Angiosperm phylogeny inferred from sequences of four mitochondrial genes

Authors


Author for correspondence. E-mail: ylqiu@umich.edu; Tel.: 1-734-764-8279; Fax: 1-734-763-0544.

Abstract

Abstract  An angiosperm phylogeny was reconstructed in a maximum likelihood analysis of sequences of four mitochondrial genes, atp1, matR, nad5, and rps3, from 380 species that represent 376 genera and 296 families of seed plants. It is largely congruent with the phylogeny of angiosperms reconstructed from chloroplast genes atpB, matK, and rbcL, and nuclear 18S rDNA. The basalmost lineage consists of Amborella and Nymphaeales (including Hydatellaceae). Austrobaileyales follow this clade and are sister to the mesangiosperms, which include Chloranthaceae, Ceratophyllum, magnoliids, monocots, and eudicots. With the exception of Chloranthaceae being sister to Ceratophyllum, relationships among these five lineages are not well supported. In eudicots, Ranunculales, Sabiales, Proteales, Trochodendrales, Buxales, Gunnerales, Saxifragales, Vitales, Berberidopsidales, and Dilleniales form a basal grade of lines that diverged before the diversification of rosids and asterids. Within rosids, the COM (Celastrales–Oxalidales–Malpighiales) clade is sister to malvids (or rosid II), instead of to the nitrogen-fixing clade as found in all previous large-scale molecular analyses of angiosperms. Santalales and Caryophyllales are members of an expanded asterid clade. This study shows that the mitochondrial genes are informative markers for resolving relationships among genera, families, or higher rank taxa across angiosperms. The low substitution rates and low homoplasy levels of the mitochondrial genes relative to the chloroplast genes, as found in this study, make them particularly useful for reconstructing ancient phylogenetic relationships. A mitochondrial gene-based angiosperm phylogeny provides an independent and essential reference for comparison with hypotheses of angiosperm phylogeny based on chloroplast genes, nuclear genes, and non-molecular data to reconstruct the underlying organismal phylogeny.

Angiosperms are the main primary producers in most modern terrestrial ecosystems, and their evolution has had a major impact on the environment of the earth and the evolution of animals, fungi, and other plants (Friis et al., 1987; Dilcher, 2000; Algeo et al., 2001; Berner, 2001; Schneider et al., 2004; Moreau et al., 2006; Heinrichs et al., 2007; Newton et al., 2007; Hibbett & Matheny, 2009). Knowledge of their phylogeny is essential in the study of structure, function, and evolution of this important group of plants, and hence, has always been an important goal of research in botany and evolutionary biology (Takhtajan, 1969; Cronquist, 1988). Over the last two decades, unprecedented progress has been made in reconstructing angiosperm phylogeny, thanks to a large number of phylogenetic studies analyzing molecular and non-molecular data. Several large-scale analyses of chloroplast (atpB, matK, and rbcL) and nuclear (18S rDNA) gene sequences from all major angiosperm lineages have played an especially significant role in establishing the main framework of angiosperm phylogeny (Chase et al., 1993; Soltis et al., 1997, 2000; Savolainen et al., 2000a; Hilu et al., 2003). These analyses and many others focusing on specific groups of angiosperms together have clarified the following major issues. First, Amborella, Nymphaeales (including Hydatellaceae), and Austrobaileyales (the so-called ANITA grade) are established as the earliest divergent lineages of extant angiosperms (Mathews & Donoghue, 1999; Parkinson et al., 1999; Qiu et al., 1999, 2001, 2005; Soltis et al., 1999, 2000; Barkman et al., 2000; Graham & Olmstead, 2000; Savolainen et al., 2000a; Zanis et al., 2002; Borsch et al., 2003; Hilu et al., 2003; Stefanovic et al., 2004; Leebens-Mack et al., 2005; Jansen et al., 2007; Moore et al., 2007; Saarela et al., 2007; Goremykin et al., 2009). In retrospect, several pre-molecular systematic studies and pioneering molecular phylogenetic analyses had identified some members of ANITA as potentially the basalmost living angiosperms before the 1999–2000 wave of discoveries (Upchurch, 1984; Endress, 1986; Donoghue & Doyle, 1989; Martin & Dowd, 1991; Hamby & Zimmer, 1992; Qiu et al., 1993; Soltis et al., 1997). Second, eudicots are recognized as a monophyletic group (Chase et al., 1993; Savolainen et al., 2000a; Soltis et al., 2000; Hilu et al., 2003). Again, comparative analyses of palynological data from extant and fossil plants had suggested such a hypothesis earlier (Walker & Doyle, 1975; Wolfe et al., 1975; Brenner, 1976; Doyle et al., 1977), and recognition of the deep division between angiosperms with monosulcate pollen and those with tricolpate pollen dated to an even earlier time (Wodehouse, 1935, 1936; Bailey & Nast, 1943; Hu, 1950). In fact, three phylogenetic analyses of mostly morphological data (Dahlgren & Bremer, 1985; Donoghue & Doyle, 1989; Loconte & Stevenson, 1991) recovered the monophyly of eudicots several years before the first large scale analysis of molecular data (Chase et al., 1993). Third, the general phylogenetic outlines of rosids and asterids are now well circumscribed (Chase et al., 1993; Soltis et al., 1997, 2000; Nandi et al., 1998; Savolainen et al., 2000a; Hilu et al., 2003). Fourth, a deep split within monocots is identified between Alismatales (including Araceae) and Petrosaviidae (sensu Cantino et al., 2007) (Chase et al., 1993, 2006; Savolainen et al., 2000a; Soltis et al., 2000; Hilu et al., 2003; Qiu & Palmer, 2004), with Acorus placed as the sister to all other monocots (Chase et al., 1993; Duvall et al., 1993). Finally, Caryophyllales are placed as a close relative of asterids (Soltis et al., 1997, 2000; Hilu et al., 2003; Burleigh et al., 2009).

Despite this tremendous progress, some important problems remain, particularly in regard to relationships among major lineages within some large groups such as rosids and asterids, as well as the composition and placement of several key lineages related to diversification of the following large clades: angiosperms overall; eudicots; rosids; and asterids. Further, as indicated above, only three chloroplast genes and one nuclear gene have been used in the large-scale analyses, that is, of hundreds of taxa. The plant cell contains a third DNA-containing organelle, the mitochondrion, and several mitochondrial genes have been used at different levels for plant phylogenetic studies (Malek et al., 1996; Beckert et al., 1999, 2001; Parkinson et al., 1999; Qiu et al., 1999, 2005, 2006a, 2006b, 2007; Vangerow et al., 1999; Barkman et al., 2000, 2007; Bowe et al., 2000; Chaw et al., 2000; Anderberg et al., 2002; Davis et al., 2004; Dombrovska & Qiu, 2004; Wikstrom & Pryer, 2005; Forrest et al., 2006; Petersen et al., 2006; Zhu et al., 2007; Jian et al., 2008; Ran et al., 2009; Wurdack & Davis, 2009). Hence, it is important to explore the potential of mitochondrial genes for a full-scale angiosperm phylogeny reconstruction, so that an accurate organismal phylogeny, with information from all three genomes and non-molecular data, can be reconstructed (Doyle, 1992; Qiu & Palmer, 1999; Delsuc et al., 2005).

An angiosperm-wide phylogenetic study using mitochondrial gene sequences can serve several purposes. First, it can help to assess how accurately the phylogeny reconstructed with the chloroplast and nuclear genes represents the underlying organismal phylogeny. Theoretically, it can be argued that the true organismal phylogeny can never be known. In practice, this phylogeny can be inferred using multiple sources of information from organisms, such as gene sequences from all three genomes in the plant cell, morphology, and other non-molecular data. The more similar the phylogenetic hypotheses inferred from different data sources, the more likely that the true underlying plant phylogeny has been reconstructed. Even though the studies using the three chloroplast genes and one nuclear gene have established the main framework of angiosperm phylogeny, it is always desirable to assess the results with data from the mitochondrial genome, which represents a under-utilized and independent source of information. The recent short history of molecular systematics has provided some examples on gene- or genome-specific biases in phylogenetic reconstruction. The enigmatic genus Ceratophyllum has been shown to have different positions in analyses of different genes (Les et al., 1991; Chase et al., 1993; Soltis et al., 1997; Savolainen et al., 2000a; Hilu et al., 2003; Qiu et al., 2006a) or different portions of the same set of chloroplast genes (Goremykin et al., 2009). An analysis of animal mitochondrial genes reveals that there can be genome-wide noise in phylogenetic reconstruction, which is likely generated by over-representation of proteins with hydrophobic domains (Naylor & Brown, 1997). Chloroplast and mitochondrial genomes in plants both encode a disproportionately large set of trans-membrane domain-rich proteins (Jobson & Qiu, 2008), and are thus likely to be under selection of special evolutionary forces and contain certain genome-wide phylogenetic noise. Second, it is possible that mitochondrial genes may offer some insights for problems that have not been solved with chloroplast and nuclear genes. The generally lower point mutation rates of mitochondrial genes compared to chloroplast genes (Wolfe et al., 1987; Palmer & Herbon, 1988) should make them more suitable for unraveling more ancient diversification patterns. Finally, two interesting molecular evolutionary phenomena, namely, horizontal transfer (Bergthorsson et al., 2003; Won & Renner, 2003; Davis & Wurdack, 2004; Mower et al., 2004) and dramatic evolutionary rate acceleration (Cho et al., 2004; Parkinson et al., 2005) have been reported for some mitochondrial genes and lineages over the last few years. These phenomena can potentially affect the usefulness of mitochondrial genes in phylogenetic studies, and a broad survey like this one is likely to provide a realistic estimate as to how widespread these phenomena are across angiosperms.

1 Material and methods

A total of 380 species level operational taxonomic units (OTUs) were included in this study, which represented 376 genera and 296 families of all APG III (Angiosperm Phylogeny Group, 2009) orders and non-ordinal families except Petrosaviales, Picramniales, and Dasypogonaceae (Table 1). Eight diverse gymnosperms were included as the outgroup. This taxon sampling scheme was designed to reconstruct an overall angiosperm phylogeny, to resolve relationships among major clades of angiosperms, and to identify the composition and placement of some key clades involved in the origins of angiosperms, eudicots, rosids, and asterids.

Table 1.  Taxa and sequences used in this study
SpeciesFamilyatp1matRnad5rps3Voucher/DNA number
  1. Vouchers with numbers Qiu 1–Qiu 93999 are deposited in NCU, Qiu 94001–Qiu 97999 in IND, Qiu 98001–Qiu 99999 in Z, and Qiu 00001–Qiu 10999 in MICH. Numbers in parentheses are DNA numbers (no voucher or a voucher by someone without a number). Vouchers by collectors other than Y.-L. Qiu are indicated with the herbaria where they have been deposited. All vouchers by A.A. Reznicek are deposited in MICH. All sequences with GenBank accession numbers GUxxxxxx and HM357127 were generated in this study. All others were retrieved from GenBank. †Taxa removed from the 380 taxon analysis because of highly divergent sequences or lacking data for two or three genes. ND, no data.

Acanthus mollis L.AcanthaceaeGU350950AY289667, A. ebracteatus VahlGU351329GU351549Qiu 05019
Acorus calamus L.AcoraceaeAF197621NDDQ406951NDNo new data
Acorus gramineus Soland.AcoraceaeAF197622NDDQ406994GU351550Qiu 97131
Actinidia arguta Miq.ActinidiaceaeGU350951AY163745, A. rubricaulis DunnGU351330GU351551Qiu 95102
Aextoxicon punctatum Ruiz & Pav.AextoxicaceaeGU350952GU351143GU351331GU351552Th Borsch 3459 (BONN)
Agave attenuata Salm-DyckAgavaceaeAY299703, A. ghiesbreghtii K. KochDQ401408DQ407009GU351553Qiu 96067
Ailanthus altissima SwingleSimaroubaceaeGU350953GU351144NDGU351554Qiu 96141
Albizia julibrissin Durazz.FabaceaeGU350954GU351145GU351332GU351555A.A. Reznicek 11735/(Qiu 05026)
Alisma plantago-aquatica L.AlismataceaeAF197717AF197815DQ406947GU351556Qiu 96177
Allium cepa L.AlliaceaeDQ401321DQ401400DQ407007GU351557Qiu 94060
Alluaudia ascendens DrakeDidiereaceaeGU350955AF520129, A. humbertii ChouxAF520129, A. humbertii ChouxNDQiu 97036
Alnus rugosa Spreng.BetulaceaeGU350956GU351146GU351333GU351558Qiu 05007
Alseuosmia macrophylla A. Cunn.AlseuosmiaceaeGU350957GU351147GU351334GU351559Morgan 2141 (WS)
Altingia excelsa NoronhaAltingiaceaeEF370686EF370708EF370726EF370747Qiu 93006
Amborella trichopoda Baill.AmborellaceaeDQ007412AF197813AY832180GU351560Qiu 97123
Ancistrocladus tectorius Merr.AncistrocladaceaeGU350958GU351148GU351335GU351561Y.P. Hong 99394 (PE)
Androsace samentosa Wall.PrimulaceaeGU350959GU351149GU351336GU351562A.A. Reznicek 11751/(Qiu 05029)
Anisophyllea sp.AnisophylleaceaeGU350960GU351150GU351337GU351563P. Boyce 758 (K)
Anisoptera marginata Korth.DipterocarpaceaeGU350961GU351151GU351338GU351564Chase 2486 (K)
Annona muricata L.AnnonaceaeAF197695AF197766DQ406917GU351565Qiu 90031
Antirrhinum majus L.PlantaginaceaeGU350962GU351152GU351339GU351566Qiu 05011
Aphanopetalum resinosum Endl.AphanopetalaceaeEF370687EF370709EF370727GU351567Bradford 845
Aphloia theiformis Benn.AphloiaceaeGU350963GU351153GU351340GU351568REU 10012 (REU)
Apium graveolens L.ApiaceaeGU350964GU351154GU351341GU351569Qiu 05032
Arabidopsis thaliana Schur.BrassicaceaeNC_001284Y08501NC_001284NC_001284No new data
Arbutus canariensis DuhamelEricaceaeGU350965GU351155GU351342GU351570Albach 241 (K)
Aristolochia macrophylla Lam.AristolochiaceaeAF197669AF197732GU351343GU351571Qiu 91019
Ascarina sp.ChloranthaceaeAF197667AF197755DQ406865GU351572Thien 500 (NO)
Asparagus officinalis L.AsparagaceaeAF197713AF197736DQ407000GU351573Qiu 94063
Atherosperma moschatum LabillAtherospermaceaeAF197683AF197799DQ406929GU351574Qiu 92007
Austrobaileya scandens C.T. WhiteAustrobaileyaceaeAF197664AF197742DQ406986GU351575Qiu 90030
Barbeuia madagascariensis Steud.BarbeuiaceaeGU350966GU351156GU351344GU351576J.L. Zarucetti 7407 (K)
Basella alba L.BasellaceaeGU350967GU351157GU351345GU351577Qiu 02055
Batis maritima L.BataceaeGU350968GU351158GU351346GU351578(Qiu 96206)
Begonia sp.BegoniaceaeGU350969GU351159GU351347GU351580Qiu 05009
Berberidopsis beckleri VeldkampBerberidopsidaceaeDQ401303DQ401394DQ406898GU351581Qiu 98040
Berzelia lanuginoseBruniaceaeGU350970GU351160GU351348GU351582Kirstenbosch 75–89
Bistorta sp.PolygonaceaeNDGU351161GU351349GU351583A.A. Reznicek 11752/(Qiu 06003)
Bixa orellana L.BixaceaeNDGU351162NDNDQiu 97032
Blandfordia grandiflora R. Br.BlandfordiaceaeAY299727DQ401412DQ406966GU351584Qiu 97016
Borago officinalis L.BoraginaceaeGU350971GU351163GU351350GU351585Chase 2746 (K)
Bougainvillea albaNyctaginaceaeAY818932, B. glabra ChoisyNDGU351351GU351586(Qiu M67)
Brasenia schreberi J. GmelinNymphaeaceaeAF197640AF197728DQ406956GU351587Qiu 91031
Brexia madagascariensisCelastraceaeGU350972GU351164GU351352GU351588Chase 17719 (K)
Bruguiera gymnorhiza Sav.RhizophoraceaeGU350973GU351165GU351353GU351589(Qiu 96188)
Brunellia sp.BrunelliaceaeGU350974DQ110330, B. acutangula Humb. & Bonpl.GU351354GU351590G.P. Lewis 3366 (K)
Bursera sp.BurseraceaeGU350975GU351166GU351355GU351591Qiu 94206
Buxus sempervirens L.BuxaceaeAF197636AF197786DQ406879GU351592Qiu 97057
Byblis liniflora Salisb.ByblidaceaeGU350976GU351167GU351356GU351593Qiu 95128–2
Cabomba sp.NymphaeaceaeAF197641AF197729DQ406957GU351594Qiu 97027
Calceolaria integrifolia Murr.CalceolariaceaeGU350977GU351168GU351357GU351595Chase 2850 (K)
Calycanthus floridus L.CalycanthaceaeAF197678AF197777DQ406922GU351596Qiu 94155
Calyptrotheca somalensis GilgPortulacaceaeGU350978GU351169GU351358GU351597(Qiu M257), no voucher
Camellia japonica L.TheaceaeAF420952, C. sinensis KuntzeAF421034, C. sinensis KuntzeDQ406870GU351598Qiu 90999
Campanula rotundidoliaCampanulaceaeAY741815, C. garganica Ten.GU351170GU351359GU351599A.A. Reznicek 11819/(Qiu 05033)
Cananga odorata Hook. F. & ThomsonAnnonaceaeAF197700AF197763GU351360GU351600Chase 219 (NCU)
Canella winteriana Gaertn.CanellaceaeAF197676AF197757DQ406920GU351601Qiu 90017
Cannabis sativa L.CannabaceaeGU350979GU351171GU351361GU351602Chase 2992 (K)
Capparis cynophallophora L.BrassicaceaeGU350980GU351172GU351362GU351603(Qiu 96210)
Carica papaya L.CaricaceaeGU350982GU351173GU351363GU351604Qiu 94050
Carludovica palmata Ruiz & PavonCyclanthaceaeAF197707AF197734DQ406948GU351605Qiu 97021
Catalpa fargeseii Bur.BignoniaceaeAY741840, C. bignonioides WalterGU351174GU351364GU351606Qiu 95099
Caulophyllum thalictroides RegelBerberidaceaeGU350983GU351175GU351365NDA.A. Reznicek 11733/(Qiu 05023)
Ceanothus sp.RhamnaceaeGU350984GU351176GU351366GU351607Qiu 96098
Celosia cristata L.AmaranthaceaeGU350985GU351177GU351367GU351608Qiu 94153
Celtis yunnanensis C.K. Schneid.CannabaceaeGU350986GU351178GU351368GU351609Qiu P90002
Cephalotus follicularis Labill.CephalotaceaeGU350987GU351179GU351369NDR.W. Jobson CU-1023
Ceratophyllum demersum L.CeratophyllaceaeAF197627AF197730DQ406988GU351610Qiu 95003
Ceratophyllum submersum L.CeratophyllaceaeAF197628NDDQ406989GU351611Qiu 98088
Cercidiphyllum japonicum Siebold & Zucc.CercidiphyllaceaeEF370688EF370710EF370728EF370748Qiu 93013
Chamaedorea tenella H. Wendl.ArecaceaeDQ401295DQ401392DQ407003GU351612Qiu 95075
Chloranthus multistachys PeiChloranthaceaeAF197665AF197753DQ406864GU351613K. Wurdack 92–0010
Choristylis rhamnoides Harv.IteaceaeGU350988GU351180GU351370GU351614Chase 9646 (K)
Chrysolepis sempervirens Hjelmq.FagaceaeGU350990GU351182GU351372GU351616Qiu P90007
Cinnamodendron ekmanii Sleum.CanellaceaeAF197677AF197758DQ406921GU351617T. Zanoni & F. Jimenez 47067
Citrus limon Burm.f.RutaceaeGU350991GU351183GU351373GU351618Qiu 94085
Clavija eggersiana MezTheophrastaceaeAF420918, C. domingensis Urb. & EkmanAF420995, C. domingensis Urb. & EkmanGU351374GU351619Chase 216
Claytonia virginica L.PortulacaceaeGU350992GU351184GU351375GU351620Qiu 06001
Clethra barbinervis Siebold & Zucc.ClethraceaeGU350993AF520204, C. alnifolia L.GU351376GU351621Qiu 95103
Clidemia petiolaris TrianaMelastomaceaeGU350994GU351185GU351377GU351622Chase 2534 (K)
Clusia rosea Jacq.ClusiaceaeGU350995GU351186GU351378GU351623Qiu 05042
Connarus championii ThwaitesConnaraceaeGU350996GU351187GU351379GU351624Chase 15937 (K)
Coriaria myrtifolia L.CoriariaceaeGU350997GU351188GU351380GU351625Chase 245 (NCU)
Cornus florida Hook.CornaceaeAF420915, C. suecica L.AF420990, C. suecica L.DQ407012GU351626Qiu 96142
Corokia cotoneaster RaoulArgophyllaceaeGU350998GU351189GU351381GU351627Chase 2752 (K)
Corylopsis glabrescens Franch. & Sav.HamamelidaceaeEF370689EF370711EF370729EF370749Qiu 94158
Couroupita guianensis Aubl.LecythidaceaeAY725907GU351190GU351382GU351628Qiu 97025
Croomia pauciflora Miq.StemonaceaeAF197708AF197735DQ406939GU351629Qiu 97096
Crossosoma californicum Nutt.CrossosomataceaeGU350999GU351191GU351383GU351630Beier (UPS)
Crypteronia paniculata BlumeCrypteroniaceaeGU351000GU351192GU351384GU351631Y.P. Hong 99203/(Qiu 05071) (PE)
Cryptocarya meissneriana FrodinLauraceaeAF197702AF197804DQ406932GU351632Qiu 98048
Cucurbita pepo Lour.CucurbitaceaeGU351001GU351193GU351385GU351633Qiu 05018
Cupaniopsis anacardioides Radlk.SapindaceaeGU351002GU351194GU351386GU351634Qiu 98053
Curtisia dentata C.A. Sm.CurtisiaceaeGU351003GU351195GU351387GU351635(Qiu M299), no voucher
Cycas revoluta Thunb.CycadaceaeAF197623AF197720AJ130743AY345867No new data
Cyrilla racemiflora L.CyrillaceaeAF420922AY725892GU351388GU351636Qiu 95109
Dampiera diversifolia de VrieseGoodeniaceaeNDGU351196GU351389NDQiu 97144
Daphnandra micrantha Benth.AtherospermaceaeAF197684AF197800DQ406977GU351637Qiu 97015
Daphniphyllum sp.DaphniphyllaceaeEF370691EF370712EF370730EF370750Qiu 94162
Datisca cannabina L.DatiscaceaeGU351004GU351197GU351390GU351638Qiu 97102
Decaisnea fargesii Franch.LardizabalaceaeGU351005GU351198GU351391GU351639Qiu 02094
Degeneria vitiensis I.W. Bailey & A.C. Sm.DegeneriaceaeAF293752AF197771DQ406991GU351640John J. Miller 1189–63
Desfontainia spinosa Ruiz & Pav.ColumelliaceaeGU351006GU351199GU351392GU351641Chase 6419 (K)
Dicentra sp.FumariaceaeAF197649AF197796DQ406890GU351642Qiu 95026
Didymeles perrieri OlivierBuxaceaeAF197637AF197811DQ406993GU351643O. Andrianantoanina 387 (MO)
Dillenia indica L.DilleniaceaeDQ401306AY163747DQ406882GU351644Qiu 95129
Dioncophyllum tholloni Baill.DioncophyllaceaeGU351007AF520129, A. humbertii ChouxGU351393GU351645A.F. Bradley et al. 1107 (MO)
Dioscorea sp.DioscoreaceaeAF197709AF197737DQ406959GU351646Qiu 94044
Diospyros virginiana L.EbenaceaeGU351008AF520202, D. mollifolia Rehder & WilsonGU351394GU351647Qiu 94106
Dipentodon sinicus DunnDipentodontaceaeGU351009AY121494GU351395NDForrest 26561 (K)
Dipsacus sp.DipsacaceaeAY741814, D. fullonum S.G. Gmel.AY453093, D. fullonum S.G. Gmel.GU351396GU351648Qiu 95111
Donatia fascicularis Forst.StylidiaceaeGU351010GU351200GU351397NDMorgan 2142 (WS)
Drimys winteri J.R. Forster & G. ForsterWinteraceaeAF197673AF197781DQ406919GU351649Qiu 90016
Drosera regia StephensDroseraceaeGU351011GU351201GU351398GU351650Steve Williams D18
Drosophyllum lusitanicum Link.DrosophyllaceaeGU351012GU351202GU351399GU351651Steve Williams D100
Drypetes perreticulata Gagnep.PutranjivaceaeGU351013GU351203GU351400NDY.P. Hong 99310/(Qiu 05058) (PE)
Ehretia anacua I.M. Johnst.BoraginaceaeGU351014GU351204NDGU351652(Qiu 96209)
Elaeagnus sp.ElaeagnaceaeGU351015NDGU351401GU351653Qiu 95028
Elaeocarpus obovatus G. DonElaeocarpaceaeGU351016GU351205GU351402GU351654Qiu 98054
Elatine hexandra DC.ElatinaceaeNDAY674507, E. triandra SchkuhrNDGU351655Qiu 99051
Eremosyne pectinata Endl.EscalloniaceaeGU351017GU351206GU351403NDAnnels & Hearn 4795 (PERTH)
Escallonia rubra spp. macranthaEscalloniaceaeGU351018GU351207GU351404GU351656Qiu 02081
Eschscholzia californica Cham.PapaveraceaeGU351019GU351208GU351405GU351657Qiu 05049
Eucommia ulmoides OliverEucommiaceaeDQ401311DQ401387DQ406872GU351658Qiu 91024
Euonymus sp.CelastraceaeGU351020GU351209GU351406GU351659Qiu 94190
Euphorbia milii var. splendens Desmoul.EuphorbiaceaeDQ401317AY674512, E. polychroma Kern.DQ406908GU351660Qiu 94056
Eupomatia bennettii F. Muell.EupomatiaceaeAF197692AF197772DQ406927GU351661Qiu 90022
Euptelea polyandra Sieb. & Zucc.EupteleaceaeAF197650AF197787DQ406873GU351662Qiu 95098
Exbucklandia longipetala H.T. ChangHamamelidaceaeEF370692EF370713EF370731EF370751Qiu 93004
Floerkea proserpinicoideseLimnanthaceaeGU351021GU351210GU351407GU351663A.A. Reznicek 11750/(Qiu 06004)
Frankenia pulverulenta L.FrankeniaceaeGU351022GU351211GU351408NDCollenette 6/93 (K)
Gaiadendron sp.LoranthaceaeDQ110147, G. punctatum G.DonGU351212GU351409GU351664N. Munoz et al. 102 (MO)
Galax urceolata BrummittDiapensiaceaeAF420929AF421007GU351410GU351665Qiu 02069
Galbulimima belgraveana SpragueHimantandraceaeAF197693AF197773GU351411GU351666Qiu 90034
Galium sp.RubiaceaeGU351023GU351213GU351412GU351667Qiu 95025
Garrya elliptica Lindl.GarryaceaeGU351024AY453095GU351413GU351668Chase 1098 (K)
Geissois biagiana F. Muell.CunoniaceaeGU351025GU351214GU351414GU351669Qiu 97011
Geissoloma marginata A. Juss.GeissolomataceaeGU351026NDGU351415NDSavolainen GMA1 (G)
Gelsemium sp.GelsemiaceaeAY741816, G. sempervirens J.St.-Hil.GU351215GU351416GU351671Qiu 95096
Gentiana macrophylla Pall.GentianaceaeGU351027GU351216NDGU351672Qiu 96090
Geranium sanguinieum L.GeraniaceaeNDAY121488, G. wilfordiiGU351417NDA.A. Reznicek 11731/(Qiu 05036)
Ginkgo biloba L.GinkgoaceaeAF197625AF197722AJ409109GU351673Qiu 94015
Glaucidium palmatum Siebold & Zucc.RanunculaceaeGU351028GU351217GU351418GU351674A.A. Reznicek 10719/(Qiu 05025)
Gnetum gnemon L.GnetaceaeAF197617AF197718AJ409110NDno new data
Gomortega keule Baill.GomortegaceaeNDNDGU351419GU351675M.F. Doyle III-6–1986/(Qiu 05048)
Gossypium arboreum Vell.MalvaceaeGU351029GU351218GU351420GU351676Qiu 05015
Griselinia littoralis RaoulGriseliniaceaeGU351030AY453096, G. racemosa Taub.GU351421GU351677Strybing Arboretum xy-2609
Guaiacum officinale L.ZygophyllaceaeDQ401291AY674517, G. sanctum L.DQ406954GU351678Qiu 97035
Gunnera monoica RaoulGunneraceaeDQ401302DQ401383DQ406897GU351679Qiu 98071
Gyrocarpus sp.HernandiaceaeAF197701AF197805DQ406931, G. americanus Jacq.GU351680Chase 317 (NCU)
Halophytum ameghinoi Speg.HalophytaceaeGU351031GU351219GU351422NDQiu (M244), Tortosa, Bartoli, Chubut, nv
Haloragis erecta Schindl.HaloragaceaeEF370693EF370714EF370732EF370752No new data
Hamamelis mollis Forb. & Hemsl.HamamelidaceaeDQ401289AY453082, H. vernalis Sarg.DQ407011EF370753Qiu 91035
Hedera helix L.AraliaceaeDQ401310DQ401390DQ406955GU351681Qiu 98085
Hedycarya arborea J. R. & G. Forst.MonimiaceaeAF197689AF197806DQ406909GU351682Qiu 90028
Hedyosmum arborescens Sw.ChloranthaceaeAF197668AF197756DQ406863GU351683Chase 338 (NCU)
Heisteria parvifolia Sm.OlacaceaeGU351032GU351220GU351423GU351684Qiu 99018
Helianthemum grandiflorum DC.CistaceaeGU351033GU351221GU351424NDA.A. Reznicek 11775/(Qiu 05028)
Helianthus annuus L.AsteraceaeX55963GU351222GU351425AF319170(Qiu MD34), no voucher
Helwingia japonica C. Morren & Decne.HelwingiaceaeGU351034GU351223GU351426GU351685Qiu 99031
Hernandia ovigera L.HernandiaceaeDQ007413DQ007424DQ406930GU351686Qiu 01007
Heuchera sp.SaxifragaceaeDQ401290DQ401398DQ406953EF370754, H. micrantha Lindl.Qiu 95076
Hibbertia cuneiformis GilgDilleniaceaeGU351035GU351224GU351427GU351687Qiu 97020
Hirtella jamaicensis Urb.ChrysobalanaceaeGU350989GU351181GU351371GU351615Qiu 01010
Houttuynia cordata Thunb.SaururaceaeAF197632AF197749DQ406980GU351688Qiu 92016
Hua gabonii De Wild.HuaceaeGU351036GU351225GU351428GU351689J. J. Wiernga 3177 (WAG)
Humiria balsamifera Aubl.HumiriaceaeGU351037GU351226GU351429GU351690W. R. Anderson 13654 (MICH)
Hydnocarpus anthelminthica Pierre & Gagnep.AchariaceaeGU351038GU351227GU351430GU351691Y.P. Hong H001/(Qiu 05070) (PE)
Hydrangea arborescens L.HydrangeaceaeGU351039AY453091, H. macrophylla Ser.GU351431GU351692Qiu 95021
Hydrastis canadensis Poir.RanunculaceaeGU351040GU351228GU351432GU351693Z.D. Chen 2002016/(Qiu 05066) (PE)
Hydrolea ovata Nutt. ex ChoisyHydroleaceaeGU351041GU351229GU351433GU351694Olmstead 89–009 (COLO)
Hydrophyllum virginianum L.BoraginaceaeGU351042GU351230GU351434GU351695A.A. Reznicek 7887/(Qiu 05031)
Hypecoum imberbe Sibth. & Sm.FumariaceaeGU351043GU351231GU351435GU351696Chase 528 (K)
Hypericum sp.HypericaceaeGU351044GU351232GU351436GU351697Qiu 95082
Icacina mannii Oliv.IcacinaceaeGU351045GU351233GU351437GU351698Chase 2244 (K)
Idiospermum australiense S.T. BlakeCalycanthaceaeAF197680AF197779DQ406974GU351699Qiu 91042
Idria columnaria KelloggFouquieriaceaeGU351046GU351234GU351438GU351700Qiu 95065
Ilex sp.AquifoliaceaeAY741812, I. verticillata A.GrayAY453090, I. aquifolium Lour.DQ406884GU351701Qiu 94038
Illicium floridanum EllisSchisandraceaeAF197663AF197740DQ406985GU351702Qiu 61
Impatiens pallida Nutt.BalsaminaceaeAF420933, I. parviflora DC.AF421011, I. parviflora DC.DQ406952GU351703Qiu 95124
Ipomoea batatas Poir.ConvolvulaceaeAY596672GU351235GU351439GU351704Qiu 96152
Iris sp.IridaceaeDQ401300DQ401386DQ407006GU351705Qiu 95091
Itea virginica L.IteaceaeEF370696EF370716EF370735EF370755No new data
Ixerba brexioides A. CunnStrasburgeriaceaeGU351047GU351236GU351440GU351706P.J. de Lange 5809 (AK285208)
Ixonanthes icosandra var. cuneataIxonanthaceaeGU351048GU351237GU351441GU351707Chase 1301 (K)
Juglans cinerea L.JuglandaceaeGU351049GU351238GU351442GU351708Qiu 96022
Kadsura japonica DunalSchisandraceaeAF197661AF197738DQ406971GU351709Qiu 94159
Kalanchoe pinnata Pers.CrassulaceaeEF370697EF370717EF370736EF370756Qiu 94118
Krameria lanceolata Torr.KrameriaceaeGU351050GU351239GU351443GU351710Simpson 88–05-1–1 (MICH)
Lactoris fernandeziana Phil.LactoridaceaeAF197710AF197812DQ406910NDChase 1014 (K)
Lamium sp.LamiaceaeDQ401312DQ401385DQ406871GU351711Qiu 95019
Lampranthus emarginatus N.E. Br.AizoaceaeNDGU351240GU351444NDQiu 94115
Lardizabala biternata Ruiz & PavonLardizabalaceaeAF197643Qiu97135DQ406867GU351712Qiu 97135
Laurus nobilis L.LauraceaeAF197682AF197798DQ406923GU351713Qiu 94209
Leea guineensis G. Don.VitaceaeDQ401304AY674530DQ406899GU351714Qiu 97034
Lilium sp.LiliaceaeAY394729, L. tigrinum Ker Gawl.DQ401403DQ407002GU351715Qiu 96072
Limeum africanum Moq.LimeaceaeGU351051GU351241GU351445GU351716Goldblatt et al. 11512
Limonium tartaricumPlumbaginaceaeGU351052GU351242GU351446GU351717Qiu 96151
Linnaea borealis L.LinnaeaceaeGU351053GU351243GU351447GU351718Qiu 05035
Liquidambar styraciflua L.HamamelidaceaeEF370698EF370718EF370737EF370757Qiu 95089
Liriodendron chinense Sarg.MagnoliaceaeAF197690AF197774DQ406926GU351719Qiu 28
Lomandra obliqua J.F. Macbr.LaxmanniaceaeDQ401296DQ401380DQ406942GU351720Qiu 98016
Lonicera sp.CaprifoliaceaeGU351054AY453088GU351448GU351721Qiu 05010
Luculia intermedia Hutch.RubiaceaeGU351055NDNDNDHowick, Lord, & McNamara HOMC1524 (K)
Lythrum salicaria L.LythraceaeNDGU351244GU351449NDA.A. Reznicek 11741/(Qiu 05040)
Maesa japonica Zoll.MaesaceaeAF420937, M. tenera MezGU351245GU351450GU351722A.M. Lu 2073/(Qiu 05060) (PE)
Magnolia tripetala L.MagnoliaceaeAF197691AF197770DQ406916GU351723Qiu 3
Malpighia glabra L.MalpighiaceaeGU351056AF520187GU351451GU351724Qiu 95044–1
Manilkara zapota P.RoyenSapotaceaeAF420938AF421016GU351452GU351725Chase 129 (NCU)
Maranta leuconeura E. Morr.MarantaceaeAY299801DQ401410DQ406943GU351726Qiu 95081
Marcgravia rectiflora Triana & Planch.MarcgraviaceaeAF420939, M. sp.AF421017, M. sp.GU351453GU351727Qiu 01014
Mauloutchia chapelieri Warb.MyristicaceaeAF197699AF197769DQ406960GU351728Qiu 99019
Medicago sativa L.FabaceaeGU351057GU351246GU351454GU351729(Qiu M61), no voucher
Melianthus major L.MelianthaceaeGU351058GU351247GU351455GU351730Qiu 97029
Meliosma squmulata Hance.SabiaceaeAF197656DQ007426DQ406896GU351731Qiu 99002
Menispermum canadense Pall.MenispermaceaeGU351059GU351248GU351456GU351732A.A. Reznicek 11732/(Qiu 05024)
Mentzelia floridana Torr. & A.GrayLoasaceaeGU351060GU351249GU351457NDQiu 96179
Menyanthes trifoliata L.MenyanthaceaeGU351061GU351250GU351458GU351733A.A. Reznicek 11748/(Qiu 05034)
Metasequoia glyptostroboides Hu & ChengTaxodiaceaeAF197619NDDQ406973NDQiu 95084
Mirabilis jalapa L.NyctaginaceaeEU280980GU351251GU351459GU351734Qiu 05022
Mollugo verticillata L.MolluginaceaeGU351062GU351252GU351460ND(Qiu M111), no voucher
Montinia caryophyllacea Thunb.MontiniaceaeAY596706GU351253GU351461GU351735Williams 2833 (MO)
Morina longifolia Wall.MorinaceaeGU351063GU351254GU351462GU351736Qiu 97121
Morus alba L.MoraceaeGU351064GU351255GU351463GU351737Qiu 96020
Myodocarpus involucratus Dubard & R.Vig.MyodocarpaceaeGU351065GU351256GU351464NDP. Lowry 4710 (MO)
Myrica cerifera L.MyricaceaeGU351066GU351257GU351465GU351738Qiu 91036
Myriophyllum sp.HaloragaceaeEF370699EF370719EF370738EF370758Qiu 95020
Myristica maingayi Hook. f.MyristicaceaeAF197698, M. fragrans Houtt.AF197768, M. fragrans Houtt.DQ406967GU351739A.R. Khalit 15762/(Qiu M142) (Z)
Myrothamnus flabellifolia Welw.MyrothamnaceaeGU351067GU351258GU351466GU351740P. Winter 72 (RAU, JHB)
Myrtus communis L.MyrtaceaeGU351068GU351259GU351467GU351741Qiu 05043
Nandina domestica Thunb.BerberidaceaeGU351069GU351260GU351468GU351742Qiu 05014
Nelumbo nucifera GaertnerNelumbonaceaeAF197654AF197795DQ406894GU351743Qiu 91028
Nepenthes× kosobeNepenthaceaeDQ401307DQ401379DQ406900GU351744Qiu 94164
Nerium oleander L.ApocynaceaeGU351070GU351261GU351469GU351745Qiu 95048
Nicotiana tabacum L.SolanaceaeAY596704AY453113, N. sylvestris Speg.NC_006581BA000042No new data
Nitraria retusa Asch.NitrariaceaeGU351071GU351262GU351470GU351746Chase 597 (K)
Nolina recurvata Hemsl.AsparagaceaeDQ401301DQ401405DQ407008GU351579Qiu 96043
Nothofagus moorei MaidenNothofagaceaeDQ401292DQ401401DQ406905GU351747Qiu 98036
Nuphar sp.NymphaeaceaeAF197638AF197726DQ406982GU351748Qiu M114, no voucher
Nymphaea sp.NymphaeaceaeAF197639AF197727DQ406981GU351749Qiu 91029
Nyssa sylvatica MarshallCornaceaeGU351072GU351263GU351471GU351750Qiu 94156
Ochna serrulata Walp.OchnaceaeGU351073GU351264GU351472GU351751Qiu 97059
Oenothera berteroana SpachOnagraceaeX04023, O. biennis L.AY453083, O. biennis L.X07566X69140No new data
Olinia emarginata Burtt DavyPenaeaceaeGU351074GU351265GU351473GU351752Chase 6413 (K)
Oncidium sphacelatum Lindl.OrchidaceaeDQ401299DQ401393DQ407005GU351753Qiu 94134
Oncotheca balansae Baill.OncothecaceaeGU351075GU351266GU351474GU351754Chase 2392 (K)
Opilia amentacea Roxb.OpiliaceaeNDNDGU351475GU351755Chase 1902 (K)
Opuntia sp.CactaceaeGU351076GU351267GU351476NDQiu 05020
Orontium aquaticum L.AraceaeAF197705AF197745DQ406996GU351756Qiu 97112
Oryza sativa L.PoaceaeNC_007886DQ401382BA000029BA000029Qiu 01094
Oxalis sp.OxalidaceaeDQ401314AY453111, O. corniculata L.DQ406907GU351757Qiu 94028
Pachysandra terminalis Siebold & Zucc.BuxaceaeAF197634, P. procumbens Michx.AF197784, P. procumbens Michx.DQ406887, P. procumbens Michx.GU351758Qiu QL99028
Paeonia sp.PaeoniaceaeGU351077GU351268GU351477NDQiu 95090
Paeonia tenuifolia L.PaeoniaceaeEF370700EF370720EF370739NDK. Kron 447 (NCU)
Paracryphia alticola SteenisParacryphiaceaeGU351078GU351269GU351478GU351759J.C. Pintaud 561 (K)
Parnassia grandiflora Raf.CelastraceaeGU351079GU351270GU351479GU351760A.A. Reznicek 11734/(Qiu 05027)
Passiflora suberosa L.PassifloraceaeDQ401315AY453071, P. edulis SimsDQ406902GU351761Qiu 95030
Peltanthera floribunda Benth.GesneriaceaeGU351080GU351271GU351480GU351762L.D. Vargas et al. 329 (MO)
Pennantia corymbosa J.R. Forst. & G. Forst.PennantiaceaeGU351081GU351272GU351481GU351763C. Gemmill s.n.
Pentaphragma sp.PentaphragmataceaeGU351082GU351273GU351482GU351764Duangjai 049 (BRUN)
Penthorum sedoides L.PenthoraceaeEF370701EF370721EF370740EF370760Qiu 97114
Peperomia obtusifolia A. Dietr.PiperaceaeAF197629AF197814DQ406924GU351765Qiu 94135
Pereskia grandifolia Haw.CactaceaeGU351083GU351274GU351483NDQiu 94203
Peridiscus lucidus Benth.PeridiscaceaeEF370702AY674550AY674550EF370761Soares 205/(Qiu 05069)
Petrophile canescens R. Br.ProteaceaeAF197653AF197807DQ406983GU351766Qiu 98018
Peumus boldus MolinaMonimiaceaeAF197686AF197803DQ406990GU351767Royal Bot. Gard. Edinburgh 19870707
Phelline comosa Labill.PhellinaceaeGU351084GU351275GU351484GU351768P. D. Ziesing 289 (CBG)
Philydrum lanuginosum Gaertn.PhilydraceaeAY299824DQ401406GU351485GU351769Qiu 98102
Phyllanthus angustifolius Sw.PhyllanthaceaeGU351085GU351276GU351486GU351770Qiu 05041
Phyllonoma laticuspus Engl.PhyllonomaceaeGU351086NDGU351487GU351771Morgan 2124 (WS)
Physena madagascariensis Steud.PhysenaceaeGU351087NDNDNDMiller et al. 8817 (MO)
Phytolacca americana L.PhytolaccaceaeGU351088GU351277GU351488GU351772Qiu 94109
Pilea fontana Rydb.UrticaceaeGU351089GU351278NDNDQiu 96119
Pinguicula vulgaris L.LentibulariaceaeGU351090GU351279GU351489GU351773Qiu 96115
Pinus sp.PinaceaeAF197626AF197723AY832181, P. thunbergii Parl.GU351774Qiu 94013
Piper betle L.PiperaceaeAF197630AF197750DQ406925GU351775Qiu 91048
Pittosporum tobira Dryand.PittosporaceaeGU351091AF520127, P. glabratum Lindl.GU351490GU351776Qiu 95031
Platanus occidentalis L.PlatanaceaeAF197655AF197793AY832177GU351777Qiu 94152
Pleea tenuifolia MichauxTofieldiaceaeAF197703AF197743DQ406995GU351778(Qiu 96128)
Podocarpus macrophyllus SweetPodocarpaceaeAF197620DQ007425DQ406962GU351779Qiu 95006
Polygala cruciata L.PolygalaceaeGU351092GU351280GU351491GU351780Chase 155 (NCU)
Polygonum sp.PolygonaceaeGU351093GU351281GU351492GU351781Qiu 94110
Polyosma sp.PolyosmaceaeGU351094GU351282GU351493GU351782Johns 9558 (BO, FREE, K, MAN)
Populus sp.SalicaceaeNDGU351283GU351494NDQiu 05021
Portulaca oleracea L.PortulacaceaeGU351095GU351284GU351495NDQiu 94111
Potamogeton berchtoldii FieberPotamogetonaceaeAF197715AF197724DQ406938GU351783Qiu 96063
Qualea sp.VochysiaceaeGU351096GU351285GU351496GU351784Chase 168 (NCU)
Quercus alba L.FagaceaeGU351097GU351286GU351497GU351785Qiu 95115
Quillaja saponaria Poir.QuillajaceaeGU351098GU351287GU351498GU351786Chase 10931 (K)
Quintinia verdonii F. Muell.QuintiniaceaeGU351099GU351288GU351499GU351787Y. Pillon et al. 379 (NOU)
Ranunculus sp.RanunculaceaeAF197714AF197759DQ406876GU351788Qiu 95024
Reinwardtia trigyna Dalzell & A. GibsonLinaceaeGU351100GU351289GU351500GU351789Y.P. Hong H103/(Qiu 05056) (PE)
Reseda alba Delile.ResedaceaeGU351101GU351290GU351501GU351790Qiu 97070
Rhabdodendron amazonicum HuberRhabdodendraceaeGU351102GU351291GU351502GU351791E. Ribeiro (K)
Rhodoleia championii Hook.HamamelidaceaeEF370703EF370722EF370742EF370762Royal Bot. Gard. Edinburgh, no voucher
Ribes sp.GrossulariaceaeEF370704EF370723EF370743EF370763Qiu 95022
Rivina humulis L.PhytolaccaceaeGU351103GU351292GU351503GU351792D. Soltis 2643 (FLAS)
Roupala macrophylla PholProteaceaeGU351104GU351293GU351504GU351793Douglas 131 (MEL)
Roussea simplex Sm.RousseaceaeGU351105GU351294GU351505GU351794Mauritius Sugar Res. Inst.
Ruptiliocarpon caracolito Hammel & N. ZamoraLepidobotryaceaeGU351106GU351295GU351506GU351795Pennington 631 (K)
Sabia sp.SabiaceaeAF197657AF197780DQ406895GU351796Qiu 91025
Saintpaulia magungensis E.P. RobertsGesneriaceaeGU351107GU351296GU351507GU351797Chase 696 (K)
Santalum album L.SantalaceaeGU351108GU351297GU351508GU351798Chase 1349 (K)
Sarcandra chloranthoides GardnerChloranthaceaeAF197666AF197754DQ406866GU351799Qiu 92002
Sarcobatus vermiculatus Torr.SarcobataceaeGU351109GU351298GU351509GU351800King and Garvey 13892 (MO)
Sargentodoxa cuneata Rehder & WilsonLardizabalaceaeAF197644AF197790DQ406875GU351801X. Pan 93001 (Qiu M178) (NCU)
Sarracenia flava L.SarraceniaceaeAF420947AF421028GU351510GU351802Qiu 94141
Saruma henryi Oliv.AristolochiaceaeAF197672AF197752DQ406912GU351803Qiu 91018
Saururus cernuus L.SaururaceaeAF197633AF197748DQ406934GU351804Qiu 97098
Saxifraga sarmentosa L.f.SaxifragaceaeEF370705EF370724EF370744EF370764Qiu 95074
Scaevola aemula R.Br.GoodeniaceaeGU351110AY453118NDNDQiu 97058
Schinus molle L.AnacardiaceaeGU351111GU351299GU351511GU351805Z.D. Chen KEN073/(Qiu 05065) (PE)
Schisandra sphenanthera Rehder & WilsonSchisandraceaeAF197662AF197739DQ406972GU351806Qiu 94165
Schoepfia schreberi J.F. Gmel.SchoepfiaceaeGU351112GU351300GU351512GU351807Nickrent 2599 (ILL)
Scrophularia marilandica L.ScrophulariaceaeGU351113GU351301NDGU351808A.A. Reznicek 11737/(Qiu 05037)
Sedum humifusum RoseCrassulaceaeEF370706AF520100EF370745EF370765Qiu 05017
Sesamum triphyllum Asch.PedaliaceaeGU351114GU351302GU351513GU351809Chase 5710 (K)
Simmondsia chinensis C.K. Schneid.SimmondsiaceaeDQ401309DQ401397DQ406903GU351810Qiu 96120
Siparuna brasiliensis A. DC.SiparunaceaeAF197687AF197809DQ406976GU351811Qiu 02003
Smilax sp.SmilacaceaeAF039251DQ401391DQ406940GU351812Qiu 95117
Sparganium americanum Nutt.SparganiaceaeAY124509, S. eurycarpum Engelm.DQ401396DQ407010GU351813Qiu 96108
Spathiphyllum clevelandiiAraceaeAF197706AF197746DQ406997, S. wallisii Hort.GU351814Qiu 94140
Sphenoclea zeylanica Gaertn.SphenocleaceaeGU351115GU351303GU351514GU351815(Qiu M253), Bot. Gard. Bonn, no voucher
Sphenostemon lobosporus L.S. Sm.SphenostemonaceaeGU351116GU351304GU351515GU351816Chase 1900 (K)
Spinacia oleracea L.AmaranthaceaeDQ401287AY453110DQ406883GU351817Qiu 94059
Spiraea sp.RosaceaeGU351117GU351305GU351516GU351818Qiu 05008
Stachyurus chinensis Franch.StachyuraceaeGU351118GU351306GU351517GU351819Z.D. Chen JGS 005/(Qiu 05062) (PE)
Staphylea trifolia L.StaphyleaceaeDQ401294AY453105DQ406906GU351820Qiu 95106
Stegnosperma halimifolium Benth.StegnospermataceaeGU351119GU351307GU351518GU351821Martin et al. s.n. (MO)
Stegolepis sp.RapateaceaeAY124535, S. parvipetala Steyerm.DQ401411DQ407004GU351822Qiu 97132
Stellaria sp.CaryophyllaceaeGU351120NDGU351519NDQiu 95015
Sterculia balanghas L.MalvaceaeDQ401316DQ401402DQ406869GU351823Qiu 97056
Strasburgeria robusta GuillauminStrasburgeriaceaeGU351121GU351308GU351520GU351824Y. Pillon et al 60 (NOU, K)
Strelitzia reginae AitonStrelitziaceaeAY299843, S. nicolai Regel & K.KochAY453112DQ406965GU351825Qiu 96045
Strychnos spinosa Lam.LoganiaceaeAY741818GU351309GU351521ND(Qiu 96187)
Stylobasium spathulatum Desf.SurianaceaeGU351122GU351310GU351522GU351826G. Brummitt et al 21242 (K)
Styrax americanus Lam.StyracaceaeAF420950, S. officinalis L.AF520205, S. grandiflora Griff.GU351523GU351827K. Kron 521 (NCU)
Swietenia macrophylla KingMeliaceaeGU351123GU351311GU351524GU351828(Chris W. Dick 646)/(Qiu M294), no voucher
Syringa sp.OleaceaeAY741821, S. vulgaris L.GU351312GU351525GU351829Qiu 95037
Tacca chantrieri AndreDioscoreaceaeAF039252, T. pinnatifida J.R. Forst & G. ForstDQ401377DQ406941HM357127Qiu 01015
Takhtajania perrieri M. Baranova & J. LeroyWinteraceaeDQ007416DQ007427DQ406913GU351830J. Rabenantoandro 219 (MO)
Talinum patens Juss.TalinaceaeGU351124GU351313GU351526NDK.X. Xu 015/(Qiu 05064) (PE)
Tamarix sp.TamaricaceaeGU351125GU351314GU351527NDQiu 95034
Tapiscia sinensis Oliv.TapisciaceaeGU351126GU351315GU351528GU351831Chase 1021 (K)
Tasmannia insipida DC.WinteraceaeAF197674AF197782DQ406970GU351832Qiu 90032
Terminalia catappa L.CombretaceaeGU351127NDGU351529GU351833US Natl Trop Bot Gard # 731222022
Ternstroemia stahlii Krug & Urb.PentaphylaceaeAY725909AY163754, T. gymnanthera SpragueGU351530GU351834Chase 360 (K)
Tetracarpaea tasmanica Hook.f.TetracarpaeaceaeEF370707EF370725EF370746EF370766No new data
Tetracentron sinense Oliv.TrochodendraceaeAF197647AF197791DQ406874GU351835Qiu 90009
Tetracera asiatica HooglandDilleniaceaeGU351128AF520094GU351531GU351836Chase 1238 (K)
Tetramerista sp.TetrameristaceaeAF420958GU351316GU351532GU351837M. Coode 7925 (K)
Thottea tomentosa Ding HouAristolochiaceaeAF197670AF197733DQ406914GU351838Chase 2086 (K)
Thymelaea hirsuta Endl.ThymelaeaceaeGU351129GU351317GU351533GU351839(Qiu M284)
Tinospora sagittata Gagnep.MenispermaceaeGU351130GU351318GU351534GU351840Y.P. Hong 99258/(Qiu 05061) (PE)
Tofieldia calyculata Wahlenb.TofieldiaceaeAF197704AF197744DQ406935GU351841Qiu 97041
Tradescantia sp.CommelinaceaeDQ401320AY453108, T. ohiensis RafDQ406950GU351842Qiu 96059
Triglochin maritima L.JuncaginaceaeAF197716AF197725DQ406998NDQiu 97106
Trillium sp.MelanthiaceaeAF039253, T. grandiflorum Salisb.DQ401407DQ406949GU351843Qiu 95016
Trimenia moorei W.R. PhilipsonTrimeniaceaeAY009428, Trimenia sp.AF197741DQ406987GU351844Australia Natl. Bot. Gard. 701680
Triphyophyllum peltatum Airy ShawDioncophyllaceaeGU351131GU351319GU351535GU351845Chase 663 (K)
Trithuria inconspicua CheesemanHydatellaceaeGU351132NDGU351536NDP.D. Chapman s.n., NSW 428712
Trithuria lanterna D.A. CookeHydatellaceaeGU351133NDGU351537NDT.D. Macfarlane et al. 4321
Trochodendron aralioides Sieb. & Zucc.TrochodendraceaeAF197648AF197792DQ406880GU351846Qiu 90026
Tropaeolum peltophorum Benth.TropaeolaceaeGU351134GU351320GU351538GU351847Peter Kuhlman s.n./(Qiu 96150)
Urtica dioica L.UrticaceaeGU351135NDGU351539NDA.A. Reznicek 11740/(Qiu 05039)
Vahlia capenis Thunb.VahliaceaeGU351136GU351321GU351540GU351848Van Wyk 10–579 (PUR)
Valeriana officinalis L.ValerianaceaeGU351137GU351322GU351541NDPeter Kuhlman s.n./(Qiu 96013)
Verbena bonariensis L.VerbenaceaeAY741828GU351323GU351542GU351849Qiu 05012
Viburnum sp.AdoxaceaeGU351138GU351324GU351543GU351850Qiu 95083
Viola sp.ViolaceaeGU351139GU351325GU351544GU351851Qiu 95018
Vitis sp.VitaceaeDQ401305AY453123, V. riparia Michx.DQ406881GU351852Qiu 94046
Viviania marifolia Cav.VivianiaceaeGU351140GU351326GU351545NDM. Ackermann 543 (B)
Vriesea splendens Lem.BromeliaceaeDQ401298DQ401378DQ406945GU351853Qiu 96073
Welwitschia mirabilis Hook.f.WelwitschiaceaeAF197618AF197719DQ406958NDNo new data
Wendtia gracilis MeyenLedocarpaceaeNDNDGU351546NDKubitzki & Feuerer 990–68 (HBG)
Xanthorrhoea quadrangulata F. Muell.XanthorrhoeaceaeAF039250, X. australis R.Br.DQ401384DQ406946GU351854Qiu 97039
Xanthosoma mafaffa SchottAraceaeDQ401318DQ401376DQ406936GU351855Qiu 95063
Ximenia americana L.OlacaceaeGU351141GU351327GU351547GU351856(Qiu 96204)
Zamia floridana A. DC.ZamiaceaeAF197624AF197721DQ406961, Z. integrifolia Rich.GU351857Qiu 95035
Zelkova serrata MakinoUlmaceaeGU351142GU351328GU351548GU351858A.A. Reznicek 11739/(Qiu 05038)

Four mitochondrial genes, atp1 (ATPase subunit 1), matR (a group II intron-encoded maturase), nad5 (NADH dehydrogenase subunit 5), and rps3 (ribosomal protein S3), were selected for sequencing, with approximately 1.0, 1.6, 1.1, 1.4 kb sequenced, respectively. Among these genes, atp1 and matR have been widely used over the last 10 years, but nad5 and rps3 have only recently received attention from plant systematists (Qiu et al., 2006a; Jian et al., 2008; Ran et al., 2009; Wurdack & Davis, 2009). Analyses on how these four genes performed in an angiosperm-wide phylogenetic study will be described below.

The methods of DNA extraction, gene amplification, and sequencing are the same as reported before (Qiu et al., 2006a), the only modification being that nested PCR was used in some cases to improve amplification success rate. All primers were newly designed (Table 2) except those for nad5, which were published in Qiu et al., 2006a.

Table 2.  Primer sequences of atp1, matR, and rps3 used in this study
  1. The primer position is indicated by the coordinate number in the corresponding gene of Arabidopsis thaliana (NC_001284) given in parentheses. The primer melting temperature, estimated in Oligo v. 6 (Molecular Biology Insights, Cascade, CO, USA), is given in parentheses after the sequence.

Primers for atp1
 Aatp1-F1 (64–85)Aatp1-R1 (1369–1388)
 TAC RCG AAW TTK CAA GTG GAT G (62 °C)CT GTC TAG KGG CAT TYG RTC (60 °C)
 Aatp1-F2 (461–479)Aatp1-R2 (1001–1018)
 CG GTR GAT AGC CTN GTT CC (60 °C)A GGC CGA YAC GTC TCC NG (58 °C)
 Aatp1-F3 (978–997)Aatp1-R3 (619–638)
 S TTA CCC GTS ATT GAA ACA C (58 °C)CG TTT CTG TCC AAT YGC NAC (60 °C)
Primers for matR
 AmatR-F1 (61–80)AmatR-R1 (1835–1852)
 ATC AGA AYG GTA CYC GAA TC (56 °C)T GTG CTT KTG GGC WRG GG (58 °C)
 AmatR-F2 (598–616)AmatR-R2 (1395–1411)
 TCC CTT GTT TYG TCR TKG C (60 °C)G CCG GAT GTG CTK KAC G (60 °C)
 AmatR-F3 (1087–1105)AmatR-R3 (957–975)
 RTA RYT GCA CGG AGT ACG G (60 °C)TRA GTC RTC GGC RTA TCG C (58 °C)
 AmatR-F4 (1395–1411)AmatR-R4 (716–733)
 C GTC AAG CAC ATM HGG C (56 °C)C GGC GMA AAG RAR GCT CG (60 °C)
 AmatR-R5 (353–372)
 TAG GGC CRA TAG TAR TAC AC (60 °C)
 MmatR-F2 (388–406)MmatR-R2 (1398–1415)
 CTA MRC AAG CTC GAT CAG G (58 °C)YCT TGC CGG ATG TGC TTG (58 °C)
 MmatR-F3 (850–866)MmatR-R3 (1093–1111)
 CHK ATA GAG CTG GGC GG (56 °C)ATT CTA CCG TAC TCC GTG C (58 °C)
 MmatR-F4 (1182–1200)MmatR-R4 (716–733)
 GCG TCT ACG GGT AAA GCA C (60 °C)CGC MGC AAA RGA RGC TCG (62 °C)
 MmatR-F5 (1470–1487)MmatR-R5 (353–371)
 CGT TCA ACA GRC AGT CTC (56 °C)AGS GCC GAT AGT AGT ACA C (58 °C)
Primers for rps3
 rps3-F1 (128–146)rps3-R1 (1643–1662)
 GT TCG ATA CGT CCA CCT AC (58 °C)GTA CGT TTC GGA TAT RGC AC (58 °C)
 rps3-F12 (161–179)rps3-R12 (1640–1658)
 GC TTT CGY CTC GGT AGG TG (60 °C)GT TTC GGA TAT RGC ACG TC (56 °C)
 rps3-F2 (382–401)rps3-R2 (1270–1289)
 GCA GGG AAA ASW GTC RAG TC (60 °C)CT ATT AGA CAA NAA AGA TCG (54 °C)
 rps3-F3 (528–545)rps3-R3 (910–928)
 C GKG GCC TWC AAG CAT CC (60 °C)A CCT CTT TTT GKC TYS GGC (56 °C)
 rps3-F4 (997–1016)rps3-R4 (469–488)
 TTT CCW TTC TTC GGT GCT AC (58 °C)GG TGA TCG GTC ATG GTA TCC (60 °C)
 rps3-F5 (1343–1360) 
 GT GCT TCT CYR ATT GCT C (58 °C) 

A total of 900 new sequences were generated in this study; the rest were retrieved from GenBank. Their accession numbers and voucher information are provided in Table 1. Sequences were aligned using ClustalX (Thompson et al., 1997) followed by manual adjustment. Because point substitution rate was low in these genes (see below), it is relatively easy to locate mis-aligned regions and bring them to proper positions by aligning neighboring regions that share high levels of sequence identity. For each of the four genes, a single gene analysis was carried out using the parsimony method implemented in PAUP*4.0b10 (Swofford, 2003), to ensure that no contaminated or fundamentally incongruent sequences due to horizontal gene transfer were present in the dataset. The data were then combined to construct two matrices. The first contained 380 OTUs, and most OTUs had all four genes except for a small number with only one, two, or three genes. Because the OTUs with a significant amount of missing data or highly divergent sequences could artificially lower bootstrap (BS) values of the clades to which these OTUs belonged (Felsenstein, 2004), they were removed and a second matrix was constructed (see Table 1 for removed taxa). This matrix contained 356 OTUs, with each OTU having at least three genes.

Maximum likelihood analyses were carried out using a web version of RAxML 7.0.4 (Stamatakis et al., 2008) on the CIPRES cluster at the San Diego Supercomputer Center (Miller et al., 2009). The matrices were analyzed as a single partition under the GTR+G model of nucleotide evolution. Maximum likelihood BS analyses were carried out with 500 replicates of character resampling. The automatic estimation of BS replicate number in RAxML showed that for both matrices 150 replicates were sufficient. Comparison of the 150 and 500 replicate BS analysis trees showed that BS values were indeed similar.

To understand how evolutionary rates and homoplasy levels of these four mitochondrial genes may have influenced their performance in reconstructing angiosperm phylogeny, especially relative to those of the four genes (chloroplast atpB, matK, and rbcL, and nuclear 18S rDNA) that have been used in the previous large-scale angiosperm phylogenetic analyses, two more analyses were carried out after the phylogenetic analyses. To make the results comparable, a total of 272 OTUs that have sequence for each of the eight genes were selected. For the four mitochondrial genes, the sequence accession numbers are shown in Table 1. For chloroplast atpB, matK, and rbcL, and nuclear 18S rDNA, the data were retrieved from GenBank and are available from the corresponding author upon request. Eight single gene matrices were assembled accordingly. A consensus angiosperm phylogeny was drawn based on the results of this study and the previous large-scale analyses of chloroplast atpB, matK, and rbcL, and nuclear 18S rDNA (Chase et al., 1993; Soltis et al., 1997, 2000; Savolainen et al., 2000a; Hilu et al., 2003), and this tree is shown in Fig. S1.

The first analysis was to calculate the evolutionary rate of each gene. Enforcing the consensus topology of angiosperms, a phylogram was inferred from each gene matrix assuming a GTR+I+G model of nucleotide substitution in PAUP*v4.0b10 (Swofford, 2003). The phylogram was then fit to a molecular clock using the Langley–Fitch method in r8s v1.7.1 (Sanderson, 2003). Because taxon complements are identical across genes, fixing the root age to 1.0 in each of these analyses ensures that molecular rate estimates are directly comparable.

The other analysis aimed to estimate the homoplasy level in each of the eight genes. A parsimony search was carried out on each matrix under the constraint consensus angiosperm phylogeny using PAUP*v4.0b10. These searches were not run to completion because: (i) they took too long to complete; and (ii) consistency and homoplasy index values varied little on trees that differed by a small percentage of parsimony length. Consistency and homoplasy indexes were then output from the tree.

Finally, we examined the effect of RNA editing on reconstructing angiosperm phylogeny using mitochondrial gene sequences. Previously, RNA editing was shown to have some effect on phylogenetic analysis when the editing level was high in a slowly evolving gene, such as nad5 (Qiu et al., 2006a). However, this effect can be minimized through combined analyses with less edited genes (Petersen et al., 2006; Qiu et al., 2006a). In this study, we removed all RNA editing sites according to the information available in GenBank for these four genes in Arabidopsis thaliana (NC_001284) (Giege & Brennicke, 1999), Brassica napus (NC_008285) (Handa, 2003), Beta vulgaris (NC_002511) (Mower & Palmer, 2006), and Oryza sativa Japonica Group (NC_011033) (Notsu et al., 2002). As a result, 2, 8, 17, and 17 sites were deleted from the 356 OTU matrix in atp1, matR, nad5, and rps3, respectively. The matrix was then analyzed using RAxML 7.0.4 in the same way as the regular 356 OTU matrix except that only 100 BS replicates were run.

2 Results and discussion

2.1 Mitochondrial gene-based angiosperm phylogeny

A generally well resolved angiosperm phylogeny, with many major nodes moderately to strongly supported, was reconstructed from the 356 OTU matrix (−ln L = 150192.44). A schematic version of this tree is presented in Fig. 1 and a detailed version is shown in Fig. 2. A topologically highly similar phylogenetic tree, with BS support on some nodes slightly to significantly lower, was obtained from the 380 OTU matrix (−ln L = 164322.98), and it is shown in Fig. S2.1, 2.2. A topologically similar phylogenetic tree, with similar BS support values on most nodes, was obtained from the 356 OTU matrix with RNA editing sites removed (−ln L = 146408.591300), and it is shown in Fig. S3.

Figure 1.

Schematic cladogram of angiosperm phylogeny inferred from four mitochondrial genes atp1, matR, nad5, and rps3 from 356 seed plants, and a detailed version is shown in Fig. 2. Bootstrap values >50% are shown above branches. All angiosperm orders have >50% bootstrap support except the three labeled with asterisks (Saxifragales has 97% bootstrap support if Peridiscus is not included).

Figure 2.

Maximum likelihood tree of 356 seed plants inferred from nucleotide sequences of mitochondrial genes atp1, matR, nad5, and rps3. Bootstrap values are shown above the branches, with those for key nodes shown in a larger font size and boldface. The branch length is indicative of the divergence level among taxa except those in thick or dashed lines, with the scale bar shown at the bottom of the first part of the tree. Thick lines represent one-sixth of the real length of the branches. Dashed lines are near-zero length branches, and are expanded for the presentation purpose. (Fig. 2-1)
(Fig. 2-2)
(Fig. 2-3)
(Fig. 2-4)
(Fig. 2-5)
(Fig. 2-6)

2.1.1 Basalmost angiosperms  The first diverging lineage of angiosperms consists of Amborellales and Nymphaeales (Figs. 1, 2-1). Hydatellaceae, a recently identified member of this lineage based on chloroplast and nuclear phytochrome gene sequences as well as morphology (Saarela et al., 2007), also fall in this group and are sister to Nymphaeaceae–Cabombaceae (Fig. S2.1). This family is not included in the 356 OTU matrix, but is included in the 380 OTU matrix because both sampled species (Trithuria inconspicua and T. lanterna) have only two of the four genes amplified, and they are rather divergent (Table 1, Fig. S2.2). The monophyly of this group has moderate BS support (77 or 79% when Hydatellaceae are included). The sister relationship of this clade to the rest of the angiosperms has 99–100% BS support in the two analyses.

The lineage that follows Amborellales + Nymphaeales in diversification of extant angiosperms is Austrobaileyales. Both the monophyly of this order and its relationship to the remaining angiosperms are strongly supported (Figs. 1, 2-1).

With the exception of all rbcL analyses and some chloroplast genome analyses, all other analyses of molecular data have identified Amborellales, Nymphaeales, and Austrobaileyales as the basalmost extant angiosperms (Mathews & Donoghue, 1999; Parkinson et al., 1999; Qiu et al., 1999, 2005, 2006a; Barkman et al., 2000; Graham & Olmstead, 2000; Soltis et al., 2000; Zanis et al., 2002; Borsch et al., 2003; Hilu et al., 2003; Stefanovic et al., 2004; Leebens-Mack et al., 2005; Jansen et al., 2007; Moore et al., 2007; Saarela et al., 2007; Goremykin et al., 2009). In all analyses of rbcL sequences (Les et al., 1991; Chase et al., 1993; Qiu et al., 1993; Savolainen et al., 2000a) and the first and related chloroplast genome analyses (Goremykin et al., 2003), Ceratophyllum was placed as the sister to all other angiosperms, albeit with low to moderate support in cases where support values were obtained. Despite this consensus, there has been considerable controversy regarding whether the first diverging lineage of angiosperms consists of Amborella alone or Amborella and Nymphaeales (now also including Hydatellaceae) together. In fact, this controversy emerged almost as soon as the ANITA lineages were identified as the basalmost extant angiosperms. An alternative topology test using a combined dataset of chloroplast atpB and rbcL, mitochondrial atp1 and matR, and nuclear 18S showed that the topologies of Amborella alone, Nymphaeales alone, or Amborella and Nymphaeales together being sister to all other angiosperms were statistically indistinguishable (Qiu et al., 2000). An analysis of 17 chloroplast genes obtained similar results through alternative topology testing (Graham & Olmstead, 2000). However, an analysis of a somewhat different dataset than that of Qiu et al. (2000), using a different method and an orthologous copy of atp1, suggested that the Amborella+ Nymphaeales basal topology was the preferred hypothesis (Barkman et al., 2000). In a recent analysis of mitochondrial atp1, matR, and nad5 from 162 of mostly basal angiosperms, at least 91% BS support was found to support the topology with Amborella and Nymphaeales together being sister to all other angiosperms (Qiu et al., 2006a). Most recently, in an analysis of a chloroplast genomic dataset, it was shown that removal of a significant percentage of highly variable sites changed the placement of Amborella to Amborella+ Nymphaeales as the first diverging lineage of angiosperms (Goremykin et al., 2009), although the issue of removing fast-evolving sites needs to be explored further (Graham & Iles, 2009). The facts that the mitochondrial genes used in all the mentioned studies (atp1, cox1, matR, and nad5[Barkman et al., 2000; Qiu et al., 2000, 2006a]) have lower or significantly lower substitution rates than the commonly used chloroplast genes (atpB, matK, rbcL) (Table 3), and that the divergence gap between gymnosperms and angiosperms is large, increase the likelihood that the Amborella-basal topology seen in the studies that used fast-evolving chloroplast genes is an artifact. In Qiu et al., 2006a and this study, analyses of three or four mitochondrial genes from 162 or 356 seed plants with two different maximum likelihood methods both recovered the topology with Amborella and Nymphaeales together as the sister group of all other angiosperms with moderate to strong BS support. This level of consistency in the results of the two studies, as well as those cited above, supports the hypothesis that the basalmost extant angiosperm lineage includes Amborella, Hydatellaceae, and Nymphaeales.

Table 3.  Molecular clock rates of the eight genes estimated from the matrix of 272 seed plants under a constraint tree shown in Fig. S1
GeneLength (bp)Molecular clock rate (substitution/site/unit time)
  1. cp, chloroplast; mt, mitochondrial; nu, nuclear.

cp-matK18760.1430
cp-rbcL14060.1080
mt-rps323800.0838
nu-18S rDNA17320.0774
cp-atpB14820.0675
mt-matR36730.0441
mt-atp111690.0405
mt-nad511330.0292

2.1.2 Major lineages of mesangiosperms  Mesangiosperms, as defined by Cantino et al. (2007), include five groups: Chloranthaceae, Ceratophyllum, magnoliids, monocots, and eudicots. In this study, Chloranthaceae and Ceratophyllum form a clade with 63% BS support. All magnoliid taxa form a monophyletic group with 66% BS support. Both monocots and eudicots receive 100% BS support. The relationships among these four clades have virtually no BS support (Figs. 1, 2-1, 2-2).

Upon identification of the ANITA lines as the basalmost extant angiosperms, it was realized that resolving relationships among five mesangiosperm lineages was the next major challenge in the study of phylogenetic patterns among basal angiosperms (Doyle & Endress, 2000; Qiu et al., 2000). Four types of analyses can be categorized among all studies that have attempted to resolve these relationships or have them as part of their study questions, and some promising results have been obtained. First, in three large-scale analyses, which sampled one, three, and five genes, respectively, from a large number of angiosperms, especially monocots and eudicots, only the sister relationship between Ceratophyllum and eudicots received >50% BS support (Soltis et al., 2000; Hilu et al., 2003; Burleigh et al., 2009). Second, a number of medium-scale analyses were carried out, in which several genes from two or three plant genomes were analyzed from a densely sampled set of taxa. In these analyses, Chloranthaceae were sister to magnoliids (Barkman et al., 2000; Saarela et al., 2007) or magnoliids + eudicots (Zanis et al., 2002, 2003); Ceratophyllum was sister to monocots (Zanis et al., 2002, 2003; Qiu et al., 2005), eudicots (Qiu et al., 2005, 2006a; Saarela et al., 2007; Qiu & Estabrook, 2008), or Chloranthaceae (Antonov et al., 2000; Duvall et al., 2006, 2008; Qiu et al., 2006a); magnoliids were sister to eudicots (Zanis et al., 2002, 2003), or Ceratophyllum+ eudicots (Qiu & Estabrook, 2008); and monocots were sister to Ceratophyllum+ eudicots (Saarela et al., 2007). Perhaps because different phylogenetic methods were used to analyze highly diverse sets of data in these studies, the results were also very variable. Third, a series of analyses of largely morphological data with topological constraints derived from some molecular studies placed Ceratophyllum as the sister to Chloranthaceae (Doyle et al., 2008; Endress & Doyle, 2009; Doyle & Endress, 2010). Finally, four recent chloroplast phylogenomic analyses resolved relationships among the five mesangiosperm lineages with moderate to strong BS support, with monocots being sister to eudicots, and magnoliids being sister to monocots + eudicots (Jansen et al., 2007; Moore et al., 2007, 2010; Goremykin et al., 2009). Ceratophyllum, when included in the analyses, changed its position between being sister to eudicots or magnoliids, depending on the portion of the genome sequences analyzed (Moore et al., 2007, 2010; Goremykin et al., 2009). Chloranthaceae were consistently sister to magnoliids when they were included in analyses (Jansen et al., 2007; Moore et al., 2007, 2010).

Despite the heterogeneity of these analyses and the diverse results obtained, there seems to be an emerging consensus on the relationship of Ceratophyllum, as a sister to either eudicots or Chloranthaceae. Its placement as the sister to monocots was only seen in the studies that sampled insufficient number of monocots, and further, several monocots such as Acorus and alismatids had highly divergent mitochondrial genes in those studies (Zanis et al., 2002, 2003; Qiu et al., 2005). Hence, this result may be an artifact. The placement of Ceratophyllum as the sister to Chloranthaceae emerged relatively late in the studies of basal angiosperm phylogeny, but deserves some consideration. Two mitochondrial gene-based analyses, with extensive sampling of all five mesangiosperm lineages, consistently identified this relationship, even though the BS values were only over 60% (Qiu et al., 2006a) (this study, Figs. 2-1, S2.1, S3.1). Placement of Ceratophyllum and Chloranthaceae together in the morphological cladistic analyses was mostly due to their simple flowers (Doyle et al., 2008; Endress & Doyle, 2009; Doyle & Endress, 2010). At present, it is difficult to determine whether these simple flowers reflect a common ancestry or independent reduction due to adaptation to anemophily and hydrophily. Excavation of chloranthoid and ceratophyllaceous flowers or fruits from the Lower Cretaceous indicates that these simple flowers have had a long history (Friis et al., 1986; Dilcher & Wang, 2009). Given that the sister relationship between Ceratophyllum and Chloranthaceae has been recovered in two well-designed mitochondrial gene phylogenetic studies and a series of analyses of carefully constructed morphological matrices, it may be premature to dismiss this result, as was done in the recently published APG III (Angiosperm Phylogeny Group, 2009).

All other relationships among the major lineages of mesangiosperms have been resolved only in three recent chloroplast phylogenomic analyses (Jansen et al., 2007; Moore et al., 2007, 2010). While it is encouraging to see the stable results from these studies, some caution is needed in interpreting the phylogenetic meaning of the reported high BS values, as this type of analysis tends to produce high support values for whatever relationships the particular taxon sampling scheme leads to, probably because of character over-sampling and taxon under-sampling (Delsuc et al., 2005; Hedtke et al., 2006; Heath et al., 2008). Three examples published over the last decade on phylogenetic analyses of some key green alga and land plant lineages serve as a sober reminder of this effect: Amborella (Goremykin et al., 2003, 2009; Soltis et al., 2004; Stefanovic et al., 2004; Leebens-Mack et al., 2005; Qiu et al., 2005; Jansen et al., 2007; Moore et al., 2007, 2010); liverworts, mosses, and hornworts (Nishiyama et al., 2004; Goremykin & Hellwig, 2005; Qiu et al., 2006b, 2007); and Mesostigma (Lemieux et al., 2000, 2007; Qiu & Lee, 2000). In a recent chloroplast phylogenomic study, it was shown that the relationships among the five mesangiosperm lineages, despite having moderate BS support, were statistically indistinguishable in an alternative topology test (Moore et al., 2007). One should also bear in mind that BS values only measure the fit between the data and the resulting tree, and that if the data contain any bias that undercuts representativeness of the data for the whole data space of the investigated group, phylogenetic informativeness of BS values, high or low, may be compromised (Sanderson, 1995). In the case of chloroplast phylogenomic analyses, because taxon sampling is usually sparse relative to the taxonomic scope covered, any amount of bias can be amplified through sampling of a large number of characters. Compact organellar genomes, because they have coded for highly specialized functions during a long period of evolution, are known for molecular evolutionary oddities such as hydrophobicity bias, RNA editing, and GC content skew (Naylor & Brown, 1997; Jobson & Qiu, 2008). They can lead phylogenetic algorithms astray if improper attention is paid to these complicating and potentially confounding factors.

2.1.3 Magnoliids  The monophyly of magnoliids, and the sister relationships between their two pairs of member lineages, Magnoliales/Laurales and Canellales/Piperales, are weakly to moderately supported (Fig. 2-1). This is the second large-scale angiosperm phylogenetic analysis that has recovered these results; the first was the matK analysis by Hilu et al. (2003). Most medium-scale analyses focusing on basal angiosperms have also recovered these relationships with various degrees of support (Mathews & Donoghue, 1999; Qiu et al., 1999, 2000, 2005, 2006a; Barkman et al., 2000, 2007; Graham & Olmstead, 2000; Nickrent et al., 2002; Zanis et al., 2002, 2003; Borsch et al., 2003; Lohne & Borsch, 2005; Qiu & Estabrook, 2008). This is one of the cases where moderate support in many studies that sample a wide variety of genes from chloroplast, mitochondrial, and nuclear genomes has led to a consensus that the true underlying plant phylogeny has probably been reconstructed.

2.1.4 Monocots  The monophyly of monocots is strongly supported, regardless of inclusion or exclusion of two Acorus species, which have divergent sequences for the three mitochondrial genes used in this study (Figs. 2-2, S2.1, Table 1; the matR sequences of Acorus were even more divergent and thus were not used). Within monocots other than Acorus, a deep split between Alismatales and all the remaining groups (Petrosaviidae of Cantino et al., 2007) is strongly supported. This deep split was previously identified with moderate to strong support in two large-scale analyses of angiosperms (Soltis et al., 2000; Hilu et al., 2003). Three medium-scale analyses, which had relatively dense taxon sampling and included seven genes (chloroplast atpB, matK, ndhF, and rbcL; mitochondrial atp1; nuclear 18S and 26S rDNA) (Chase et al., 2006), four genes (chloroplast matK and rbcL; mitochondrial atp1 and cob) (Davis et al., 2006), or 16 kb of chloroplast DNA sequences (Graham et al., 2006), also identified this deep split with strong support. This result is further corroborated by the discovery of a trans-spliced group II intron in the mitochondrial gene nad1 in 94 genera of petrosaviid monocots, which occurs rarely in the land plant mitochondrial genome (Qiu & Palmer, 2004).

The relationships among four strongly supported clades within petrosaviid monocots (Asparagales, Pandanales + Dioscoreales, Liliales, and commelinids) have only weak to moderate support. These results are comparable to or better than those obtained in previous large-scale angiosperm analyses (Savolainen et al., 2000a; Soltis et al., 2000; Hilu et al., 2003). In three medium-scale analyses focusing on monocots (Chase et al., 2006; Davis et al., 2006; Graham et al., 2006), relationships among Asparagales, Pandanales + Dioscoreales, Liliales, and commelinids were resolved differently, with generally moderate support. However, Asparagales and commelinids were sister to each other in the analyses of Chase et al. (2006) and Graham et al. (2006), with moderate to strong BS support. This relationship seems to be supported by distribution of silica bodies and floral zygomorphy resulting from organ suppression in these two groups (Prychid et al., 2004; Rudall & Bateman, 2004), but some homoplasy in these characters suggest that they may not be synapomorphic. The placement of Asparagales as the sister to other petrosaviid monocots in this study may be an analytical artifact caused by long branches in several alismatalean taxa and insufficient taxon sampling (the critical Petrosaviales were not sampled here).

2.1.5 Basal eudicots  The monophyly of eudicots is strongly supported (Fig. 1). The Ranunculales, Sabiales, Proteales, Trochodendrales, and Buxales form a series of diverging lineages at the base of eudicot phylogeny, with Buxales being sister to the group of eudicots that have been named Gunneridae (Cantino et al., 2007). All of these relationships, except those of Sabiales and Proteales, whose arrangement is effectively unresolved, have strong BS support (Figs. 1, 2-2). These results are similar to those obtained in a matK analysis of angiosperms (Hilu et al., 2003). The combined analyses of atpB/rbcL (Savolainen et al., 2000a), atpB/rbcL/18S rDNA (Soltis et al., 2000), and atpB/matK/rbcL/18S and 26S rDNAs (Burleigh et al., 2009) for angiosperms recovered similar, but less resolved relationships with lower support. The most recently reported chloroplast phylogenomic analysis also showed similar resolution for these relationships, but again the arrangement of Sabiales and Proteales was resolved differently (Moore et al., 2010). Ancient losses of two ribosomal protein genes in the angiosperm mitochondrial genome support the relationships reconstructed here: the loss of rps11 in the common ancestor of Buxales and Gunneridae, and the loss of rps2 in the common ancestor of Trochodendrales, Buxales, and Gunneridae (note, however, that both genes have been lost a few times separately in angiosperms) (Adams et al., 2002).

The Ranunculales are the only group of basal eudicots with substantial living diversity. In this analysis (Fig. 2-2) and an earlier mitochondrial gene analysis with similar sampling of basal eudicots (Qiu et al., 2006a), Menispermaceae were shown to be sister to the rest of the order with moderate support. In contrast, most other molecular phylogenetic studies have shown that Papaveraceae or Eupteleaceae either singly or together were sister to the rest of Ranunculales (Hoot et al., 1999; Qiu et al., 1999, 2005, 2006a; Savolainen et al., 2000a; Soltis et al., 2000; Hilu et al., 2003; Wang et al., 2009b). Gynoecium structure suggests that Menispermaceae and Lardizabalaceae are closely related (Endress & Igersheim, 1999). These families are distantly separated in the mitochondrial gene trees (Fig. 2-2; Qiu et al., 2006a), whereas the results of other molecular phylogenetic studies are more compatible with the relationship suggested by the gynoecium evidence. The long branch leading to Menispermaceae (Fig. 2-2), corroborated by observation of a number of unique mutations in the gene alignment, may have caused misplacement of the family in the mitochondrial gene trees shown here.

2.1.6 Basal gunnerids  Five groups, Gunnerales, Saxifragales, Vitales, Berberidopsidales, and Dilleniales, all except Saxifragales having little taxonomic diversity, form another series of diverging lineages before eudicots differentiate into well supported rosids and asterids. The support for these relationships is generally low (Fig. 2-3). These taxa have also been difficult to place in previous large-scale analyses of angiosperms (Savolainen et al., 2000a; Soltis et al., 2000; Hilu et al., 2003). The Gunnerales diverge first in this series, with low support. Previously, an analysis of atpB, rbcL, and 18S and 26S rDNA sequences from 201 eudicots obtained moderate support for the same placement (Soltis et al., 2003). The Saxifragales, often treated as a quasi-rosid member (Angiosperm Phylogeny Group, 2009), are essentially a member of a trichotomy that also includes an expanded rosid clade and an expanded asterid clade. The BS support for monophyly of Saxifragales is only 49% when the enigmatic genus Peridiscus (Davis & Chase, 2004) is included, but is high (97%) when Peridiscus is excluded. The relationships within this difficult order are fairly well resolved, and are in general agreement with the results of a recent study that sampled 16 chloroplast, mitochondrial, and nuclear genes (Jian et al., 2008) (the rps3 sequence for Paeonia tenuifolia used in that study was found to be a contaminant during the course of this study).

The Vitales are sister to rosids, with only 44% BS support. This result is in agreement with the results of two previous large-scale analyses of angiosperms (Savolainen et al., 2000a; Soltis et al., 2000). The Berberidopsidales and Dilleniales form a weakly supported clade, which is sister to a monophyletic group consisting of Santalales, Caryophyllales, and asterids, but with only 49% BS support. The three previous large-scale analyses of angiosperms with dense taxon sampling have not been able to resolve these relationships (Savolainen et al., 2000a; Soltis et al., 2000; Hilu et al., 2003).

2.1.7 Rosids  The monophyly of rosids has a BS support of 74% (Figs. 1, 2-3). The current rosid concept was formed in the first large-scale analysis of angiosperms using molecular data (Chase et al., 1993). Other analyses of this kind have recovered the monophyly of rosids with various degrees of support, depending on the number and evolutionary rates of the genes sampled: 99% jackknife (JK) value in the atpB/rbcL/18S rDNA analysis (Soltis et al., 2000), 95% JK value in the matK analysis (Hilu et al., 2003), 99% BS value in the atpB/matK/rbcL/18S/26S rDNA analysis (Burleigh et al., 2009), and 61% BS value in the atpB/rbcL analysis (Savolainen et al., 2000a). In an analysis of non-molecular data, the rosids were also clustered together, although the group contained a few non-rosid taxa (Nandi et al., 1998).

Within rosids, Geraniales, Myrtales, and Crossosomatales/Zygophyllales are placed as a series of successively closer outgroups to the remaining rosids, or eurosids (Soltis et al., 2000) (Zygophyllales were among eurosids in that study), all with weak BS support (Fig. 2-3). The eurosids form a monophyletic group with only 43% BS support, and contain three large clades: (i) the nitrogen-fixing clade, as identified in a previous study (Soltis et al., 1995), which includes Fabales, Rosales, Cucurbitales, and Fagales; (ii) the malvid clade (Malvidae of Cantino et al., 2007), first recognized in Chase et al. (1993) as rosid II, which contains Huerteales, Sapindales, Malvales, and Brassicales; and (iii) the COM clade, as termed in two recent studies (Endress & Matthews, 2006; Zhu et al., 2007), which comprises Celastrales, Oxalidales, and Malpighiales (Fig. 2-4). These relationships are in general agreement with the results from previous large-scale analyses of angiosperms or eudicots (Chase et al., 1993; Savolainen et al., 2000a, 2000b; Soltis et al., 2000; Hilu et al., 2003; Burleigh et al., 2009), but with one major exception.

The major result obtained in this study that differs from those of all previous large-scale analyses of angiosperms concerns the monophyly of fabids (Fabidae of Cantino et al., 2007; called rosid I in Chase et al., 1993). The weakly supported COM clade is sister to the strongly supported malvid clade, with 99% BS support (Fig. 2-4). Upon inclusion of Bixa, Elatine, and Populus, which had missing data for two or three of the four genes used in the analysis (Table 1), an 87% BS value was still recovered for this relationship (Fig. S2.1). In all previous large-scale analyses of angiosperms or eudicots, the COM clade was placed as a sister to the nitrogen-fixing clade (Chase et al., 1993; Savolainen et al., 2000a, 2000b; Soltis et al., 2000; Hilu et al., 2003; Burleigh et al., 2009), but never with strong support, the highest being 89% BS in Burleigh et al. (2009). It should be further pointed out that these analyses were based either entirely or mostly on chloroplast genes. Thus far, only one large-scale analysis of angiosperms has been carried out on nuclear gene data, 18S rDNA (Soltis et al., 1997), and in that study, no clearly defined fabids or malvids could be recognized.

Two medium-scale analyses focusing on rosids have been published recently, one using mitochondrial matR, chloroplast atpB and rbcL, and nuclear 18S rDNA (Zhu et al., 2007), and the other sampling 10 chloroplast genes (atpB, matK, ndhF, psbBTNH region, rbcL, rpoC2, and rps4) and two nuclear (18S and 26S) rDNAs (Wang et al., 2009a). In the former, a matR analysis generated 54% maximum likelihood BS support for the COM and malvid sister relationship, and a combined analysis of atpB/rbcL/matR/18S rDNA recovered fabid monophyly, with 70% or 85% BS support in parsimony or likelihood analyses, respectively. In the latter, 100% likelihood BS support was obtained for fabid monophyly. It is perhaps worth pointing out that most of the 10 chloroplast genes used in the latter study are fast-evolving. Although these chloroplast genes are certainly good for resolving rapid radiations, they are also susceptible to accumulation of homoplasious changes at deep nodes. These two medium-scale analyses indicate again that the support for fabid monophyly lies in chloroplast genes, as seen in the large-scale analyses.

One interesting piece of evidence that supports the newly identified sister relationship between the COM and the malvid clades comes from a recent broad survey of floral structural characters (Endress & Matthews, 2006). It was shown that 22 COM families and 18 malvid families share a type of ovule with a thicker inner integument than the outer one, which is otherwise very rare in eudicots (with only one other occurrence, in Trochodendrales). There are some other features that may indicate a close relationship between the COM and the malvid clades: contort petals, and a tendency towards polystemony and polycarpelly (Endress & Matthews, 2006). Notably, that survey did not find any feature supporting the monophyly of fabids.

We also examined our own data to see what type of mutations were behind the strong BS support of the sister relationship between the COM and the malvid clades. A total of six synapomorphic mutations were detected, two in matR (one synonymous and one non-synonymous) and four in rps3 (one synonymous and three non-synonymous), but none in atp1 or nad5. Two of these mutations (one a T→A change, the other T→C) were “perfect” synapomorphies, without any reversal in the identified monophyletic group (the COM–malvid clade) or independent evolution of the same apomorphic state in any other group. The third synapomorphy was a G→A change, which again did not have any reversal within the COM–malvid clade but had one independent evolution of the apomorphic state in Potamogeton, which showed accelerated evolution in matR and all other mitochondrial genes that had been examined (Y.-L. Qiu, unpublished observation, 2010). The other three synapomorphies involved G→A or C→T changes, and also had relatively low levels of homoplasy. The fact that six synapomorphies involved five types of substitutions, four being transitions and one being a transversion, indicated that there was not any special molecular evolution mechanism such as RNA editing or GC content skew that could generate these changes. Furthermore, the extremely low point mutation rates in the plant mitochondrial genome in general (Wolfe et al., 1987; Palmer & Herbon, 1988) and in the specific genes used in this study (Table 3) are most likely to have contributed to the low levels of homoplasy observed here. Indeed, the four mitochondrial genes show significantly (21% on the average) lower levels of homoplasy than chloroplast atpB, matK, and rbcL and nuclear 18S rDNA across angiosperms in our analyses (Table 4). We also examined alignment of the four genes to see if there were any types of mutations that supported the fabid monophyly hypothesis, and we did not find any.

Table 4.  Consistency index (CI) and homoplasy index (HI) values of the eight genes estimated from the matrix of 272 seed plants in parsimony searches under a constraint tree shown in Fig. S1
 mt-atp1mt-matRmt-nad5mt-rps3
  1. cp, chloroplast; mt, mitochondrial; nu, nuclear.

CI:0.2840.4130.3250.334
HI:0.7160.5870.6760.666
 
 cp-atpBcp-matKcp-rbcLnu-18S
 
CI:0.1630.1640.1480.185
HI:0.8370.8360.8520.815

Finally, as we assembled 272 OTU matrices for chloroplast atpB, matK, and rbcL and nuclear 18S rDNA in the rate and homoplasy analyses, we checked these matrices to see what type of mutations supported the fabid monophyly hypothesis, and if there was any evidence in these four genes that would support the sister relationship between the COM and the malvid clades. As a result, we found that in atpB, matK, and rbcL, there were one (G→A), five (two A→G, three T→C), and five (three C→T, one T→C, one G→A) changes, respectively, in favor of the fabid monophyly hypothesis, and two (one T→C, one G→A), two (both C→T), and one (C→T) changes supporting the sister relationship between the COM and the malvid clades. We wish to point out that in comparison to synapomorphic changes in mitochondrial matR and rps3, these chloroplast synapomorphies contained a higher level of homoplasy, and several were extremely homoplasious. This observation is consistent with the results of our homoplasy analyses of these chloroplast genes across angiosperms (Table 4). The nuclear 18S rDNA contained no synapomorphy for either hypothesis.

2.1.8 Santalales and Caryophyllales  Both Santalales and Caryophyllales are strongly supported (99 and 100% BS values, respectively) as monophyletic groups. They and asterids effectively form a trichotomy, as the support for the sister relationship between the two orders is negligibly low (Fig. 2-5). The expansion of Caryophyllales to include several traditionally non-caryophyllalean taxa, such as Nepenthaceae, Droseraceae, Dioncophyllaceae, Ancistrocladaceae, Frankeniaceae, Tamaricaceae, Simmondsiaceae, and Rhabdodendraceae (Albert et al., 1992; Fay et al., 1997), has been supported by previous large-scale analyses of angiosperms (Savolainen et al., 2000a, 2000b; Soltis et al., 2000; Hilu et al., 2003) as well as a cladistic analysis of morphological data (Nandi et al., 1998). In the previous large-scale analyses of angiosperms, Santalales and Caryophyllales have been placed as two successively closer outgroups to (or sometimes merely as close relatives of) asterids, usually with weak support (Soltis et al., 1997, 2000; Hilu et al., 2003; Burleigh et al., 2009). The most recent chloroplast phylogenomic analysis has reconstructed these relationships with strong BS support (Moore et al., 2010). Until recently, however, no clear morphological or other non-molecular data have been found to support these relationships. In a broad survey of floral structural and embryological characters across eudicots, it was noticed that Santalales and Caryophyllales tend to have a relatively thin nucellus in their ovules, thus conforming to asterids (Endress, in press). It has also been suggested that a shift from palmate leaf venation in basal eudicots (Ranunculales, Proteales, Trochodendrales) to pinnate venation seen in Berberidopsidales, Dilleniales, Santalales, Caryophyllales and asterids may represent a synapomorphy of the five latter groups (Doyle, 2007). Within Caryophyllales and Santalales, the relationships resolved here are in general agreement with those reconstructed in two medium-scale multigene analyses focusing on each of the two orders (Cuenoud et al., 2002; Malecot & Nickrent, 2008).

2.1.9 Asterids  The asterids form a monophyletic group with 82% BS support (Figs. 2-5, 2-6). Although the subclass Asteridae was recognized in the pre-molecular systematics era (Cronquist, 1981), it was significantly expanded in the first wave of molecular phylogenetic studies (Downie & Palmer, 1992; Olmstead et al., 1992, 1993; Chase et al., 1993), to include a number of groups that were previously placed in the Rosidae and the now defunct Dilleniidae and Hamamelidae (Cronquist, 1981). All large-scale analyses of angiosperms have obtained strong support for the monophyly of asterids (Savolainen et al., 2000a; Soltis et al., 2000; Hilu et al., 2003; Burleigh et al., 2009).

Within asterids, Ericales, Curtisia, and Cornales are placed as a series of successively closer outgroups to a weakly supported monophyletic group that has been called euasterids (Soltis et al., 2000). Among euasterids, Apiales, Dipsacales, Asterales and several seemingly poorly defined orders (Bruniales, Escalloniales, and Paracryphiales) form a strongly supported monophyletic group, the campanulids (Campanulidae of Cantino et al., 2007). Boraginaceae, Vahliaceae, Solanales, Gentianales, and Lamiales form another major clade, the lamiids (Lamiidae of Cantino et al., 2007), but with only weak support. Oncothecaceae, Icacinaceae, Garryales, and Aquifoliales represent basal euasterid lineages whose relationships to the two large clades are essentially unresolved. These relationships agree fairly well with what have been reconstructed in the previous phylogenetic studies of asterids (Olmstead et al., 1993, 2000; Albach et al., 2001; Bremer et al., 2002). Because three of the four mitochondrial genes (atp1, matR, and nad5) used in this study are slow- or very slow-evolving in comparison to most chloroplast genes used in angiosperm phylogenetic studies, and the fourth gene, rps3, is still not very fast (Table 3), relationships among the major clades within Ericales, campanulids, and lamiids are not well supported. Interestingly, in two studies that sampled several fast-evolving chloroplast genes such as ndhF, matK, and rps16 intron, lack of resolution among these clades was also observed (Bremer et al., 2002; Schonenberger et al., 2005). Hence, the asterids may well represent a case of rapid radiation of a major clade of angiosperms, and the late appearance of sympetalous flowers relative to other floral types in the Late Cretaceous (Friis, 1985; Martinez-Millan, 2010) is consistent with such a hypothesis.

2.2 Mitochondrial genes and angiosperm phylogeny

The rate analyses show that the eight genes are ranked in the following order, from the fastest to the slowest: matK > rbcL > rps3 > 18S rDNA > atpB > matR > atp1 > nad5 (Table 3). The homoplasy analyses show that the four mitochondrial genes show significantly (on average 21%) lower levels of homoplasy than the three chloroplast genes and nuclear 18S rDNA, as measured by the homoplasy index (Table 4). The consistency indexes of the four mitochondrial genes were almost twice as high as those of the other four genes (Table 4). These data provide some theoretical underpinning to support the empirical results presented above and show that for angiosperm-wide phylogeny reconstruction, these mitochondrial genes are at least as good as, if not better than, those four genes that have been used widely. It is not surprising to see the correlation between the evolutionary rate and the homoplasy level, because homoplasy is after all determined by at least four factors: (i) rate (if no variation, no homoplasy); (ii) factors that cause character state changes in similar directions since common ancestry; (iii) limited character evolution space such as that of DNA with only four possible nucleotides (states); and (iv) rate heterogeneity.

Several general results can be summarized here. First, the mitochondrial gene-based angiosperm phylogeny agrees with the chloroplast and nuclear gene-based angiosperm phylogeny to a great extent. This indicates that the hypotheses developed for angiosperm phylogeny thus far are likely to be largely correct. The congruence of the results among these different studies is especially significant, as the several large-scale analyses of angiosperms have used different methods (parsimony or likelihood) for searching optimal trees and also used either BS or JK analyses for evaluating robustness of the topology. Until now, our knowledge of higher level relationships within angiosperms has been derived largely from three chloroplast genes, atpB, matK, and rbcL (Chase et al., 1993; Savolainen et al., 2000a; Soltis et al., 2000; Hilu et al., 2003). The angiosperm phylogeny inferred from the sole nuclear gene, 18S rDNA, was inconclusive on some issues (Soltis et al., 1997). The other nuclear gene that has been used for reconstructing higher level relationships within angiosperms, 26S rDNA, had too much missing data to permit a critical evaluation (Burleigh et al., 2009). Use of non-molecular data for reconstructing an angiosperm-wide phylogeny has only been explored experimentally (Nandi et al., 1998), and much needs to be done after morphological and other non-molecular characters have been surveyed critically across all major groups, as has been done for basal angiosperms (Doyle & Endress, 2000, 2010; Endress & Igersheim, 2000; Endress & Doyle, 2009), basal eudicots (Endress & Igersheim, 1999), monocots (Prychid et al., 2004; Rudall & Bateman, 2004), rosids (Endress & Matthews, 2006), and eudicots (Endress, in press). Although the chloroplast gene-based or gene-dominated studies have significantly improved our understanding of angiosperm phylogeny, it is always desirable and perhaps necessary to obtain information from mitochondrial and nuclear genomes as well as non-molecular sources to reconstruct angiosperm phylogeny. Until such comparison and mutual corroboration is carried out, the phylogenetic hypothesis derived from the chloroplast genome remains a genome phylogeny, and cannot be equated to the organismal phylogeny (Doyle, 1992; Qiu & Palmer, 1999). The results discussed below provide some examples to illustrate this point.

Second, in certain areas where chloroplast gene-based or gene-dominated large-scale analyses of angiosperms failed to resolve or obtained lower support for relationships among member lineages, such as magnoliids, basal eudicots, Saxifragales, and the COM clade of rosids (Savolainen et al., 2000a; Soltis et al., 2000; Hilu et al., 2003), this mitochondrial gene analysis appears to provide more resolution and/or better support. The agreement of these results with those obtained from other studies (Endress & Matthews, 2006; Qiu et al., 2006a; Jian et al., 2008; Burleigh et al., 2009) may be evidence that the results from this mitochondrial gene analysis are likely to be correct.

Third, this study produced different results in several critical areas than the previous chloroplast gene-based or gene-dominated large-scale analyses of angiosperms (Savolainen et al., 2000a; Soltis et al., 2000; Hilu et al., 2003). The most notable example is the newly identified rosid clade that contains the COM and the malvid clades, which is independently supported by some morphological data (Endress & Matthews, 2006). In addition, this study also provided alternative hypotheses that deserve further investigation with respect to the basalmost angiosperm lineage and the placement of Ceratophyllum. This result and the second one discussed above may be attributed to the facts that the mitochondrial genome has a lower overall substitution rate than the chloroplast genome (Wolfe et al., 1987; Palmer & Herbon, 1988), and that the mitochondrial genes used in this study are generally more slowly evolving and less homoplasious than the chloroplast genes that have been used in previous large-scale analyses of angiosperms (Savolainen et al., 2000a; Soltis et al., 2000; Hilu et al., 2003) (Tables 3, 4). These are also the parts of angiosperm phylogeny where reconstruction has been likely hampered by extinction. In this case, the less homoplasious information provided by the more slowly evolving genes may have contributed less conflicting signals within the dataset (Table 4), which is what a statistic resampling procedure like bootstrapping is designed to measure (Felsenstein, 2004).

Finally, several potential problems of mitochondrial genes for phylogenetic use, rate heterogeneity among lineages (Palmer et al., 2000; Cho et al., 2004; Parkinson et al., 2005), horizontal transfer (Bergthorsson et al., 2003; Won & Renner, 2003; Davis & Wurdack, 2004; Mower et al., 2004), and RNA editing (Bowe & dePamphilis, 1996; Petersen et al., 2006; Qiu et al., 2006a), turned out not to have serious negative effects on phylogenetic reconstruction of an angiosperm-wide phylogeny, probably for the following reasons. For rate heterogeneity, only a small number of lineages showed significant rate acceleration for the four genes sequenced here: Metasequoia, Podocarpus, Acorus, Alismatales, Geraniaceae, Urticaceae, Viscum, Phaulothamnus, Asteropeia, Galenia, Phlox, Phryma, Plantaginaceae, Hydrolea, and Goodeniaceae. Some of these were already detected in a previous large-scale Southern hybridization survey of mitochondrial genes and introns throughout land plants (Qiu et al., 1998; Adams et al., 2002; Qiu & Palmer, 2004). This problem was dealt with by either choosing a different member of the group or omitting the problematic taxa. The extent of horizontal gene transfer has probably been over-stated, at least in non-parasitic plants, as only one such case was encountered in this study. The parasitic plant Cynomorium songaricum, depending on the genes analyzed, was placed with either Saxifragales (matR and nad5) or Sapindales (atp1 and rps3). It was thus not included in the final analyses. Previously, this genus has been suggested to be related to Saxifragales in an analysis of matR (Nickrent et al., 2005) or Rosales in a study of the inverted repeat region of the chloroplast genome (Zhang et al., 2009). The level and distribution across lineages of RNA editing are clearly not high or widespread enough to have affected phylogenetic reconstruction in this analysis (Fig. S3), and this result is consistent with what was found in an earlier smaller-scale analysis of 162 seed plants (Qiu et al., 2006a). Other characteristics of plant mitochondrial DNA suggested previously that might affect its performance in phylogenetic analysis, such as slow rate and presence of introns (Palmer, 1992), can be exploited from different angles for phylogenetic uses, as plant molecular systematists become more skillful and knowledgeable of molecular techniques and the plant mitochondrial genome (Qiu et al., 1998, 2006b; Qiu & Palmer, 2004; Barkman et al., 2007; Ran et al., 2009; Wurdack & Davis, 2009). The slow rate, in fact, turns out to be a merit of plant mitochondrial DNA when it is used for resolving ancient phylogenetic relationships, as it contains less homoplasy. Therefore, the mitochondrial genome has great potential for investigating phylogenetic relationships of angiosperms as well as non-flowering land plants.

3 Conclusions and future prospects

An angiosperm phylogeny was reconstructed with information extracted from nucleotide sequence variation of four slowly evolving mitochondrial genes, atp1, matR, nad5, and rps3. It is largely congruent with the phylogeny of angiosperms that have been reconstructed from analyzing the chloroplast genes atpB, matK, rbcL, and nuclear 18S rDNA (Chase et al., 1993; Soltis et al., 1997, 2000; Savolainen et al., 2000a; Hilu et al., 2003). The most prominent difference is that the COM clade is sister to the malvid clade, instead of to the nitrogen-fixing clade. This relationship is not only supported by highly conservative mutations identified in this study, but also independently corroborated by one embryological character and perhaps a few other floral structural and embryological characters (Endress & Matthews, 2006). Hence, the long recognized monophyly of fabids (Chase et al., 1993) needs to be re-evaluated with evidence from all three plant genomes and non-molecular data. Other major differences between the results of this study and those of the earlier large-scale analyses of angiosperms include placement of Amborella, Hydatellaceae, and Nymphaeales together as the clade sister to all other angiosperms, and Ceratophyllum as the sister group of Chloranthaceae. This study shows that mitochondrial genes are informative markers for resolving relationships among genera, families, or higher rank taxa across angiosperms. Their slow evolutionary rates are particularly beneficial for reconstructing ancient phylogenetic relationships, as they have been shown to be less homoplasious than typically faster-evolving chloroplast genes (Table 4). Several potential problems of mitochondrial genes such as rate heterogeneity, horizontal transfer, and RNA editing have been somewhat exaggerated, and can be effectively dealt with by selective taxon sampling and analysis of combined multigene datasets. Otherwise, one would not expect such a high level of congruence between the angiosperm phylogeny based on the mitochondrial genes and those based on the three chloroplast genes and nuclear 18S rDNA.

This study also shows that, despite the tremendous progress in our understanding of angiosperm phylogeny made through analyses of chloroplast genes, it is essential to develop independent hypotheses on angiosperm phylogeny using information from mitochondrial and nuclear genes as well as non-molecular data, so that the underlying organismal phylogeny, not just the phylogeny of a single genome, is reconstructed. The case of rosids uncovered in this study serves as a sober reminder that it is better not to rely on just a few genes to reconstruct phylogenetic relationships among major clades even in large-scale taxon dense analyses. Because nuclear genes often experience duplication over a long evolutionary time in large taxonomic groups, such as angiosperms and land plants, mitochondrial genes may be more suitable for developing hypotheses for large-scale phylogenies. The low substitution rates and low levels of homoplasy of the mitochondrial genes, relative to most chloroplast genes, make them particularly useful for reconstructing ancient phylogenetic relationships, and several studies have shown that the efforts of exploiting mitochondrial genes have produced insightful results (Beckert et al., 1999, 2001; Qiu et al., 1999, 2006a, 2006b, 2007; Vangerow et al., 1999; Bowe et al., 2000; Chaw et al., 2000; Wikstrom & Pryer, 2005; Barkman et al., 2007; Wurdack & Davis, 2009). More mitochondrial markers are thus needed in addition to the four genes used here. A preliminary analysis shows that there are at least four other mitochondrial genes that show variation at the levels of atp1 and matR: atp6 (ATPase subunit 6, 0.8 kb), ccmFN (cytochrome c biogenesis FN, 1.2 kb), cox3 (cytochrome c oxidase subunit 3, 0.8 kb), and mttB (transport membrane protein, also called tatC, 0.8 kb) (Y.-L. Qiu, unpublished data, 2010). These genes are suitable for reconstructing phylogenies of both angiosperms and non-flowering land plants, as they are widely present throughout land plants (Li et al., 2009). Thus, it is entirely feasible that, in the near future, a well resolved and robustly supported mitochondrial gene-based phylogeny for angiosperms or land plants could be developed, and compared with a chloroplast gene-based phylogeny in order to reconstruct the underlying organismal phylogeny.

Acknowledgments

Acknowledgements  We thank James A. DOYLE, Peter K. ENDRESS, and Jeffrey D. PALMER for discussion, Douglas E. SOLTIS and Pamela S. SOLTIS for leading the Angiosperm Tree of Life project, Mark A. MILLER at San Diego Super Computer for help with data analyses, Jeremy J. BRUHL, Mark W. CHASE, Laszlo CSIBA, Chris W. DOCK, Peter K. ENDRESS, Sean W. GRAHAM, Brett HALL, Khidir W. HILU, Richard W. JOBSON, Kathleen A. KRON, Clifford R. PARKS, Anton A. REZNICEK, Douglas E. SOLTIS, Pamela S. SOLTIS, Jan J. WIERINGA, Kenneth WURDACK, Qiu-Yun Jenny XIANG, and the South African National Biodiversity Institute for DNA samples or plant materials, and Noor AL-BASSAM, Youyou DUANMU, Hong-Qiang HAN, Pete L. HAYNES, Mona L. VEKARIA, and Adam M. WHITE for technical assistance. This work was supported by a National Science Foundation Assembly the Tree of Life grant (DEB 0431239) to YLQ.

Ancillary