UC Davis UC Davis Previously Published Works Title Recognising angiosperm clades in the Early Cretaceous fossil record Permalink https://escholarship.org/uc/item/2td719gp Journal Historical Biology, 27(3-4) ISSN 0891-2963 Author Doyle, JA Publication Date 2015-04-03 DOI 10.1080/08912963.2014.938235 Peer reviewed eScholarship.org Powered by the California Digital Library University of California This article was downloaded by: [Professor James Doyle] On: 19 December 2014, At: 11:56 Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK Historical Biology: An International Journal of Paleobiology Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/ghbi20 Recognising angiosperm clades in the Early Cretaceous fossil record James A. Doylea a Department of Evolution and Ecology, University of California, Davis, CA 95616, USA Published online: 17 Dec 2014. Click for updates To cite this article: James A. Doyle (2014): Recognising angiosperm clades in the Early Cretaceous fossil record, Historical Biology: An International Journal of Paleobiology, DOI: 10.1080/08912963.2014.938235 To link to this article: http://dx.doi.org/10.1080/08912963.2014.938235 PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http:// www.tandfonline.com/page/terms-and-conditions HistoricalBiology,2014 http://dx.doi.org/10.1080/08912963.2014.938235 Recognising angiosperm clades in the Early Cretaceous fossil record James A.Doyle* DepartmentofEvolutionandEcology,UniversityofCalifornia,Davis,CA95616,USA (Received29November2013;accepted21June2014) StudiesoftheearliestCretaceousangiospermsinthe1970smadeonlybroadcomparisonswithlivingtaxa,butdiscoveries of fossil flowers and increasingly robust molecular phylogenies of living angiosperms allow more secure recognition of extantclades.ThemiddletolateAlbianriseoftricolpatepollenandthefirstlocaldominanceofangiospermleavesmarkthe influxofnear-basallinesofeudicots.Associatedflowersindicatethatpalmatelylobed‘platanoids’andSapindopsisareboth stem relatives of Platanus, while Nelumbites was related to Nelumbo (also Proteales) and Spanomera to Buxaceae. MonocotsareattestedbyAptianLiliaciditespollenandAcaciaephyllumleavesandAlbianaraceousinflorescences.Several Albian–Cenomanian fossils belong to Magnoliidae in the revised monophyletic sense, including Archaeanthus in MagnolialesandVirginianthusandMauldiniainLaurales,whilelateBarremianpollentetrads(Walkeripollis)arerelatedto 4 Winteraceae. In the basal ANITA grade, Nymphaeales are represented by Aptian and Albian flowers and whole plants 1 0 (Monetianthus,CarpestellaandPluricarpellatia).EpidermalsimilaritiesoflowerPotomacleavestowoodymembersofthe 2 r ANITA grade are consistent with Albian flowers assignable to Austrobaileyales (Anacostia). Aptian to Cenomanian e b mesofossilsrepresent bothcrown groupChloranthaceae (Asteropollisplant)and stemrelativesofChloranthaceae and/or m e Ceratophyllum(Canrightia,Zlatkocarpus,PennipollisplantandpossiblyAppomattoxia). c e D Keywords: angiosperms;Cretaceous;palaeobotany;palynology;phylogeny 9 1 6 5 1: 1. Introduction modern plants, it was taken by Doyle (1969) and Muller 1 at The past five decades have seen a dramatic reawakening (1970)asevidencethatthemainradiationofangiosperms e] ofinterestintheearlyfossilrecordofangiosperms,which was going on during this interval, and as support for yl o with some notable exceptions (e.g. Teixeira 1948; existing ideas on the evolution of angiosperm pollen and D s Vakhrameev 1952) had been neglected following the the related ‘magnoliid’ theory of angiosperm evolution. e m first studies of Early Cretaceous floras around the turn of This was considered evidence against the older view that a r J the last century. A major question in this field is what angiosperms originated and began to diversify much so living clades are represented among the oldest unques- earlier and were simply invading lowland basins of s fe tioned angiosperms. Early palaeobotanists such as deposition in the Early Cretaceous (e.g. Axelrod 1952, o Pr Fontaine (1889) and Berry (1911), working on the 1970). [ y Potomac Group of the Atlantic Coastal Plain of the In the 1970s, reinvestigations of the Potomac Group b d United States (Aptian to early Cenomanian, Virginia and angiosperm leaf record (Wolfe et al. 1975; Doyle and e d Maryland),didnothesitatetoassignangiospermleavesto Hickey 1976; Hickey and Doyle 1977) showed that early a nlo living genera, such as Ficus or Sassafras, or to fossil angiospermleaves were alsolessmodern than previously w genera with names intended to suggest affinities with thought and showed a pattern of temporal diversification o D living taxa, such as Ficophyllum, Proteaephyllum, similar to that seen in the pollen record (Figure 1). In Vitiphyllum and Sapindopsis. However, serious doubt general, Doyle and Hickey (1976) avoided comparing was cast on these identifications in the 1960s and 1970s. these leaves with living taxa, except to show that ThiswaspartlyduetopalynologicalstudiesonCretaceous morphological features seen in the fossil leaves occur in rocks (e.g. Brenner 1963; Pacltova´ 1966; Muller 1968), livingtaxathathavepollentypesknownatthesamelevel, which showed that the first recognisable angiosperm but Hickey and Doyle (1977) did make some broad pollenwasmuchlessdiverseandmodernthanimpliedby comparisonswithmajorangiospermgroups.Forexample, the old leaf identifications and showed a progressive palmately lobed ‘platanoid’ fossils from the middle and increase in the diversity of morphological types from the late Albian (Figure 1(p)) resemble leaves of Platanus, Barremian to the Cenomanian – first monosulcate, then which has reticulate tricolpate pollen ofthetype thatwas tricolpate, tricolporate and finally triporate (Figure 1). As becoming abundant at that time. In the lower Potomac this pattern paralleled the sequence of pollen evolution (Aptian?),thereareleaveswithmonocot-likevenationand inferred from comparative morphological studies of monosulcate pollen with monocot-like sculpture (Doyle *Email:[email protected] q2014Taylor&Francis 2 J.A. Doyle 4 1 0 2 r e mb Figure1. StratigraphicsequenceofmajorangiospermpollenandleaftypesinthePotomacGroupofeasternNorthAmerica(modified e fromDoyleandHickey1976),withcorrelationsofplantlocalitiesinothergeographicareasandstratigraphicpositionsofotherpollenand c e floralfossils. D 9 1 6 5 1973; Figure 1(b,g)). Wolfe et al. (1975), Doyle and the results have held up remarkably well with studies of 1: 1 Hickey (1976) and Hickey and Doyle (1977) argued that many more genes (Soltis et al. 2005; Jansen et al. 2007; at mostoftheearliestleaveshadfeaturesfoundinwhatwere Moore et al. 2007; APG III 2009). These analyses have ] yle then called ‘magnoliids’, in the paraphyletic sense of resolvedmanylong-standingproblems,suchaswherethe o Takhtajan (1969) and others, meaning those ‘dicots’ that angiosperm phylogenetic tree is ‘rooted’, or which living D s retain monosulcate and related pollen types. These lines branched off first at the base of the tree: the New e m included several leaf types (e.g. Ficophyllum; Figure 1 Caledonian genus Amborella, Nymphaeales (water lilies) a r J (l)) with pinnate venation, like ‘woody magnoliids’ (now andAustrobaileyales, allclassifiedasmagnoliidsinolder o s assigned to Magnoliales, Laurales, Canellales and more systems, informally designated the ‘ANITA grade’. The s e f basal groups such as Amborella and Austrobaileyales), remaining 99.9% of angiosperm species, named Mesan- o r P and a few with palmate venation (e.g. Proteaephyllum giospermae (Cantino et al. 2007), belong to five major [ y reniforme;Figure1(i)),like‘herbaceousmagnoliids’(now clades: eudicots, including all groups with tricolpate and b d placed inNymphaealesand Piperales). derivedtypesofpollen;monocots;acladeincludingmost e ad Since 1977, there have been several major advances traditional magnoliids, designated Magnoliidae in a new o nl that allow more secure identification of early angiosperm sense (Cantino et al. 2007); the rootless aquatic genus w o lines. First was the discovery of fossil flowers, preserved Ceratophyllum;andChloranthaceae,whicharecharacter- D ascharcoalorlignite,byFriis,Craneandothers,working ised by unusually simple flowers. All five clades and first on the Potomac Group and later on Cretaceous orders within them have strong statistical support, as sedimentsinPortugal(Craneetal.1986;Friisetal.1986, measured by bootstrap analysis, but relationships among 2011;Friis, Pedersen, etal. 1994).Many ofthese flowers them are less strongly supported, and they vary from are in the millimetre or ‘mesofossil’ size range. Not only analysistoanalysis (Soltis et al. 2005;Qiu et al. 2010). do flowers usually have more characters than pollen and Recent studies have attempted to clarify the phyloge- leaves, they can often be connected with the dispersed netic position of early angiosperms by integrating pollen record through pollen in stamens or adhering to mesofossil and molecular evidence through phylogenetic stigmas.Secondwasthedevelopmentofcladisticmethods analysis. The ideal procedure might be to combine for the analysis of phylogenetic relationships, and third morphological and molecular data in a ‘total evidence’ was the applicationof these and related methods toDNA analysis, with fossils scored as unknown for molecular sequence data, which has resulted in a vast flood of new characters, but this has not been attempted because of information on relationships among living angiosperms. problemsinrepresentationoflivingtaxainmolecularand Since the first large-scale analyses of sequences from morphological data-sets (by species and by a mixture of several genes combined (Mathews and Donoghue 1999; species and higher taxa, respectively) and choice of Parkinson et al. 1999;Qiu et al. 1999;Soltis et al. 1999), molecular data. Instead, Doyle et al. (2008), Endress and Historical Biology 3 Doyle (2009) and Doyle and Endress (2010, 2014) have phyllum is linked with eudicots and Chloranthaceae with useda‘molecularscaffold’approach(Springeretal.2001; the magnoliids. In the D&E backbone tree, based on Hermsen and Hendricks 2008), where a morphological combined analyses of morphological and molecular data data-set of living and fossil taxa is analysed with the by Doyle and Endress (2000) and Endress and Doyle arrangementoflivingtaxafixedtoa‘backbonetree’based (2009),CeratophyllumandChloranthaceaearelinkedwith onmoleculardata.Inthecaseofangiosperms,thismaybe each other and sister to the remaining mesangiosperms. an acceptable approach, as many relationships are so The Ceratophyllum–Chloranthaceae clade has also been strongly supportedby molecular data that they are highly found in analyses of mitochondrial genes (Duvall et al. unlikely to be overturned by morphological data from 2006;Qiu et al.2010);slowly evolvingchloroplastgenes fossils. Variously modified versions of the same (Moore et al. 2011), which should be most reliable for morphologicaldata-sethavebeenusedinsimilaranalyses detecting ancient splits; and low-copy nuclear genes by other authors (von Balthazar et al. 2008; Friis et al. (Zhang et al. 2012). 2009;Friis and Pedersen 2011;Mohr et al. 2013). The purpose of the present article is to review the Because the arrangement of the five mesangiosperm resultsoftheseanalyses,beginningnotwiththeoldesttaxa cladesismoreuncertain,thesestudiesusedtwobackbone but with eudicots from the middle and late Albian, for trees that cover the range of current hypotheses, which which the results are clearest, and ending with taxa that 4 differ most significantly in the position of Ceratophyllum pose more problems but are equally or more interesting. 1 20 and Chloranthaceae. In the J/M backbone tree, based on Mostparsimoniouspositions ofmostofthe fossilson the er analyses of nearly complete chloroplast genomes by D&EbackbonetreeareshowninFigure2;whereafossil b m Jansen et al. (2007) and Moore et al. (2007), Cerato- has several most parsimonious positions, one has been e c e D 9 1 6 5 1: 1 at ] e yl o D s e m a J r o s s e f o r P [ y b d e d a o nl w o D Figure2. MostparsimoniouspositionsofEarlyCretaceousandCenomanianfossilsonaphylogenyoflivingangiosperms(Doyleetal. 2008; Endress and Doyle 2009; Friis et al. 2009, 2011; Doyle and Endress 2010, 2014). When a fossil has more than one most parsimoniousposition,oneofthesehasbeenselectedforpurposesofdiscussion.CouperitesandArchaefructusnotincludedbecauseof their excessively uncertain phylogenetic positions. Nymph, Nymphaeales; Aust, Austrobaileyales; Chlor, Chloranthaceae; Piper, Piperales;Ca,Canellales;Magnol,Magnoliales;plat,platanoid. 4 J.A. Doyle selectedfor purposesofdiscussion.Agesof the localities shape of the leaf blade, which is like that of Nelumbo in containing these fossils are discussed critically in Doyle varyingfromroundtowiderthanlong,whereastheblade etal.(2008),DoyleandEndress(2010,2014)andMassoni in Brasenia is longer than wide (cf. Wang and Dilcher etal.(inpress).Manyoftheseresultsarealsodiscussedin 2006). DoyleandUpchurch(inpress),butwithspecialemphasis Significantly, the two most common Potomac on progress in understanding the Potomac Group leaf angiosperm leaf types are related to another member of record. Proteales,namelyPlatanus(¼ Platanaceae,sycamoreor plane tree). These include the palmately lobed platanoid leaves already mentioned (Figure 1(p)), which are a 2. Eudicots conspicuousfeatureoflateAlbianandCenomanianfloras around the Northern Hemisphere, including the classic One of the first major findings of molecular systematics Dakota flora of Kansas (Lesquereux 1892) and similar was that plants with tricolpate and derived pollen types, floras in Kazakhstan (Vakhrameev 1952). Their platanac- which include nearly 75% of angiosperm species, form a eous affinity was suspected based on similarities in clade called eudicots (Doyle and Hotton 1991; Magallo´n venation and co-occurrence with inflorescence axes etal.1999;Cantinoetal.2007).Thisconclusioncontrasts bearing Platanus-like heads (e.g. Hickey and Doyle withtheolderviewthatsuchplantswerepolyphyletic(e.g. 4 1977).Subsequently,Craneetal.(1986),Friisetal.(1988) 1 Takhtajan 1969) but was anticipated by some morpho- 20 andPedersenetal.(1994)showedthattheheadswerelike er logical cladistic analyses (Dahlgren and Bremer 1985; those of Platanus in consisting of unisexual flowers with b Donoghue and Doyle 1989). This means that the oldest m peltate stamens and free carpels, but apparently more e known tricolpate pollen, for example from the late c primitive (most notably in having a better-developed De BarremianofEngland(HughesandMcDougall1990)and perianth). A more surprising result was the discovery of 9 Africa(Doyleetal.1977;Doyle1992),isevidenceforthe 1 Crane et al. (1993) that pinnately dissected (nearly 6 existence of this clade or the stem lineage leading to it. 5 compound) leaves known as Sapindopsis (Figure 1(o)) 1: Furthermore, studies of fossil flowers confirm and extend 1 were from plants with heads made up of similar flowers e] at eanargliioesrpesrumggleesatifontyspetshaitnmthaenymiodfdletheandfirsltatecoAmlmbioann (Figure3(b)).TherehadbeensuggestionsthatSapindopsis yl and the platanoids were related, based on similarities in o belongtolinesthatdivergenearthebaseoftheeudicotsin D leaf architecture (Hickey and Wolfe 1975; Doyle and s molecular trees, below the vast bulk of eudicots, which Hickey1976)andcuticleanatomy(Upchurch1984a)and e m typically have pentamerous flowers and well-differen- thepresenceofheadsinthesamebeds(HickeyandDoyle a J tiatedsepalsandpetalsandhavebeencalledcoreeudicots r 1977).However,HickeyandWolfe(1975)andHickeyand o orPentapetalae (Cantino et al. 2007). ss Doyle (1977) had compared Sapindopsis with Rosidae, e A widespread Albian leaf type is Nelumbites of where pinnatelycompoundleavesaremostcommon,and Pr (Figures 1(n) and 3(a)), assigned by some earlier authors argued that this supported a common origin of the now- [ y to Menispermites, which resembled leaves of modern defunct subclass Hamamelididae (to which Platanus was d b Nelumbo (¼ Nelumbonaceae, lotus) inbeingpeltate and assigned) and Rosidae. This hypothesis would be e d palmatelyveinedbutdifferedinbeingsmallerandhaving a inconsistent with the finding that Sapindopsis had heads nlo less regular venation (Berry 1911; Vakhrameev 1952; ofunisexualflowers,whichwouldnotbeexpectedonthe w Hickey and Doyle 1977; Upchurch et al. 1994). In many o lineleadingtoRosidae.TheanalysisofDoyleandEndress D pre-molecular classifications, Nelumbo was placed in (2010) strongly confirmed that both Sapindopsis and the Nymphaeales,becauseofitsaquatichabitandmultiparted platanoidswereextinctrelativesofthePlatanus,eitheras flowers, but it has tricolpate pollen, and molecular data two successive outgroups (with Sapindopsis diverging indicate that it belongs to Proteales, a clade that diverges first) ora sister clade. oneortwonodesabovethebaseoftheeudicots.Upchurch According to molecular analyses, the sister group of etal.(1994)reportedmouldsoffloralreceptaclesinabed Platanus is the Southern Hemisphere family Proteaceae, dominatedbyNelumbitesthatresembledthoseofNelumbo which has triporate pollen of a type that appears in the in being covered with pits for the carpels but differed in Cenomanian of Africa and Brazil (Triorites africaensis: beingroundratherthanflat-topped.AnanalysisbyDoyle Jardine´ and Magloire 1965; Ward and Doyle 1994; and Endress (2010; Figure 2) confirmed that Nelumbites Dettmann and Jarzen 1996; Sauquet et al. 2009). As the wasmostcloselyrelatedtoNelumbo,butitwasalmostas twootherlinesinProtealesoccurinthemiddleAlbian,the parsimonious to place it in Nymphaeales with Brasenia, treeimpliesthatthelineleadingtoProteaceaealsoexisted whichalsohaspeltateleaves.Thisuncertaintyislargelya by thistime, butpresumablyit still hadtricolpatepollen. function of the small number of characters preserved. A Another eudicot line is represented by Spanomera, characternotusedbyDoyleandEndress(2010)thatwould from the late Albian and early Cenomanian of Maryland favour a relationship of Nelumbites to Nelumbo is the (Drinnanetal.1991),whichhadreducedunisexualflowers Historical Biology 5 4 1 0 2 r e b m e c e D 9 1 6 5 1: 1 at ] e yl o D s e m a Figure 3. Cretaceous angiosperm fossils discussed in the text (images reproduced with permission from the publications cited). (a) J r Nelumbites extenuinervis, leaf (Doyle and Hickey 1976); (b) Friisicarpus brookensis, female flowers associated with Sapindopsis o s variabilis leaves (Friis et al. 2011); (c) Spanomera mauldinensis, reconstruction of inflorescence with lateral male flower above and s e terminalfemaleflowerbelow(Friisetal.2011);(d)Archaeanthuslinnenbergeri,reconstructionofleafandflowerinfruitstage(Dilcher f ro andCrane1984);(e)Virginianthuscalycanthoides,flower(Friisetal.2011);(f)Mauldiniamirabilis,reconstructionofinflorescenceunit P [ withfiveflowers(Friisetal.2011);(g)Walkeripollisgabonensis,pollentetrad,SEM(Doyleetal.1990);(h)Monetianthusmirus,flower by (Friisetal.2011);(i)Pluricarpellatiapeltata,rhizomewithroots,leavesandflowerinfruitstage(Friisetal.2011);(j)Anacostiateixeiri, d carpelwithenclosedseedandbrokenportionofseedcoat,SEM(Friisetal.2011);(k)Asteropollisplant,femaleflower(Friisetal.2011); e d (l)Canrightiaresinifera,reconstructionofflower(Friisetal.2011);(m)Pennipollisplant:Pennistemonportugallicus,fragmentofmale a o inflorescence,andPennicarpustenuis,carpel(Friisetal.2011);(n)Appomattoxiaancistrophora,carpelandseed(Friisetal.2011). nl w o D grouped into bisexual inflorescences (Figure 3(c)), as in DissectedleavessuchasVitiphyllum(Figure1(h))and livingBuxaceae,andstriatetricolpatepollen(Striatopollis Potomacapnos (Jud and Hickey 2013) from the lower paraneusandStriatopollisvermimurus).Similarbutmore Potomac Group (Zone I, Aptian to early Albian?), fragmentary remains are known from the early Albian of LeefructusfromtheBarremian–AptianYixianFormation Portugal(Pedersenetal.2007).TheanalysisofDoyleand of NE China (Sun et al. 2011) and morphotype ‘LC- Endress (2010) confirmed that Spanomera was related to Microphyll trifoliate’ from the Aptian of Argentina Buxaceae, although it could almost as well be located a (Puebla 2009) are suggestive of Ranunculales, the sister node lower. These results also fit the view that the oldest groupofallothereudicots,intheirternateorganisation,a known eudicots belonged to lines that diverged near the featureofmanymembersofthisorder(HickeyandWolfe beginningoftheradiationoftheclade.Spanomerahasnot 1975; Doyle 2007). However, similar but more finally beenassociatedwithleaves,butbyplottingleafcharacters dissected leaves also occur in Archaefructus, an aquatic in the Doyle and Endress (2010) data-set on the tree, we plant of highly debated affinities from the Yixian canpredictthatithadovateleaveswithpalmatevenation Formation (Sun et al. 1998, 2002; Friis et al. 2003; and chloranthoid teeth, like ‘Populus’ potomacensis Endress and Doyle 2009), discussed further below. The (Figure 1(m)) in the middle–late Albian part of the fact that Leefructus and the Argentine leaves (though not Potomac Group. Vitiphyllum) predate the appearance of definite tricolpate 6 J.A. Doyle pollen in their geographic areas suggests the possibility representative of Araceae is Spixianthus, based on aroid- that they represent either stem relatives of eudicots or like leaves and a stem bearing roots and leaves from the some unrelated extinct group(s), as discussed by Doyle late Aptian Crato Formation of Brazil (Coiffard et al. (2012). Parsimony analysis implies that dissected leaves 2013b). arose within the eudicots (Doyle 2007), but a likelihood- based analysis by Geeta et al. (2012) indicated that they were mostlikely ancestral in the clade. 4. Magnoliids Core eudicots or Pentapetalae, with basically penta- All groups of angiosperms other than eudicots and merous flowers, are probably represented among the monocots were formerly classified as magnoliids (e.g. tricolporatepollentypesthatappearandbegintodiversify Takhtajan 1969), a name no longer used in this sense inthelateAlbian.Suchpollenappearstobebasicforthe becauseitwoulddesignateaparaphyleticgrade.However, two main subclades of Pentapetalae, Rosidae and moleculardataindicatethatmostoftheseplantsdobelong Asteridae,althoughbothtricolpateandtricolporatepollen to one clade, called Magnoliidae in a revised sense occurs in some of the more basal lines, such as (Cantino et al. 2007), which includes Magnoliales, Saxifragales and Caryophyllales (Doyle 2005; Furness Laurales,CanellalesandPiperales.Anincreasingnumber etal.2007).Atypical pentamerous flowerfromthelatest of these groups have been recognised in the Cretaceous 4 1 Albian Rose Creek locality in the Dakota Formation of floral record, implying that they were actively radiating 0 r 2 Nebraska(BasingerandDilcher1984)probablybelongsto alongsideearly eudicotsand monocots. be crown group Pentapetalae, but it has not been analysed Several mid-Cretaceous fossils appear to represent m e phylogenetically. Magnoliales. The most securely placed of these is c De Archaeanthus (Figure 3(d)), from the latest Albian or 9 earliest Cenomanian of the Dakota Formation of Kansas 1 6 3. Monocots (Dilcheretal.1976;DilcherandCrane1984).Thisplantis 5 1: The next-largest mesangiosperm clade is the monocots. known from isolated leaves with a bilobed apex, bracts, 1 at Two kinds of fossils from the lower Potomac Group tepals,andfloralaxesbearingnumerouscarpelsandscars e] (Aptian)havebeeninterpretedasmonocots(Doyle1973; that correspond in size and position to the numerous yl o HickeyandDoyle1977;WalkerandWalker1984):boat- stamens,threewhorlsofperianthpartsandsheathingfloral D s shaped monosulcate pollen of the Liliacidites type bractinMagnoliaceae.TheanalysisofDoyleandEndress e m (Figure 1(b)), with ‘graded’ reticulate sculpture that is (2010) strongly associated Archaeanthus with Magnolia- a J finerattheendsofthegrainandcoarserinthemiddle;and ceae, either attached to the stem lineage of the family or r o Acaciaephyllum(Figure1(g)),bestknownasaleafystem nested within it, linked with one or the other of its two s es bearingnarrowleaveswithapicallyfusingvenation.These subgroups (Liriodendron and Magnolioideae¼Magnolia f o Pr fossils were questioned as monocots by Gandolfo et al. s.l.). A recent analysis by Romanov and Dilcher (2013) [ y (2000), in part because of cases where the name placedArchaeanthusinthecrowngroupofMagnoliaceae, b d Liliacidites was used for pollen without the typical asthesistergroupofLiriodendron.However,thisanalysis e d sculpture pattern and leaf architectural similarities treatedlobationoftheleafapexandtheleafbladeastwo a nlo between Acaciaephyllum and living and fossil Gnetales, characters,whentheyrepresentonlyonechangefromthe w but their relationship to monocots was reaffirmed by ancestral unlobed leaf, and it incorrectly identified the o D phylogeneticanalysesofDoyleetal.(2008).Sculptureof whorled perianth phyllotaxis of Archaeanthus and the Liliacidites type is restricted to monocots today Liriodendron as a synapomorphy because it used the (although far from universal among them), and the distantly related genus Illicium (Austrobaileyales) as combination of spiral phyllotaxis, narrowblade, presence outgroup rather than other Magnoliales, most of which ofamedianvein,andapicallyfusing,low-anglesecondary also have awhorled perianth (Massoni et al. inpress). veins distinguishes Acaciaephyllum both from other Older fossils that probably belong to Magnoliales but angiospermswithpalmatevenationandfromsuperficially arelesspreciselylocatedwithintheorderareEndressinia similarleavesin Gnetales. (Mohr and Bernardes-de-Oliveira 2004) and Schenker- Although not yet analysed phylogenetically, inflor- iphyllum (Mohr et al. 2013) from the late Aptian Crato escencesofsessile,ebracteateanddimerousflowersfrom flora of NE Brazil, both represented bystems with leaves the probable early Albian of Portugal (Friis et al. 2010b; and incompletely preserved flowers. Doyle and Endress Figure 1) provide evidence that monocots had begun to (2010) placed Endressinia below or within the clade diversifybythistime,becausethefloralfeatureslistedare consisting of Degeneria, Galbulimima, Eupomatia and apomorphiesofAraceae,intheAlismatales,anorderthat Annonaceae,supportedinpartbysimilarinnerstaminodes diverges one node above the base of the monocots (after in these taxa, but an analysis by Mohr et al. (2013) Acorus) in molecular trees. Another probable early associated Endressinia and Schenkeriphyllum with each Historical Biology 7 other as the sister group of Magnoliaceae, based on their wood, and the small tropical family Canellaceae. This sheathingleafbasesanddryfruit.Theseconflictingresults clade is represented by tetrads of monoporate grains that have not been resolved, but the two studies agree in closely resemble pollen of living Winteraceae in their indicatingtheexistenceofcrowngroupMagnolialesinthe apertureandexinestructurebuthavefinersculpture,from late Aptian. Two Potomac Group fossils that have been the late Barremian to Cenomanian of Gabon, Israel and compared with Magnoliales are large, boat-shaped Argentina(Walkeretal.1983;Doyleetal.1990;Barreda monosulcate pollen grains with granular infratectal andArchangelsky2006;Schrank2013).Thephylogenetic structure (Lethomasites, Aptian: Ward et al. 1989) and analysis of Doyle and Endress (2010) supported the laminar stamens containing smooth, boat-shaped mono- interpretation of the oldest pollen (Walkeripollis gabo- sulcate pollen (dispersed stamen type 1, Albian: Crane nensis; Figure 3(g)) as representing stem relatives of et al. 1994), but these comparisons are speculative given Winteraceae. These Early Cretaceous fossils were more the small number of characters involved. tropicalintheirdistributionthanLateCretaceous,Tertiary, Laurales are represented in the Albian–Cenomanian and living Winteraceae, suggesting a later shift from byseveralfloraltypes,whichcorrespondtobothrelatively tropical totemperateclimatic preferences (Doyle 2000). plesiomorphic and more derived morphologies in the SofarnoEarlyCretaceousfossilshavebeendescribed order. Virginianthus, from the middle Albian of the that clearly represent the fourth magnoliid order, 14 Potomac Group (Figure 3(e)), was assigned by Friis, Piperales. Friis et al. (1995, 2011) suggested that Albian 20 Eklund et al. (1994) to ‘stem group Calycanthaceae’ fruits described as Appomattoxia were piperalean, but as ber because it had a deep floral cup with numerous spiral discussed below phylogenetic analysis indicates that they m tepals,stamens,andstaminodesontherimandnumerous e are more likely related to Amborella or Chloranthaceae c carpelsinside,asinCalycanthaceae,thesistergroupofthe e (Doyle and Endress2014). D remaining Laurales, but monosulcate rather than dis- 9 1 ulculate pollen. In the analysis of Doyle et al. (2008), 6 5 Virginianthus was sister to Calycanthaceae in one most 1: 5. ANITA lines 1 parsimonious tree, but in the other it was attached to the e] at line to the remaining Laurales. At the other extreme is MococluercruedlarinttrheeesEaprrlyedCicrtetathcaetoutsh,eanbdatshaelreAiNsIiTncArealsininegs yl Mauldinia(Drinnanetal.1990;Figure3(f)),knownfrom o fossilevidencethattheydid.Thebest-documentedlineis D inflorescences made up of dichasial cymes of trimerous s flowers with one carpel and stamens with paired basal Nymphaeales, consisting of Cabombaceae, Nymphaea- e m ceae and the highly reduced Hydatellaceae (Trithuria), glands and flap dehiscence, as in Lauraceae, and a J whichuntilrecentlywerethoughttobemonocots(Saarela r associated wood (Paraphyllanthoxylon: Herendeen o et al. 2007). The phylogenetic position of Nymphaeales, ss 1991), from the early Cenomanian of Maryland. Fossils e one node above the base of the angiosperms, implies that of namedPriscafromtheDakotaFormationofKansas(latest Pr Albian or early Cenomanian) that Retallack and Dilcher they are an ancient line that went from the land into the [ y (1981) interpreted as elongate floral axes are probably water, but it does not say when this shift occurred – it b d inflorescences of similar or identical plants. Despite the could have been in the Early Cretaceous, earlier or e d much later. a nearlylauraceousfloralmorphology,theanalysisofDoyle nlo and Endress (2010) indicated that Mauldinia did not The most informative nymphaealean fossil is Mon- w etianthus,aflowerfromtheearlyAlbianofPortugal(Friis o belongtoLauraceaebutwasinsteadanextinctsistergroup D ofLauraceaeandtherelatedfamilyHernandiaceae,based et al. 2009; Figure 3(h)). It had numerous tepals and on its retention of ancestral features such as a more stamens and a ring of 12 carpels, and X-ray microtomo- differentiatedperianthandendosperminthematureseed. graphy showed that the carpels had the typical laminar Similar but less complete flowers from Virginia, placentation of Nymphaeales. Using the data-set of Potomacanthus and Cohongarootonia (von Balthazar Saarela et al. (2007), Friis et al. (2009) concluded that et al. 2007, 2011), appear to extend this line back to the Monetianthus was nested within Nymphaeaceae, as the middle Albian. Leaves assigned to Laurales by Upchurch sistergroupofBarclayaandNymphaeoideae(Nymphaea, and Dilcher (1990) are common in the late Albian of the Victoria and Euryale), with which it shares an inferior Dakota Formation, including both simple, pinnately ovary. With a larger data-set, Doyle and Endress (2014) veined leaves (Pandemophyllum) and palmately lobed obtained a similar result, with additional most parsimo- leaves (Pabiania), formerly confused with platanoids, nious positions sister to either Barclaya or Nymphaeoi- which are superficially similar to Sassafras but more deae. Von Balthazar et al. (2008) described a similar plesiomorphicthanextantLauraceaeintheirfinevenation but less well-preserved flower from the middle Albian of (Doyle and Upchurch inpress). VirginiaasCarpestella.Itspositionwaspoorlyresolvedin The third magnoliid order, Canellales, includes theirphylogeneticanalysis,butDoyleandEndress(2014) Winteraceae, which are famous for their vesselless found that it had the same three most parsimonious 8 J.A. Doyle positions as Monetianthus. Because all living Nym- bearing loosely spaced stamens and carpels from the phaealesareaquatic(exceptforanobviousreversaltowet BarremianorAptianYixianFormationofNEChina(Sun terrestrial habitats in Barclaya rotundifolia: Feild et al. et al. 1998, 2002). A phylogenetic analysis by Sun et al. 2004),thisevidencethatAlbianfossilswerecrowngroup (2002)attachedArchaefructustothestemlineageofliving membersimplies that the nymphaealean line had become angiosperms, but Friis et al. (2003) interpreted the fertile aquatic bythis time. axes as inflorescences of reduced flowers and argued that More direct evidence for an aquatic habit is seen in Archaefructus was a crown group angiosperm that had Pluricarpellatia (Mohr et al. 2008; Figure 3(i)), from the undergonefloralreductionduringadaptationtoanaquatic lacustrine Crato Formation of Brazil (late Aptian: habitat.AnanalysisbyEndressandDoyle(2009)placedit Heimhofer and Hochuli 2010), which had cordate to inNymphaeales,sistertoHydatellaceae,whichshoweven peltateleaveswithlongpetiolesattachedtoanherbaceous more extreme floral reduction. Wang and Zheng (2012) rhizome and flowers (in the fruit stage) with several free reported that ovules of Archaefructus were borne on the carpels. An analysis of vegetative characters by Taylor midrib of the carpel, as in Cabombaceae, which might (2008) indicated that Pluricarpellatia could be sister to support this hypothesis. However, the Endress and Doyle either Cabombaceae or Nymphaeales as a whole (not (2009) result depended on the interpretation of structures includingTrithuria),butDoyleandEndress(2014)found extracted from the stamens as monosulcate pollen grains, 4 that it could be sister to Cabombaceae, Brasenia (one of whichwasquestionedbyFriisetal.(2003,2011)because 1 20 the two genera of Cabombaceae) or Nymphaeaceae; the of their irregular size and shape. If pollen characters are er ambiguity was due in part to uncertainty on the removedfromtheEndressandDoyle(2009)analysis,itis b m arrangement of the carpels. An analysis by Coiffard more parsimonious to place Archaefructus in Ranuncu- e c et al. (2013a) concluded that other Crato fossils with lales, near the base of the eudicots, which would be e D stems, roots and palmately veined, cordate leaves but no consistentwithitsternatelydissectedleaves.However,as 9 1 flowers(Jaguariba)werenestedwithinNymphaeaceae.A discussedbyDoyle(2012),apositionincrowneudicotsis 6 5 variety of other cordate and peltate leaves that probably hard to reconcile with the lack of tricolpate pollen in the 1: 1 represent Nymphaeales have been described from the Yixian and other floras at the same latitude. Another ] at AlbianofJordan(Tayloretal.2008)andthelateAlbianor possibility, mentioned above, is that Archaefructus and e yl early Cenomanian of Kansas (Wang and Dilcher 2006). other ternately dissected Early Cretaceous leaves such as o D It is possible that lower Potomac (Aptian–early Albian) Vitiphyllum (Figure 1(h)) represent stem relatives of s reniform leaves with palmate venation (Proteaephyllum eudicots. e m reniforme:Figure 1(i))alsorepresent Nymphaeales. The other two ANITA lines, Amborella and Austro- a J r As several authors have noted (Nixon 2008; Coiffard baileyales,arewoodyplantswithpinnateleafvenation.In o ss et al. 2013a; Doyle and Endress 2014), Early Cretaceous thelowerPotomacGroup(Aptian–earlyAlbian),thereare e of Nymphaealespresentacaseofconflictbetweenthefossil pinnatelyveinedleavessuchasFicophyllumandRogersia r P record and molecular dating analyses, which use the that have unusually irregular venation, with variable [ y divergence of DNA sequences and fossil calibrations to spacing,branchingandloopingofthesecondaryveinsand b d estimate the age of clades. Because these analyses are a continuum of finer vein orders (Figure 1(k,l)). Wolfe e d a based on DNA from living taxa, they necessarily provide et al. (1975), Doyle and Hickey (1976) and Hickey and o nl agesforcrowncladesonly.SuchananalysisbyYooetal. Doyle(1977)pointedoutsimilarvenationinWinteraceae, w o (2005)concludedthatthecrowngroupofNymphaeaceae now known to belong in the magnoliid clade, but it also D and Cabombaceae (Hydatellaceae had not yet been occurs in ANITA groups such as Illicium (Upchurch and associated with the order) was less than a third the age Dilcher 1990). In studying cuticle anatomy of the early of the angiosperms, or Eocene (44.6Ma) on the Potomac leaves, Upchurch (1984b) concluded that the assumption that angiosperms were not much older than most similar living taxa were Amborella and Austrobai- the first unquestioned fossil pollen of angiosperms in the leya, 15 years before molecular data showed that these Hauterivian. However, the analyses of Monetianthus, weretwoofthemostbasalangiospermlines.Somecuticle Carpestella and Pluricarpellatia imply that this crown types have features that suggest a position nested within group had originated by the end of the Aptian (ca. Austrobaileyales (Doyle and Upchurch in press). Other 112Ma),whichwouldputtheoriginofangiospermsinthe leavesfromthelateAlbianoftheDakotaFormation may Carboniferous. This could be taken to mean that also represent Austrobaileyales (as ‘Illiciales’: Upchurch angiosperms do go back to the Palaeozoic, or that the and Dilcher 1990). fossils are incorrectly placed, but it could also reflect Reproductive evidence for Austrobaileyales comes problems of molecular dating methods in dealing with from uniovulate fruits and multicarpellate floral axes unequalrates ofevolution(cf. Nixon 2008). named Anacostia from the early and middle Albian of A more controversial fossil is Archaefructus, an Portugal and the Potomac Group (Friis et al. 1997; aquatic plant with finely dissected leaves and fertile axes Figure 3(j)). Pollen on the stigmas is of the type called
Description: