. 2, 4: 1062–1066.

Butterworth CA, Wallace RS. 2004. Phylogenetic studiesof Mammillaria (Cactaceae) – insights from chloroplastsequence variation and hypothesis testing using the para-metric bootstrap. American Journal of Botany 91: 1086–1098.

Chamisso A. 1832. De plantis in Expeditione Romanzoffianaobservatis dicunt. Verbenaceae. Linnaea 7: 105–128, 213–272.

Chatrou LW, Pirie MD, Erkens RHJ, Couvreur TLP,Neubig KM, Abbott JR, Mols JB, Maas JW, SaundersRMK, Chase MW. 2012. A new subfamilial and tribalclassification of the pantropical flowering plant familyAnnonaceae informed by molecular phylogenetics. BotanicalJournal of the Linnean Society 169: 5–40.

Choi SC, Hey J. 2011. Joint inference of population assign-ment and demographic history. Genetics 189: 561–577.

Crisp MD, Arroyo MTK, Cook LG, Gandolfo MA, JordanGJ, McGlone MS, Weston PH, Westoby M, Wilf P,

Linder P. 2009. Phylogenetic biome conservatism on aglobal scale. Nature 458: 754–756.

Degnan JH, Rosenberg NA. 2009. Gene tree discordance,phylogenetic inference, and the multispecies coalescent.Trends in Ecology and Evolution 24: 332–340.

Doyle JJ, Doyle JL. 1987. A rapid DNA isolation procedurefor small quantities of fresh leaf tissue. Phytochemical Bul-letin, Botanical Society of America 19: 11–15.

Edwards SV. 2009. Is a new and general theory of molecularsystematics emerging? Evolution 63: 1–19.

Eiserhardt WL, Pintaud J-C, Asmussen-Lange C,Hahn WJ, Bernal R, Balslev H, Borchsenius F. 2011.Phylogeny and divergence times of Bactridinae (Arecaceae,Palmae) based on plastid and nuclear DNA sequences. Taxon60: 485–498.

Gadakgar SR, Rosenberg MS, Kumar S. 2005. Inferringspecies phylogenies from multiple genes: concatenatedsequence tree versus consensus gene tree. Journal of Experi-mental Zoology 304B: 64–74.

Gentry AH. 1982. Neotropical floristic diversity: phytogeo-graphical connections between Central and South America,Pleistocene climatic fluctuations, or an accident of theAndean orogeny? Annals of the Missouri Botanical Garden69: 557–593.

Givnish TJ, Barfuss MHJ, Van Ee B, Riina R, Schulte K,Horres R, Gonsiska PA, Jabaily RS, Crayn DM, SmithJAC, Winter K, Brown GK, Evans TM, Holst BK,Luther H, Till W, Zizka G, Berry PE, Sytsma K. 2011.Phylogeny, adaptive radiation, and historical biogeographyin Bromeliaceae: insights from an eight-locus plastid phyl-ogeny. American Journal of Botany 98: 872–895.

González D, Vovides AP, Bárcenas C. 2008. Phylogeneticrelationships of the Neotropical genus Dioon (Cycadales,Zamiaceae) based on nuclear and chloroplast DNA sequencedata. Systematic Botany 33: 229–236.

Grose SO, Olmstead RG. 2007. Evolution of a charismaticNeotropical clade: molecular phylogeny of Tabebuia s.l.,Crescentieae, and allied genera (Bignoniaceae). SystematicBotany 32: 650–659.

Heled J, Drummond AJ. 2010. Bayesian inference ofspecies trees from multilocus data. Molecular Biology andEvolution 27: 570–580.

Katoh K, Misawa K, Kuma K, Miyata T. 2002. MAFFT: anovel method for rapid multiple sequence alignment basedon fast Fourier transform. Nucleic Acids Research 30: 3059–3066.

Kluge AG. 1989. A concern for evidence and a phylogenetichypothesis of relationships among Epicrates (Boidae, Ser-pentes). Systematic Zoology 38: 7–25.

Knowles LL. 2009. Estimating species trees: methods ofphylogenetic analysis when there is incongruence acrossgenes. Systematic Biology 58: 463–467.

Kubatko LS, Carstens BC, Knowles LL. 2009. STEM:species tree estimation using maximum likelihood for genetrees under coalescence. Bioinformatics 25: 971–973.

Kubatko LS, Degnan JH. 2007. Inconsistency of phyloge-netic estimates from concatenated data under coalescence.Systematic Biology 56: 17–24.



Lavin M, Schrire BP, Lewis G, Pennington RT, Delgado-Salinas A, Thulin M, Hughes CE, Matos AB, Wojcie-chowski MF. 2004. Metacommunity process rather thancontinental tectonic history better explains geographicallystructured phylogenies in legumes. Philosophical Transac-tions of the Royal Society B: Biological Sciences 359: 1509–1522.

Lavin M, Thulin M, Labat J-N, Pennington RT. 2000.Africa, the odd man out: molecular biogeography of dalber-gioid legumes (Fabaceae) suggests otherwise. SystematicBotany 25: 449–467.

Levin RA, Whelan AP, Miller JS. 2009. The utility of nuclearconserved ortholog set II (COSII) genomic regions for species-level phylogenetic inference in Lycium (Solanaceae). Molecu-lar Phylogenetics and Evolution 53: 881–890.

Liu L. 2008. BEST: Bayesian estimation of species treesunder the coalescent model. Bioinformatics 24: 2542–2543.

Lohmann LG. 2006. Untangling the phylogeny of Neotropi-cal lianas (Bignonieae, Bignoniaceae). American Journal ofBotany 93: 304–318.

López-Palacios S. 1991. Aspectos críticos de la sistemáticade Verbenaceae de Venezuela. Addenda delenda et corri-genda a Verbenaceae Flora de Venezuela (1977). Pittieria19: 40–52.

Ludeña B, Chabrillange N, Aberlenc-Bertossi F,Adam H, Tregear JW, Pintaud J-C. 2011. Phylogeneticutility of the nuclear genes AGAMOUS 1 and phytochromeB in palms (Arecaceae): an example within Bactridinae.Annals of Botany 108: 1433–1444.

Maddison WP. 1997. Gene trees in species trees. SystematicBiology 46: 523–536.

Maddison WP, Maddison DR. 2011. Mesquite: a modularsystem for evolutionary analysis. Version 2.75. Available at:http://mesquiteproject.org

Marx HE, O’Leary N, Yuan YW, Lu-Irving P, Tank DC,Mulgura ME, Olmstead RG. 2010. A molecular phylogenyand classification of Verbenaceae. American Journal ofBotany 97: 1647–1663.

Miller MA, Pfeiffer W, Schwartz T. 2010. Creating theCIPRES gateway for inference of large phylogenetic trees.In: Proceedings of the Gateway Computing EnvironmentsWorkshop (GCE 2010). New Orleans, LA: Institute of Elec-trical and Electronics Engineers (IEEE), 1-8.

Miller MR, Dunham JP, Amores A, Cresko WA,Johnson EA. 2007. Rapid and cost-effective polymorphismidentification and genotyping using restriction site associ-ated DNA (RAD) markers. Genome Research 17: 240–248.

Moldenke HN. 1940. Novelties in the Avicenniaceae andVerbenaceae. Phytologia 1: 411–412.

Moldenke HN. 1959. A resume of the Verbenaceae, Avicen-niaceae, Stilbaceae, Symphoremaceae, and Eriocaulaceaeof the world as to valid taxa, geographic distribution andsynonymy. Yonkers: published by the author.

O’Leary N, Múlgura ME. 2012. A taxonomic revision of thegenus Phyla (Verbenaceae). Annals of the Missouri Botani-cal Garden 98: 578–596.

O’Leary N, Denham SS, Salimena F, Múlgura ME. 2012.Species delimitation in Lippia section Goniostachyum (Ver-

benaceae) using the phylogenetic species concept. BotanicalJournal of the Linnean Society 170: 197–219.

O’Leary N, Yuan YW, Chemisquy A, Olmstead RG. 2009.Reassignment of species of paraphyletic Junellia s.l. to thenew genus Mulguraea (Verbenaceae) and new circumscrip-tion of genus Junellia: molecular and morphological congru-ence. Systematic Botany 34: 777–786.

Olmstead RG. 2012. Phylogeny and biogeography inSolanaceae, Verbenaceae, and Bignoniaceae: a comparisonof continental and intercontinental diversification patterns.Botanical Journal of the Linnean Society 171: doi: 10.1111/j.1095-8339.2012.01306.x.

Olmstead RG, Bohs L, Abdel Migid H, Santiago-ValentinE, Garcia VF, Collier SM. 2008. A molecular phylogeny ofthe Solanaceae. Taxon 57: 1159–1181.

Olmstead RG, Sweere JA. 1994. Combining data in phylo-genetic systematics: an empirical approach using threemolecular data sets in the Solanaceae. Systematic Biology43: 467–481.

Olmstead RG, Zjhra ML, Lohmann LG, Grose SO,Eckert AJ. 2009. A molecular phylogeny and classificationof Bignoniaceae. American Journal of Botany 96: 1731–1743.

Pennington RT, Lavin M, Oliveira-Filho A. 2009.Woody plant diversity, evolution, and ecology in the tropics:perspectives from seasonally dry tropical forests. AnnualReview of Ecology, Evolution, and Systematics 40: 437–457.

Posada D. 2008. jModelTest: phylogenetic model averaging.Molecular Biology and Evolution 25: 1253–1256.

Ronquist F, Huelsenbeck JP. 2003. MRBAYES 3: Bayesianphylogenetic inference under mixed models. Bioinformatics19: 1572–1574.

Salimena FRG. 2002. New synonyms and typifications inLippia sect. Rhodolippia (Verbenaceae). Darwiniana 40:121–125.

Sanders RW. 1987. Taxonomic significance of chromosomeobservations in Caribbean species of Lantana (Verben-aceae). American Journal of Botany 74: 914–920.

Sanders RW. 2001. The genera of Verbenaceae in thesoutheastern United States. Harvard Papers in Botany 5:303–358.

Sanders RW. 2006. Taxonomy of Lantana sect. Lantana(Verbenaceae): I. Correct application of Lantana camaraand associated names. Sida 22: 381–421.

Santos IEM. 2002. A taxonomic revision of Lantana sect.Lantana (Verbenaceae) in the Greater Antilles. Willdenowia32: 285–301.

Sass C, Specht CD. 2010. Phylogenetic estimation of thecore bromelioids with an emphasis on the genus Aechmea(Bromeliaceae). Molecular Phylogenetics and Evolution 55:559–571.

Schauer JC. 1847. Verbenaceae. In: De Candolle AP, ed.Prodromus, Vol. 11. Lehre: Cramer. 522–700.

Schulte K, Barfuss MHJ, Zizka G. 2008. Phylogeny ofBromelioideae (Bromeliaceae) inferred from nuclear andplastid DNA loci reveals the evolution of the tank habitwithin the subfamily. Molecular Phylogenetics and Evolu-tion 51: 327–339.



Seelanan T, Schnabel A, Wendel JF. 1997. Congruenceand consensus in the cotton tribe (Malvaceae). SystematicBotany 22: 259–290.

Shaw J, Lickey EB, Schilling EE, Small RL. 2007. Com-parison of whole chloroplast genome sequences to choosenoncoding regions for phylogenetic studies in Angiosperms:the Tortoise and the Hare III. American Journal of Botany94: 275–288.

Shimodaira H, Hasegawa M. 1999. Multiple comparisons oflog-likelihoods with applications to phylogenetic inference.Molecular Biology and Evolution 16: 1114–1116.

Siedo SJ. 2008. Systematics of Aloysia (Verbenaceae).Doctoral thesis, University of Texas at Austin.

Silva TRS. 1999. Redelimitação e revisão taxonômica dogenero Lantana L. (Verbenaceae) no Brasil. Doctoral thesis,Universidade de São Paulo.

Silva TRS, Salimena FRG. 2002. Novas combinaçõese novos sinônimos em Lippia e Lantana (Verbenaceae).Darwiniana 40: 57–59.

Simon MF, Grether G, de Queiroz LP, Skema C,Pennington RT, Hughes CE. 2009. Recent assembly ofthe cerrado, a Neotropical plant diversity hotspot, by in situevolution of adaptations to fire. Proceedings of the NationalAcademy of Sciences of the United States of America 106:20359–20364.

Slowinski JB, Page RDM. 1999. How should species phyl-ogenies be inferred from sequence data? Systematic Biology48: 814–825.

Small RL, Cronn RC, Wendel JF. 2004. Use of nucleargenes for phylogeny reconstruction in plants. AustralianSystematic Botany 17: 145–170.

Steele PR, Guisinger-Bellian M, Linder R, Jansen RK.2008. Phylogenetic utility of 141 low-copy nuclear regions intaxa at different taxonomic levels in two distantly relatedfamilies of rosids. Molecular Phylogenetics and Evolution48: 1013–1026.

Swofford DL. 2000. PAUP*: phylogenetic analysis usingparsimony (*and other methods). Version 4b.10. Sunderland:Sinauer Associates.

Taberlet P, Gielly L, Pautou G, Bouvet J. 1991. Universalprimers for amplification of three non-coding regions ofchloroplast DNA. Plant Molecular Biology 17: 1105–1109.

Than C, Nakhleh L. 2009. Species tree inference by mini-mizing deep coalescences. PLoS Computational Biology 5:e1000501.

Tippery NP, Philbrick CT, Bove CP, Les DH. 2011. Sys-tematics and phylogeny of Neotropical riverweeds(Podostemaceae: Podostemoideae). Systematic Botany 36:105–118.

Torke BM, Schaal BA. 2008. Molecular phylogenetics ofthe species-rich Neotropical genus Swartzia (Leguminosae–Papilionoideae) and related genera of the Swartzioid clade.American Journal of Botany 95: 215–228.

Troncoso NS. 1974. Los géneros de Verbenaceas de Suda-merica extra-tropical (Argentina, Chile, Bolivia, Paraguai,Uruguay y sur de Brasil). Darwiniana 18: 295–412.

Verdcourt B. 1992. Verbenaceae. In: Polhill RM, ed. Flora ofTropical East Africa. Rotterdam: A.A. Balkema, 1–155.

Whittall JB, Medina-Marino A, Zimmer EA, Hodges SA.2006. Generating single-copy nuclear gene data for a recentadaptive radiation. Molecular Phylogenetics and Evolution39: 124–134.

Wilgenbusch JC, Warren DL, Swofford DL. 2004. AWTY:a system for graphical exploration of MCMC convergencein Bayesian phylogenetic inference. Available at: http://ceb.csit.fsu.edu/awty

Yuan YW, Liu C, Marx HE, Olmstead RG. 2009a. Thepentatricopeptide repeat (PPR) gene family, a tremendousresource for plant phylogenetic studies. New Phytologist182: 272–283.

Yuan YW, Liu C, Marx HE, Olmstead RG. 2009b. Anempirical demonstration of using pentatricopeptide repeat(PPR) genes as plant phylogenetic tools: phylogeny of Ver-benaceae and the Verbena complex. Molecular Phylogeneticsand Evolution 54: 23–35.

Yuan YW, Olmstead RG. 2008a. A species-level phylogeneticstudy of the Verbena complex (Verbenaceae) indicates twoindependent intergeneric chloroplast transfers. MolecularPhylogenetics and Evolution 48: 23–33.

Yuan YW, Olmstead RG. 2008b. Evolution and phylogeneticutility of the PHOT gene duplicates in the Verbena complex(Verbenaceae): dramatic intron size variation and footprintof ancestral recombination. American Journal of Botany 95:1166–1176.

Zwickl DJ. 2006. Genetic algorithm approaches for thephylogenetic analysis of large biological sequence datasetsunder the maximum likelihood criterion. PhD dissertation,The University of Texas at Austin. Available at: http://garli.googlecode.com

SUPPORTING INFORMATION

Additional Supporting Information may be found in the online version of this article:

Table S1. Voucher information for the specimens from which DNA was obtained, and GenBank identificationnumbers for the sequence data generated in this study.Table S2. Sequences of primers used in this study.Table S3. Results of SH tests (P-values) on trees inferred from different data sets.



Investigating the evolution of Lantaneae (Verbenaceae...

Documents

Transcript of Investigating the evolution of Lantaneae (Verbenaceae...