Molecular Phylogenetics and Evolution 95 (2016) 116–136

Contents lists available at ScienceDirect

Molecular Phylogenetics and Evolution

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

Divergence times, historical biogeography, and shifts in speciation ratesof Myrtalesq

http://dx.doi.org/10.1016/j.ympev.2015.10.0011055-7903/� 2015 Elsevier Inc. All rights reserved.

q This paper was edited by the Associate Editor Akiko Soejima.⇑ Corresponding author at: Department of Biological Sciences, St. John’s Univer-

sity, 8000 Utopia Parkway, Queens, NY 11432, USA. Fax: +1 608 262 4490.E-mail address: brent.a.berger@gmail.com (B.A. Berger).

Brent A. Berger a,b,⇑, Ricardo Kriebel b, Daniel Spalink b, Kenneth J. Sytsma b

aDepartment of Biological Sciences, St. John’s University, 8000 Utopia Parkway, Queens, NY 11432, USAbDepartment of Botany, University of Wisconsin-Madison, 430 Lincoln Dr., Madison, WI 53706, USA

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

Article history:Received 9 June 2015Revised 3 September 2015Accepted 4 October 2015Available online 14 November 2015

Keywords:BEASTBioGeoBEARSBAMMDiversificationGondwanaPhylogenetics

We examine the eudicot order Myrtales, a clade with strong Gondwanan representation for most of itsfamilies. Although previous phylogenetic studies greatly improved our understanding of intergenericand interspecific relationships within the order, our understanding of inter-familial relationships stillremains unresolved; hence, we also lack a robust time-calibrated chronogram to address hypotheses(e.g., biogeography and diversification rates) that have implicit time assumptions. Six loci (rbcL, ndhF,matK, matR, 18S, and 26S) were amplified and sequenced for 102 taxa across Myrtales for phylogeneticreconstruction and ten fossil priors were utilized to produce a chronogram in BEAST. Combretaceae isidentified as the sister clade to all remaining families with moderate support, and within the latter clade,two strongly supported groups are seen: (1) Onagraceae + Lythraceae, and (2) Melastomataceae + theCrypteroniaceae, Alzateaceae, Penaeaceae clade along with Myrtaceae + Vochysiaceae. Divergence timeestimates suggest Myrtales diverged from Geraniales �124 Mya during the Aptian of the EarlyCretaceous. The crown date for Myrtales is estimated at �116 Mya (Albian–Aptian). BioGeoBEARSshowed significant improvement in the likelihood score when the ‘‘jump dispersal” parameter wasadded. South America and/or Africa are implicated as important ancestral areas in all deeper nodes.BAMM analyses indicate that the best configuration included three significant shifts in diversificationrates within Myrtales: near the crown of Melastomataceae (�67–64 Mya), along the stem of subfamilyMyrtoideae (Myrtaceae; �75 Mya), and along the stem of tribe Combreteae (Combretaceae; �50–45 Mya). Issues with conducting diversification analyses more generally are examined in the contextof scale, taxon sampling, and larger sets of phylogenetic trees.

� 2015 Elsevier Inc. All rights reserved.

1. Introduction

Resolving deep-level phylogenetic relationships has often beendifficult due to confounding issues of ancient, rapid lineage diver-sification, a lack of clear morphological synapomorphies, a propen-sity for morphological and molecular homoplasy, and organismalextinction (e.g., Davis et al., 2005; Deng et al., 2015; Givnishet al., 2009; Schönenberger et al., 2005). A lack of a well-resolvedphylogenetic framework has made it difficult to estimate biogeo-graphic ancestral ranges and assess shifts in species diversificationrates within such groups. However, increased amounts of phyloge-netic/phylogenomic data and improved analytical tools to evaluatethese data are providing more robust phylogenetic frameworks for

hypothesis testing (Davis et al., 2014; Ruhfel et al., 2014; Soltiset al., 2011; Zeng et al., 2014; Magallón et al., 2015). In turn,model-based approaches have been developed to more rigorouslyestimate biogeographic history (Lagrange: Ree et al., 2005; BEAST:Drummond et al., 2006; BioGeoBEARS: Matzke, 2013) and to assessshifts in diversification rates (MEDUSA: Alfaro et al., 2009; BAMM:Rabosky, 2014; Rabosky et al., 2014). The use of these advances hasallowed more detailed insight into clades lacking resolution ofearly diverging lineages (e.g., Asteraceae: Panero et al., 2014), withhistories shaped by ancient intercontinental disjunctions (e.g.,campanulids; Beaulieu et al., 2013), and with lineage specific shiftsin speciation and/or extinction rates (e.g., Bromeliaceae: Givnishet al., 2014; hummingbirds: McGuire et al., 2014).

1.1. Biogeographic hypotheses of Southern Hemisphere disjunct groups

Of particular interest for re-examination using these improvedphylogenetic approaches are groups exhibiting a Southern

B.A. Berger et al. /Molecular Phylogenetics and Evolution 95 (2016) 116–136 117

Hemisphere affinity and whose current distributions, shifts intodifferent habitat or biome types, and thus rates of speciesdiversification may be influenced by the continental breakup ofGondwana during the Cretaceous (Crisp et al., 2009; Donoghueand Edwards, 2014). The distribution patterns and implied biogeo-graphical events of Southern Hemisphere organisms have beenvigorously debated since the late 1960s, but a strong vicariancevoice (e.g., Axelrod, 1970; Cracraft, 1988; Raven and Axelrod,1974; Rosen, 1978) had emerged for explaining present-day dis-junct patterns in a wide diversity of plant and animal groups(e.g., Edwards and Boles, 2002; Haddrath and Baker, 2001;Murphy et al., 2001; Sequeira and Farrell, 2001; Swenson et al.,2001; Vinnersten and Bremer, 2001). Recent molecular phyloge-netic studies that have included dating methods have providedsupport for the vicariance model for some Southern Hemispheregroups (e.g., Bukontaite et al., 2014; Korall and Pryer, 2013; Maoet al., 2012; Murienne et al., 2013; Wilf and Escapa, 2014). How-ever, others have questioned the sole reliance on Gondwananvicariance as explanatory for many plant and animal clades show-ing Southern Hemisphere disjunct patterns (Crisp and Cook, 2013;Sanmartín and Ronquist, 2004). These studies have included sup-port for the Boreotropics Hypothesis (Lavin and Luckow, 1993) orLaurasian rather than Gondwanan origin of some of these plantclades (Baker and Couvreur, 2013; Davis et al., 2002; Renneret al., 2001; Zerega et al., 2005). Most studies, however, haverevealed clade dates that are indicative of long-distance dispersals(sometimes associated with vicariance) rather than solely vicari-ance for Southern Hemisphere disjunctions of both plants and ani-mals (Armstrong et al., 2014; Barker et al., 2007; Beaulieu et al.,2013; Chacón and Renner, 2014; Christenhusz and Chase, 2013;Cook and Crisp, 2005; Friedman et al., 2013; Gallaher et al.,2014; Gamble et al., 2011; Givnish and Renner, 2004; Givnishet al., 2000, 2004; Knapp et al., 2005; Müller et al., 2015; deQueiroz, 2005; Rowe et al., 2010; Sytsma et al., 2004; Thomaset al., 2014).

1.2. Myrtales as a model group for assessing Southern Hemispherebiogeography and rates of species diversification

Here we examine the eudicot order Myrtales, a clade that hasstrong Gondwanan representation for most of its families, althoughone family (Onagraceae) is most diverse in Laurasia. APGIII (2009)recognizes nine families in Myrtales, including: Melastomataceae(188 genera/4618 species), Myrtaceae (131/5638), Onagraceae(22/667), Lythraceae (31/522), Combretaceae (14/570),Vochysiaceae (7/217), Penaeaceae (9/29), Crypteroniaceae (3/10),and Alzateaceae (1/2). Four large families (Melastomataceae,Myrtaceae, Lythraceae, and Combretaceae) are pantropical in distri-bution, while the Vochysiaceae possess an amphi-Atlantic disjunctdistribution pattern. The order is the thirdmost species rich lineageof the Superrosidae clade of angiosperms (12,264 spp.), variestremendously in habit (including herbaceous herbs, lianas, trees,and mangroves), floral form, and fruit type (berry, capsule, drupeand samara), and exhibits high species diversifications in severalfleshy-fruited and dry-capsular clades (e.g., within Melastomat-aceae and Myrtaceae) (Dahlgren and Thorne, 1984). Despite thisecological and morphological variation, as well as the presence ofspecies rich subclades within Myrtales, no detailed analysis ofdiversification rates, shifts in these rates, or correlation with evolu-tionary traits has been performed across the order or within familyclades.

A well-resolved and temporally calibrated phylogenetic frame-work of Myrtales is essential to address questions and hypothesesrelating to ancestral range estimation, importance of vicarianceand/or dispersal models, and character evolution relative to diver-sification. All previous molecular phylogenetic studies, either

focusing on the order or including placeholder taxa, support amonophyletic Myrtales regardless of the genic region used (seeSoltis et al., 2011; Sytsma et al., 2004). Although previous phyloge-netic studies greatly improved our understanding of intergenericand interspecific relationships within the order, our understandingof inter-familial relationships still remains unresolved, and thus,we also lack a strong time-calibrated chronogram to addresshypotheses (e.g., biogeography and diversification rates) that haveimplicit time assumptions. An earlier r8s (Sanderson, 2002) analy-sis for the order used multiple fossil calibration points and a two-genome data set, but did not account for time estimate variation atnodes and was based on a single fixed topology (Sytsma et al.,2004). Advances in the last decade for estimating chronograms,availability of considerably more molecular data, and documenta-tion of many more Myrtales fossils mandate a more thorough anal-ysis of the Myrtales as done here.

Especially problematic in Myrtales has been the placement ofthe pantropical Combretaceae (Conti et al., 1996, 1997; Sytsmaet al., 2004). Two alternative hypotheses exist: (1) Combretaceaeis sister to the clade comprising Onagraceae + Lythraceae, whichis then sister to the rest of Myrtales (Conti et al., 1996, 1997); or(2) Combretaceae is sister to all other members of Myrtales(Sytsma et al., 2004). The placement of Combretaceae is of consid-erable importance because its position directly influences cladeages, hypotheses addressing dispersal/vicariance scenarios, speciesdiversification rates, and character reconstructions. Since Contiet al. (1996, 1997) proposed the first interfamilial relationshipsof Myrtales based on rbcL, most ordinal level phylogenetic studieshave continued to utilize a limited number of taxa and generegions (often the same GenBank accessions) as placeholders forCombretaceae (e.g., Magallón, 2010; Rutschmann et al., 2007;Sytsma et al., 2004; Wang et al., 2009). The continued use of thesame 2–5 taxa and 1–2 plastid regions has limited our ability toplace Combretaceae and confidently address other hypotheses.To circumvent this issue (as summarized by Sanderson et al.,2010), our analysis increases sampling, both in terms of numberof taxa and breadth across the family.

Thus, we develop here a more rigorous time-calibrated phylo-genetic framework for Myrtales based on a three-genomeapproach with nearly 98% coverage of gene regions for each sam-pled taxon. Taxa sampling is designed to cover all major cladeswithin each family. We then use this new historical frameworkof Myrtales to: (1) evaluate phylogenetic relationships withinand among the nine families; (2) estimate when and where majorlineages of Myrtales originated with BEAST and BioGeoBEARS anal-yses; (3) examine the biogeographic processes that may have con-tributed to extant distributions, especially several differentdisjunct patterns in the Southern Hemisphere; and (4) test forshifts in speciation and extinction rates across Myrtales and withineach of the five major family clades using BAMM analyses.

2. Material and methods

2.1. Taxon and gene sampling, and phylogenetic analyses

Sampling was performed in an effort to maximize diversityacross the nine currently recognized families (APGIII, 2009); thus,we based our sampling of 102 taxa on available sequences and pre-vious phylogenetic studies (see Supplementary InformationTable S1). Outgroups (15 species) were selected from Vitales (Vitis),Crossosomatales (Crossosoma), Malvales (Thymelaea), Brassicales(Arabidopsis, Carica), and Geraniales (California, Erodium, Francoa,Geranium, Hypseocharis, Melianthus, Monsonia, Pelargonium, Vivia-nia). Geraniales is the sister order to Myrtales, with the Crossoso-matales, Malvales, and Brassicales representing three orders of

118 B.A. Berger et al. /Molecular Phylogenetics and Evolution 95 (2016) 116–136

the Malvid clade within the Rosidae sister to these two orders(APGIII, 2009; Soltis et al., 2011). Vitis was used as the ultimateoutgroup in all phylogenetic reconstructions. When possible, thesame species, and often the same DNA samples from previousstudies, were used to expand gene sampling. When unavailable,sequences of closely related species were generated or obtainedfrom GenBank to serve as generic placeholders. Species, voucher/-collection information, and GenBank accession numbers are pro-vided in SI Table S1.

Six loci (rbcL, ndhF, matK, matR, 18S, and 26S) were amplifiedand sequenced for taxa across Myrtales and Geraniales using pri-mers shown in SI Table S2. Each locus was selected based on utilityin large-scale studies across angiosperms (see Soltis et al., 2011),the Rosidae (Wang et al., 2009; Zhu et al., 2007), and family-wide studies in Myrtales (Biffin et al., 2010; Bult and Zimmer,1993; Clausing and Renner, 2001; Conti et al., 1993, 1996, 1997,2002; Gadek et al., 1996; Huang and Shi, 2002; Graham et al.,2005; Levin et al., 2003, 2004; Lucas et al., 2005; Maurin et al.,2010; Rutschmann et al., 2004, 2007; Schönenberger and Conti,2003; Sytsma et al., 2004; Tan et al., 2002; Thornhill et al., 2012;Wilson et al., 2005). 307 sequences from GenBank were added tothe dataset to improve sampling and to serve as placeholders forkey lineages essential for biogeographic reconstruction and assess-ing rates of species diversification.

Total genomic DNA was extracted from silica-dried leaf mate-rial and herbarium specimens using a DNeasyTM Plant Mini Kit (Qia-gen, Valencia, California). All samples received an extended heattreatment (30 min at 65 �C instead of 10 min) to reduce secondarycompound co-precipitation. PCR methods followed those describedpreviously (Wilson et al., 2001; Huang and Shi, 2002; Sytsma et al.,2004). PCR products were sequenced using BigDye Terminator kits(Applied Biosystems, Foster City, California) on an Applied Biosys-tems 3730xl automated DNA sequencer at the University ofWisconsin-Madison Biotechnology Center. Contiguous alignmentswere edited and assembled using Geneious Pro v.6.1.8 (http://www.geneious.com, Kearse et al., 2012). Sequences were alignedinitially using MAFFT v.6.864 (Katoh and Toh, 2008) with subse-quent adjustments made by eye in Geneious. Gaps were treatedas missing data.

Best fitting models of sequence evolution for each locus weredetermined using the Bayesian Information Criterion (BIC) injModeltest v.2.1.4 (Darriba et al., 2012; Guindon and Gascuel,2003). The BIC was used for model selection based on its abilityto outperform other model-selection criteria (Luo et al., 2010).Models selected included TVM+I+G for rbcL, TVM+G formatK,matR,and ndhF, GTR+I+G for 26S, and SYM+I+G for 18S. Maximum likeli-hood (ML) and Bayesian inference (BI) analyses were performedusing the CIPRES Science Gateway v.3.3 (www.phylo.org; Milleret al., 2010). ML analyses were conducted using default parametersin GARLI v.2.01 (Zwickl, 2006). One thousand bootstrap (BS) repli-cates were conducted using the same parameters applied for MLsearches. BI was performed using MrBayes v.3.2.3 (Alteker et al.,2004; Huelsenbeck and Ronquist, 2001; Ronquist andHuelsenbeck, 2003). Preliminary BI analyses revealed topologicalvariation in the placement and relationships of outgroup taxa (datanot shown), especially with regard to Geraniales; therefore, the sis-ter relationship of Myrtales to Geraniales was constrained (APGIII,2009; Ruhfel et al., 2014; Soltis et al., 2011). All BI analyses wererun for 15,000,000 generations with four chains in two parallel runssampling every 1500 generations. Proper mixing was determinedusing Tracer v.1.6 (Rambaut et al., 2014) and 20% of trees were dis-carded as burn-in prior to constructing a majority rule consensustree using TreeAnnotator v.1.8.0 (Drummond et al., 2012).

Congruence among plastid (rbcL, ndhF, matK), mitochondrial(matR), 18S rDNA, and 26S rDNA was assessed by estimating con-cordance factors (CF; Baum, 2007) using BUCKy v.1.4.3 (Larget

et al., 2010). BUCKy uses the Bayesian concordance approach(Ane et al., 2007), which integrates over gene tree uncertaintywithout speculating what causes discordance. For the BUCKy anal-ysis, we used the posterior distribution of the chloroplast, mito-chondrial, and nuclear ribosomal gene trees produced withMrBayes. The 18S and 26S data sets were analyzed separately asa conservative approach given the differing best models ofsequence evolution. The primary concordance tree and concor-dance factors were estimated with the discordance parameter (a)of 1 and 100. Both analyses were run with two Markov chainMonte Carlo searches (MCMC) of 100,000 generations with10,000 generations burn-in. Both runs converged on the same pri-mary concordance tree with identical concordance factors. Thefinal analysis was conducted with a = 1 and two MCMC chains of1,000,000 generations following a burn-in of 100,000 generations.

2.2. Fossil priors and BEAST analyses

Divergence time estimates were performed in BEAST v.1.8.0(Drummond et al., 2012), incorporating an uncorrelated lognor-mal clock and Yule speciation process. In order to accommodaterate heterogeneity, chloroplast markers and matR were parti-tioned into three codon positions. We ran two independent anal-yses of 50,000,000 generations each. To verify effective samplingof all parameters and to assess convergence of independentchains, we examined output log files in Tracer v.1.6 (Rambautet al., 2014). After removing 20% of samples as burn-in, indepen-dent runs were combined and a maximum clade-credibility(MCC) tree was constructed using TreeAnnotator v.1.8.0(Drummond et al., 2012). We offset the minimum ages of 14nodes across the phylogeny using a combination of fossil (seebelow) and secondary priors (Table 1). All fossil priors wereplaced under a lognormal distribution with a mean of 1.5 and astandard deviation of 1, allowing for the possibility that thesenodes are considerably older than the fossils themselves. Second-ary priors were placed under a normal distribution and the rootunder a uniform distribution. Because of the aforementionedtopological variation with regard to Myrtales and Geraniales, weconstrained the topology using the most likely tree obtained fromGARLI. No age offset was placed on this prior.

Eleven fossil priors were utilized in this study. In Melastomat-aceae, we used a leaf fossil (Hickey, 1977) to offset the most recentcommon ancestor (MRCA) of Melastomataceae s.s. (minus tribeKibessieae) and a seed fossil as a Melastomeae crown prior(Clausing and Renner, 2001; Renner et al., 2001; Renner, personalcommunication; Rutschmann et al., 2007; Sytsma et al., 2004;Wehr and Hopkins, 1994). The placement of the second fossilwas supported by morphological characters, as extant Melas-tomeae exhibit cochleate seeds with a round operculum and tuber-culate testa, whereas all other Melastomataceae have straight orcuneate seeds that lack operculi and differ in ornamentation(Renner et al., 2001). In Myrtaceae, we used two Myrtaceidites(=Syncolporites) fossilized pollen grains to place priors on or nearthe crown of the family. Myrtaceidites lisimae is the oldest fossilin Myrtaceae (from Gabon, Africa; Boltenhagen, 1976; Muller,1981). While the assignment ofM. lisimae to the Myrtaceae is clear,its affinity to either subfamily is ambiguous and likely represents astem lineage placement (Rutschmann et al., 2007; Sytsma et al.,2004); therefore, we placed it conservatively on the crown of thefamily. We placedMyrtaceidites mesonesus on the first node in fromthe crown of subfamily Myrtoideae (i.e., excluding Lophostemon;M.mesonesus) following the suggestion of Thornhill et al. (2012). Thethird prior across Myrtaceae, placed on the stem node of Eucalyp-tus, represents the earliest record of meso-fossils attributed tothe eucalypt clade from South America (Gandolfo et al., 2011;Thornhill et al., 2012). In Lythraceae, we used fossilized pollen

Table 1Details of fossil calibrations used in BEAST analysis. Age estimates are based on 2009 Geologic Time Scale (Walker and Geissman, 2009). Clades designated with (⁄) indicatesecondary calibrations obtained from the literature.

Fossil (clade) Minimum age(Ma)

Coalescent node Reference(s) Prior

Root⁄ 100 Vitis + Osbeckia See Sytsma et al. (2014) Uniform 125–100Malvidae crown⁄ 109.5 Crossosoma + Osbeckia Wang et al. (2009) Normal 109.5

(5.33)Caricaceae/Brassicaceae crown⁄ 98.65 Arabidopsis + Carica Cardinal-McTeague et al.

(unpublished)Normal 98.65(3.7)

Geraniales crown⁄ 87 Viviania + Geranium Sytsma et al. (2014) Normal 87.0 (5.6)Pelargonium crown⁄ 28.4 Pelargonium a + Pelargonium

hSytsma et al. (2014) Lognormal 1.5

(1.0)Eucalyptus frenguelliana Gandolfo & Zamaloa, sp. n. 51.69 Stockwellia + Eucalyptus Gandolfo et al. (2011) Lognormal 1.5

(1.0)Myrtaceidites lisamae van der Hammen ex

Boltenhagen83.5 Heteropyxis +Myrtus See Sytsma et al. (2004) Lognormal 1.5

(1.0)Myrtaceidites mesonesus Cookson & Pike 61.7 Osbeckia +Myrtus Couper (1960) Lognormal 1.5

(1.0)Melastomeae seeds 23.0 Rhexia + Osbeckia See Renner et al. (2001) Lognormal 1.5

(1.0)Acrovena laevis Hickey 48.6 Bellucia + Osbeckia See Renner et al. (2001) Lognormal 1.5

(1.0)Koninidites aspis Pocknall & Mildenhall 21.4 Fuchsia pa + Fuchsia c Berry et al. (1990); Lee et al. (2013) Lognormal 1.5

(1.0)Lythrum elkensis sp. n./Peplis eaglensis sp. n. 81.0 Lythrum + Peplis Grímsson et al. (2011) Lognormal 1.5

(1.0)Lagerstroemia patelii Lakhanpal & Guleria 56.0 Lagerstroemia + Duabanga See Graham (2013) Lognormal 1.5

(1.0)Punicoxylon eocenicum Privé-Gill 40.4 Punica + Galpinia See Graham (2013) Lognormal 1.5

(1.0)Sonneratiaoxylon preapetalum Awasthi 63.8 Sonneratia + Trapa See Graham (2013) Lognormal 1.5

(1.0)Dilcherocarpon combretoides gen. et. sp. n. 93.5 Laguncularia + Terminalia k Manchester and O’Leary (2010) Lognormal 1.5

(1.0)

B.A. Berger et al. /Molecular Phylogenetics and Evolution 95 (2016) 116–136 119

grains attributed to Lythrum and Peplis as an offset for the crown ofthe two lineages (Grímsson et al., 2011). We also used the oldestdescribed Lagerstroemia leaf impression fossil, the silicified woodof Punicoxylon, and the oldest confirmed Sonneratiaoxylon woodas conservative offsets on the stem nodes of Lagerstroemia, Punica,and Sonneratia, respectively (Graham, 2013). In Onagraceae, fossilpollen of Koninidites aspis (the type palynomorph of Diporites aspiswas recently found to be a fungal sporophyte, thus, new typologi-cal nomenclature was assigned; Berry et al., 1990; Lee et al., 2013)was used to offset the split between Old World and New WorldFuchsia (Berry et al., 2004; Rutschmann et al., 2007; Sytsmaet al., 2004).

Although previous large-scale studies (Bell et al., 2010; Smithet al., 2011; Wikström et al., 2001) have included putative Combre-taceous fossils assigned to Esgueiria (Friis et al., 1992; Takahashiet al., 1999) to offset the Combretaceae node, the fossil differs sig-nificantly from extant taxa, especially with regard to stylarbranches, and its placement cannot be assigned confidently (E.M.Friis, personal communication; Friis et al., 1992; Magallón et al.,1999). In lieu of Esgueiria, we use the fin-winged fruits of Dilchero-carpon combretoides (Manchester and O’Leary, 2010) as an offset onthe stem node of tribe Combreteae. Morphologically, the epigynousfruits resemble those of extant Combretum and some Terminalia(Gere, 2014; Maurin et al., 2010), possessing four lateral wingsarising in two perpendicular planes. The presence of abundant par-allel veins and the absence of a fimbrial vein on the fruit is anextremely rare combination of characters outside of the family(Manchester and O’Leary, 2010). Although winged fruits arereported in all genera of Combretaceae (Johnson and Briggs,1984), the occurrence of winged fruits in the genus Strephonema(subfamily Strephonematoideae) has not been confirmed and thefruits of tribe Laguncularieae appear to be secondarily derivedfrom bracteoles (Stace, 2007).

2.3. Inserting additional terminals for biogeographical anddiversification analyses

To improve lineage sampling for biogeographical and diversifi-cation analyses, we added five tips to the MCC tree from BEAST.These five taxa were too incomplete for our gene sampling, buttheir phylogenetic placements within families are well known.The five new tips added included: (1) Xylonagra Donn. Sm. & Rosein Onagraceae (Levin et al., 2003, 2004); (2) Syncarpia Ten. and (3)Xanthostemon F. Muell. in Myrtaceae to represent tribes Syncar-pieae and Xanthostemoneae, respectively (Thornhill et al., 2012);and (4) Astronia Blume and (5) Merianthera Kuhlm. in Melastomat-aceae to represent the Astronieae and the clade comprised of‘‘Merianthera and its allies” + ‘‘Bertolonieae 2” + Cyphostyleae(Goldenberg et al., 2012). The taxa were added to the BEAST treeusing the bind.tip function of the phytools package (Revell, 2012)for R (R Development Core Team, 2014). This function rescales anultrametric tree after new tips are added so the tree remains ultra-metric. Similarly, we added these five tips to a random subset of100 trees obtained from the posterior distribution of post-burninBEAST trees for further biogeographical analyses.

2.4. Ancestral range estimation in BioGeoBEARS

Ancestral range estimation (ARE) for Myrtales was done usingthe nested DEC and DECj models in BioGeoBEARS (Matzke, 2013,2014) in R v3.1.1. Similar to the Dispersal-Extinction-Cladogenesis(DEC) program LaGrange (Ree and Smith, 2008; Ree et al., 2005),the DEC model in BioGeoBEARS evaluates ML parameters for ana-genetic events involving range expansion and extinction and forcladogenetic events involving sympatry and vicariance. UnlikeLaGrange, however, BioGeoBEARS with the DECj model can alsoparameterize cladogenetic ‘‘founder-events” (Templeton, 1980)

120 B.A. Berger et al. /Molecular Phylogenetics and Evolution 95 (2016) 116–136

by incorporating the J parameter for ‘‘jump-dispersals”. This Jparameter allows for a daughter lineage to immediately occupyvia long-distance dispersal a new area that is different from theparental lineage. DECj models have been shown to be significantlybetter than DEC models for island groups (Matzke, 2014) and forinter-continental distributions, but not for some more localizedcontinental groups (e.g., western North American Salvia; Walkeret al., 2015).

We identified six broad geographic areas, modified after Buerkiet al. (2011), important in the context of the distributions of thefamilies in Myrtales (see Fig. 3): (1) Eurasia (from western Europe,Mediterranean Africa, to temperate Asia); (2) Sub-Saharan Africaincluding Madagascar; (3) Southeast Asia, including India, Indo-China, the Malaysian Peninsula, the Philippines, Sumatra, Borneoand the Inner Banda Arc; (4) Australia, including New Guinea,New Caledonia and New Zealand, as well as the Pacific Islands(e.g. Hawaii); (5) North America; and (6) South America, includingmost of Mexico, Central America and the West Indies. Because ourstudy focused on family-level relationships, all tip taxa withinMyrtales were coded as present or absent for each of the six areasbased on extensive literature available for each family. In instanceswhere a tip taxon represented a more diverse and widespread lin-eage, we coded these tips to cover the maximal distribution of thelineage. Due to the difficulty of scoring under-sampled outgrouporders for geographic areas, we restricted outgroup taxa only tothe Geraniales as the sister clade to Myrtales. Geraniales has anancestral area of Africa, South America, or combined Africa + SouthAmerica (Fiz et al., 2008; Palazzesi et al., 2012; Sytsma et al., 2014).Our sampling of three families in the order included all possibili-ties: African Melianthaceae, South American Vivianiaceae, andlikely combined African + South American Geraniaceae.

In BioGeoBEARS we allowed the inferred ancestors to occupy upto all six areas. Dispersal probabilities between pairs of areas andareas allowed were specified for four separate time slices (SITable S3) based on known geological events that have been simi-larly analyzed elsewhere (e.g., Buerki et al., 2011; Drew andSytsma, 2012; Sessa et al., 2012; Sulman et al., 2013). We esti-mated ancestral ranges on the MCC tree from BEAST, pruned ofall outgroups except Geraniales, with the addition of five terminals.The resulting ML score for the more parameter-rich DECj modelwas tested for significance against the resulting ML score of theDEC model. We explored the degree to which differences in treetopology and branch lengths yield different ARE for nodes bydeveloping a script to run BioGeoBEARS on the 100 randomlyselected PP trees from the BEAST analysis (with five species addedfor each) under both DEC and DECj models. Additionally, we con-ducted biogeographical stochastic mapping analyses in BioGeo-BEARS on the single best BEAST tree (with five species added)under both DEC and DECj models. This permitted measuring theprobability of each class of cladogenetic event (vicariance, sympa-try, subset-sympatry, and jump dispersals) given the DEC and DECjmodels, distribution data, and phylogeny.

2.5. Estimating speciation and extinction rates and identifying rateshifts in species diversification in BAMM

We used the program BAMM (Rabosky, 2014; Rabosky et al.,2014) to: (1) estimate rates of speciation, extinction, and net diver-sification for clades, (2) conduct rate-through-time analysis ofthese rates, and (3) identify and visualize shifts in species ratesacross the Myrtales phylogeny. BAMM is a Bayesian approach thatuses reversible jump Markov chain Monte Carlo (RJMCMC) sam-pling to explore shifts in macro evolutionary regimes assumingthey occur across the branches of a phylogeny under a compoundPoisson process, and explicitly accommodates diversification ratevariation through time and among lineages. BAMM is both time-

sensitive and diversity-dependent, allowing rate shifts to occuranywhere on a branch based on the posterior tree density. BAMMaccounts for non-random and incomplete taxon sampling in thephylogenetic trees by allowing all non-sampled species to be asso-ciated with a particular tip or more inclusive clade. Terminals andproportion of extant species sampled for each are provided in SITable S4. For example, only one of eight species of Circaea (Ona-graceae) was sampled, and thus BAMM recognized sampling at thistip as only 0.125. Tips were assigned to the smallest possible tax-onomic unit, which often corresponded to genera or tribes withinfamilies. For example, all 873 species representing a number ofgenera of the tribe Eucalypteae (Myrtaceae) were identified asbelonging to a clade defined by the most recent common ancestorof our three sampled genera (Eucalyptus, Angophora, and Stock-wellia). Species numbers were obtained from published sources(Forster, 1994; Jordaan et al., 2011; Maurin et al., 2010; Renner,1993; Stace, 2007; Stevens, 2001; The Plant List, 2014). Terminalsand proportion of extant species sampled for each are provided inSI Table S4. A total of 12,264 species in Myrtales are recognized,updating previous numbers from Stevens (2001) and Soltis et al.(2011).

Priors for BAMM were generated using the R package BAMM-tools v.2.0.2 (Rabosky, 2014) by providing the MCC tree fromBEAST and total species numbers across the order (12,264). Twoindependent MCMC chains of 100,000,000 generations were runin BAMM and convergence was assessed by computing the effec-tive sample sizes of log likelihoods, as well as the number of shiftevents present in each sample using the R package coda v. 0.16-1(Plummer et al., 2006). After removing 10% of trees as burn-in,we analyzed the BAMM output using BAMMtools and computedthe 95% credible rate shift configurations using the Bayes factorscriterion for including nodes as core shifts set to 5. This procedurecomputes Bayes factors evidence associated with a rate shift forevery branch in the phylogeny and excludes all unimportant nodesusing the Bayes factor criterion (see BAMM documentation). LargeBayes factors values are taken as strong evidence for a shift on abranch in the tree. We also estimated the rate shift configurationwith the highest maximum a posteriori (MAP) probability (‘‘thebest shift configuration”) after having excluded all none core shifts.

Additionally, we ran independent BAMM analyses on each ofthe five largest families within Myrtales (Combretaceae, Lythra-ceae, Melastomataceae, Myrtaceae, and Onagraceae) to test ifBAMM identified different shifts when each family was analyzedseparately. The same parameters and analytical procedures wereused for family-specific analyses except the number of generationswas reduced to 50,000,000. Family level phylogenies wereobtained by extracting the corresponding families from the Myr-tales phylogeny.

3. Results

3.1. Phylogenetic relationships and timing of clade formation inMyrtales

A total of 362 sequences were generated for this study: 31 rbcL,28 ndhF, 80matK, 63 18S, 62 26S and 98matR (see SI Table S1). Thealigned, partitioned data set consisted of 7337 characters, with1626 being parsimony-informative. ML and BI trees highlight sev-eral important phylogenetic results at the broad scale within Myr-tales (Fig. 1). First, Combretaceae is identified as the sister clade toall remaining families with moderate support (ML-BS/BI-PP = 69/0.84). Second, within the latter clade, two strongly sup-ported groups are observed: (1) Onagraceae + Lythraceae form asubclade with strong support (ML-BS/BI-PP = 100/1.0), and (2)Melastomataceae + the CAP clade and Myrtaceae + Vochysiaceae

Clarkia

Bellucia

Ammania

Syzygium

Galpinia

Terminalia a

Tococa

Erisma

Macrocentrum

Ruizterania

Peplis

Strephonema m

Rhynchocalyx

Erismadelphus

Adenaria

Lopezia

Leptospermum

Miconia

Viviania

Trapa

Tibouchina

Lavoisiera

Osbornia

Olinia

Eugenia

Bucida

Fuchsia pa

Blakea

Fuchsia c

Combretum in

Ludwigia

Melaleuca

Mouriri

Vochysia

Laguncularia

Sonneratia

Conocarpus

Combretum f

Combretum iv

Punica

Lagerstroemia

Pterolepis

Decodon

Strephonema p

Geranium

Terminalia p

Terminalia c

Heteropyxis

Terminalia k

Memecylon

Clidemia

Heimia

Osbeckia

Combretum c

Penaea

Leandra

Crypteronia

Tristania

Epilobium

Oenothera

Angophora

Circaea

Lawsonia

Stockwellia

Combretum r

Melianthus

Terminalia b

Metrosideros c

Lythrum

Fuchsia pr

Medinilla

Lindsayomyrtus

Pternandra e

Backhousia

Meriania

Bertolonia

Chamelaucium

Buchenavia

Graffenrieda

Eucalyptus

Qualea

Terminalia u

Tristaniopsis

Metrosideros m

Alzatea

Kunzea

Hauya

Heterocentron

Melastoma

Arthrostemma

Macropteranthes

Rhynchanthera

Uromyrtus

Gongylocarpus

LophostemonPsiloxylon

Terminalia i

Cuphea

Pternandra c

Henriettea

Duabanga

Callistemon

Myrtus

Rhexia

Lumnitzera

Francoa

CaliforniaErodiumMonsoniaPelargonium aPelargonium h

CaricaArabidopsisThymelaeaCrossosomaVitis

Hypseocharis

Mel

asto

mat

acea

eM

yrta

ceae

CA

PV

och

ys.

Lyt

hra

ceae

On

agra

ceae

Co

mb

reta

ceae

Ger

ania

les

69/0.84

0.2 substitutions/site

71/0.9881/1.0

85/1.0

89/1.0

83/1.0

*/0.99

93/1.0

86/1.0

81/1.0

*/1.090/1.0

*/1.0

*/0.92*/0.93

*/1.0

77/0.99

*/0.90

72/0.9389/1.0

85/0.97

91/0.99*/0.94 95/1.0

89/1.0*/0.99

*/0.99

*/0.90

77/0.97*/1.0

*/1.0

89/1.092/0.99

77/0.97

Fig. 1. Maximum likelihood tree of Myrtales and Geraniales based on six loci shown as a cladogram (phylogram inset). Tree is rooted with four other Rosidae orders (Vitis isthe ultimate outgroup). Support values are provided for each branch: (ML-BS/BI-PP). Bold black branches are ML-BSP 95% and BI-PP = 1.0. Aside from the node designatingthe divergence of Combretaceae from all other Myrtales, support values 670% ML-BS are designated by a (⁄). Support values 670% ML-BS and 60.9 BI-PP are collapsed.Families are color-coded (as in other figures) to assist comparisons.

B.A. Berger et al. /Molecular Phylogenetics and Evolution 95 (2016) 116–136 121

122 B.A. Berger et al. /Molecular Phylogenetics and Evolution 95 (2016) 116–136

are both highly supported (ML-BS/BI-PP = 100/1.0), as is the entiresubclade (ML-BS/BI-PP = 100/1.0). Third, all families within Myr-tales are monophyletic with high support (ML-BS/BI-PP = 100/1.0). And fourth, our sub-sampling scheme corroboratesmost large-scale studies addressing intergeneric relationships(see Section 4). Bayesian concordance analysis using BUCKyresulted in a primary concordance tree with the same overalltopology as the BI and ML phylogenies (SI Fig. S1). Concordancefactors were moderate to high for all families (>0.5) and these val-ues decreased toward the backbone of the tree.

Divergence time estimates suggest Myrtales diverged fromGeraniales �123.6 Mya (Table 2; Fig. 2) during the Aptian of theEarly Cretaceous. The crown date for Myrtales is estimated at�116.4 Mya (Albian–Aptian). Stem lineages of the four majorclades of Myrtales are all present by the end of the Albian(�109.3 Mya): (1) Combretaceae, (2) Lythraceae + Onagraceae, (3)Melastomataceae + the CAP clade, and (4) Myrtaceae + Vochysi-aceae. Stem lineages of all extant families are present by�90.7 Mya (mid-Turonian). Crown dates of modern families rangefrom �102.6 Mya for the Combretaceae (Albian) to �38.8 Mya forthe Vochysiaceae (late Eocene).

3.2. Biogeographical patterns of diversification in Myrtales

In BioGeoBEARS, significant improvement in the likelihoodscore of the model was seen when the ‘‘jump dispersal” parameterwas added (DECj) vs. without (DEC), as indicated by a likelihoodratio test (DEC LnL �341.97, DECj LnL = �287.70, df = 1,P = 2.05e�25). Because of the significant improvement in the like-lihood of the model when the jump dispersal parameter is added,only ARE under the DECj model is presented. The resulting param-eters of the DECj models included: anagenetic dispersal rated = 0.0049; extinction rate e = 1.0e�08; cladogenetic dispersal ratej = 0.096. Biogeographical stochastic mapping, given the parame-ters of the DECj model, indicated that 60% of all cladogeneticevents are sympatric, 32% are subset-sympatry, 7% are vicariance,and 1% involve founder-event dispersals.

ARE for Myrtales and Geraniales using BioGeoBEARS is shown inFig. 3 (area or combined area with highest frequency shown foreach node and corner). The chronogram illustrates ARE in the con-text of geological age and continental drift during the Cretaceousand Paleogene. The MCC tree indicates a combined South Amer-ica + Africa ARE for the ancestor of Myrtales + Geraniales and forthe crowns of both Myrtales and Geraniales, with these diversifica-tions occurring in the early Cretaceous when the two continentswere attached (Fig. 3). The only difference in ARE between theBEAST tree and the 100 PP BEAST trees is in the crown node of Myr-tales (South America only 85% of the trees, as shown in Fig. 3).Thus, ARE of Myrtales is robust at least in the context of likelyalternative topologies or branch lengths. The frequencies of otherpossible areas (or area combinations) at each node and cornerare provided in SI Fig. S2.

Table 2Estimated ages (Mya) of nodes within Myrtales using BEAST analysis and comparable noder8s analysis. Combretaceae and Lythraceae are only family crowns not comparable betwecrown ages) than sampled in the r8s analysis (younger crown ages).

Node Clade BEAST age (95% HPD)

Stem

1 Myrtales 123.6 (124.8–122)2 Combretaceae 116.4 (118.8–113.7)3 Onagraceae 104.6 (109.1–100.2)4 Lythraceae 104.6 (109.1–100.2)5 Vochysiaceae 101.1 (107.0–95.1)6 Myrtaceae 101.1 (107.0–95.1)7 CAP clade 90.7 (98.8–82.4)8 Melastomataceae 90.7 (98.8–82.4)

Most of the major family clades also have either South America(Melastomataceae, Vochysiaceae, Onagraceae, Lythraceae) orAfrica (Myrtaceae, CAP clade) as ancestral. Australia, includingislands east of the Wallace Line and Polynesia, is an important bio-geographic area early in the Late Cretaceous or Paleocene radiationof subfamily Myrtoideae (Myrtaceae), Melastomataceae, Cryptero-niaceae, and Combretaceae. Eurasia and Southeast Asia appear toplay more minor and recent roles in most families, with the excep-tion of an early diversification (Late Cretaceous) in Lythraceae inEurasia. More recent migrations in the Paleogene over adjacentor still connected areas are seen between South America and Aus-tralia (e.g., Melastomataceae, Onagraceae, and Lythraceae) andbetween Australia, Southeast Asia, and Eurasia (Melastomataceae,Myrtaceae, and Combretaceae). Long distance dispersal and estab-lishment (LDDE) events between South America and Africa andover oceanic water barriers during the Paleogene occur in theCAP clade, Vochysiaceae, and Combretaceae.

3.3. Speciation and extinction rates and identification of three rateshifts in species diversification in Myrtales

After discarding the burn-in, we confirmed convergence of theMCMC chains in the BAMM analyses, as well as effective samplessizes >900 for both the number of shifts and log likelihoods. BAMManalyses strongly supported a diversity-dependent speciation pro-cess across Myrtales with a net diversification rate (speciationminus extinction) of 0.065 species/Myr (Table 3). The highest spe-ciation rates are seen at nodes within Combretaceae (0.69 species/Myr), Melastomataceae (0.45 species/Myr), and Myrtaceae(0.32 species/Myr) (Table 3). These three families also show thehighest extinction rates (0.54, 0.25, 0.15 species/Myr, respectively).Onagraceae, Lythraceae, and the CAP clade exhibit the lowest spe-ciation and extinction rates within the order. Rate-through-timeplots for the �120 Myr history of Myrtales revealed a decreasingspeciation rate until �60–70 Mya, at which point speciation beganto increase, with the highest average rate (�0.28 species/Myr) seenat present (Fig. 4). Extinction rates were low (0.06 species/Myr)until 60–70 Mya and steadily climbed until the present (0.19 spe-cies/Myr).

The 95% credible set of rate shift configurations sampled withBAMM included 271 distinct configurations of which the configu-rations with the highest probability included three to four shifts.Fig. 5 represents the single best configuration and includes threeshifts detected within Myrtales: (1) near the crown of Melastomat-aceae (�63 Mya) after the divergence of Pternandra, (2) along thestem of subfamily Myrtoideae (Myrtaceae; �76 Ma), and (3) alongthe stem of tribe Combreteae (Combretaceae; �43 Mya). Althoughthe specific placements are at different times, all three rate shiftsoccur on branches that span the Cretaceous–Paleogene boundary(K–Pg = 66 Mya; formerly K–T boundary). Bayes factor evidencewas strongest (>800) for the shifts in Combretaceae (BF = 22,921)and Melastomataceae (BF = 3240). The shift at the base of the Myr-

dates based on r8s (Sytsma et al., 2004). Stem of Myrtales constrained to 120 Mya inen the two analyses as BEAST analysis included earlier diverging lineages (thus older

r8s age (Sytsma et al., 2004)

Crown Stem Crown

116.4 (118.8–113.7) 120 111102.6 (106.5–98.9) 111 4685.4 (96.2–73.3) 93 8295.5 (99.7–91.7) 93 6038.9 (51.8–28.0) 93 3285.0 (87.6–83.6) 93 7652.7 (66.4–40.4) 82 5264.5 (74.81–56.1) 82 56

Fig. 2. Maximum clade credibility (MCC) tree of Myrtales obtained from BEAST analysis. Mean divergence time estimates are shown with 95% highest posterior density (HPD;blue bars). Clades I–VII represent crown radiations of extant families. Mean age plus 95% HPD are as follows: I – 64.5 Ma (74.8–56.1 Ma); II – 52.7 Ma (66.4–40.4 Ma); III –85.0 Ma (87.6–83.6 Ma); IV – 38.8 Ma (51.8–28.0 Ma); V – 95.5 Ma (97.7–91.7 Ma); VI – 85.4 Ma (96.2–73.3 Ma); and VII – 102.6 Ma (106.5–98.9 Ma). Black circles indicatecalibration points 1–7 (see Table 2 for additional information). The shaded box corresponds to the maximum/minimum age range for the divergence of all extant Myrtaleanstem lineages (based on 95% confidence intervals).

B.A. Berger et al. /Molecular Phylogenetics and Evolution 95 (2016) 116–136 123

Mel

asto

mat

acea

eM

yrta

ceae

CA

PV

och

ys.

Lyt

hra

ceae

On

agra

ceae

Co

mb

reta

ceae

Ger

ania

les

Fig. 3. Ancestral range estimation (ARE) on the Myrtales chronogram with BioGeoBEARS (DEC+J model). Areas of tip species (or genera) are shown left of taxa names, color-coded for the six biogeographical areas shown on the map inset. Boxes at each node and corner are color coded for the area (or combined area, up to four allowed) with thehighest ML probability. SI Fig. S2 provides all ARE per node and corner with pies designating probability of each area (or combined area). Numbers next to nodes indicate theproportion out of 100 PP trees (if less than 1.0) that the area (or combined area) had the highest ML probability. Inset globes depict continental positions at the beginning ofthe Late Cretaceous and the K–Pg boundary (modified after Scotese, 1998).

124 B.A. Berger et al. /Molecular Phylogenetics and Evolution 95 (2016) 116–136

Table 3Estimated speciation, extinction, and net diversification rates (with 95% CI) for selected crown nodes in Myrtales based on BAMM analysis. Ages are Mya. Rates are species/Myr.Clades are ordered by magnitude of speciation rate. Three clades with significant shifts in speciation rates are indicated in bold (see Figs. 4–6). Symbols ‘‘ + ” or ‘‘�” indicate oneyounger or older node, respectively.

Clade Age Speciation Extinction Net diversification

Median 95% CI Median Median Median 95% CI

Combretaceae shift + 17.0 0.69 0.38–1.08 0.54 0.20–0.94 0.15 0.14–0.19Combretaceae shift 33.2 0.57 0.33–0.87 0.44 0.16–0.76 0.12 0.11–0.16Miconieae (Melastomataceae) 27.1 0.45 0.23–0.72 0.25 0.05–0.51 0.20 0.18–0.21Melastomataceae shift + 42.4 0.40 0.22–0.65 0.25 0.04–0.51 0.15 0.11–0.15Melastomataceae shift 62.2 0.39 0.21–0.65 0.24 0.04–0.51 0.15 0.14–0.17Melastomataceae 64.5 0.36 0.20–0.59 0.23 0.04–0.48 0.13 0.11–0.15Combretaceae shift � 94.5 0.32 0.19–0.48 0.24 0.10–0.41 0.07 0.07–0.09Myrteae (Myrtaceae) 18.4 0.30 0.12–0.60 0.15 0.01–0.39 0.15 0.11–0.21Combretaceae 102.6 0.28 0.17–0.41 0.21 0.09–0.36 0.07 0.05–0.08Myrtaceae shift 67.3 0.22 0.12–0.39 0.13 0.02–0.31 0.09 0.08–0.10Myrtaceae shift + 64.1 0.22 0.12–0.40 0.13 0.01–0.32 0.09 0.09–0.09Syzigieae (Myrtaceae) 22.5 0.22 0.11–0.40 0.13 0.01–0.33 0.09 0.08–0.11Myrtaceae 85.0 0.21 0.11–0.37 0.12 0.02–0.30 0.08 0.07–0.09Myrtales 116.4 0.19 0.14–0.26 0.13 0.07–0.20 0.06 0.06–0.07Onagreae (Onagraceae) 45.7 0.14 0.04–0.35 0.08 0.01–0.21 0.06 0.03–0.14Onagraceae 85.4 0.10 0.05–0.17 0.07 0.01–0.14 0.03 0.03–0.03Lythraceae 95.5 0.09 0.04–0.16 0.07 0.01–0.14 0.02 0.02–0.03CAP clade 52.7 0.08 0.03–0.14 0.07 0.01–0.14 0.02 0.01–0.03Fuchsia (Onagraceae) 23.0 0.08 0.030.16 0.06 0.01–0.14 0.02 0.02–0.02

120 100 80 60 40 20 0

0.0

0.1

0.2

0.3

0.4

time before present (Mya)

spec

iatio

n ra

te (λ)

Com

bret

acea

e

Mel

asto

mat

acea

e

Myr

tace

ae

120 100 80 60 40 20 0

0.0

0.1

0.2

0.3

0.4

time before present (Mya)

extin

ctio

n ra

te (

µ)

120 100 80 60 40 20 0

0.0

0.1

0.2

0.3

0.4

time before present (Mya)

net d

iver

sific

atio

n ra

te (

r)

Fig. 4. Rates-through-time plots of speciation, extinction, and net diversification in Myrtales utilizing BAMM. Curved black lines represent the median values with the 95%confidence intervals shown in gray. Arrows point to the three significant shifts in rates of speciation identified in Fig. 5.

B.A. Berger et al. /Molecular Phylogenetics and Evolution 95 (2016) 116–136 125

taceae (excluding Heteropyxis and Psiloxylon) received a Bayes fac-tor of 300, which does not closely approach the 800 value of over-whelming support for a rate shift.

The family level analyses (Figs. 5 and 6) recovered shifts in spe-ciation rates that are similar but not identical to the ordinal levelanalyses. Like in the large analyses, shifts were detected in the Com-bretaceae, Myrtaceae, and Melastomataceae. In Combretaceae, the95% credible set of rate shift configurations sampled with BAMMincluded 2 distinct configurations of which the configurations withthe highest probability included 1–2 core shifts. The best shift con-figuration included a single core shift on the same branch as theordinal level analyses (Fig. 6). Bayes factor evidence for this shiftwas high (BF = 3158). In the Melastomataceae, the 95% credibleset of rate shift configurations sampled included 3 distinct configu-rations of which the configurations with the highest probabilityincluded one or no core shifts. The best configuration detected a sin-gle shift within the tribe Miconieae, on the branch leading to the

clade comprised of Clidemia,Miconia and Tococa (Fig. 6). Bayes factorevidence for this shift was low (BF = 116). Lastly in Myrtaceae, the95% credible set of rate shift configurations sampled included 23distinct configurations of which the configurations with the highestprobability included one, two, or no core shifts. The best configura-tion included a single core shift on the crown of the Myrteae clade(Fig. 6). Bayes factor evidence for this shift was 84.

4. Discussion

4.1. Myrtales: a model Cenozoic clade for understanding the role ofdispersal, vicariance, cladogenesis, and extinction in formation ofSouthern Hemisphere biodiversity

Our results, which are based on the largest assemblage of bothtaxa and loci for the order Myrtales, provide strong evidence for

Fig. 5. Speciation rates in Myrtales utilizing BAMM. Chronogram of Myrtales with branch lengths colored according to speciation rate. Red circles indicate significantincreases in speciation rates present in the best shift configuration. Right panel shows speciation rates in seven family clades when each are evaluated separately. The bluecurve is the speciation rate for the entire Myrtales, the respective colored curve is for the specific family.

126 B.A. Berger et al. /Molecular Phylogenetics and Evolution 95 (2016) 116–136

Strephonema mStrephonema pLagunculariaLumnitzeraMacropteranthesCombretum fCombretum rQuisqualisCombretum cCombretum iBucidaConocarpusBuchenaviaTerminalia aTerminalia bTerminalia uTerminalia pTerminalia iTerminalia cTerminalia k

Pternandra cPternandra eMemecylonMouririAstroniaBelluciaHenrietteaMerianiaGraffenriedaMacrocentrumLeandraClidemiaMiconiaTococaBertoloniaBlakeaMedinillaMeriantheraLavoisieraRhynchantheraRhexiaArthrostemmaHeterocentronTibouchinaPterolepisMelastomaOsbeckia

HeteropyxisPsiloxylonLophostemonXanthostemonOsborniaCallistemonMelaleucaChamelauciumKunzeaLeptospermumSyncarpiaStockwelliaAngophoraEucalyptusLindsayomyrtusUromyrtusEugeniaMyrtusTristaniopsisTristaniaBackhousiaSyzygiumMetrosideros mMetrosideros c

100 80 60 40 20 0 60 50 40 30 20 10 0 80 60 40 20 0

CBA

0 0.8

Fig. 6. Speciation rates for three families in Myrtales utilizing BAMM. Chronograms of each family with branch lengths colored according to speciation rate. Red circlesindicate significant increases in speciation rates present in the best shift configuration. (A) Combretaceae. (B) Melastomataceae. (C) Myrtaceae.

B.A. Berger et al. /Molecular Phylogenetics and Evolution 95 (2016) 116–136 127

phylogenetic relationships among and within the nine families, forthe ancestral areas, timing, and biogeographical mode of thisremarkable radiation, and for the specific subclades that haveundergone significant shifts in speciation. We show for the firsttime, stronger support for the placement of Combretaceae as sisterto all remaining Myrtales. This relationship and other early clado-genetic events within Myrtales signal the importance of both SouthAmerica and Africa (western Gondwana) as ancestral areas in theearly biogeographic diversification of the order during the earlyCretaceous. This biogeographic diversification involves both vicari-ant and dispersal (including jump or founder) events among conti-nents. Lastly, we find three significant shifts in speciation rateswithin three families (Myrtaceae, Melastomataceae, Combre-taceae) and all on branches spanning the K–Pg boundary. Withthe wealth of previously published phylogenetic, fossil, biogeo-graphical, ecological, and evolutionary studies, the Myrtales isnow one of the better-known Southern Hemisphere dominant lin-eages of organisms.

4.2. Phylogenetic perspective of evolution within Myrtales

Our results provide new insight into phylogenetic relationshipswithin Myrtales and support an ancient radiation of the order clo-ser to the diversification of the eudicot angiosperms than has pre-viously been suggested (Bell et al., 2010; Magallón and Castillo,2009; Moore et al., 2010; Smith et al., 2010; Sytsma et al., 2004;Wang et al., 2009; Wikström et al., 2001). Seven major clades rep-resenting the major families (the CAP clade is considered a singlelineage) are resolved with high support. Aside from the positionof Combretaceae, all other interfamilial relationships are congruentwith previous studies.

Prior analyses of relationships within Myrtales have providedan improved understanding of both the phylogenetic relationshipsamong the families and the biogeography of the order; however,the placement of Combretaceae has remained unresolved (Contiet al., 1996, 1997, 2002; Soltis et al., 2011; Sytsma et al., 2004).The lack of adequate taxon sampling of the Combretaceae (Figs. 1and 2) contributed to the uncertainty in its phylogenetic placementin these previous studies. In our study, we explicitly improvedsampling in Combretaceae by including representatives of Stre-phonema (subfam. Strephonematoideae; Stace, 2007), the ‘man-grove’ clade (subfam. Combretoideae tribe Laguncularieae; Stace,

2007), and multiple representatives from each of the largest sub-tribes (subfam. Combretoideae tribe Combreteae subtribe Com-bretinae and subtribe Terminaliinae; Stace, 2007). Based on theML and BI analyses, Combretaceae is sister to the rest of the orderwith moderate support (Fig. 1). In addition to molecular phyloge-netic results, the position of Combretaceae is supported by recentfossil evidence considered the earliest occurrence of Myrtales(Manchester and O’Leary, 2010).

Importantly, this new phylogenetic framework for Myrtales (orthat based on sets of plausible PP trees) will be critically importantin re-evaluating the prodigious wealth of morphological andanatomical data that has accumulated for the order, such as thatfound in the hallmark work of Johnson and Briggs (1984) whoexamined ‘‘cladistically” 77 characters across Myrtales. For exam-ple, they highlight the unusual and distinctive presence of fibrousseed exotegmen as an example of convergence between the Com-bretaceae and the Lythraceae + Onagraceae clades based on a mor-phology derived phylogram. However, under the new DNA-basedframework with Combretaceae sister to the remaining families,this feature now can be interpreted as the plesiomorphic stateretained in these three families but subsequently lost in the largerclade sister to Lythraceae + Onagraceae. We are now using updatedMyrtales relationships to better understand morphospace evolu-tion of quantitative features such as pollen size and shape(Kriebel et al., 2015).

4.3. Origin and biogeography of Myrtales

The biogeographical scenario presented here is the first for theentire order Myrtales. Several families or closely related cladeshave been examined in more detail, notably Myrtaceae, Vochysi-aceae, Combretaceae, Melastomataceae, and the CAP clade(Berger, 2012; Conti et al., 2002; Gere, 2014; Maurin et al., 2010;Renner et al., 2001; Rutschmann et al., 2004; Sytsma et al.,2004). ARE at greater taxonomic depth presents both opportunitiesto gain biogeographical insight concerning early cladogeneticevents but also issues due to the lack of sampling within familiesand thus loss of biogeographical signal. This trade-off is clearlyobserved with Myrtales where on one hand, ARE provides a consis-tent geographical area (South America and/or Africa) for the stem,crown, and other deeper nodes of the order (and sister order Gera-niales), while on the other hand, a lack of dense taxonomic sam-

128 B.A. Berger et al. /Molecular Phylogenetics and Evolution 95 (2016) 116–136

pling within families, despite adequate coverage of major within-family subclades, generates ARE of some family crown nodes thatare at odds with previous family-level studies (see below for fur-ther discussion). One possible solution to this biogeographic chal-lenge, which we are now exploring, is to use a supermatrix-derivedchronogram that is framed on the well-supported Myrtales topol-ogy presented here that uses a smaller data set but with nearlycomplete taxon-gene coverage. Additionally, it is increasingly rec-ognized that extant taxa indicate persistence in time only, not nec-essarily in area, and that many clades have been subject toconsiderable extinction without an attendant fossil record (Crispet al., 2011; Kooyman et al., 2014). Thus, the following discussionof biogeographic history in Myrtales concedes that limitations infull taxon sampling and in our knowledge of extinction events donot permit unambiguous interpretations.

4.3.1. Early historical biogeography of MyrtalesThe biogeography of Myrtales is often attributed to a Gond-

wanan origin (Conti et al., 1997; Johnson and Briggs, 1984;Muller, 1981; Raven and Axelrod, 1974; Rutschmann et al., 2007;Sytsma et al., 2004). However, phylogenetic uncertainty (Contiet al., 1996, 1997; Sytsma et al., 2004), particularly with respectto the position of Combretaceae, has been a hindrance. Improvedsupport for the placement of Combretaceae and other cladesshown by this study provides a good phylogenetic framework forhypothesis testing and ancestral range estimation. Importantly,despite the use of a larger number of fossils, different genes, anddifferent analytical approaches to estimate a chronogram, thedates we infer here are largely consistent with previous dates forMyrtales evolution (Table 2) based on fixed fossil constraints inr8s (Sytsma et al., 2004). Most of the stem and crown datesobtained with r8s, other than clades not comparably sampled attheir bases, fall within the 95% credibility ages obtained withBEAST (Table 2). Thus, we argue that downstream analyses utiliz-ing this Myrtales chronogram(s) (e.g., biogeography, species diver-sification, character evolution) have a stronger phylogenetic andtemporal framework from which to interpret those results.

The BioGeoBEARS analyses support a west Gondwana (SouthAmerica, Africa) origin of the Myrtales (stem age �124 Mya, crownage �116 Mya) and not east Gondwana (Antarctica, Australia,India, Madagascar). ARE does suggest other possible areas (or com-bined areas) at the oldest nodes within Myrtales, but the dominantARE shown in Fig. 3 have fairly high likelihood scores. The westGondwana origin for Myrtales has previously been suggested byRaven and Axelrod (1974), as well as Johnson and Briggs (1984),and is also seen in the Geraniales, which is the sister group to Myr-tales (Sytsma et al., 2014). Diversification of all the major stemclades within Myrtales are also west Gondwana in origin and dateto no younger than 85–90 Mya.

Although dates obtained in our analysis are all younger than ageestimates for the initial breakup of Gondwana (180–150 Mya; Aliand Aitchison, 2008; Chatterjee et al., 2013; Jokat et al., 2003;McLoughlin, 2001; Scotese et al., 1988; White et al., 2013) intowest Gondwana and east Gondwana, subsequent continental con-nections, separations, and tectonic shifts allowed for persistent butintermittent contact and floral exchange into the Paleogene(Hallam, 1994). Raven and Axelrod (1974) argued that it appearslikely that upwards of 80 large angiosperm families were widelydistributed in tropical regions before the Gondwanan continentsseparated. According to Pitman et al. (1993), the beginning of theseparation of Africa from South America occurred just before135 Mya in the Early Cretaceous (Valanginian). The early phaseof rifting lasted from 135 to about 106 Mya (middle Albian) but,due to the rotation of Africa away from South America, tropicalWest Africa was still in contact with northeastern South America.Complete separation of the two continents was achieved sometime

between 106 and 84 Mya. Thus, the early diversification of Myr-tales, including all the major stem clades, are west Gondwana inorigin, date to no younger than 85–90 Mya, and are largely sup-portive of vicariant explanations.

Recent fossil evidence assigned to Combretaceae (Manchesterand O’Leary, 2010) represents the earliest occurrence of the orderand establishes a minimum date for tribe Combreteae that is olderthan the crown date used for the family in several previous studies(Bell et al., 2010; Magallón et al., 1999; Smith et al., 2010;Wikström et al., 2001). Our analyses estimate the stem age of Com-bretaceae (=crown of Myrtales) at �116 Mya, which is closelyaligned with the recent estimate for the crown of Myrtalesobtained when examining flowering plant diversification(�117 Mya; Magallón et al., 2015). Of the remaining Myrtales lin-eages, two major clades were established shortly thereafter in theearly Albian–late Aptian (�115 Mya): (1) Onagraceae + Lythraceaeand (2) Melastomataceae s.l. + CAP + Myrtaceae + Vochysiaceae.Separation of Melastomataceae s.l. + CAP from Myrtaceae+ Vochysiaceae occurred mid-Albian (�109 Mya) with all majorextant lineages represented by the Turonian (�91 Mya). As in pre-vious studies (APGIII, 2009; Conti et al., 1996, 1997; Rutschmannet al., 2007; Soltis et al., 2011; Sytsma et al., 2004), Onagraceae+ Lythraceae form a well-supported sister clade, which diverged�105 Mya in the Albian. This date represents the oldest estimatefor the clade by 12 Myr (93 Mya in Sytsma et al., 2004) and maybe attributed to new fossil findings, particularly of Lythrum andPeplis pollen from the Lower Campanian (82–81 Mya; Grímssonet al., 2011). These estimates agree with ages obtained in otherfossil-calibrated studies of major clades of eudicots, including theCampanulids (Beaulieu et al., 2013), Saxifragales (Jian et al.,2008), Malpighiales (Davis et al., 2005), Geraniales (Sytsma et al.,2014), Sapindales (Buerki et al., 2010), and Dipsacales (Bell andDonoghue, 2005).

4.3.2. Historical biogeography of CombretaceaeCombretaceae began differentiating �116 Mya with the split

into two extant subfamilies occurring �104 Mya. While the stemdate of the Combretaceae is older than previously proposed agesfor the crown of Myrtales (88.2 Mya, Magallón and Sanderson,2001; 107 Mya, Wang et al., 2009; 107 Mya, Wikström et al.,2001; 108 Mya, Magallón and Castillo, 2009; 111 Mya, Sytsmaet al., 2004), the estimates presented here better approximatethe ages of the order and the family for three reasons: (1) the inclu-sion of the West African Strephonema (subf. Strephonematoideae)allows more accurate estimation of the stem lineage of Combre-taceae (Conti et al., 1996, 1997; Sytsma et al., 2004); (2) improvedsampling provides support for Combretaceae as sister to the rest ofthe order; and (3) recent fossil descriptions attributed to Combre-taceae from the Albian–Cenomanian (112–93.5 Mya; Manchesterand O’Leary, 2010) pushes back the crown date of the family toat least �103 Mya.

Based on BioGeoBEARS analysis, the crown of Combretaceaewas in an ancestral area comprising Africa + South America(Fig. 2), which is consistent with a Gondwanan vicariant origin.Subsequent diversification within the family involved repeatedarea shifts, often involving more recent dispersals based on conti-nental separations (Fig. 3). We note that the long branch leading tothe core Combretaceae (minus the African trees of Strephonemaand the mangrove clade) has an ARE of South America for boththe stem and crown nodes. Recent studies with denser samplingwithin the core Combretaceae (Berger, 2012; Gere, 2014; Maurinet al., 2010) have ARE of Africa for the same nodes, suggesting thatbiogeographic signal for these nodes is impacted by taxon sam-pling. Family level ARE such as within Combretaceae, however,often cannot, or do not, take into account the third branch leadinginto the crown node – that coming from the sister clade. This might

B.A. Berger et al. /Molecular Phylogenetics and Evolution 95 (2016) 116–136 129

be especially problematic in biogeographic studies centering onCombretaceae as our results provide support that the sister cladeis the remainder of Myrtales.

4.3.3. Historical biogeography of Onagraceae + LythraceaeWhile multiple phylogenetic studies have corroborated most

tribal and generic relationships across Onagraceae (Bult andZimmer, 1993; Conti et al., 1993; Levin et al., 2003, 2004), the bio-geographical history of the family has remained speculative. AGondwanan origin was suggested by Raven and Axelrod (1974)and Raven (1988) based on the pattern of extant plant distribu-tions and plate tectonics. Conti et al. (1997) provided the first phy-logenetic context invoking a West Gondwanan origin, whileSytsma et al. (2004) first estimated divergence times using Penal-ized Likelihood (PL) and suggested a Late Cretaceous origin forOnagraceae + Lythraceae during the Albian (99 Mya) with subse-quent divergence of the two families by 93 Mya (Table 2). In con-trast to Sytsma et al. (2004), our data support an earlier EarlyCretaceous origin (�115 Mya) for the stem lineage of Ona-graceae + Lythraceae with subsequent splitting of the families bythe end of the Albian (�105 Mya). Both the timing of these eventsand the pattern of cladogenesis provide credence for a west Gond-wanan origin of Onagraceae with LDDE to the humid Boreotropicsof North America from South America during the Late Cretaceousto Paleocene likely playing a role in diversification of the clade(Wolfe, 1975). Within the family, two dispersal events back toSouth America from North America coincide with the cooling anddegradation of the Boreotropical flora during the Early Eocene Cli-matic Optimum/Paleocene–Eocene Thermal Maximum (EECO/PETM), 55–50 Mya (Zachos et al., 2001). Additionally, as coolingof the planet and subsequent drying intensified with the onset ofglaciation during the Oligocene �33 Mya (McLoughlin, 2001),new lineages, including Circaea and Fuchsia, evolved from a SouthAmerican lineage (�29 Mya) and moved northward and south-ward, respectively. Unlike previous studies (Berry et al., 2004;Xie et al., 2009), we did not constrain the age of Circaea + Fuchsia(41.5 Mya; based on Sytsma et al., 2004) and obtained an age esti-mate more than 10 Myr younger, but much more in agreementwith bioclimatic events. Consistent with previous studies, Circaeaexhibits a North American–Eastern Asian disjunct pattern (Xieet al., 2009), while Fuchsia demonstrates a pattern of LDDE fromSouth America to the South Pacific (Berry et al., 2004).

Previous phylogenetic studies of Lythraceae resolved most rela-tionships among extant genera and supported an Old World originof the family (Conti et al., 1997; Graham et al., 2005; Huang andShi, 2002; Morris, 2007). However, deep-level relationships of thefamily have remained unresolved, likely reflecting an early, rapidradiation (Morris, 2007). While our sampling represents a subsetof major lineages from the family, we do find evidence of rapiddiversification near the base of Lythraceae. Lythraceae divergedfrom Onagraceae by the end of the Albian (�105 Mya) with thecrown date for the family estimated at �96 Mya with two majorclades established within 5 Myr. The fossil record of Lythraceae isdiverse (wood, leaves, flowers, pollen and seeds) and widespreadin the northern hemisphere (Graham, 2013; Graham andGraham, 1971, 2014) indicative of a Boreotropical distribution. Bio-geographical estimation provides strong support for a South Amer-ican ancestor of the Onagraceae + Lythraceae clade, suggestingmovement from South America into North America early in thehistory of the family (�90 Mya). Fossils attributed to the familysupporting this distribution include (reviewed in Graham, 2013):Late Cretaceous (82–81 Mya) fossil pollen grains from Park County,Wyoming, USA, and western Siberia, Russia; leaf impressions andwood from the Deccan intertrappean beds of western India (LateCretaceous or Early Paleocene 66–63.7 Mya); pollen grains fromthe Paleocene in France; fruits from the Eocene London Clay Flora;

and seeds from the lower Eocene beds in England and Eocene Rus-sia. While one of the major clades (Trapa, Sonneratia, Duabanga,Lagerstroemia, Ammania, Lawsonia, Lythrum, Peplis, and Decodon)provides clear signal of movement from North America into Africa,Eurasia, and eventually into the South Pacific for some lineages, theother (Galpinia, Punica, Cuphea, Adenaria, and Hemia) indicatesLDDE from South America into Sub-Saharan African with move-ment northward in Eurasia during the Eocene by a single lineage(Punica).

4.3.4. Historical biogeography of Melastomataceae + CAP cladeRenner et al. (2001) suggested a Paleogene (Paleocene/Eocene),

non-Gondwana (north of Tethys Sea) diversification for Melastom-ataceae s.s. based exclusively on ndhF. They estimated the crowndiversification �53 Mya with substantial diversification withinthe family beginning �30 Ma. While an alternative biogeographichypothesis based on plate tectonics was put forward suggestinga much older age of Melastomataceae (Morley and Dick, 2003),Renner (2004) provided additional evidence refuting most claims.Sytsma et al. (2004) further supported the findings of Renneret al. (2001) using rbcL + ndhF and suggested a crown date forMelastomataceae of 56 Mya. In both Renner et al. (2001) andSytsma et al. (2004), Early Eocene fossil leaves (Hickey, 1977) wereused to constrain the crown of Melastomataceae s.s. including tribeKibessieae. However, based on the leaf description (Hickey, 1977)and personal correspondence (S. Renner, personal communica-tion), we used the fossils to offset the crown of Melastomataceaes.s. without tribe Kibessieae; hence the older dates for the crownMelastomataceae s.l. should be viewed as minimum ages asopposed to maximum ages.

Although some phylogenetic uncertainty remains with regardto the earliest diverging lineages of the family (Pternandra orMour-iri +Memecylon), biogeographic reconstruction with just extanttaxa included supports a South American origin for the family withits crown diversification at �64 Mya. Across Melastomataceae, aminimum of five dispersal events occurred. The shift from SouthAmerica to Australasia of the Pternandra clade occurred between64 and 8 Mya. Whether this migration was facilitated by SouthernHemisphere connections of South America, Antarctica, and Aus-tralia or by more direct LDDE is unclear. The spread of Pternandraacross tropical Australasia during the Paleogene was likely via dis-persal of fleshy capsules by frugivorous bats (Bolmgren andEriksson, 2010; Hodgkison et al., 2003; Tan et al., 2000), a mam-malian lineage undergoing diversification at this time (Gunnelland Simmons, 2005; O’Leary et al., 2013). During the Eocene(�38 Mya), widespread dispersal of the Medinilla lineage intosub-Saharan Africa, Madagascar, Eurasia, and Southeast Asiaoccurred, but limited sampling prevents any type of dispersal pat-tern from being recognized. Near the Oligocene–Miocene border(�27–24 Mya), movement of three lineages out of South Americaoccurred, including Rhexia moving into North America(�27 Mya), a clade within tribe Melastomeae dispersing to theOld World (Sub-Saharan Africa, Eurasia vs. Southeast Asia arerecovered equally as the most likely area), and LDDE of Memecylonfrom South America to the South Pacific (�21 Mya).

Evidence previously presented by Conti et al. (2002, 2004) andRutschmann et al. (2004, 2007) described the Gondwanan breakupas critical to the diversification of the CAP clade, with emphasis onthe isolation and rafting of the Indian subcontinent and subsequentdispersal of Crypteroniaceae out of India. Our analyses stronglysupport (ML-BS/BI-PP = 100/1.0) the relationship between theSoutheast Asian Crypteroniaceae with the west Gondwanan clade,as well as the split between the South American Alzateaceae andSouth African Penaeaceae (including Rhynchocalycaceae and Olini-aceae; ML-BS/BI-PP = 100/1.0). While the ideal for calibration is toassign a fossil as close to the node to be estimated as possible, no

130 B.A. Berger et al. /Molecular Phylogenetics and Evolution 95 (2016) 116–136

current fossil evidence can confidently be assigned to the CAP clade(Mid-Miocene heterocolpate pollen assigned to Crypteroniaceae isdifficult to distinguish from Melastomataceae s.l. and Penaeaceae;Muller, 1975). In lieu of this limitation, we view that the use ofmultiple fossils across Myrtales provides a robust context to betterapproximate the stem (�91 Mya) and crown (�53 Mya) dates ofthis lineage. Although our study used a subset of taxa from theCAP clade, the estimated stem age (�91 Ma) is compatible withpreviously published estimates that utilized more intensive taxonsampling (141–106 Mya, Conti et al., 2002; 109–62 Mya,Rutschmann et al., 2004; 87.25–72.15 Mya, Rutschmann et al.,2007) and thus supports the ‘‘out-of-India” hypothesis previouslyproposed.

4.3.5. Historical biogeography of MyrtaceaeStrong support (ML-BS/BI-PP = 100/1.0) for the inclusion of

Heteropyxis and Psiloxlon within Myrtaceae (now considered sub-family Psiloxyloideae) agrees with previous results (Biffin et al.,2010; Conti et al., 1996, 1997; Gadek et al., 1996; Sytsma et al.,2004; Thornhill et al., 2012, 2015; Wilson et al., 2005). The splitof subfamily Psiloxyloideae from the rest of Myrtaceae (subfamilyMyrtoideae) occurred �85 Mya, which is in accordance with theestimate of �83 Mya obtained by Sytsma et al. (2004). Similaritybetween the estimates for the crown date is likely due to the place-ment of the fossil Myrtaceidites lisimae (=Syncolporites) as a mini-mum date for the split between the two subfamilies; however,where Sytsma et al. (2004) used a fixed date with a tight confi-dence interval, we used a lognormal distribution to allow the pos-sibility for the clade to be much older. We note that the use of M.lisimae at the crown of Myrtaceae or M. mesonesus at the crown ofsubf. Myrtoideae as suggested by Thornhill et al. (2012) (or bothtogether) provides very similar dates for the Myrtaceae radiation.

Biogeographical estimations including subf. Psiloxyloideae havebeen problematic for two reasons in Myrtaceae: (1) the restrictedranges of Heteropyxis (SE Africa) and Psiloxylon (Mauritius andRéunion; part of the volcanic Mascarene Island chain) are consid-ered west Gondwanan, while the earliest diverging lineages ofsubf. Myrtoideae are more indicative of an east Gondwanan origin(Australian); and (2) the timing of the split between the two sub-families of Myrtaceae is younger than would be expected for aGondwanan vicariance event (complete separation by 100–95 Mya; Pitman et al., 1993). The current distribution of subf.Psiloxyloideae reflects the growing challenge of biogeographicalstudies that merge molecular systematics with stochastic modelsof area change. On one hand, sampling of extant taxa from acrosstheir distribution ranges is good; on the other hand, the overall his-tory of both the lineages and the areas (especially with regard tovolcanic island chains) is potentially more complex than currentdata can reconstruct. We estimate the split between the two gen-era at �18 Mya, 20 Myr younger than estimates (�40–38 Mya)obtained by Sytsma et al. (2004). Even with the reduced divergenceage, the issue remains that Mauritius, oldest of the MascareneIslands, is probably no older than 8 Myr (Bonneville et al., 1988;Hantke and Scheidegger, 1998) suggesting that the island endemicPsiloxylon must have occupied other areas during the Paleogene(Sytsma et al., 2004). We recognize two plausible scenarios: (1)Psiloxylon dispersed directly to Mauritius sometime after its emer-gence from the Indian Ocean; and (2) Psiloxylon was perhaps situ-ated on older islands of the chain now submersed and is restrictedto the oldest emergent island. Because our evidence is no moreconclusive than Sytsma et al. (2004), we also support the scenarioof an early Gondwanan origin in Africa, followed by a dispersal ofunknown age to the Mascarenes (no earlier than �18 Mya), withsubsequent island hopping and extinction events. The presenceof old lineages on young islands is a repeated and perplexing phe-

nomenon in the Mascarenes (Heads, 2011; Le Pechon et al., 2015;Nowak et al., 2014; Renner et al., 2010).

The difficulties surrounding the biogeographic history of subf.Psiloxyloideae and the likely extinction of early lineages of Gond-wanan Myrtaceae elsewhere (Crisp et al., 2011; Kooyman et al.,2014; Thornhill et al., 2015) may impact our ability to preciselydetermine the area of origin for the crown of the Myrtaceae andthe subsequent spread of the family whether via vicariance or dis-persal. BioGeoBEARS interprets, with our sampling, an origin ofMyrtaceae in Africa (consistent perhaps with location of the oldestknown fossil) and subsequent LDDE to Australia. However, we can-not explicitly exclude the Antarctic, Australian, or South Americanparts of Gondwana from the ancestral area of Myrtaceae, with theevidence for this subsequently removed by extinction. As withother Gondwanan plant families, the biogeographic history of theMyrtaceae family may be more complex than an all-LDDE or all-vicariance explanation (Crisp et al., 2011; Thornhill et al., 2015).Our results support a K–Pg boundary origin �67 Mya for subf. Myr-toideae in the part of Gondwana that broke off to form Australia.This date is very similar to that found (�64 Mya) by Thornhillet al. (2012) based on another set of fossils, although later thanthe �75 Mya date from Biffin et al. (2010) and Thornhill et al.(2015). Dispersal events to Africa, the Mediterranean, and theAmericas occurred much more recently during the Miocene.Although our phylogenetic analyses did not fully resolve relation-ships along the backbone of subf. Myrtoideae (Paleocene–Eocene),all nodal inferences clearly support an Australian/South Pacific ori-gin. The rapid and extensive radiation of Myrtaceae within Aus-tralia (often unresolved in phylogenetic analyses) occurs fromthe Eocene into the Miocene as cooling and aridification intensified(Byrne et al., 2008; Crisp and Cook, 2013; McLoughlin, 2001). Oursampling of South American taxa of fleshy-fruited myrtoids is lim-ited or includes genera of wider geographical distribution (e.g.,multiple areas attributed to a single placeholder); hence, there isa lack of strong dispersal signal to the Neotropics relative to thatseen in Sytsma et al. (2004), which sampled broadly across thefamily, and in Thornhill et al. (2015), which recognized up to nineLDDE events in lineages with succulent fruits. The twomain fleshy-fruited lineages (Syzygieae and Myrteae) appear to achieve theirpantropical distributions via dispersal during the Miocene. Phylo-genetic relationships found strong support for tribal associationsfor Eucalypteae + (Leptospermeae + Chamelaucieae), Osbornieae+ Melaleuceae, and Metrosidereae + Backhousieae, all of whichare also supported by Thornhill et al. (2012).

4.3.6. Historical biogeography of VochysiaceaeVochysiaceae is one of the more intriguing groups within Myr-

tales because it is so morphologically distinct that it was not placedin the order until Quirk (1980) and van Vliet and Baas (1984)described the presence of bicollateral vascular bundles and ves-tured pits—two anatomical synapomorphies of Myrtales. Thisanatomical association was subsequently corroborated by earliermolecular phylogenetic studies (Conti et al., 1996, 1997; Sytsmaet al., 2004). The amphi-atlantic distribution of the family has beenof interest for decades and has prompted many discussions ofvicariance (e.g., Axelrod, 1970) versus dispersal (e.g., Smith,1973; Thorne, 1972, 1973) as explanatory for the disjunct pattern.Sytsma et al. (2004) provided the first phylogenetic context using amatK + ndhF chronogram including six American taxa and one spe-cies of the African genus Erismadelphus. Those results indicated aseparation from the Myrtaceae �93 Mya with a crown group radi-ation for extant taxa �36–33 Mya. Our data demonstrate a similarpattern of an ancient divergence from the Myrtaceae (�101 Mya)followed by a long period of time until the crown radiation�39 Mya.

B.A. Berger et al. /Molecular Phylogenetics and Evolution 95 (2016) 116–136 131

With regard to phylogenetic relationships, Sytsma et al. (2004)indicated a moderately supported sister relationship betweenAmerican Erisma and African Erismadelphus (ML-BS = 85), whichwas then sister to a clade comprised of the remaining five Neotrop-ical species (ML-BS < 50). The origin of African taxa out of a SouthAmerican clade was consistent with previous morphological andmolecular studies (Litt, 1999, 2003; Litt and Stevenson, 2003). Inthis study with more limited taxon sampling but expanded locisampling, we found the African genus Erismadelphus sister to allNeotropical Vochysiaceae (BI-PP = 0.92). However, the African/American split at the base of the family (rather than more internal)is still in agreement with previous studies as the stem lineage ofthe family supports a South American origin. The timing of theSouth American – African split in the late Eocene indicates a LDDEevent in Vochysiaceae across an Atlantic Ocean that was thenalmost 2/3rd formed, one of a number of similar South Ameri-can–African disjunct patterns best explained by LDDE rather thanvicariance (Givnish and Renner, 2004; Givnish et al., 2000, 2004).

4.4. Myrtales exhibits three significant shifts in species diversificationrates

An important question now being addressed more critically inevolutionary biology is the nature of the processes that lead to sig-nificant shifts in speciation and/or extinction rates within clades(e.g., Alfaro et al., 2009; Antonelli et al., 2015; Biffin et al., 2010;Givnish et al., 2014; Linder et al., 2014; McGuire et al., 2014;Rabosky et al., 2014; Smith et al., 2011). Relevant issues in detect-ing significant rate shifts include incorporating extinction, phylo-genetic uncertainty, phylogenetic scale, sampling density,correlation and/or causality of biotic or niche attributes drivingthe rate shifts. The program BAMM, as now implemented, canaddress a number of these issues. The largely Southern Hemi-sphere Myrtales is an ideal candidate for such studies consideringits relatively old origins in the Cretaceous, a widespread biogeo-graphical distribution, occupation of diverse habitats, considerablemorphological diversity, the presence of both species-rich anddepauperate clades, and a now well-supported and dated phy-logeny. Our analyses were conducted at two scales: (1) across Myr-tales with 102 species, but carefully chosen to represent all themajor distinct clades within families, and (2) within individualfamilies or family groups.

The baseline or average rate of net diversification for Myrtales isnot striking (0.06–0.07 species/Myr; Table 3). This rate is very sim-ilar to those estimated by Magallón and Sanderson (2001) usingdifferent approaches (0.079–0.097 species/Myr), and comparableto average rates they report for Eudicots and Angiosperms.Antonelli et al. (2015) provided estimates for net diversificationin Myrtales (and 16 other lineages) from a supermatrix Angios-perm tree, and while not strictly comparable to our study as theyseparated out rates based on biogeographical area, their resultsare in line with those presented here (in species/Myr): Africantropics (0.111), Asian tropics (0.051), American tropics (0.009). Ingeneral, and demonstrated in rates-through-time analysis(Fig. 4), Mytales exhibited long-term speciation rate decline forthe first �60 Myr and extinction rate increase throughout its entire�116 Myr diversification. However, at �65 Mya speciation (andnet diversification) increases rapidly (Fig. 4). Around this timethree significant rate shifts are seen within the Myrtales (Fig. 5)and, though they are placed at different ages (�76, �64,�43 Mya), all occur on branches that span the K–Pg boundary(66 Mya). Antonelli et al. (2015) presented intriguing evidence,based on their angiosperm-wide meta-analysis of net diversifica-tion rates in the context of biogeography, that a peak in angios-perm speciation and in biogeographical shifts between tropicalregions is correlated to another climatic event – the Paleocene–

Eocene Thermal Maximum at �56.3 Mya. Whether either of thesetwo major climatic events with profound vegetation changes(Jaramillo et al., 2010; McElwain and Punyasena, 2007) influencedthe shifts in diversification within Myrtales remains unknown.

The first and oldest shift in speciation rates within Myrtales,which occurred �76 Mya on the branch leading to the subf. Myr-toideae (after divergence of the African lineage comprisingHeteropyxis and Psiloxylon), coincides with the likely LDDE of itsancestors to Australia from Africa (Fig. 3). A paraphyletic assem-blage of largely Australian, capsular-fruited, and arid-adaptedtribes (Xanthostemoneae, Lophostemoneae, Osbornieae, Melaleu-ceae, Lindsayomyrteae, Syncarpieae, Eucalypteae, Leptospermeae,and Chamelaucieae) are the first diverging clades within subf. Myr-toideae. Notably, a large, Southern Hemisphere clade containing allother tribes, including the fleshy-fruited and mesic-adapted taxa,diverged in the Paleocene and rapidly diversified starting at theEocene/Oligocene boundary. Thus, the shift in species diversifica-tion in Myrtaceae is correlated with a shift to a new biogeographicarea and seemingly to one of the strongest biome/niche shifts inthe Myrtales – tropical forest to arid communities. This niche shift,a pattern seen in other major radiations (Crisp et al., 2009;Donoghue and Edwards, 2014; Kissling et al., 2012; Linder et al.,2014), would be possible however only in the mid-Miocene(15 Mya) when initial aridification of Australia occurred afterSouth America broke free of Antarctica and initiated the Antarcticcircumpolar current (Byrne et al., 2008). Thus, our dates (late Cre-taceous/early Paleocene) and even earlier dates of Ladiges et al.(2003), Thornhill et al. (2015) for the crown radiation of subf. Myr-toideae surprisingly suggest a rapid radiation of taxa all now lar-gely adapted and confined to an arid biome that would notappear in Australia for at least another 40 Myr.

What biotic features of the Myrtaceae may be correlated withthis shift in species diversification? Ancestral state reconstructionof fruit type onto the phylogeny of Myrtales (unpublished data)allowed us to test the hypothesis that birds or bats may haveplayed a role in this dispersal event, which was then followed bythe shift in speciation rate. We found no support for this hypothe-sis since the common ancestor of subf. Myrtoideae unambiguouslyhad dry fruit, with fleshy fruits evolving much later some 30 mil-lion years ago. Other studies have also suggested the commonancestor and early divergent lineages of Myrtaceae to be dryfruited (Biffin et al., 2010). In that study focused on the subcladeof the family with two, probably unrelated, fleshy fruit tribes,increased rates of speciation were correlated with the acquisitionof fleshy, bird-dispersed fruits. A second set of factors that maybe involved in the diversification of Myrtaceae include the evolu-tion and radiation of the most diverse group of bees, the corbicu-lates (Martins et al., 2014), and subsequently �50 Mya theallodapines in Australia (Chenoweth and Schwarz, 2011). Beesare the most common pollinators of Australian Myrtaceae today(Beardsell et al., 1993) and further studies should explicitly testthe impact of bee pollinators on subf. Myrtoideae. Interestingly,the shift in speciation rate detected in Myrtaceae when examiningthe entire order was not detected in the Myrtaceae-only analysis(Fig. 6). This is likely due to its occurrence toward the base of theMyrtaceae-only phylogeny. A supermatrix chronogram of Myr-taceae may be necessary to obtain sufficient taxon sampling tomore rigorously test the number and placement(s) of significantrate shifts and examine the potential role of these and other bioticand abiotic factors in promoting diversification within the family.

The second shift in species rates within Myrtales occurred�64 Mya (branch age of 64.5–62.2 Mya) at the base of Melastom-ataceae, after the split between the genus Pternandra (tribe Kibes-sieae) and the rest of the family (Fig. 4), and clearly occurred in aSouth American lineage based on our phylogenetic estimate andARE (Fig. 3). This shift corresponds with the suggested synapomor-

132 B.A. Berger et al. /Molecular Phylogenetics and Evolution 95 (2016) 116–136

phy for the subfamily Melastomatoideae of libriform fibers(Renner, 1993). This shift was not detected in the Melastomataceaeonly BAMM analysis that shows support for a shift within tribeMiconieae, after the divergence from Leandra. Both of these result-ing shifts in speciation rates are somewhat unexpected becausefleshy-fruited clades such as the Miconieae are so diverse thatwe would have predicted an association between diversificationrates and fleshy fruits within the Melastomataceae. This resultcould also be due to the effects of scaling, sampling and phyloge-netic resolution discussed below, which can affect the placementof shifts such as the one we expected in the common ancestor ofthe tribe Miconieae. Future studies with higher taxon samplingin the Melastomataceae are needed to test this hypothesis.

A second possible explanation for the shift in speciation ratetoward the base of the Melastomataceae is the evolution of porici-dal anthers that provide a pollen reward via buzz-pollination andare visited by female bees (Buchmann, 1983; De Luca andVallejo-Marín, 2013). Most of the 6000 species of Melastomataceaeare thought to be buzz-pollinated and poricidal anthers wereinferred to have evolved early in the evolution of the family(Clausing and Renner, 2001; Renner, 1989, 1993). Since onlyfemale bees can access the pollen reward offered by flowers inthe family, buzz-pollination is considered a highly specializedmechanism of pollination and likely contributed to the radiationof the family. Such is the strength of the relationship betweenthese melastomes and their bee pollinators, that the Melastomat-aceae have been suggested to be trapped in an adaptive peak withtheir bee pollinators, comparable to what is suggested for theMalpighiaceae and Solanaceae (Biffin et al., 2010; Renner, 1989).Whether this close relationship between bees and the Melastomat-aceae has resulted in the observed species rate shift is unclear.Buzz pollination has rarely evolved within Myrtales aside fromthe Melastomataceae (but see Proença, 1992). For this reason, wepresently lack statistical power to test for the relationship betweenporicidal anthers/buzz-pollination and an increase in speciationrate.

A recent study looking at shifts in diversification rates associ-ated with the transition from the forest to the fynbos habitat inSouth Africa found a shift within the CAP clade that was notdetected in our study (Onstein et al., 2014). The difference in shiftsis likely due to the different strategies in taxon and gene samplingemployed by both studies and highlights the sensitivity in detect-ing shifts in diversification rates associated with scale.

The final and third shift in speciation rate within Myrtales, andthe best supported with Bayes factor evidence, occurred withinCombretaceae at �43 Mya. The shift occured after the earlier diver-gence of the west African tree genus Strephonema and the largelypantropical mangrove clade. Although placed at �43 Mya, thebranch exhibiting the rate shift spans from 94.5 to 33.2 Mya(Fig. 5). Our biogeographical reconstruction indicated that the shiftoccurred in South America, although Africa was important earlierin family differentiation. Greater taxonomic sampling of thepantropical Combretum and Terminalia clade (subtribes Combreti-nae and Terminaliinae) derived from this shift in speciation ratesuggests Africa as a more likely ARE for this branch (Berger,2012; Gere, 2014; Maurin et al., 2010). If so, the three significantshifts within Myrtales at this scale involved three different South-ern Hemisphere continents: Africa, South America, and Australia.

The Combretaceae occupy a diversity of ecological niches andexhibit striking habit diversity (liana, mangrove, shrub, tree). Theshift in speciation rate occured after the tree genus Strephonemaand a mainly mangrove clade had already diverged. The cladedemonstrating the increased rate of speciation, the Combretum(272 spp.) and Terminalia (232 spp.) lineages, are rarely fleshy

fruited (as in the Melastomataceae and Myrtaceae clades showingsignificant speciation rate shifts). However, one feature this cladeshares is winged fruits, although lacking in some members andconvergent with Macropteranthes (Manchester and O’Leary, 2010)in the mangrove clade. Additionally, this clade shows pronouncedlevels of dispersal events between the three main tropical areas(Berger, 2012) and these events may have spurred repeated diver-sification in newly occupied areas.

The analyses presented here, both at the ordinal and familylevels in Myrtales, reinforce the importance of scale, sampling,and phylogenetic resolution when examining and interpreting pat-terns of diversification. First, we demonstrate that scale influencesthe best shift configuration despite the same taxa being analyzed.For example, the rate shifts seen in Melastomataceae are differentthan that seen when the Melastomataceae are examined in thecontext of the entire Myrtales (Fig. 6). Second, increased taxonsampling (and thus larger numbers of terminals that can be scoredfor extant species richness) should provide more precise identifica-tion of rate shift placements. And third, most diversification anal-yses assume a single tree as representative of the group beingstudied. Specifically, Biffin et al. (2010) acknowledge this issue asthey evaluated rate shifts in speciation among the fleshy-fruitedMyrtaceae, two main lineages whose relationships to each otherare not yet known with certainty. Future studies should focus ontwo approaches. One is taking a supermatrix approach (e.g.,Antonelli et al., 2015), although this approach entails assumptions(and thus trade-offs) using a single tree and estimating branchlengths with a high percentage of missing data. A second approachshould examine PP distributions of trees (as we did in biogeo-graphic analyses with BioGeoBEARS) and summarizing signal inrate shifts across many trees. We note that a BAMM analysis con-ducted on one of the most different 100 PP trees examined in thisstudy (Combretaceae sister to Onagraceae + Lythraceae), the samethree significant rate shifts in speciation were uncovered withinMyrtales although at different points on their respective branches(data not shown). Thus, these results should be viewed as coarse-grained, and future studies using supermatrix approaches (e.g.,Kriebel et al., 2015) and suites of alternative but likely trees mayprovide refinement on the placement of and processes leading tosignificant shifts in speciation rates within Myrtales.

Acknowledgments

The authors thank Reinaldo Aguilar, Eladio Cruz, Roger BlancoSegura, Álvaro Herrera, Aarón Rodríguez Contreras, Francisco JavierRendon Sandoval, Shirley Graham, Tim Curran, Craig Costion andthe Organization for Tropical Studies in aiding with field collec-tions; Olivier Maurin, Michelle van der Bank, Elena Conti, FrankRutschmann, Fabian Michelangeli, and the Royal Botanical Gar-dens, Kew, for DNA samples; Jim Solomon, Jeremy Bruhl, and IanTelford for curatorial assistance with herbarium samples; BillHaber for the photograph of Alzatea verticillata; and Sarah Friedrichfor graphical art assistance. The first author also thanks DavidBaum, Cécile Ané, Tom Givnish, and Ken Cameron for helpful com-ments and guidance toward completion of this dissertationresearch, and the Botany Department at the University ofWisconsin-Madison for financial support.

Appendix A. Supplementary material

Supplementary data associated with this article can be found, inthe online version, at http://dx.doi.org/10.1016/j.ympev.2015.10.001.

B.A. Berger et al. /Molecular Phylogenetics and Evolution 95 (2016) 116–136 133

References

Alfaro, M.E., Santini, F., Brock, C., Alamillo, H., Dornburg, A., Rabosky, D.L., Carnevale,G., Harmon, L.J., 2009. Nine exceptional radiations plus high turnover explainspecies diversity in jawed vertebrates. Proc. Natl. Acad. Sci. 106, 13410–13414.

Ali, J.R., Aitchison, J.C., 2008. Gondwana to Asia: plate tectonics, paleogeography andthe biological connectivity of the Indian sub-continent from the Middle Jurassicthrough latest Eocene (166–35 Ma). Earth Sci. Rev. 88, 145–166.

Alteker, G., Dwarkadas, S., Huelsenbeck, J.P., Ronquist, F., 2004. Parallel Metropolis-coupled Markov chain Monte Carlo for Bayesian phylogenetic inference.Bioinformatics 20, 407–415.

Ane, C., Larget, B., Baum, D.A., Smith, S.D., Rokas, A., 2007. Bayesian estimation ofconcordance among gene trees. Mol. Biol. Evol. 24, 412–426.

Antonelli, A., Zizka, A., Silvestro, D., Scharn, R., Cascales-Miñana, B., Bacon, C.D.,2015. An engine for global plant diversity: highest evolutionary turnover andemigration in the American tropics. Front. Genet. 6, 130.

APGIII, 2009. An update of the angiosperm phylogeny group classificationfor the orders and families of flowering plants: APG III. Bot. J. Linn. Soc. 161,105–121.

Armstrong, K.E., Stone, G.N., Nicholls, J.A., Valderrama, E., Anderberg, A.A.,Smedmark, J., Gautier, L., Naciri, Y., Milne, R., Richardson, J.E., 2014. Patternsof diversification amongst tropical regions compared: a case study inSapotaceae. Front. Genet. 5, 362.

Axelrod, D.I., 1970. Mesozoic paleogeography and early angiosperm history. Bot.Rev. 36, 277–319.

Baker, W.J., Couvreur, T.L.P., 2013. Global biogeography and diversification of palmssheds light on the evolution of tropical lineages. I. Historical biogeography. J.Biogeogr. 40, 274–285.

Barker, N.P., Weston, P.H., Rutschmann, F., Sauquet, H., 2007. Molecular dating ofthe ‘‘Gondwanan” plant family Proteaceae is only partially congruent with thetiming of the break-up of Gondwana. J. Biogeogr. 34, 2012–2027.

Baum, D.A., 2007. Concordance trees, concordance factors, and the exploration ofreticulate genealogy. Taxon 56, 417–426.

Beardsell, D.V., O’Brien, S.P., Williams, E.G., Know, R.B., Calder, D.M., 1993.Reproductive biology of Australian Myrtaceae. Austr. J. Bot. 41, 511–526.

Beaulieu, J.M., Tank, D.C., Donoghue, M.J., 2013. A Southern Hemisphere origin forcampanulid angiosperms and traces of the break-up of Gondwana. BMC Evol.Biol. 13, 80.

Bell, C.D., Donoghue, M.J., 2005. Dating the Dipsacales: comparing models, genes,and evolutionary implications. Am. J. Bot. 92, 284–296.

Bell, C.D., Soltis, D.E., Soltis, P.S., 2010. The age and diversification of theangiosperms re-revisited. Am. J. Bot. 97, 1296–1303.

Berger, B., 2012. Myrtales: Molecules, Mangroves and Metrosideros. Ph.D. Thesis,University of Wisconsin, Madison, Wisconsin.

Berry, P.E., Skvarla, J.J., Partridge, A.D., Macphail, M.K., 1990. Fuchsia pollen from theTertiary of Australia. Aust. Syst. Bot. 3, 739–744.

Berry, P.E., Hahn, W.J., Sytsma, K.J., Hall, J.C., Mast, A., 2004. Phylogeneticrelationships and biogeography of Fuchsia (Onagraceae) based on noncodingnuclear and chloroplast DNA data. Am. J. Bot. 94, 601–614.

Biffin, E., Lucas, E., Craven, L., Ribeiro da Costa, I., Harrington, M., Crisp, M., 2010.Evolution of exceptional species richness among lineages of fleshy-fruitedMyrtaceae. Ann. Bot. 106, 79–93.

Bolmgren, K., Eriksson, O., 2010. Seed mass and the evolution of fleshy fruits inangiosperms. Oikos 119, 707–718.

Boltenhagen, E., 1976. Pollens et Spores Senoniens du Gabon. CahiersMicropaleontol. 3, 1–21.

Bonneville, A., Barriot, J.P., Bayer, R., 1988. Evidence from geoid data of a hotspotorigin for the southern Mascarene Plateau and Mascarene Islands (IndianOcean). J. Geophys. Res. 93 (B5), 4199–4212.

Buchmann, S.L., 1983. Buzz pollination in angiosperms. In: Jones, C.E., Little, R.J.(Eds.), Handbook of Experimental Pollination Biology. van Nostrand ReinholdCompany, Inc., New York, pp. 73–113.

Buerki, S., Forest, F., Alvarez, N., Nylander, J.A.A., Arrigo, N., Sanmartín, I., 2010. Anevaluation of new parsimony-based versus parametric inference methods inbiogeography: a case study using the globally distributed plant familySapindaceae. J. Biogeogr. 38, 531–550.

Buerki, S., Forest, F., Alvarez, N., 2011. An evaluation of new parsimony-basedversus parametric inference methods in biogeography: a case study using theglobally distributed plant family Sapindaceae. J. Biogeogr. 38, 531–550.

Bukontaite, R., Miller, K.B., Bergsten, J., 2014. The utility of CAD in recoveringGondwanan vicariance events and the evolutionary history of Aciliini(Coleoptera: Dytiscidae). BMC Evol. Biol. 14, 5.

Bult, C.J., Zimmer, E.A., 1993. Nuclear ribosomal RNA sequences for inferring tribalrelationships within Onagraceae. Syst. Bot. 18, 48–63.

Byrne, M., Yeates, D.K., Joseph, L., Kearney, M., Bowler, J., Williams, M.A.J., Cooper, S.,Donnellan, S.C., Keogh, J.S., Leys, R., Melville, J., Murphy, D.J., Porch, N., Wyrwoll,K.-H., 2008. Birth of a biome: insights into the assembly and maintenance of theAustralian arid zone biota. Mol. Ecol. 17, 4398–4417.

Cardinal-McTeague, W.M., Sytsma, K.J., Hall, J.C., unpublished. Biogeography anddiversification of Brassicales: a 103 million year old tale. Mol. Phylogenet. Evol.(unpublished)

Chacón, J., Renner, S.S., 2014. Assessing model sensitivity in ancestral areareconstruction using Lagrange: a case study using the Colchicaceae family. J.Biogeogr. 41, 1414–1427.

Chatterjee, S., Goswami, A., Scotese, C.R., 2013. The longest voyage: tectonic,magmatic, and paleoclimatic evolution of the Indian plate during its northwardflight from Gondwana to Asia. Gondwana Res. 23, 238–267.

Chenoweth, L.B., Schwarz, M.P., 2011. Biogeographical origins and diversification ofthe exoneurine allodapine bees of Australis (Hymenoptera, Apidae). J. Biogeogr.38, 1471–1483.

Christenhusz, M.J., Chase, M.W., 2013. Biogeographical patterns of plants in theNeotropics – dispersal rather than plate tectonics is most explanatory. Bot. J.Linn. Soc. 171, 277–286.

Clausing, G., Renner, S.S., 2001. Molecular phylogenetics of Melastomataceae andMemecylaceae: implications for character evolution. Am. J. Bot. 88, 486–498.

Conti, E., Fischbach, A., Sytsma, K.J., 1993. Tribal relationships in Onagraceae:implications from rbcL sequence data. Ann. Mo. Bot. Gard. 80, 672–685.

Conti, E., Litt, A., Sytsma, K.J., 1996. Circumscription of Myrtales and theirrelationships to other rosids: evidence from rbcL sequence data. Am. J. Bot.83, 221–233.

Conti, E., Litt, A., Wilson, P.G., Graham, S.A., Briggs, B.G., Johnson, L.A.S., Sytsma, K.J.,1997. Interfamilial relationships in Myrtales: molecular phylogeny and patternsof morphological evolution. Syst. Bot. 22, 629–647.

Conti, E., Eriksson, T., Schönenberger, J., Sytsma, K.J., Baum, D.A., 2002. Early Tertiary‘‘Out-of-India” dispersal of Crypteroniaceae: evidence from phylogeny andmolecular dating. Evolution 56, 1931–1942.

Conti, E., Rutschmann, F., Eriksson, T., Sytsma, K.J., Baum, D.A., 2004. Calibration ofmolecular clocks and the biogeographic history of Crypteroniaceae: a reply toMoyle. Evolution 58, 1874–1876.

Cook, L.G., Crisp, M.D., 2005. Not so ancient: the extant crown group of Nothofagusrepresents a post-Gondwanan radiation. Proc. R. Soc. Lond. Ser. B 272, 2535–2544.

Couper, R.A., 1960. New Zealand Mesozoic and Cainozoic Plant Microfossils. R.E.Owen, Government Printer, Wellington.

Cracraft, J., 1988. Deep-history biogeography: retrieving the historical pattern ofevolving continental biotas. Syst. Zool. 37, 221–236.

Crisp, M.D., Cook, L.G., 2013. How was the Australian flora assembled over the last65 million years? A molecular phylogenetic perspective. Annu. Rev. Ecol. Evol.Syst. 44, 303–324.

Crisp, M.D., Arroyo, M.T.K., Cook, L.G., Gandolfo, M.A., Jordan, G.J., Mcglone, M.S.,Weston, P.H., Westoby, M., Wilf, P., Linder, P.H., 2009. Phylogenetic biomeconservatism on a global scale. Nature 458, 754–756.

Crisp, M.D., Trewick, S.A., Cook, L.G., 2011. Hypothesis testing in biogeography.Trends Ecol. Evol. 26, 66–72.

Dahlgren, R., Thorne, R., 1984. The order Myrtales: circumscription, variation, andrelationships. Ann. Mo. Bot. Gard. 71, 633–699.

Darriba, D., Taboada, G.L., Doallo, R., Posada, D., 2012. JModelTest 2: more models,new heuristics and parallel computing. Nat. Methods 9 (8), 772.

Davis, C.C., Bell, C.D., Mathews, S., Donoghue, M.J., 2002. Laurasian migrationexplains Gondwanan disjunctions: evidence from Malpighiaceae. Proc. Natl.Acad. Sci. 99, 6833–6837.

Davis, C.C., Webb, C.O., Wurdack, K.J., Jaramillo, C.A., Donoghue, M.J., 2005.Explosive radiation of Malpighiales supports a mi-Cretaceous origin ofmodern tropical rain forests. Am. Nat. 165, E36–E65.

Davis, C.C., Xi, Z., Mathews, S., 2014. Plastid phylogenomics and green plantphylogeny: almost full circle but not quite there. BMC Biol. 12, 1–4.

De Luca, P.A., Vallejo-Marín, M., 2013. What’s the ‘‘buzz” about? The ecology andevolutionary significance of buzz-pollination. Curr. Opin. Plant Biol. 16, 429–435.

de Queiroz, A., 2005. The resurrection of oceanic dispersal in historicalbiogeography. Trends Ecol. Evol. 20, 68–73.

Deng, J.-B., Drew, B.T., Mavrodiev, E.V., Gitzendanner, M.A., Soltis, P.S., Soltis, D.E.,2015. Phylogeny, divergence times, and historical biogeography of theangiosperm family Saxifragaceae. Mol. Phylogenet. Evol. 83, 86–98.

Donoghue, M.J., Edwards, E.J., 2014. Biome shifts and niche evolution in plants.Annu. Rev. Ecol. Evol. Syst. 45, 547–572.

Drew, B.T., Sytsma, K.J., 2012. Phylogenetics, biogeography, and staminal evolutionin the tribe Mentheae (Lamiaceae). Am. J. Bot. 99, 933–953.

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

Drummond, A.J., Suchard, M.A., Xie, D., Rambaut, A., 2012. Bayesian phylogeneticswith BEAUti and the BEAST 1.7. Mol. Biol. Evol. 29, 1969–1973.

Edwards, S.V., Boles, W.E., 2002. Out of Gondwana: the origin of passerine birds.Trends Ecol. Evol. 17, 347–349.

Fiz, O., Varga, P., Alarcón, M., Aedo, C., García, J.L., Aldasoro, J.J., 2008. Phylogeny andhistorical biogeography of Geraniaceae in relation to climate changes andpollination ecology. Syst. Bot. 33, 326–342.

Forster, P.I., 1994. Notes on Dansiea Byrnes and Macropteranthes F. Muell.(Combretaceae). Austrobaileya 4, 149–153.

Friedman, M., Keck, B.P., Dornburg, A., Eytan, R.I., Martin, C.H., Hulsey, C.D.,Wainwright, P.C., Near, T.J., 2013. Molecular and fossil evidence place the originof cichlid fishes long after Gondwanan rifting. Proc. R. Soc. Lond. Ser. B 280,20131733.

Friis, E.M., Pedersen, K.R., Crane, P.R., 1992. Esgueiria gen. nov., fossil flowers withcombretaceous features from the Late Cretaceous of Portugal. Biol. Skrift. 41, 1–45.

Gadek, P.A., Wilson, P.G., Quinn, C.J., 1996. Phylogenetic reconstruction inMyrtaceae using matK, with particular reference to the position of Psiloxylonand Heteropyxis. Aust. Syst. Bot. 9, 283–290.

134 B.A. Berger et al. /Molecular Phylogenetics and Evolution 95 (2016) 116–136

Gallaher, T., Callmander, M.W., Buerki, S., Keeley, S.C., 2014. A long distancedispersal hypothesis for the Pandanaceae and the origins of the Pandanustectorius complex. Mol. Phylogenet. Evol. 83, 20–32.

Gamble, T., Bauer, A.M., Colli, G.R., Greenbaum, E., Jackman, T.R., Vitt, L.J., Simons, A.M., 2011. Coming to America: multiple origins of New World geckos. J. Evol.Biol. 24, 231–244.

Gandolfo, M.A., Hermsen, E.J., Zamaloa, M.C., Nixon, K.C., González, C.C., Wilf, P.,Cúneo, N.R., Johnson, K.R., 2011. Oldest known Eucalyptus macrofossils are fromSouth America. PLoS ONE 6, e21084.

Gere, J., 2014. Combretaceae: Phylogeny, Biogeography and DNA Barcoding. Ph.D.(Botany), University of Johannesburg. <https://ujdigispace.uj.ac.za> (accessed26.08.14) (Unpublished).

Givnish, T.J., Renner, S.S., 2004. Tropical intercontinental disjunctions: Gondwanabreakup, immigration from the boreotropics, and transoceanic dispersal. Int. J.Plant Sci. 165, S1–S6.

Givnish, T.J., Evans, T.M., Zjhra, M.L., Berry, P.E., Sytsma, K.J., 2000. Molecularevolution, adaptive radiation, and geographic diversification in theamphiatlantic family Rapateaceae: evidence from ndhF sequence data.Evolution 54, 1915–1937.

Givnish, T.J., Millam, K.C., Evans, T.M., Hall, J.C., Pires, J.C., Berry, P.E., Sytsma, K.J.,2004. Ancient vicariance or recent long-distance dispersal? Inferences aboutphylogeny and South American–African disjunctions in Rapateaceae andBromeliaceae based on ndhF sequence data. Int. J. Plant Sci. 165, S35–S54.

Givnish, T.J., Millam, K.C., Mast, A.R., Paterson, T.B., Theim, T.J., Hipp, A.L., Henss, J.M., Smith, J.F., Wood, K.R., Sytsma, K.J., 2009. Origin, adaptive radiation, anddiversification of the Hawaiian lobeliads (Asterales: Campanulaceae). Proc. R.Soc. Lond. Ser. B 276, 407–416.

Givnish, T.J., Barfuss, M.H.J., Van Ee, B., Riina, R., Schulte, K., Horres, R., Gonsiska, P.A.,Jabaily, R.S., Crayn, D.M., Smith, J.A.C., Winter, K., Brown, G.K., Evans, T.M., Holst,B.K., Luther, H., Till, W., Zizka, G., Berry, P.E., Sytsma, K.J., 2014. Adaptiveradiation, correlated and contingent evolution, and net species diversification inBromeliaceae. Mol. Phylogenet. Evol. 71, 55–78.

Goldenberg, R., de Fraga, C.N., Fontana, A.P., Nicolas, A.N., Michelangeli, F.A., 2012.Taxonomy and phylogeny of Merianthera (Melastomataceae). Taxon 61, 1040–1056.

Graham, S.A., 2013. Fossil records in the Lythraceae. Bot. Rev. 28, 410–420.Graham, A., Graham, S.A., 1971. Geologic history of Lythraceae. Brittonia 23, 335–

346.Graham, S.A., Graham, A., 2014. Ovary, fruit, and seed morphology of the

Lythraceae. Int. J. Plant Sci. 175, 202–240.Graham, S.A., Hall, J.C., Sytsma, K.J., Shi, S., 2005. Phylogenetic analysis of the

Lythraceae based on four gene regions and morphology. Int. J. Plant Sci. 166,995–1017.

Grímsson, F., Zetter, R., Hofmann, C.-C., 2011. Lythrum and Peplis from the LateCretaceous and Cenozoic of North America and Eurasia: new evidencesuggesting early diversification within the Lythraceae. Am. J. Bot. 98, 1801–1815.

Guindon, S., Gascuel, O., 2003. A simple, fast and accurate method to estimate largephylogenies by maximum-likelihood. Syst. Biol. 52, 696–704.

Gunnell, G.F., Simmons, N.B., 2005. Fossil evidence and the origin of bats. J.Mammal. Evol. 12, 209–246.

Haddrath, O., Baker, A.J., 2001. Complete mitochondrial DNA genome sequences ofextinct birds: ratite phylogenetics and the vicariance biogeography hypothesis.Proc. R. Soc. Lond. Ser. B 268, 939–945.

Hallam, A., 1994. An Outline of Phanerozoic Biogeography. Oxford Univ. Press,Oxford.

Hantke, R., Scheidegger, A.E., 1998. Morphotectonics of the Mascarene Islands. Ann.Geofis. 41, 165–181.

Heads, M., 2011. Old taxa on young islands: a critique of the use of island age to dateisland-endemic clades and calibrate phylogenies. Syst. Biol. 60, 204–218.

Hickey, J.L., 1977. Stratigraphy and Palaeobotany of the Golden Valley Formation(Early Tertiary) of Western North Dakota. Geological Society of America,Memoir 150, Boulder, Colorado, USA.

Hodgkison, R., Balding, S.T., Zubaid, A., Kunz, T.H., 2003. Fruit bats (Chiroptera:Pteropodidae) as seed dispersers and pollinators in a lowland Malaysian rainforest. Biotropica 35, 491–502.

Huang, Y., Shi, S., 2002. Phylogenetics of Lythraceae sensu lato: a preliminaryanalysis based on chloroplast rbcL gene, psaA-ycf3 spacer, and nuclear rDNAInternal Transcribed Spacer (ITS) sequences. Int. J. Plant Sci. 163, 215–225.

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

Jaramillo, C., Ochoa, D., Contreras, L., Pagani, M., Carvajal-Ortiz, H., Pratt, L.M.,Krishnan, S., Cardona, A., Romero, M., Quiroz, L., Rodriguez, G., Rueda, M.J., de laParra, F., Morón, S., Green, W., Bayona, G., Montes, C., Quintero, O., Ramirez, R.,Mora, G., Schouten, S., Bermudez, H., Navarrete, R., Parra, F., Alvarán, M., Osorno,J., Crowley, J.L., Valencia, V., Vervoort, J., 2010. Effects of rapid global warming atthe Paleocene–Eocene boundary on neotropical vegetation. Science 330, 957–961.

Jian, S., Soltis, P.S., Gitzendanner, M.A., Moore, M.J., Li, R., Hendry, T.A., Qiu, Y.-L.,Dhingra, A., Bell, C.D., Soltis, D.E., 2008. Resolving an ancient, rapid radiation inSaxifragales. Syst. Bot. 57, 38–57.

Johnson, L.A.S., Briggs, B.G., 1984. Myrtales and Myrtaceae – a phylogenetic analysis.Ann. Mo. Bot. Gard. 71, 700–756.

Jokat, W., Boebel, T., König, M., Meyer, U., 2003. Timing and geometry of earlyGondwana breakup. J. Geophys. Res. 108, 2428.

Jordaan, M., Van Wyk, A.E., Maurin, O., 2011. Generic status of Quisqualis(Combretaceae), with notes on the taxonomy and distribution of Q. parviflora.Bothalia 41, 161–169.

Katoh, K., Toh, H., 2008. Recent developments in the MAFFT multiple sequencealignment program. Br. Bioinform. 9, 286–298.

Kearse, M., Moir, R., Wilson, A., Stones-Havas, S., Cheung, M., Sturrock, S., Buxton, S.,Cooper, A., Markowitz, S., Duran, C., Thierer, T., Ashton, B., Mentjies, P.,Drummond, A., 2012. Geneious Basic: an integrated and extendable desktopsoftware platform for the organization and analysis of sequence data.Bioinformatics 28, 1647–1649.

Kissling, W.D., Eiserhardt, W.L., Baker, W.J., Borchsenius, F., Couvreur, T.L.P., Balslev,H., Svenning, J.-C., 2012. Cenozoic imprints on the phylogenetic structureof palm species assemblages worldwide. Proc. Natl. Acad. Sci. 109,7379–7384.

Knapp, M., Stöckler, K., Havell, D., Delsuc, F., Sebastiani, F., Lockhart, P.J., 2005.Relaxed molecular clock provides evidence for long-distance dispersal ofNothofagus (southern beech). PLoS Biol. 3, e14.

Kooyman, R.M., Wilf, P., Barreda, V.D., Carpenter, R.J., Jordan, G.J., Sniderman, J.M.K.,Allen, A., Brodribb, T.J., Crayn, D., Feild, T.S., Laffan, S.W., Lusk, C.H., Rossetto, M.,Weston, P.H., 2014. Paleo-Antarctic rainforest into the modern Old Worldtropics: the rich past and threatened future of the ‘‘southern wet forestsurvivors”. Am. J. Bot. 101, 2121–2135.

Korall, P., Pryer, K.M., 2013. Global biogeography of scaly tree ferns (Cyatheaceae):evidence for Gondwanan vicariance and limited transoceanic dispersal. J.Biogeogr. 41, 402–413.

Kriebel, R., Berger, B., Spalink, D., Reginato, M., Sytsma, K.J., 2015. The evolution ofshape and size in the pollen of Myrtales. In: Abstract of Talk Given in July 2015at Botany 2015, Edmonton, Alberta – Symposium on Morphospaces,Morphometrics, and Phylogenetics.

Ladiges, P., Udovicic, F., Nelson, G., 2003. Australian biogeographical connectionsand the phylogeny of large genera in the plant family Myrtaceae. J. Biogeogr. 30,989–998.

Larget, B.R., Kotha, S.K., Dewey, C.N., Ane, C., 2010. BUCKy: gene tree/species treereconciliation with Bayesian Concordance Analysis. Bioinformatics 26, 2910–2911.

Lavin, M., Luckow, M., 1993. Origins and relationships of tropical North America inthe context of the Boreotropics Hypothesis. Am. J. Bot. 80, 1–14.

Le Pechon, T., Dai, Q., Zhang, L.B., Gao, X.-F., Sauquet, H., 2015. Diversification ofDombeyoideae (Malvaceae) in the Mascarenes: old taxa on young islands? Int. J.Plant Sci. 176, 211–221.

Lee, D.E., Conran, J.G., Bannister, J.M., Kaulfuss, U., Mildenhall, D.C., 2013. A fossilFuchsia (Onagraceae) flower and an anther mass with in situ pollen from theearly Miocene of New Zealand. Am. J. Bot. 100, 2052–2065.

Levin, R.A., Wagner, W.L., Hoch, P.C., Nepokroeff, M., Pires, J.C., Zimmer, E.A., Sytsma,K.J., 2003. Family-level relationships of Onagraceae based on chloroplast rbcLand ndhF data. Am. J. Bot. 90, 107–115.

Levin, R.A., Wagner, W.L., Hahn, W.J., Hoch, P.C., Baum, D.A., Katinas, L., Zimmer, E.A., Sytsma, K.J., 2004. Paraphyly in tribe Onagreae: insights into phylogeneticrelationships of Onagraceae based on nuclear and chloroplast sequence DNA.Syst. Bot. 29, 147–164.

Linder, H.P., Rabosky, D.L., Antonelli, A., Wüest, R.O., Ohlemüller, R., 2014.Disentangling the influence of climatic and geological changes on speciesradiations. J. Biogeogr. 41, 1313–1325.

Litt, A., 1999. Floral Morphology and Phylogeny of Vochysiaceae. Ph.D. Dissertation,City University of New York.

Litt, A., 2003. Floral development and morphology of Vochysiaceae. II. The positionof the single fertile stamen. Am. J. Bot. 90, 1548–1559.

Litt, A., Stevenson, D., 2003. Floral development and morphology of Vochysiaceae. I.The structure of the gynoecium. Am. J. Bot. 90, 1533–1547.

Lucas, E.J., Belsham, S.R., Lughadha, E.M.N., Orlovich, D.A., Sakuragui, C.M., Chase, M.W., Wilson, P.G., 2005. Phylogenetic patterns in the fleshy-fruited Myrtaceae –preliminary molecular evidence. Plant Syst. Evol. 251, 35–51.

Luo, A., Qiao, H., Zhang, Y., Shi, W., Ho, S.Y.W., Xu, W., Zhang, A., Zhu, C., 2010.Performance of criteria for selecting evolutionary models in phylogenetics: acomprehensive study based on simulated datasets. BMC Evol. Biol. 10, 242.

Magallón, S., 2010. Using fossils to break long branches in molecular dating: acomparison of relaxed clocks applied to the origin of angiosperms. Syst. Biol. 59,384–399.

Magallón, S., Castillo, A., 2009. Angiosperm diversification through time. Am. J. Bot.96, 349–365.

Magallón, S., Sanderson, M.J., 2001. Absolute diversification rates in angiospermclades. Evolution 55, 1762–1780.

Magallón, S., Crane, P.R., Herendeen, P.S., 1999. Phylogenetic pattern, diversity, anddiversification of eudicots. Ann. Mo. Bot. Gard. 86, 297–372.

Magallón, S., Gómez-Acevedo, S., Sánchez-Reyes, L.L., Hernández-Hernández, T.,2015. A metacalibrated time-tree documents the early rise of flowering plantphylogenetic diversity. New Phytol. http://dx.doi.org/10.1111/nph.13264.

Manchester, S.R., O’Leary, E.L., 2010. Phylogenetic distribution and identification offin-winged fruits. Bot. Rev. 76, 1–82.

Mao, K., Milne, R.I., Zhang, L., Peng, Y., Liu, J., Thomas, P., Mill, R.R., Renner, S.S., 2012.Distribution of living Cupressaceae reflects the breakup of Pangea. Proc. Natl.Acad. Sci. 109, 7793–7798.

Martins, A.C., Melo, G., Renner, S.S., 2014. The honeybee/bumblebee lineagediverged from New World oil-collecting bees: implications for the origin ofpollen baskets. Mol. Phylogenet. Evol. 80, 88–94.

B.A. Berger et al. /Molecular Phylogenetics and Evolution 95 (2016) 116–136 135

Matzke, N.J., 2013. Probabilistic historical biogeography: new models for founder-event speciation, imperfect detection, and fossils allow improved accuracy andmodel-testing. Front. Biogeogr. 4, 242–247.

Matzke, N.J., 2014. Model selection in historical biogeography reveals that founder-event speciation is a crucial process in island clades. Syst. Biol. http://dx.doi.org/10.1093/sysbio/syu056.

Maurin, O., Chase, M.W., Jordaan, M., van der Bank, M., 2010. Phylogeneticrelationships of Combretaceae inferred from nuclear and plastid DNAsequence data: implications for generic classification. Bot. J. Linn. Soc. 162,453–476.

McElwain, J., Punyasena, S., 2007. Mass extinction events and the plant fossil record.TREE 22, 548–557.

McGuire, J.A., Witt, C.C., Remsen Jr., J.V., Corl, A., Rabosky, D.L., Altshuler, D.L.,Dudley, R., 2014. Molecular phylogenetics and the diversification ofhummingbirds. Curr. Biol. 24, 910–916.

McLoughlin, S., 2001. The breakup history of Gondwana and its impact on pre-Cenozoic, floristic provincialism. Aust. J. Bot. 49, 271–300.

Miller, M.A., Holder, M.T., Vos, R., Midford, P.E., Liebowitz, T., Chan, L., Hoover, P.,Warnow, T., 2010. The CIPRES Portals. <http://www.phylo.org>.

Moore, M.J., Soltis, P.S., Bell, C.D., Burleigh, J.G., Soltis, D.E., 2010. Phylogeneticanalysis of 83 plastid genes further resolves the early diversification of eudicots.Proc. Natl. Acad. Sci. 107, 4623–4628.

Morley, R., Dick, C., 2003. Missing fossils, molecular clocks, and the origin of theMelastomataceae. Am. J. Bot. 90, 1638–1644.

Morris, J., 2007. A Molecular Phylogeny of the Lythraceae and Inference of theEvolution of Heterostyly. Ph.D. Dissertation, Kent State University, Ohio.

Muller, J., 1975. Note on the pollen morphology of Crypteroniaceae s.l. Blumea 22,275–294.

Muller, J., 1981. Fossil pollen records of extant angiosperms. Bot. Rev. 47, 1–142.Müller, S., Salomo, K., Salazar, J., Naumann, J., Jaramillo, M.A., Neinhuis, C., Feild, T.S.,

Wanke, S., 2015. Intercontinental long-distance dispersal of Canellaceae fromthe New to the Old World revealed by a nuclear single copy gene andchloroplast loci. Mol. Phylogenet. Evol. 84, 205–219.

Murienne, J., Daniels, S.R., Buckley, T.R., Mayer, G., Giribet, G., 2013. A living fossiltale of Pangaean biogeography. Proc. R. Soc. Lond. Ser. B 281, 20132648.

Murphy, W.J., Eizirik, E., O’Brien, S.J., Madsen, O., Scally, M., Douady, C.J., Teeling, E.,Ryder, O.A., Stanhope, M.J., de Jong, W.W., Springer, M.S., 2001. Resolution of theearly placental mammal radiation using Bayesian phylogenetics. Science 294,2348–2351.

Nowak, M.D., Haller, B.C., Yoder, A.D., 2014. The founding of Mauritian endemiccoffee trees by a synchronous long-distance dispersal event. J. Evol. Biol. 27,1229–1239.

O’Leary, M.A., Bloch, J.I., Flynn, J.J., Gaudin, T.J., Giallombardo, A., Giannini, N.P.,Goldberg, S.L., Kraatz, B.P., Luo, Z.-X., Meng, J., Ni, X., Novacek, M.J., Perini, F.A.,Randall, Z.S., Rougier, G.W., Sargis, E.J., Silcox, M.T., Simmons, N.B., Spaulding,M., Velazco, P.M., Weksler, M., Wible, J.R., Cirranello, A.L., 2013. The placentalmammal ancestor and the post-K–Pg radiation of placentals. Science 339,6621–6667.

Onstein, R.E., Carter, R.J., Xing, Y., Linder, H.P., 2014. Diversification rate shifts in theCape Floristic Region: the right traits in the right place at the right time.Perspect. Plant Ecol. Evol. Syst. 16, 331–340.

Palazzesi, L., Gottschling, M., Barreda, V., Weigend, M., 2012. First Miocene fossils ofVivianiaceae shed new light on phylogeny, divergence times, and historicalbiogeography of Geraniales. Biol. J. Linn. Soc. 107, 67–85.

Panero, J.L., Freire, S.E., Ariza Espinar, L., Crozier, B.S., Barboza, G.E., Cantero, J.J.,2014. Resolution of deep nodes yields an improved backbone phylogeny and anew basal lineage to study early evolution of Asteraceae. Mol. Phylogenet. Evol.80, 43–53.

Pitman III, W.C., Cande, S., LaBrecque, J., Pindell, J., 1993. Fragmentation ofGondwana: the separation of Africa from South America. In: Goldblatt, P.(Ed.), Biological Relationships between Africa and South America. YaleUniversity Press, New Haven, Connecticut, pp. 15–34.

Plummer, M., Best, N., Cowles, K., Vines, K., 2006. CODA: convergence diagnosis andoutput analysis for MCMC. R News 6, 7–11.

Proença, C.E.B., 1992. Buzz-pollination – older and more widespread than we think?J. Trop. Ecol. 8, 115–120.

Quirk, J.T., 1980. Wood anatomy of the Vochysiaceae. IAWA J. 1, 172–179.R Development Core Team, 2014. R: A Language and Environment for Statistical

Computing. R Foundation for Statistical Computing, Vienna, Austria. <http://www.R-project.org/>.

Rabosky, D.L., 2014. Automatic detection of key innovations, rate shifts, anddiversity-dependence on phylogenetic trees. PLoS ONE 9, e89543.

Rabosky, D.L., Grundler, M., Anderson, C., Title, P., Shi, J.J., Brown, J.W., Huang, H.,Larson, J.G., 2014. BAMMtools: an R package for the analysis of evolutionarydynamics on phylogenetic trees. Methods Ecol. Evol. 5, 701–707.

Rambaut, A., Suchard, M.A., Xie, D., Drummond, A.J., 2014. Tracer v1.6. <http://beast.bio.ed.ac.uk/Tracer>.

Raven, P.H., 1988. Onagraceae as a model of plant evolution. In: Gottlieb, L.D., Jain, S.K. (Eds.), Plant Evolutionary Biology: A Symposium Honoring G. LedyardStebbins. Chapman and Hall, London, UK, pp. 85–107.

Raven, P.H., Axelrod, D.I., 1974. Angiosperm biogeography and past continentalmovements. Ann. Mo. Bot. Gard. 61, 539–673.

Ree, R.H., Smith, S.A., 2008. Maximum likelihood inference of geographic rangeevolution by dispersal, local extinction, and cladogenesis. Syst. Biol. 57, 4–14.

Ree, R.H., Moore, B.R., Webb, C.O., Donoghue, M.J., 2005. A likelihood framework forinferring the evolution of geographic range on phylogenetic trees. Evolution 59,2299–2311.

Renner, S.S., 1989. A survey of reproductive biology in NeotropicalMelastomataceae and Memycelaceae. Ann. Mo. Bot. Gard. 76, 496–518.

Renner, S.S., 1993. Phylogeny and classification of the Melastomataceae andMemecylaceae. Nord. J. Bot. 13, 519–540.

Renner, S.S., 2004. Bayesian analysis of combined chloroplast loci, using multiplecalibrations, supports the recent arrival of Melastomataceae in Africa andMadagascar. Am. J. Bot. 91, 1427–1435.

Renner, S.S., Clausing, G., Meyer, K., 2001. Historical biogeography ofMelastomataceae: the roles of Tertiary migration and long-distance dispersal.Am. J. Bot. 88, 1290–1300.

Renner, S.S., Strijk, J.S., Strasberg, D., Thebaud, C., 2010. Biogeography of theMonimiaceae (Laurales): a role for East Gondwana and long-distance dispersal,but not West Gondwana. J. Biogeogr. 37, 1227–1238.

Revell, L.J., 2012. Phytools: an R package for phylogenetic comparative biology (andother things). Meth. Ecol. Evol. 3, 217–223.

Ronquist, F., Huelsenbeck, J.P., 2003. MRBAYES 3: Bayesian phylogenetic inferenceunder mixed models. Bioinformatics 19, 1572–1574.

Rosen, D., 1978. Vicariant patterns and historical explanations in biogeography.Syst. Zool. 27, 159–188.

Rowe, D.L., Dunn, K.A., Adkins, R.M., Honeycutt, R.L., 2010. Molecular clocks keepdispersal hypotheses afloat: evidence for trans-Atlantic rafting by rodents. J.Biogeogr. 37, 305–324.

Ruhfel, B.R., Gitzendanner, M.A., Soltis, P.S., Soltis, D.E., Burleigh, J.G., 2014. Fromalgae to angiosperms – inferring the phylogeny of green plants (Viridiplantae)from 360 plastid genomes. BMC Evol. Biol. 14, 23.

Rutschmann, F., Eriksson, T., Conti, E., 2004. Did Crypteroniaceae really disperseout-of-India? Molecular dating evidence from rbcL, ndhF, and rpl16 intronsequences. Int. J. Plant Sci. 165, 569–583.

Rutschmann, F., Eriksson, T., Salim, K.A., Conti, E., 2007. Assessing calibrationuncertainty in molecular dating: the assignment of fossils to alternativecalibration points. Syst. Biol. 56, 591–608.

Sanderson, M.J., 2002. Estimating absolute rates of molecular evolution anddivergence times: a penalized likelihood approach. Mol. Biol. Evol. 19, 101–109.

Sanderson, M.J., McMahon, M.M., Steel, M., 2010. Phylogenomics with incompletetaxon coverage: the limits to inference. BMC Evol. Biol. 10, 155.

Sanmartín, I., Ronquist, F., 2004. Southern Hemisphere biogeography inferred byevent-based models: plant versus animal patterns. Syst. Biol. 53, 216–243.

Schönenberger, J., Conti, E., 2003. Molecular phylogeny and floral evolution ofPenaeaceae, Oliniaceae, Rhynchocalycaceae, and Alzateaceae (Myrtales). Am. J.Bot. 90, 293–309.

Schönenberger, J., Anderberg, A.A., Sytsma, K.J., 2005. Molecular phylogenetics andpatterns of floral evolution in the Ericales. Int. J. Plant Sci. 166, 265–288.

Scotese, C.R., 1998. Quicktime Computer Animations, PALEOMAP Project,Department of Geology, University of Texas at Arlington, Arlington, Texas.

Scotese, C.R., Gahagan, L.M., Larson, R.L., 1988. Plate tectonic reconstructions of theCretaceous and Cenozoic ocean basins. Tectonophysics 155, 27–48.

Sequeira, A.S., Farrell, B.D., 2001. Evolutionary origins of Gondwanan interactions:how old are Araucaria beetle herbivores? Biol. J. Linn. Soc. 74, 459–474.

Sessa, E.B., Zimmer, E.A., Givnish, T.J., 2012. Phylogeny, divergence times, andhistorical biogeography of New World Dryopteris (Dryopteridaceae). Am. J. Bot.99, 730–750.

Smith, A.C., 1973. Angiosperm evolution and the relationship of the floras of Africaand America. In: Meggers, B.J., Ayensu, E.S., Duckworth, W.D. (Eds.), TropicalForest Ecosystems in Africa and South America: A Comparative Review.Smithsonian Institution, Washington, DC, pp. 49–61.

Smith, S.A., Beaulieu, J.M., Donoghue, M.J., 2010. An uncorrelated relaxed-clockanalysis suggests an earlier origin for flowering plants. Proc. Natl. Acad. Sci. 107,5897–5902.

Smith, S.A., Beaulieu, J.M., Stamatakis, A., Donoghue, M.J., 2011. Understandingangiosperm diversification using small and large phylogenetic trees. Am. J. Bot.98, 404–414.

Soltis, D.E., Smith, S.A., Cellinese, N., Wurdack, K.J., Tank, D.C., Brockington, S.F.,Refulio-Rodriguez, N.F., Walker, J.B., Moore, M.J., Carlsward, B.S., Bell, C.D.,Latvis, M., Crawley, S., Black, C., Diouf, D., Xi, Z., Gitzendanner, M.A., Sytsma, K.J.,Qui, Y.-L., Hilu, K.W., Davis, C.C., Sanderson, M.J., Olmstead, R.G., Judd, W.S.,Donoghue, M.J., Soltis, P.S., 2011. Angiosperm phylogeny: 17 genes, 640 taxa.Am. J. Bot. 98, 704–730.

Stace, C.A., 2007. Combretaceae. In: Kubitzki, K. (Ed.), The Families and Genera ofVascular Plants, vol. 9. Springer, Berlin, pp. 67–82.

Stevens, P.F., 2001 (onwards). Angiosperm Phylogeny Website. Version 9, June 2008(and More or Less Continuously Updated Since). <http://www.mobot.org/MOBOT/research/APweb/>.

Sulman, J., Drew, B.T., Drummond, C., Hayasaka, E., Sytsma, K.J., 2013. Systematics,biogeography, and character evolution of Sparganium (Typhaceae):diversification of a widespread, aquatic lineage. Am. J. Bot. 100, 2023–2039.

Swenson, U., Hill, R.S., McLoughlin, S., 2001. Biogeography of Nothofagus supportsthe sequence of Gondwana break-up. Taxon 50, 1025–1041.

Sytsma, K.J., Litt, A., Zjhra, M.L., Pires, J.C., Nepokroeff, M., Conti, E., Walker, J.,Wilson, P.G., 2004. Clades, clocks, and continents: historical andbiogeographical analysis of Myrtaceae, Vochysiaceae, and relatives in theSouthern Hemisphere. Int. J. Plant Sci. 165, S85–S105.

136 B.A. Berger et al. /Molecular Phylogenetics and Evolution 95 (2016) 116–136

Sytsma, K.J., Spalink, D., Berger, B., 2014. Calibrated chronograms, fossils, outgrouprelationships, and root priors: re-examining the historical biogeography ofGeraniales. Biol. J. Linn. Soc. 113, 29–49.

Takahashi, M., Crane, P.R., Ando, H., 1999. Esgueiria futabensis sp. nov., a newangiosperm flower from the Upper Cretaceous (lower Coniacian) ofnortheastern Honshu, Japan. Paleo. Res. 3, 81–87.

Tan, K.H., Zubaid, A., Kunz, T.H., 2000. Fruit dispersal by the lesser dog-faced fruitbat, Cynopterus brachyotis (Muller) (Chiroptera: Pteropodidae). Malayan Nat. J.56, 57–62.

Tan, F., Shi, S., Zhong, Y., Gong, X., Wang, Y., 2002. Phylogenetic relationships ofCombretoideae (Combretaceae) inferred from plastid, nuclear gene and spacersequences. J. Plant. Res. 115, 475–481.

Templeton, A.R., 1980. The theory of speciation via the founder principle. Genetics94, 1011–1038.

The Plant List, 2014. Version 1.1. Published on the Internet. <http://www.theplantlist.org/> (accessed 12.10.14).

Thomas, N., Bruhl, J.J., Ford, A., Weston, P.H., 2014. Molecular dating of Winteraceaereveals a complex biogeographical history involving both ancient Gondwananvicariance and long-distance dispersal. J. Biogeogr. 41, 894–904.

Thorne, R.F., 1972. Major disjunctions in the geographic ranges of seed plants.Quart. Rev. Biol. 47, 365–411.

Thorne, R.F., 1973. Floristic relationships between tropical Africa and tropicalAmerica. In: Meggers, B.J., Ayensu, E.S., Duckworth, W.D. (Eds.), Tropical ForestEcosystems in Africa and South America: A Comparative Review. SmithsonianInstitution, Washington, DC, pp. 27–47.

Thornhill, A.H., Popple, L.W., Carter, R.J., Ho, S.Y.W., Crisp, M.D., 2012. Are pollenfossils useful for calibrating relaxed molecular-clock dating of phylogenies: acomparative study using Myrtaceae. Mol. Phylogenet. Evol. 63, 15–27.

Thornhill, A.H., Ho, S.Y.W., Külheim, C., Crisp, M.D., 2015. Interpreting the moderndistribution of Myrtaceae using a dated molecular phylogeny. Mol. Phylogenet.Evol. 93, 29–43.

van Vliet, G.J.C.M., Baas, P., 1984. Wood anatomy and classification of the Myrtales.Ann. Mo. Bot. Gard. 71, 783–800.

Vinnersten, A., Bremer, K., 2001. Age and biogeography of major clades in Liliales.Am. J. Bot. 88, 1695–1703.

Walker, J., Geissman, J., 2009. Geologic time scale. GSA Today 19, 60–61.Walker, J.B., Drew, B.T., Sytsma, K.J., 2015. Unravelling species relationships and

diversification within the iconic California Floristic Province sages (Salviasubgenus Audibertia, Lamiaceae). Syst. Bot. 40, 826–844.

Wang, H., Moore, M., Soltis, P., Bell, C., Brockington, S., Alexandre, R., Davis, C., Latvis,M., Manchester, S., Soltis, D., 2009. Rosid radiation and the rapid rise ofangiosperm-dominated forests. Proc. Natl. Acad. Sci. 106, 3853–3858.

Wehr, W.C., Hopkins, D.Q., 1994. The Eocene orchards and gardens of Republic.Wash. Geol. 22, 27–34.

White, L.T., Gibson, G.M., Lister, G.S., 2013. A reassessment of paleogeographicreconstructions of eastern Gondwana: bringing geology back into the equation.Gondwana Res. 24, 984–998.

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

Wilf, P., Escapa, I.H., 2014. Green Web or megabiased clock? Plant fossils fromGondwanan Patagonia speak on evolutionary radiations. New Phytol. http://dx.doi.org/10.1111/nph.13114.

Wilson, P.G., O’Brien, M.M., Gadek, P.A., Quinn, C.J., 2001. Myrtaceae revisited: areassessment of infrafamilial groups. Am. J. Bot. 88, 2013–2025.

Wilson, P.G., O’Brien, M.M., Heslewood, M.M., Quinn, C.J., 2005. Relationships withinMyrtaceae sensu lato based on a matK phylogeny. Plant Syst. Evol. 251, 3–19.

Wolfe, J.A., 1975. Some aspects of plant geography of the Northern Hemisphereduring the late Cretaceous and Tertiary. Ann. Mo. Bot. Gard. 62, 264–279.

Xie, L., Wagner, W.L., Ree, R.H., Berry, P.E., Wen, J., 2009. Molecular phylogeny,divergence time estimates, and historical biogeography of Circaea (Onagraceae)in the Northern Hemisphere. Mol. Phylogenet. Evol. 53, 995–1009.

Zachos, J., Pagani, M., Sloan, L., Thomas, E., Billups, K., 2001. Trends, rhythms, andaberrations in global climate 65 Ma to present. Science 292, 686–693.

Zeng, L., Zhang, Q., Sun, R., Kong, H., Zhang, N., Ma, H., 2014. Resolution of deepangiosperm phylogeny using conserved nuclear genes and estimates of earlydivergence times. Nat. Commun. 5, 1–12.

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

Zhu, X.-Y., Chase, M.W., Qui, Y.-L., Kong, H.-Z., Dilcher, D.L., Li, J.-H., Chen, Z.-D.,2007. Mitochondrial matR sequences help to resolve deep phylogeneticrelationships in rosids. BMC Evol. Biol. 7, 217.

Zwickl, D., 2006. Genetic Algorithm Approaches for the Phylogenetic Analysis ofLarge Biological Sequence Datasets under the Maximum Likelihood Criterion.Ph.D. Thesis, University of Texas, Austin, Texas.