Fisiologia Extinçao Flores

8/9/2019 Fisiologia Extinao Flores

1/26

www.newphytologist.org 383

Review

BlackwellPublishing,Ltd.

Tansley review

Form, function and environments of theearly angiosperms: merging extantphylogeny and ecophysiology withfossils

Taylor S. Feild

1

and Nan Crystal Arens

2

1

Department of Ecology and Evolutionary Biology, Dinwiddie 310, Tulane University, New Orleans, LA,

701185698, USA; 2

Department of Geosciences, Hobart and William Smith Colleges, Geneva, NY

14456, USA

Contents

Summary 383

I. Introduction 384

II. Previous images of early angiosperms and their habitats 385

III. Progress in understanding angiosperm phylogeny:extant basal relations 386

IV. Early angiosperm ecology: inferences from extantplants and reflections from the fossil record 387

V. The ecology of angiosperm diversification:gaining a roothold and subsequent diversification 397

VI. The environmental context of early angiospermevolution 399

VII. Conclusions 402

Acknowledgements 402

References 402

Author for correspondence:Taylor S. Feild

Fax: +1 504 862 8706

Email: [email protected]

Received: 2 September 2004

Accepted: 8 November 2004

Key words:

Amborella

, Archaefructus

,

angiosperm phylogeny, basal

angiosperms, Chloranthaceae,

diversification, key innovations,

phytochrome function.

Summary

The flowering plants angiosperms appeared during the Early Cretaceous period and

within 1030 Myr dominated the species composition of many floras worldwide.

Emerging insights into the phylogenetics of development and discoveries of early

angiosperm fossils are shedding increased light on the patterns and processes of

early angiosperm evolution. However, we also need to integrate ecology, in particular

how early angiosperms established a roothold in pre-existing Mesozoic plant

communities. These events were critical in guiding subsequent waves of angiosperm

diversification during the AptianAlbian. Previous pictures of the early flowering

plant ecology have been diverse, ranging from large tropical rainforest trees, weedydrought-adapted and colonizing shrubs, disturbance- and sun-loving rhizomatous

herbs, and, more recently, aquatic herbs; however, none of these images were tethered

to a robust hypothesis of angiosperm phylogeny. Here, we synthesize our current

understanding of early angiosperm ecology, focusing on patterns of functional ecology,

by merging recent molecular phylogenetic studies and functional studies on extant basal

angiosperms with the picture of early angiosperm evolution drawn by the fossil record.

New Phytologist

(2005) 166

: 383408

New Phytologist

(2005) doi

: 10.1111/j.1469-8137.2005.01333.x


2/26

Tansley review

New Phytologist

(2005) 166

: 383408

www.newphytologist.org

New Phytologist

(2005)

Research384

I. Introduction

The flowering plants angiosperms are the most ecologicallydiverse and species-dense branch on the green plant tree oflife. With roughly 250 000 extant species (Crane et al

., 1995;Wing & Boucher, 1998; Soltis et al

., 2004), angiospermshave evolved an unparalleled spectrum of growth habits

and ecologies in response to a broad range of habitats(Stebbins, 1974; Doyle, 1977; Crane et al

., 1995; Wing &Boucher, 1998). Also, angiosperms dominate the composition,biomass, and biogeochemical functioning of most terrestrialecosystems, with the exception of boreal and sometemperate rainforest zones (Bond, 1989; Wing & Boucher,1998).

In contrast to their prodigious modern presence, the fossilrecord reveals that angiosperms and their ecological ascend-ancy are geologically recent developments. For much of landplant history, terrestrial vegetation consisted of free-sporing(e.g. lycopsids, ferns, sphenopsids, mosses) and gymnosper-

mous seed plants (e.g. bennettitaleans, conifers, cycads,ginkgos, Gnetales and various seed ferns). Despite intriguingreports of pre-Cretaceous angiosperms (Cornet & Habib,1992) and suggestive molecular clock analyses (Li et al

., 1989;Martin et al

., 1989), distinctive pollen forms of Early Creta-ceous age (BerriasianValanginian, Fig. 1) mark the currently

accepted first appearance of flowering plants (Doyle, 1983;Hughes & McDougall, 1987; Brenner, 1996). By the late

Aptian (

112 Ma), flowering plants had diversified signifi-cantly and fossils representing a wide range of lineages were

well-differentiated (Doyle, 1969, 1977; Wolfe et al

., 1975;Doyle & Hickey, 1976; Hickey & Doyle, 1977; Muller, 1981;Upchurch, 1984; Friis et al

., 1994, 1999, 2000, 2001; Lupia

et al

., 1999). New fossil-calibrated molecular clock studiesalso suggest that some of the most speciose lowland tropicalangiosperm clades may have originated during Aptian timesor earlier (Bremer, 2000; Davies et al

., 2004).Ecologically, angiosperms expanded to both poles and

began to dominate the species composition in some low-latitude floras by Aptian times (Retallack & Dilcher, 1981;Romero & Archangelsky, 1986; Crane & Lidgard, 1989;Lidgard & Crane, 1990; Spicer, 1990; Dettmann, 1994;Cantrill & Nichols, 1996; Wing & Boucher, 1998; Barrett &

Willis, 2001). Many groups of Mesozoic free-sporing andseed plants became extinct or declined in abundance, as

angiosperms proliferated (Knoll, 1986; Crane & Lidgard,1989; Lidgard & Crane, 1990; Lupia et al

., 1999; McElwain

et al.

, 2005). Yet other groups pleurocarp mosses, Polypo-diaceae ferns and epiphytic lycopods may have diversifiedsecondarily as broad-leaved angiosperms restructured theecology of lowland rain forest plant communities and created

Fig. 1 Major events in the Cretaceous history of flowering plants. Sources for age data and stratigraphic nomenclature are taken from Gradsteinand Ogg (1996), Gradstein et al. (1995) and Berggren et al. (1995), as compiled by A. R. Palmer and J. Geissman for the Geological Society ofAmerica. Global temperature trends (noted by lateral line) are from Frakes (1999), Veizer et al. (2000), and Crowley and Berner (2001). Evidencefor high latitude glaciation is from Frakes and Francis (1988). First appearances of angiosperm pollen are as recorded by Doyle (1983) andMcIntyre and Brideaux (1980). Diversity, relative abundance and geographic distribution data are from Lupia et al. (1999). The age ofArchaefructusis from Swisher et al. (1999). Major intervals of oceanic anoxia events (OAEs) are based on Bralower et al. (2002a,b). In all cases,an OAE consisted of several brief periods of anoxia distributed through various ocean basins. The level of stratigraphic resolution needed topresent each individual event is outside the scope of this representation. For more detailed stratigraphies, see Heimhofer et al. (2004) and Herleet al. (2004).


3/26

Tansley review

New Phytologist

(2005)


New Phytologist

(2005) 166

: 383408

Research 385

new environmental opportunities for these lineages duringthe Late Cretaceous (Wikstrm & Kenrick, 2001; Shaw et al

.,2003; Schneider et al

., 2004).Although the Cretaceous explosion of angiosperm diversity

is well-documented, the ecological factors underlying thisradiation remain unclear. Most speculation has focused onthe competitive superiority of angiosperms (Stebbins, 1974,

1981; Regal, 1977; Burger, 1981; Bond, 1989). Despite thisemphasis, little attention has been paid either to the actualecological strategies used by early angiosperms or to theenvironmental context in which the lineage evolved. Recently,however, new data from several disparate sources are allowingus to pose and test concrete hypotheses for early angiospermecology. In particular, a spectrum of independent molecularstudies have, for the first time, produced a robust hypothesisof angiosperm phylogenetic relationships (Mathews &Donoghue, 1999; Parkinson et al

., 1999; Qiu et al

., 1999; Soltis

et al

., 1999; Doyle & Endress, 2000; Graham & Olmstead,2000; Zanis et al

., 2002), which supplies a framework for

testing ideas about early angiosperm ecology. Also, new dataand analyses from the global Ocean Drilling Project and othersources have shed new light on global climate and biogeo-chemical change during the Cretaceous (Fig. 1; e.g. Jahren

et al

., 2001; Beerling et al

., 2002; Bralower et al

., 2002b;Heimhofer et al

., 2004; Herle et al

., 2004). These perspectivesprovide a context in which to understand the response of earlyangiosperms to their changing world.

To begin to connect ecology to the early radiation of angio-sperms, we must understand the ecological starting point forangiosperm evolution. In turn, this may clarify the selectiveforces within the environment driving the evolution of

unusual character combinations (e.g. carpels, xylem vesselsand reticulate-veined leaves with freely ending veinlets all onthe same body plan) that may have opened up new ecologicalpossibilities for Cretaceous angiosperms (Feild, 2004; Feild

et al

., 2004). This combination of traits and environment wascritical in initiating waves of angiosperm diversificationduring the AptianAlbian, which ultimately drove the develop-ment of the modern angiosperm-replete flora. In this review,

we synthesize recent phylogenetic, ecophysiological andgeological discoveries that are pollinating new ideas aboutthe ecology and environment of early angiosperms.

We will begin by reviewing previous views on earlyangiosperm form and function. We will focus specifically on

vegetative functional traits because these have been onlycursorily considered in recent reviews (Crane et al

., 1995;Wing & Boucher, 1998; Feildet al

., 2003a; Soltis et al

., 2004).Next, we will synthesize recent empirical and conceptualadvances in the structure of angiosperm phylogeny, and howthese results have, or at least should be, impacting our abilityto pose testable hypotheses about how early angiospermsfunctioned within their environments. Through the discus-sion, we will explore patterns of ecophysiology and vegetativemorphology drawn from living plants and the support for

these trends in the Early Cretaceous fossil record, whilepointing out areas in the fossil record in need of futureattention. We will finish by discussing the ecology of earlyangiosperm diversification and the Cretaceous world into

which the angiosperms debuted and the processes in thephysical and biotic environment that might have allowedthe new lineage to invade, and eventually dominate, the

landscape.

II. Previous images of early angiosperms andtheir habitats

1. Ranalian rainforest tree

An early and influential idea about the ecology of the earliestangiosperms is the woody Ranales or woody magnoliidhypothesis. This view flowed from a preconception that ancestralangiosperm flowers were large and consisted of numerousspirally arranged free perianth parts, stamens and carpels

features that are retained in living Magnoliales and Winteraceae(Arber & Parkin, 1907; Bews, 1927; Axelrod, 1952; Takhtajan,1969; Thorne, 1976; Cronquist, 1988; Gottsberger, 1988).By analogy to these living taxa, the first flowering plants wereportrayed as trees or shrubs that grew and matured slowlyin tropical lowland rainforests or cloud forests. They hadthick, heavy branches plumbed with primitive tracheid-basedxylem. These plants were predicted to have large, simple,entire-margined leaves that photosynthesized at low rates(Bews, 1927; Takhtajan, 1969; Cronquist, 1988). Their seedlingsestablished on undisturbed soil in wet, low-light environmentsbelow the forest canopy. Plants such as Degeneria vitiensis

(Smith, 1949), Galbulimma belgraveana

(van Royen, 1962),

Eupomatia

, some genera in the Winteraceae (Feild et al

.,2000) and many Magnoliaceae and Myristicaceae species(Forget, 1991; Kwit et al

., 1998; Yasumura et al

., 2002)epitomize the woody Ranalian ecological stereotype. However,other Magnoliales, such as Annonaceae (126 genera, 1200species), have diverse growth habits (e.g. vines, shrubs,and trees) and resist ecological generalization (Berry et al

.,1999; Fisher et al

., 2002). A related view reconstructed earlyangiosperms as arborescent residents of lowland equablerainforests. For example, Corner (1949) suggested that thelarge-boled tropical durians (

Durio

, Bombacaceae) were theleast-modified living descendents of the first angiosperms.

2. Weedy xeric shrub

Others suggested that fast-growing, drought-adapted shrubsfrom subtropical open, disturbed habitats preceded shade-tolerant rainforest trees in angiosperm evolution (Stebbins,1965, 1974; Axelrod, 1970, 1972). This concept was basedon the belief that seasonal drought would not only favor theorigin of reproductive and vegetative hallmarks of angiospermy,but would also foster diversification (Stebbins, 1965, 1974).


4/26

Tansley review

New Phytologist

(2005) 166

: 383408


New Phytologist

(2005)

Research386

Although the woody magnoliids, Magnoliales and Winteraceaewere still viewed as among the most primitive living angiosperms,Stebbins (1974) concluded that their extant rain forest tocloud forest ecology was secondarily derived. Stebbins (1974)pointed out that modern Magnoliaceae and Winteraceaeconsist chiefly or entirely of genera with a polyploid origin(Ehrendorfer & Lambrou, 2000) and that their diploid ancestors

must be extinct, a trend supported by the fossil record ofMagnoliaceae (Masterson, 1994). Over time, Stebbins (1974)concluded that, polyploidy events, which commonly supportpopulations invading new environments, allowed the early,drought-adapted lineages to move into lowland forest museums

where they survived due to perceived low extinction rates.Meanwhile, other more advanced angiosperm lineages radiatedon a drought-disturbance theme, forcing the early drought-adapted magnoliid forms to extinction.

The arid-origin model conveniently explained the paucityof early angiosperm fossils because few plant remains can bepreserved over geologic time in dry climates. In fact, no direct

fossil evidence for xerophytic ancestral angiosperms has beenfound. However, Doyle and Hickey (Doyle & Hickey, 1976;Hickey & Doyle, 1977) suggested that the morphologyand sedimentary facies associations of most fossil leaves fromthe Early Cretaceous (AptianAlbian) Potomac Group ofeastern North America indirectly supported Stebbins semiaridhypothesis (Stebbins, 1965, 1974). This was not based onevidence of aridity in the Potomac area, but rather on the beliefthat the disturbed stream margins where fossil angiospermsleaves were preferentially preserved were the habitats mostlikely to be occupied by weedy immigrants from dry regions.More support for a drought-adapted origin came from the

tropical latitudes of northern Gondwanaland (e.g. Brazil andWest Africa), where a diversity of Early Cretaceous pollenforms was discovered in association with geologic indicatorsof seasonal aridity, such as low abundance of fern spores,pollen typical of supposedly xerophytic plants (i.e. the extinctconifer family Cheirolepidiaceae and ephedroid forms linkedto Gnetales) and salt deposits (Doyle & Hickey, 1976; Hickey& Doyle, 1977; Brenner, 1996). Subsequent studies, how-ever, showed that angiosperm pollen was equally diverse andabundant in restricted wet areas of Northern Gondwana, suchas northern South America and the Middle East, during thesame timeframe, where there were fewer Cheirolepidiaceae,more ferns, and deposits of coal (Doyle et al

., 1982; Brenner,

1996). Consequently, one climatic regime cannot be favoredover another as the initial one occupied by floweringplants.

3. Herbaceous weeds

A related view, which emphasized disturbance as a majorcatalyst for early angiosperm evolution, proposed that thefirst angiosperms were rhizomatous, semiherbaceous weedsof sunny, unstable stream-sides and flood plains (Taylor &

Hickey, 1992, 1996). In this model, early angiosperms wererapidly growing pioneers that tolerated disturbance. Theirleaves were toothed with marginal hydathodal glands and

were capable of high photosynthetic rates. This interpretationwas spurred by some early cladistic analyses that placedherbaceous magnoliids the paleoherbs, which havevariously included Nymphaeales, Piperaceae, Saururaceae,

Lactoris

, Aristolochiaceae and Chloranthaceae at the base ofthe extant angiosperm tree (Taylor & Hickey, 1992; Doyle

et al

., 1994; Nixon et al

., 1994). Proponents of a paleoherborigin also argued that the hypothesis was consistent withsmall sizes of angiosperm flowers and seeds, the rarity ofangiosperm wood in Early Cretaceous fossil record, and thepresence of early angiosperm leaf fossils in ephemeral habitatsthat were interpreted as sunny (Taylor & Hickey, 1990, 1992,1996; Wing & Boucher, 1998; Friis et al

., 1999, 2000;Eriksson et al

., 2000; but see Feild et al

., 2004).

III. Progress in understanding angiospermphylogeny: extant basal relations

Whereas the previous descriptions of angiosperm ecologywere consistent with the assumptions based on the dataavailable at the time, none were explicitly linked to a robustand independent hypothesis of phylogeny. However, recentadvances in molecular phylogenetics have converged onsimilar topologies for extant lineages near the root of theangiosperm tree (Figs 2 and 3; Mathews & Donoghue, 1999;Soltis et al

., 1999, 2004; Doyle, 2001; Endress, 2001; Zanis

et al

., 2002; Feild et al

., 2004). This congruence offers a newtool for exploring early angiosperm ecology.

The first extant basal angiosperms, which are living repre-sentatives of lineages that split from the main angiosperm linerelatively early in its history, were recognized using nuclearrRNA (Hamby & Zimmer, 1992; Doyle et al

., 1994) andchloroplast rDNA ITS sequences (Goremykin et al

., 1996).These analyses submerged the Nymphaeales at the base of theangiosperm tree. Later studies, incorporating multiple genesequence comparisons, retained Nymphaeales near the base,but placed a suite of several unexpected tropical woodyshrubs, small trees and woody vines around them. In most ofthese studies, the first branch was Amborella

(1 species), avesselless shrub from New Caledonia that had been previouslyplaced in the Laurales; a second lineage included Nymphae-

ales; and a third branch (called the Austrobaileyales by some)consisted ofAustrobaileya

(1 species), Trimenia

(9 species) andthe Illiciales Illicium

(

40 species, including star anise) andthe Schisandraceae (

Kadsura

and Schisandra

,

40 species). Analternative topology mergedAmborella

and Nymphaeales intoa single basal clade (Barkman et al

., 2000; Jansen et al

., 2004),but analyses by Zanis et al

. (2002) indicated that this unionwas not well-supported. Molecular analyses yielding thepresent phylogenetic picture incorporated various combina-tions of all three cellular genomes. These included rbcL

and


5/26

Tansley review

New Phytologist

(2005)


New Phytologist

(2005) 166

: 383408

Research 387

atpBfrom the chloroplast, 18S rDNA from the nucleus, andfive mitochondrial genes (Soltis et al., 1997, 1999; Parkinsonet al., 1999; Qiu et al., 1999; Barkman et al., 2000), duplicatedphytochrome genes (Mathews & Donoghue, 1999; Zaniset al., 2002), 17 chloroplast genes (Graham & Olmstead,2000), genes with high substitution rates (i.e. trnLand matK;Borsch et al., 2003; Hilu et al., 2003), and most recently,genome-wide analysis of the chloroplast (Jansen et al.,2004).

In the current phylogenetic consensus, basal angiospermsform a grade, referred to as ANITA (forAmborella, Nymphae-ales, Illiciales, TrimeniaandAustrobaileya) by Qiu et al. (1999).

Above the basal grade, the remaining angiosperms formeight well-supported monophyletic lineages: Ceratophyllum,Chloranthaceae, eudicots, Laurales (sans Amborella, Aus-trobaileya and Trimenia), Magnoliales, monocots, Piperales(Piperaceae, Saururaceae, Lactorisand Aristolochiaceae) and

Winterales (Canellaceae plus Winteraceae). In contrast towidespread agreement on the basal grade topology, relations

among these lineages are unresolved. In particular, it remainsunclear which group(s) diverged just above the basal grade.There are increasing indications, from multigene analyses(Qiu et al., 1999; Zanis et al., 2002), that the clade includingmonocots and Ceratophyllum, forms the next branch abovethe ANITA grade, and Chloranthaceae diverge next (Soltiset al., 2004). The remaining angiosperm lineages aggregateinto a eudicot clade and a eumagnoliid clade, which iscomposed of four strongly supported subclades: Magnolialesand Laurales are sister to Winterales and Piperales (Soltis et al.,2004). However, these placements are far from concrete.Other analyses place Ceratophyllumwith the eudicots (Hilu

et al., 2003). Chloranthaceae are also sister to monocots inmatKtrees (Hilu et al., 2003).Resolving the positions of Ceratophyllum, Chloranthaceae

and the topology of basal-branches within speciose groupssuch as monocots and eudicots remains a significant technicalchallenge. Angiosperms above the ANITA grade are highlyvariable in morphological and ecological characters, which,depending on the topology, can influence the interpretationof ancestral states at the root of the phylogenetic tree. Thus,

we must explore how different phylogenetic hypotheses affectthe patterns of character evolution.

IV. Early angiosperm ecology: inferences fromextant plants and reflections from the fossilrecord

Emerging consensus on the phylogeny of living angiospermsoffers a useful approach for elucidating evolutionary patterns(Soltis et al., 1999, 2004). For instance, the form andfunction of basal lineages can be documented and overlaidonto hypotheses of angiosperm phylogeny to reconstructtraits of the common ancestor (Forbis et al., 2002; Feildet al., 2003a, 2004; Soltis et al., 2003, 2004; Kim et al., 2004).

However, such phylogenetic character mapping has threeimportant caveats if our goal is to understand the ecologyof Early Cretaceous angiosperms. First, traits reconstructedas ancestral among living basal angiosperm lineages mayactually reflect a derived condition. This could be true if theearliest angiosperm lineages and their unique ecologies

went extinct, leaving only later-derived ecologies among the

modern flora. Indeed, the fossil record of early angiospermflowers suggests that extensive extinction and turnoveroccurred during the Early Cretaceous (BarremianAptian)

when angiosperms were initially diversifying (Friis et al.,1999, 2000, 2001). Surveys of five Barremian through

Aptian assemblages from Portugal by Frii s et al. (1999,2000, 2001) have identified 150 different angiosperm taxa,represented by flowers, fruits and seeds, of which most (85%)

were magnoliids (broadly defined to include the ANITAgrade plus eumagnoliids) or monocots. In these floras,only two of the most basal families (Nymphaeaceae andChloranthaceae) have been identified, on the basis of flowers,

fruits and/or dispersed stamens, in the Early Cretaceous (Friiset al., 1999, 2000, 2001; Eklund et al., 2004). Also, no fossilscan unequivocally place any living genera of basal clades in theEarly Cretaceous, althoughAsteropollispollen may be relatedto Hedyosmumof the Chloranthaceae (Eklund et al., 2004).Instead, extant genera of basal lineages, such as Illicium,Chloranthaceae and Nymphaeales, appear to be the productsof much younger (early to mid-Tertiary) diversification, basedon fossil-calibrated molecular clock phylogenetic studies(Oh et al., 2003; Zhang & Renner, 2003).

Second, if extant basal lineages identified by molecularstudies truly represent nodes near the start of the angiosperm

radiation, taxa at the tips of the branches, which parameterizethe character data set, may have undergone ecological modi-fication over the last 130 Ma, such that ancestral traits werelost or significantly modified. For instance, the form andfunction of basal lineages could have been re-worked bychanges in climate (e.g. Tertiary cooling), habitat use, or byshifts in community structure, from fern and gymnosperm-dominated communities to ones dominated by angiosperms

with possibly different ecosystem-level properties (Knoll &James, 1987). Evaluating the relative roles of extinction andecological modification on patterns of ecological characterevolution will remain unclear until an extensive sample ofEarly Cretaceous fossil angiosperms can be added to existing

phylogenies a goal that is significantly beyond current capa-bility and will require renewed attention to morphologicaldata (Doyle, 2001; Eklund et al., 2004).

A third caveat surrounds the phylogenetic hypothesisitself. Although todays molecular phylogenies enjoy wideconsensus among a variety of genes and, most unusually,independent research groups, it is possible that further speciesand gene sampling will again rearrange the branching of basallineages (Goremykin et al., 2003; but see Soltis & Soltis,2004).


6/26

Tansley review

New Phytologist(2005) 166: 383408 www.newphytologist.org New Phytologist(2005)

Research388

Despite these caveats, a character-mapping approach doesallow specifichypotheses to be posed and tested using the fossilrecord, a significant advance over previous understanding(Doyle & Donoghue, 1993; Doyle, 2001; Feild et al., 2003,2004; Soltis et al., 2004). Although fossils cannot be re-greened to exhibit physiology or ecology, they can preserveecomorphological, biochemical and/or contextual features

(i.e. types and chemistry of preserving sediments) that areindicative of past lifestyles (Wing & Boucher, 1998). Thus,fossils can be applied to test patterns of ecological evolutiondeduced from living basal lineages. Most importantly, usingfossil form and function, rather than taxonomic identity, totest historical hypotheses does not hinge on the unlikelypossibility of finding examples of basal lineages likeAmborellain the Early Cretaceous because ancestral characters suggestedby living plants provide predictions about the ecomorphologicaland contextual features of fossils that we should expect tosee if the hypothesis holds. Addressing solely the living endof this approach, an evolutionary analysis of extant basal

angiosperm ecology, incorporating a variety of eco-functionaltraits (e.g. preferred habitats for seedling recruitment, leafphotosynthetic rate, leaf ecomorphic traits, growth forms)has revealed two distinct possibilities for the ancestral ecologyof angiosperms: (1) one represented by Amborellaand theterrestrial members of the ANITA grade; and (2) anotherrepresented by the water lilies (Feild et al., 2003, 2004).In the next section we review the predicted ecological andfunctional characters, given these alternatives, and discusstheir relative support by the fossil record and phylogeneticdata.

1. Growth habit and habitat

The first phylogenetic hypothesis predicts woody, tropicalearly angiosperms with unusual seeding development (Figs 2and 3).Amborellaand Austrobaileyales pass through a creepingestablishment phase, where seedlings possess a pseudo-rhizomatous lignotuber and send out multiple scandentshoots from basal buds before adopting a more typical shrub/small-tree form (Feild et al., 2004; Fig. 2). Scandent develop-ment can also extend to adult plants, as indicated by evolutionof twinning liana habits inAustrobaileya, Schisandraceae andtwo Trimeniaspecies (Feild et al., 2003b; Fig. 2). Based on thephylogenetic distribution of functional characters, ancestral

habitats were reconstructed as moist, low-light understories,with frequent soil disturbance (Fig. 2; Feild et al., 2004).Modern examples of these wet, dark and disturbed habitatsinclude understory stream margins and steep soil washoutsin montane tropical and subtropical cloud forests (Feildet al., 2004). Multiple shoots and early commitment tobelow-ground carbon storage in basal angiosperm seedlingsseem to enable colonization and persistence on unstablesubstrates prone to shifting (such as brittle clay and sandysoils, in between rocks and rotting logs) by increasing whole-

plant anchorage and resilience to traumas that break branchesand roots (Feild et al., 2003a,b, 2004). Yet, to date, no directevidence indicates that early Cretaceous angiosperms were

woody or occupied wet, dark, and disturbed zones. Uplandslopes, like those favored by living members of the extantbasal grade, seldom enter the geologic record (Behrensmeyeret al., 2000). However, stream margins and channels twodisturbed habitats are common terrestrial depositionalsettings (Doyle & Hickey, 1976; Hickey & Doyle, 1977;Behrensmeyer et al., 2000). Thus, re-examination anddiscovery of new forested flood plain deposits from the EarlyCretaceous is essential. Examination of these assemblages withan eye for characters linked to wet, dark and disturbed zones

will allow this hypothesis to be tested. However, simplisticapplication of leaf and seed traits to infer growth form ispotentially fraught with error (Feild et al., 2004).

A second image of early angiosperms, represented by theNymphaeales, suggests an herbaceous, rhizomatous water plant

with submerged to floating leaves and aerial flowers (an epihydate

habit Cook, 1999). For most water lilies, two dramaticallydifferent light regimes characterize their life history. Mostgerminate and develop some leaves under extremely low lightand reduced oxygen tension in undisturbed sediment at thebottoms of lakes and ponds (Smits et al., 1990). Eventually,seedlings send up floating leaves capable of tolerating fullsunlight on the water surface (Williamson & Schneider,1993; Brewer & Smith, 1995). Some tuber-forming waterlilies, such as Barclaya and Ondinea, can establish in moreunstable sediments on the bottoms of clear, slow-movingstreams (Williamson et al., 1989).

An aquatic image for early angiosperms emerged from a

few phylogenies where Nymphaeales emerged as basal (Doyleet al., 1994; Parkinson et al., 1999; Graham & Olmstead,2000). However, if eitherAmborellaor Nymphaeales linkedtoAmborella(Parkinson et al., 1999; Barkman et al., 2000;Graham & Olmstead, 2000; Jansen et al., 2004) branch firstfrom the angiosperm root, then parsimony phylogenetic analysisreconstructs angiosperms as ancestrally woody and terrestrial(reconstruction not shown). The latter relation, although

weakly supported by molecular data (Zanis et al., 2002; butsee Jansen et al., 2004), may be substantiated by the presenceof what appear to be vestigial lacunae canals in the protoxylemof Amborella stems (Bailey & Swamy, 1948; Feild et al.,2000). Such lacunae, which foster gas exchange in submerged

stems and roots (Dacey, 1980), may be homologous with thosein many Nymphaeales. Under an aquatic origin hypothesis,angiosperm ancestors may have adopted the submergedlifestyle to escape competition on land and tap into under-exploited freshwater habitats (Sargant, 1908; Arber, 1920;Sun et al., 2002). Adaptation to seasonally ephemeral pools,

which occurs in most living water lilies (both drytropical andfrozentemperate) (Sculthorpe, 1967; Williams & Schneider,1993), may have been a ripe environment for the evolutionof rapid growth (via herbaceousness) and accelerated


7/26

Tansley review

New Phytologist(2005) www.newphytologist.org New Phytologist(2005) 166: 383408

Research 389

Fig. 2 Growth habits among extant basal angiosperms. (a) Twinning, lianaous habit of Austrobaileya scandens, Mt. Bartle Frere, Queensland,Australia. (b) Seedling of Illicium floridanum, illustrating a pseudo-rhizomatous creeping habit with a central stem axis and an additional shootemerging from the base, Apalachicola, Florida, USA. (c) Scandent shrub habit of Amborella trichopodafrom Mt. Aoupinie, New Caledonia. Notethe flattened cane-like shoots. (d) Chloranthus henryiherbs regenerating on a collapsed slope from a subtropical evergreen forest, Hunan,China. (e) Habit ofSchisandra glabra, occurring as an understory, scrambling vine, Crowleys Ridge, Arkansas, USA. (f) Sapling of Amborellatrichopoda, note the numerous weeping shoots, that are produced from a lignotuber, from Plateau de Dogny, New Caledonia. (g) Seedling ofTrimenia papuana, Mt. Gumi, Morobe Province, Papua New Guinea. (h) A vine-like sapling of Trimenia papuana, from same locality as (g).Later in ontogeny, plants become normal-looking multiple stemmed trees. (i) The scrambling vine growth form of Trimenia mooreifromStockyard Creek, New South Wales, Australia.


8/26

Tansley review


Research390


9/26

Tansley review


Research 391

reproduction two characteristics thought to underliemuch of the ecological success of angiosperms (Stebbins,1974, 1981; Burger, 1981; Bond, 1989; Robinson, 1994;Taylor & Hickey, 1996).

Although most parsimonious character reconstructionsindicate that flowering plants were terrestrial and woody(Fig. 4, Feild et al.2004), the fossil record offers someevidence for early angiosperm aquatics. A recently discoveredfossilized water-lily flower (Friis et al., 2001) showed that thelineage was present in the Early Cretaceous (Barremian toearly Aptian), but it did not provide direct evidence that theparent plant was aquatic as are all modern representatives of

the clade. More clearly aquatic forms, resembling water liliesin having palmately veined nymphaeid (long-petioled, oval-shaped) leaves, are represented by Brasenia-like leaf imprints

from the Kurnub Group (AptianAlbian) (Taylor et al., 2001)of Jordan and whole plant compression fossils deposited ina lake environment from the AptianAlbian of Brazil (Mohr& Friis, 2000). However, none of these fossils preserveenough detail to associate them unequivocally with the Nym-phaeales. Other possible early aquatic angiosperms includefossil leaves Nelumbitesfrom the Potomac group, which maybe related to water lotus, Nelumbo, a basal aquatic eudicot(Upchurch et al., 1994).

The discovery ofArchaefructusfrom lake deposits in Chinahas renewed enthusiasm for the aquatic origin hypothesis(Sun et al., 1998, 2002; Zhou et al., 2003). Archaefructus

undisputed aquatic habit is based on the character of theenclosing sediments and fish fossils preserved among theplants foliage (Sun et al., 2002). Archaefructus is also

Fig. 4 Stomatal vestibules and striatedabaxial cuticles in extant basal angiosperms.(a) Austrobaileya scandens; (b) Chloranthusjaponicus; (c) Illicium floridanum; (d)Trimenia neocaledonica. Bars, 30 m.

Fig. 3 Parsimony distribution of growth habit characters among extant basal angiosperms using MacClade (Madison & Madison, 2000) forthree different phylogenetic hypotheses. Growth habit was treated as an unordered character, and a matrix of character states and codingdecisions is available from the first author on request. Hypothesis (a) is based on the combined molecular/morphology presented by Doyle andEndress (2000), with the resolution of relations in basal eudicots, Nymphaeales, and monocots based on the molecular studies of Kim et al.(2004), Les et al. (1999) and Tamura et al. (2004), respectively. Hypothesis (b) is based on the backbone of the matK consensus tree providedby Hilu et al. (2003) with eudicot, monocot, and Nymphaeales relations expanded as hypothesis (a). Hypothesis (c) is based on the backbonemolecular tree of Zanis et al. (2002).


10/26

Tansley review


Research392

exceptional not only for its age (originally dated Late Jurassic;however, radiometric data supports an Early Cretaceous,Barremian 124 Ma age, Swisher et al., 1999; Zhou et al.,2003), but because all parts of the plant are preserved inorganic connection, allowing a complete and accurate recon-struction of the living plant. In the reconstruction of Sun et al.(2002), flowers and seeds were displayed above the water

surface, leaves were submerged (based on their highly dissectedmorphology which suggests they were analogous to thebuoyant filliform leaves of Cabomba, Ceratophyllum, andsome Ranunculusaquatics) (Sculthorpe, 1967), limbs werepartially supported by the water, and roots penetrated lakesediments. Cladistic analysis by Sun et al. (2002) placed

Archaefructus below the common ancestor of all livingangiosperms. Subsequent studies, based on the assumptionthat the outgroup had flowers, have nested Archaefructusatpositions away from the angiosperm root (i.e. as a water lilyor basal eudicot; Friis et al., 2003; M. Bowe, pers. comm.,2004). Until its relationships are resolved, it will be difficult

to evaluate the ecological significance of Archaefructus.Furthermore, if species ofArchaefructusare correctly dated asBarremian, then they are too young to represent the earliestflowering plants because angiosperm radiation was already

well-underway by that time (Friis et al., 1999, 2000, 2001;Doyle, 2001, fig. 1). In fact, basal-grade eudicots are alsodescribed from the Yixian Formation, which containsArchae-

fructus(Leng & Friis, 2003).

2. Leaf function: leaf optics and photosynthetic/water-use physiology

Under the dark-disturbed hypothesis, leaves are functionallytuned to low-light, wet habitats (Feild et al., 2001, 2003a,b,2004). Leaf cross-sections ofAmborellaand the Austrobaileyalesare dominated by spongy parenchyma tissue. The spheroidshapes of spongy mesophyll create numerous light-reflectingairwater interfaces, resulting in extensive light-scattering

within the leaf. Consequently, the average path length ofphotons moving through the leaf is increased, which canincrease the probability that they are absorbed by chlorophylland processed for CO2reduction. Such extended light pathsare important in light-limited understory habitats, but aliability in high light environments unless CO2fixation rateis correspondingly increased (Smith et al., 1997). Also, the

photosynthetic apparatus ofAmborellaand most Austrobaileyales(based on field measurements of 53 species) functionssimilarly to that of shade-adapted plants. As shade plants,

Amborellaand most Austrobaileyales are characterized bylow, steady-state maximum photosynthetic electron transportrates and light-saturation points (generally less than 20% fullsunlight), limited ability to adjust leaf physiology to increasedgrowth irradiances, as well as fruit production under low light(Feild et al., 2001, 2003b; Griffin et al., 2004). CO2uptake rateshave not been sampled widely, but reported light-saturated

rates measured under steady-state and physiologically optimalconditions (i.e. under low leaf water potentials and highhumidity) vary from 4.5 to 12 mol CO2m

2s1(Feild et al.,2003b; Griffin et al., 2004). Low photosynthetic capacity canbe advantageous in shady understory habitats by decreasingthe construction and maintenance respiratory carbon costsfor leaf production (Chazdon et al., 1996). Low photosyntheticcapacity, however, can constrain a plants ability to take advantageof greater light availability (Smith et al., 1997).

Measurements of dynamic (non-steady-state) photosyn-thetic responses are more relevant for understanding plantperformance in variably lit understories where transient sun-flecks (< 1 min) often dominate the amount of light capturedby leaves (Chazdon et al., 1996). Unfortunately, data are onlyavailable for one basal angiosperm,Austrobaileya scandens(an understory, subcanopy liana from a few mid-montanecloud forests of tropical, Queensland, Australia) (Feild et al.,2003b). Feild et al. found that stomatal conductance of A.scandensclosed very slowly (3 h, compared with 15 40 min

in temperate and tropical zone eudicots angiosperms)(Robinson, 1994; Franks & Farquhar, 1999) when transferredfrom a humid to a drier atmosphere. Thus, consistent with itstypically humid/shady habitats,A. scandensappeared unableto adjust leaf water loss rates fast enough to avoid ensuing lowleaf water potentials and possibly xylem cavitation under thehigh and fluctuating humidity conditions of large forest gapsand exposed canopy habitats. Slow stomatal response tohumidity, however, may be advantageous in understory andsubcanopy environments whereA. scandensgrows by allowingtime averaging of microenvironmental variation, such as rapidexcursions in humidity, light and leaf temperature during

sunflecks, which could trigger stomatal closure and increasestomatal limitations on carbon assimilation (Chazdon et al.,1996).

Robinson (1994) speculated that slow stomatal kineticsmay result from increases in guard cell rigidity induced ligni-fication. Although stomatal lignification was not measured,stomatal opening kinetics inAustrobaileyawere relatively fast(30 min). Thus, if increased lignification constrains stomatalaperture then it only seems to impact stomatal closure and notopening. The slow stomatal response to humidity inAustrobaileyamay be related to the anatomy of the stomatal complex.Cuticular vestibules, formed from overarching extensions ofthe epidermal cells, cover the guard cells ofAustrobaileya, result-

ing in a chamber of still air above the guard cells that maydecouple them from atmospheric humidity (Fig. 4a). Conse-quently, the guard cells may respond to the locally high watervapor concentrations within the vestibule that are built-up

when the stomata are open. Slow rates of closure in responseto decreased humidity may reflect slow rates of de-gassing ofthe vestibule. Additional functional studies on the regulationof non-steady-state gas-exchange in more basal angiosperms,taking into account responses to natural light/humidityregimes in relation to epidermal anatomy, would be highly


11/26

Tansley review


Research 393

informative. In applying these physiological responses to earlyangiosperms, more work will be needed to understand thetypes of light environments that early angiosperms mayhave occupied in the Early Cretaceous forests because forestcanopies then were conifer-dominated. Currently, few com-parative datasets are available, but light environments fromtemperate podocarp forests of New Zealand appear broadly

similar in sunfleck characteristics to angiosperm forests(McDonald & Norton, 1992).

Water lily leaves, in contrast, are specialized for an amphib-ious life, with several traits associated with photosynthesisunderwater and opportunistic gas-exchange in floating leavesexposed to air. Features of water lily leaves increasing aqueouscarbon uptake include finely dissected (Cabomba) and ruffledleaf blades (e.g. Barclaya, Ondinea, Nuphar), which probablydecrease leaf boundary layer thickness and the widespreadabsences of cuticle, stomata and intercellular air-spaces(Sculthorpe, 1967; Williamson et al., 1989; Williamson &Schneider, 1993). Photosynthetic physiologies of floating

lily pad and emergent leaves of many Nymphaeales taxa alsodiffer considerably from terrestrial basal angiosperms, withevolution of much greater leaf photosynthesis and multi-storied palisade mesophyll tissue (Feild et al., 2004). Thedevelopment of palisade cells and high photosynthetic ratesmay be correlated traits in water lilies because the tubularshape of palisade cells enhances the penetration efficiency ofdirect (collimated) sunlight into the leaf (Smith et al., 1997).Regulation of stomatal aperture is also different in water lilyleaves compared with terrestrial basal angiosperms (Feildet al., 2003b). Stomata of Nymphaeaand Nupharwere claimedto remain permanently open, owing to the absence of a

substomatal cavity, which is required to allow subsidiarycells to bend (Brewer & Smith, 1995). Consistent with leafanatomy, Brewer and Smith (1995) found that leaf stomatalconductances of floating leaves of a temperate water lily,Nuphar polysepalum, did not respond to diurnal changes inleaf-to-air vapor pressure deficit, light intensity and temper-ature responses consistent with occurrence in habitats withunlimited water (Brewer & Smith, 1995).

At present, the fossil evidence for early angiosperm leaffunction consists mainly of leaf form, cuticular morphologyand stomatal architecture. However, these features have notyet been extensively analysed for their ecophysiological signif-icance. Nonetheless, it is clear that Early Cretaceous leaves

share a variety of features with living ANITA grade angiosperms.For example, angiosperm cuticles from zone I of the PotomacGroup (Upchurch, 1984) are typically thick and possessstriated surfaces, highly variable subsidiary cell arrangements,and large-sized stomata (i.e. stomata longer than 30 m) thatare concealed by cuticular vestibules. All of these features areobserved inAmborella,Austrobaileya, Illicium, Schisandraceae,Trimeniaand some Chloranthaceae cuticles, which suggest

wet understory adaptation (Fig. 4; Bailey & Nast, 1948;Bailey & Swamy, 1948, 1949; Upchurch, 1984; Kong, 2001;

Feild et al., 2003b; Oh et al., 2003). Others have interpretedthe stomatal protection and thickened cuticles in zone Ileaves as potentially water-conserving xeromorphic features(Upchurch, 1984). Yet, paleoclimatic models suggest that thePotomac Group climate was moist and subtropical (Beerling& Woodward, 2001), and that thick cuticles are not a secureindicator of high drought tolerance (Feild et al., 2003b).

Alternatively, these cuticular features may act as compensatingfeatures (by reducing maximum leaf water loss rates) for phys-iological drought imposed at the leaf level poor hydraulicsof leaf veins/stem xylem in early angiosperms (Doyle et al.,1982; McElwain et al., 2005). However, this explanationseems unlikely because xylem hydraulic capacities are notexceptionally low among extant basal angiosperms that possessa similar leaf anatomy and occur in wet habitats (Fig. 2) (Feildet al., 2000, 2001, 2003b).

3. Xylem function

Among the vegetative features of flowering plants, origin ofthe angiosperm vessel from single-celled tracheids has longbeen viewed as a pivotal innovation (for a review, see Feild,2004). In terms of xylem conducting efficiency, angiospermvessels can dramatically increase the amount of water transportedthrough a vascular network under a given pressure gradient incomparison with a similarly sized hydraulic system composedof tracheids (Sperry, 2003). This is because evolution ofvessels, which are functionally multicellular, allows for theformation of considerably longer hydraulic compartment lengths(as short as 1 cm to up to 25 m in length), and therefore lessfrequent crossing of resistive pit membranes during water flow

as compared with shorter tracheids (0.10.7 cm long) (Sperry,2003; Feild, 2004; Sperry & Hacke, 2004). The individualcells comprising a vessel can be larger in diameter thantracheids, which greatly increases hydraulic flow (i.e. flowthrough ideal capillaries increases to the fourth power of theradius) (Sperry, 2003; Sperry & Hacke, 2004). Taking greaterhydraulic volume and diameter into account, vessels permitgreater potential hydraulic conductivity per unit conduit areathan a tracheid-based transport system. Indeed, stems ofvessel-bearing angiosperms from a variety of habitats generallypossess greater xylem hydraulic capacity, expressed in terms ofhydraulic flow relative to the amount of cross-sectional area ofxylem invested (sapwood-specific hydraulic conductivity),

than conifers and vesselless angiosperms (Brodribb & Feild,2000; Sperry, 2003). Although many tracheid-bearing speciesare capable of achieving similar or greater stem/root xylemhydraulic capacities on a leaf area basis (i.e. similar leaf specifichydraulic conductivities, kL, see Brodribb & Feild, 2000), theadvantage of vessels is in maintaining these conductivities

with less total investment in wood required to support a givenflow rate and leaf stomatal conductance (Sperry, 2003).

Although the xylem ofAmborella is composed entirely oftracheids (Bailey & Swamy, 1948; Feild et al., 2000), it is


12/26

Tansley review


Research394

unclear if the first angiosperms were vesselless or vessel-bearing because the ancestral state for extant angiosperms as a

whole is uncertain under the three phylogenetic hypotheses(Fig. 5). An equivocal ancestral state results from the fact thatNymphaeales, which lack secondary xylem and possess perfo-rated tracheary elements in the primary xylem (thus vessels,but clearly of a different developmental origin than those of

woody angiosperms), and vessel-bearing Austrobaileyalesdiverge immediately aboveAmborella(Fig. 6). The ancestralstate remains uncertain even if Amborellais linked to Nym-phaeales (reconstruction not shown). It also unclear what vas-cular character state should be used forArchaefructusbecause

Archaefructus is represented by compression fossils, whichhave not yielded cellular details of the vascular system (Sunet al., 1998, 2002).

Comparative anatomical studies (for a review, see Carlquist& Schneider, 2002) reveal that the extant basal angiospermflora exhibits a continuum of xylem morphologies, fromvesselless to large diameter vessels with simple perforations. For

instance, some vessels in basal angiosperms are relatively wide,but very short (13 mm long), whereas other can be com-posed of narrow elements that form long (up to 15 cm) andpotentially wide multicellular hydraulic compartments(Feild, 2004). Functional studies of basal angiosperm xylem(Brodribb & Feild, 2000; Feild et al., 2000, 2001, 2003b)performance show that hydraulic efficiencies range from valueson par with vesselless conifers, such as inAmborella, to capac-ities on the low end of temperate/tropical eudicot trees andshrubs (Brodribb & Feild, 2000; Feild et al., 2000, 2003b).In addition, increasing stem hydraulic efficiency seems tobe associated with tolerance to greater sunlight (Feild, 2004).

This fortunate condition may allow the piecing together ofhow different patterns of vessel development specificallyimpact functional ability to inhabit different light environ-ments. A blurred distinction between vessels and tracheids inextant basal angiosperms also suggests that vessel origin doesbring about an immediate dramatic functional shift inhydraulic efficiency because hydraulic conductivities of some

ANITA grade and Chloranthaceae members are similar tovessellessAmborella(such as Illicium, Sarcandra, Chloranthusand a few Ascarina species) (Feild et al., 2000, 2003a).Instead, vessel origin appears to allow for the exploration of anew morphospace of xylem hydraulic design, with some ofthese experiments associated with the evolution of greater

hydraulic efficiency involving subtle fine-tuning of vesselstructure, including increases in conduit shape as well aspitting and perforation plate structure.

The xylem hydraulics of the vascular systems of Nymphae-ales require future experimental investigation. The hightranspiration rates achieved by floating leaves of Nuphar(Brewer & Smith, 1995) suggest that high xylem hydrauliccapacity may characterize water lily vascular systems. How-ever, the water-conducting vascular system is dramaticallyreduced in many Nymphaeales (Schneider & Williamson,

1993; Williamson & Schneider, 1993) in consisting ofa few protoxylem/metaxylem vascular bundles with smalldiameter tracheary elements. The large, mega-porouscanals in the protoxylem of water lily stems and leaves arelikely gas-filled and are involved in the downward transportof oxygen to the root system and venting of carbon dioxideand methane (Dacey, 1980). Another possible explanation

for the high transpiration rates of floating leaves is thatwater supplies are locally delivered rather than transportedover long distances from root to shoot. For example, petiolesand the lower epidermis of floating leaves of Nymphaeapossess abundant glands (hydopotes) that actively take upions from aqueous solutions (Sculthorpe, 1967). Perhapsthese glands act at the primary sources of transpired water in

water lily leaves.Unfortunately, the Early Cretaceous fossil record is depau-

perate in angiosperm wood, a pattern that has often beenmentioned as negative evidence for the hypothesis that firstangiosperms were herbaceous (Taylor & Hickey, 1996). Cur-

rently, it is unclear if this pattern is real or if it results from ataphonomic bias in the fossil record (Behrensmeyer et al.,2000). In this regard, decomposition studies on the wood of

ANITA taxa would be informative since their low density andlack of resins may decrease their likelihood of preservation.Until suitable wood fossils can be securely identified asangiospermous, it will be difficult to test patterns of xylemstructure/function evolution in basal angiosperms using thefossil record.

4. Photosensory physiology and seed germinationecology: a common ecophysiological thread?

With respect to optical cues received during seed germination,the two ecological models for early angiosperms appear to besimilar (Feild et al., 2003b; Mathews et al., 2003). Althoughmore specific information on seedling light environments iscritically needed, seedling recruitment of most ANITA plantsoccurs under low and/or far-red enriched light habitats, suchas in deep water pond sediments or in disturbed soil ofmicrosites in forest understories (Smits et al., 1990; Barrat-Segretain, 1996; Feild et al., 2004). Consequently, Mathewset al. (2003) have suggested that the evolution of traitsenabling seeds to sense and respond to small and transientvariations in light quantity, and to the spectral composition

in a shady environment, may have been crucial for the initialestablishment of angiosperms.

Among the poss ible features, evolution in the formand function of phytochromes photoreceptors (which arechromophore-bearing proteins that transduce variations inambient light conditions into a diverse set of physiologicalresponses) (Smith, 2000) may prove to be a significant pieceof the puzzle. Based on the phylogeny of seed plant phyto-chromes, there is evidence that both gene and functionaldiversification occurred very early in the history of


13/26

Tansley review


Research 395

Fig. 5 Parsimony distribution of xylem vessel anatomy among extant basal angiosperms. Hypothesis (a) is based on the combined molecular/morphology presented by Doyle and Endress (2000), with the resolution of relations in basal eudicots, Nymphaeales, and monocots based onthe molecular studies of Kim et al. (2004), Les et al. (1999) and Tamura et al. (2004), respectively. Hypothesis (b) is based on the backbone ofthe matK consensus tree provided by Hilu et al. (2003) with eudicot, monocot, and Nymphaeales relations expanded as hypothesis (a).Hypothesis (c) is based on the backbone molecular tree of Zanis et al. (2002).


14/26

Tansley review


Research396

Fig. 6 Parsimony distribution of cambium morphology among extant basal angiosperms. Hypothesis (a) is based on the combined molecular/morphology presented by Doyle and Endress (2000), with the resolution of relations in basal eudicots, Nymphaeales, and monocots based onthe molecular studies of Kim et al. (2004), Les et al. (1999) and Tamura et al. (2004), respectively. Hypothesis (b) is based on the backbone ofthe matK consensus tree provided by Hilu et al. (2003) with eudicot, monocot, and Nymphaeales relations expanded as hypothesis (a).Hypothesis (c) is based on the backbone molecular tree of Zanis et al. (2002). Extensive cambial development refers to wood productionaccruing over several years, as compared to herbs that develop secondary xylem over less than 12 yr.


15/26

Tansley review


Research 397

angiosperms, involving two phytochromes that play a criticalrole in germination and early seedling development undershade cast by plant canopies. For example, the PHYB lineage,

which encodes the principal mediator of shade avoidance,split into two lineages some time after the origin of water liliesbut before the origin of Austrobaileyales (S. Mathews, pers.comm., 2004). Also, the PHYA lineage, which encodes the

principal mediator of very low fluence responses (VLFR) andresponses to far-red light, apparently was subject to an episodeof positive selection (Mathews et al., 2003). Shade avoidanceallows angiosperms to detect and respond to neighboringvegetation by enhancing elongation growth and acceleratingshoot development. Although shade avoidance is not endemicto angiosperms (Warrington et al., 1988), its adaptive signifi-cance has been tested in angiosperms (Schmitt et al., 1995),and diversification in the genes acting in shade avoidance mayhave been advantageous. Conversely, the VLFR and the far-red high irradiance response (FR-HIR) may be important inallowing shade tolerant angiosperms to germinate and de-

etiolate in dimly lit environments and to counteract earlyshade avoidance responses that could be counterproductivefor young seedlings (Mathews et al., 2003). Outside ofangiosperms, FR-HIRs are rudimentary (Burgin et al., 1999;Christensen et al., 2002), and VLFRs have not been docu-mented. For this reason, Mathews et al. (2003) suggested thatinnovation in phyA, the photoreceptor that mediates theseresponses, may have enhanced the ability of early angiospermsto colonize the forest understory. However, much more workis needed on seedling responses to basal angiosperms tocontrolled light environments to see how they compare toeudicot and monocot model plant systems (i.e.Arabidopsis

and Oryza) as well as other seed plant lineages (Mathews et al.,2003).Little is known about the phylogeny and function of other

photoreceptors that influence seedling development, such as theblue light receptors, cryptochrome and phototropin, in extantbasal angiosperms. However, two forms of each of these pho-toreceptors occur inArabidopsis: one that is responsive to highlight intensity and one that is responsive to low light intensity(Briggs & Christie, 2002). If this condition originated in theearliest angiosperms, innovation in the blue light receptorsthat play roles in seedling de-etiolation (cryptochrome) anddirectional responses to light (phototropin), may also havecontributed to the establishment of early angiosperms in the

forest understory (Liscum et al., 2003).

V. The ecology of angiosperm diversification:gaining a roothold and subsequentdiversification

When angiosperms first appear in the fossil record during theBerriasian (i.e. based on pollen) (Doyle, 1983; Hughes &MacDougall, 1987), they are represented by only a few speciesand are of very low abundance. Not until the Aptian and

Albian, some 30 Myr later, do flowering plants becomesignificant players in the ecosystems they inhabit (Lupia et al.,1999). Darwin (1859) blamed this pattern on the paucity ofthe fossil record. Today, with intense effort focused on findingthe first fossil angiosperm (e.g. Sun et al., 1998; Friis et al.,2001) and microfossil as well as macrofossil patterns of theirdiversification (Lidgard & Crane, 1990; Lupia et al., 1999),this argument no longer seems credible. Clearly, the processesthat promoted the origin of the lineage are largely decoupledfrom those that allowed early angiosperms to successfullyinvade established Mesozoic plant communities and tospeciate. What were these processes and how did they play outon the Cretaceous stage?

1. Origin, invasion and diversification in three acts

The origin of the flowering plants remains as much anabominable mystery today as it was when Darwin ruminatedon the topic to his friend J. D. Hooker in July 1879. The fossil

record of the earliest angiosperms is indeed sparse, with theearliest undisputed records simply a few grains of pollen(Doyle, 1983). Some pre-Cretaceous macrofossils suggestangiospermy (e.g. Sanmigueliafrom the Triassic) (Cornet& Habib, 1992), but these are commonly poorly preservedor can be linked to other Mesozoic seed plant groups (e.g.Furculafrom Triassic) (Scott et al., 1960).

The search for the earliest angiosperm if such a quest canbe undertaken may be complicated by several factors. First,the environments in which the early flowering plants likelygrew seldom leave a good fossil record. The montane forestsfavored by modern ANITA lineages are sites of erosion, not

sedimentary deposition. Forests associated with lowland riverscan leave excellent records, but few such systems have beenextensively studied from the Jurassic and Cretaceous. Fresh-

water lakes are one exception. These systems offer excellentpreservations and have been extensively studied because oftheir extraordinary vertebrate fossil records (e.g. the YixianFormation of China from which the renowned feathereddinosaurs have been uncovered) (Qiang et al., 1998). Second,many of the worlds latest Jurassic and Early Cretaceousterrestrial rocks have yet to be extensively studied. Depositsin the Middle East, China and New Zealand hold promisefor future fieldwork.

Third, the interpretation of many fossils from this key

interval has been complicated by our preconceptions aboutthe form of the original angiosperm. We expect the firstangiosperm to have the recognizable constellation of synapo-morphies associated with the crown group. This may explain

why pollen has been so useful in documenting the firstappearance of angiosperm. The tectate collumelate form ofthe angiosperm pollen wall is a unique trait that appears earlyin the history of the lineage. Macrofossils have been morechallenging. In the modern world, flowers are easy to recog-nize. However, many Mesozoic lineages (e.g. Sanmiguelia,


16/26

Tansley review


Research398

Cornet & Habib, 1992; Pentoxylon/Carnoconites, Bose et al.,1985; Cycadeoidea, Crepet, 1974; Williamsonia/Weltrichia,Delevoryas, 1991) produced flower-like reproductive struc-tures. Net-veined leaves, another feature of angiospermy, alsoappeared in Gnetumand a variety of late Paleozoic seed fernfoliage including Linopteris, Reticulopteris and Lonchopteris(Taylor & Taylor, 1993). Syndetocheilic stomata (guard and

subsidiary cells originate from one epidermal cell) characterizeangiosperms, but also the Bennettitales and possibly otherMesozoic seed ferns. Xylem vessels also present problems.Some basal angiosperms lack vessels (e.g.Amborella), whereassome nonangiosperm seed plants possess them (i.e. Gnetales,gigantopterids). However, vessels in nonangiosperm lineagesdiffer subtly in structure and ontogeny (for a review, see Feild,2004). Even the hallmarks of flowering plants ovulessurrounded by sporophytic tissue (closed carpel) andpseudo-double fertilization appear in other lineages (e.g.Caytoniales, Dilcher, 1979 and Gnetales, Friedman, 1992,respectively). Such widespread convergence among co-

occurring Mesozoic lineages and the early angiosperms thatreplaced them likely points toward environmental factorsfavoring such traits (Feild, 2004). However, until this recordis reanalysed with an eye toward convergence, and until theorigin of the features can be timed and linked to localenvironmental conditions, rather than the unique superiorityof angiosperms, such factors will remain elusive. In the end,conventional wisdom in paleontology can be shattered with asingle stroke of the rock hammer and a lucky find.

The low species richness and ecological similarity amongextant ANITA lineages, combined with the rarity ofangiosperm fossils before the BarremianAptian, suggest that

the traits promoting angiosperm diversification arose wellafter the origin of the lineage. Thus, the dark and disturbedecology did not trigger the angiosperms mid-Cretaceousradiation. Instead, this lifestyle may have been a way forangiosperms to gain a roothold in well-established Mesozoicplant communities. Architectural flexibility and dispersion ofmeristems may have allowed early angiosperms to flourish indisturbed understory habitats. Multi-stemmed growth, char-acterized by large bud banks for shoot iteration followingdamage, allows long-term persistence in habitats that experi-ence a wide variety of disturbances. This contrasts with mod-ern understory cycads that recover slowly from disturbanceand are vulnerable to loss of their fewer apical meristems.

With the evolution of vessels, carbon costs for shoot and rootreiteration may have decreased, making early angiospermseven more efficient in their recovery from disturbance underlow light. In addition, rhizomatous and lianoid habits, andextensive vegetative propagation increase the discovery andexploitation of ephemeral, resource-rich patches in the forestunderstory. In contrast, the squat, unbranched, sparselyleaved forms of many Bennettitales (e.g. Cycadeoidea andWilliamsonia) and cycads were probably less able to use patchyunderstory resources. If Early Cretaceous forests experienced

significant increases in disturbance frequency or intensity, darkand disturbed angiosperms may have seized the opportunityto invade.

If the dark and disturbed ecology did not promoteexplosive diversification, what did? Most researchers havefocused on single traits when evaluating the key innovationscommonly associated with spurring increases in speciation

rate (e.g. Sanderson & Donoghue, 1994; Donoghue, 2004).However, it may be combinations of traits, occurringtogether, that spurred angiosperm radiation into a wide rangeof new habitats. The combination of vessels, reticulate leafvenation and a relatively more flexible photosynthetic mech-anism needed to handle subcanopy sun flecks rather thanany of these traits alone may have formed the trait complexthat prompted angiosperms to move into new habitats anddiversify. None of these functional traits were necessarilyunique to Mesozoic angiosperms, but their combination was.Modifications to vessels (e.g. longer functional lengthproduced by simple perforation plates) increased hydraulic

conductivity. This vascular system, combined with increasingvein density in leaves, could support the higher transpirationrates needed to control thermal load in sunny environments,

which, in turn, allowed the evolution of higher photosyn-thetic rate. Thus, co-occurrence of vessels, reticulate venationin leaves and high photosynthetic rates may have allowed

Aptian angiosperms to move into sunnier and dryer habitats.In turn, new habitats exposed angiosperms to novel pollina-tion, predation and dispersal pressures that could set offexplosive diversification. While testable, this hypothesisawaits a more detailed analysis of relationships amongangiosperms at the cusp of the mid-Cretaceous radiation.

This, in turn, awaits further fossil discoveries and advances intechniques for analysing morphological data in a phylogeneticcontext.

2. Aquatic origin or early aquatic invasion?

Although some phylogenetic studies have suggested anaquatic ancestor for angiosperms, this seems unlikely froma variety of ecological and developmental perspectives.First, once submerged, the angiosperm lineage would havefaced significant barriers to subsequent speciation andmorphological innovation. Fewer than two per cent ofliving angiosperm species are aquatic and many of these rep-

resent monotypic genera (Cook, 1999). This suggests thatdiversification is curtailed in lineages that move from land to

water (Cook, 1999; Donoghue, 2004). This claim requiresadditional phylogenetic scrutiny, however, because someaquatic lineages are more speciose than their terrestrial sister-groups (e.g. Utricularia).

Second, an aquatic lifestyle imposes radically differentbiophysical and biomechanical demands on plants compared

with life in air. The diffusivity of O2and CO2in water is sub-stantially lower than that in air. A thousand-fold increase in


17/26

Tansley review


Research 399

fluid density also means that aquatic plants operate withgreater diffusion constraints and experience predominantlytensile forces of waves and currents rather than the compres-sive effects of gravity (Pedersen, 1993; Niklas, 1997). Whereasthe shift to an aquatic environment would initially favor theexploration of a new set of vegetative designs in response to anovel environment, extant angiosperm aquatics show that

these experiments leave little room for morphological diversi-fication. A strong convergence in form and function amongextant aquatic angiosperms from disparate clades suggests thatonly a narrow range of solutions to the special problems ofaquatic life is contained within the angiosperm developmen-tal repertoire (Sculthorpe, 1967; Cook, 1999).

Third, once aquatic morphologies are adopted, the lineagesseem unlikely to reinvade land. Out of approximately 200terrestrial-to-aquatic transitions in flowering plants (Cook,1999), monocots, which were either ancestrally aquatic ororiginated with a high physiological dependence on water ashyperhydates (i.e. lake-margin plants that grow in water-

saturated soils or in the water itself with emergent leaves andflowers based onAcorusand the Alismatales as early diverginglines) (Les & Schneider, 1995; Tamura et al., 2004; S. Graham,pers. comm., 2004), appear to be the only group that re-evolved numerous terrestrial forms (Fig. 6; Les & Schneider,1995). This one-way pattern likely stems from a difficulty inre-acquiring key traits useful for life on land. For example,

water lilies, and most likelyArchaefructus, lack a cambium. Anorigin for angiosperms among these lineages would requirethree to five re-inventions of the simple bifacial cambiumobserved in most lignophytes (i.e. seed plant out-groups) andmany angiosperms (Fig. 6). Considering angiosperms as a

whole, the cambium seems easy to lose but harder (if notimpossible) to re-evolve without a distinctly different struc-ture and function. For instance, monocots have evolved anovel type of unifacial cambium (etagen cambium) (Tomlinson,1995). This cambium is situated near the periphery of thestem, and produces derivatives that differentiate into newvascular bundles scattered in the stem, each containing bothxylem and phloem (Tomlinson, 1995). To date, no such tell-taledifferences in cambium structure or development haveemerged among the woody basal angiosperms. Further com-parative data on the development and genomes of these lineagesmight help to close the book on the aquatic origin hypothesis.

If angiosperms did not originate in aquatic habitats, then the

lineage certainly explored wetlands early in its history. Beyondthe all-aquatic Nymphaeales, ancestrally aquatic monocots,Nelumbo(2 species, sister-group to Platanus), and some Ranun-culus are other examples of near-basal aquatic angiosperms.Ceratophyllum(6 species of rootless submersed herbs), if unre-lated to monocots, may represent another aquatic experiment.

What fostered the early aquatic exploration? First, someearly angiosperm lineages may have escaped crowded habitatson land by evolving toward water. This process has beensuggested as justification for an aquatic origin (Sargant, 1908;

Arber, 1920; Sun et al., 2002). However, it may be morelogical in this context because an early aquatic invasion is con-sistent with Early Cretaceous diversification among bothterrestrial and aquatic lineages without requiring the reacqui-sition of traits, such as a bifacial cambium, lost in the transi-tion to a submerged lifestyle. Second, a rich exaptive trait pool(sensuGould, 2002, traits that can be co-opted for new func-tional roles) inherited from ancestors growing in the disturbedforest understory may have primed lineages for the aquatictransition. For example, phytochrome mechanisms necessaryfor germination in low light and root pressure with hydath-odes needed for submerged transpiration (Pedersen, 1993)

were already in place in the common ancestor of extantangiosperms (Feild et al., 2003a; Mathews et al., 2003).Third, the aquatic transition may have allowed early lineagesto explore sunny environments without extensive modifica-tions to water-use physiology because they were already accus-tomed to shady environments where water is not limiting.Thus, early angiosperms could move to open water by up-

regulating photosynthesis and losing the cambium.

VI. The environmental context of earlyangiosperm evolution

The origin of new traits and unique trait combinations innewly branched clades have little long-term evolutionarysignificance in the absence of environmental circumstancesthat allow novel clades to become ecologically important anddiversify (Knoll & Carroll, 1999). Such key environmentalopportunities combine with traits key innovations to giverise to significant evolutionary radiations (for a recent review,

see Donoghue, 2004). Similarly, environmental opportunitiesare essential to allow a new clade, such as the Early Cretaceousangiosperms, to gain a roothold in established plantcommunities of the time (von Hagen & Kadereit, 2003). Wecan consider such environmental opportunities at two spatialand temporal scales. Forces such as tectonics, changes inatmospheric composition and climate act at regional to globalscales and over intervals of tens of thousands to millions ofyears. Such large-scale and long-term changes alter the abioticcontext in which plant communities live. In contrast, a varietyof biotic factors such as pollination, dispersal and disturbanceact at local spatial scales and over very short time spans. Bothsmall and large-scale environmental changes characterized the

Early Cretaceous and formed the environmental context forthe angiosperm invasion.

1. Large-scale abiotic factors

The world into which angiosperms emerged has traditionallybeen assumed to be uniformly warm and equable (Brooks,1906; Schwarzbach, 1963). However, reports of short-termglobal seal-level fluctuations (Abreu et al., 1998), temperatureinference from oxygen isotope data (Hochuli et al., 1999;


18/26

Tansley review


Research400

Veizer et al., 2000) and sediments related to glaciation(Frakes et al., 1992; Alley & Frakes, 2003) suggest that theEarly Cretaceous may have been cooler and more climaticallyvariable than previously believed. For example, detailedanalyses of phytogeography and the distribution of climatesensitive sediments show significant spatial and temporalclimate variability (Rees et al., 2002). Rees et al. combineddetailed reconstructions of tectonic plate configurations

with the distribution of climate sensitive sediments (e.g. coaland evaporates) and fossil floras to reconstruct the globaldistribution of ancient climates for the latest Jurassic. Duringthis interval highlands in equatorial Gondwana, to whichsome of the earliest angiosperm pollen fossil have beenlinked (Brenner, 1996), experienced cool (Veizer et al., 2000),summer-wet seasonal (Rees et al., 2002) climates. Thisclimate mode continued into the Berriasian and Valanginian(Krassilov, 1973; Allen et al., 1998; Eberth et al., 2001).Hauterivian time witnessed a brief but dramatic cooling thatis observed in both low and mid latitudes (Price et al., 2000;

Veizer et al., 2000). Another short-lived climate perturbationcoincided with the Early Aptian oceanic anoxic event(OAE) ( Jenkyns, 1980; Bralower et al., 2002a). OAEs are ofrelatively short duration (lasting 0.51 Myr) and producemajor changes in marine faunas (Bralower et al., 2002a).Calculations by Beerling et al. (2002) suggest that asubstantial release of methane clatharate, likely due to achanging tectonic regime, resulted in a short-lived globaltemperature increase of up to 3C. Methane hydrate occurs

where conditions of temperature and pressure allow methaneproduced by sedimentary bacteria to become trapped in thecrystal lattice of ice (Haq, 1998). Once thought to occur only

in deep space, extensive deposits have been found on Earthscontinental shelves (Dickens, 2001). When temperature andpressure conditions change, large reservoirs of gas hydrate canmelt, releasing methane into the water column, where it isoxidized to produce CO2(Dickens et al., 1997; Haq, 1998;Dickens, 2001; Jahren et al., 2001). This sudden release ofCO2has been linked to rapid climate change (Dickens et al.,1997; Haq, 1998), periods of ocean anoxia and massextinction ( Jahren et al., 2001; de Wit et al., 2002). TheCretaceous greenhouse began during the late Aptian and

Albian periods and prevailed through the end of theCretaceous (Barron, 1983; Frakes & Francis, 1988; Veizeret al., 2000; Huber et al., 2002). During the greenhouse

interval, global climate varied from warm in the late Albianthrough to the late Cenomanian to hot during the latestCenomanian through early Campanian, with cooling towardthe end of the period (Huber et al., 2002). Although themagnitude of climate variation estimated from marine oxygenisotope data (e.g. Huber et al., 2002) has been questionedbecause of uncertainties in paleosalinity and assumptionsabout the presence or absence of glacial ice (Fassell &Bralower, 1999), data from fossil floras have produced similarpatterns (Frakes, 1999), which lends support to the trends.

Proposing a direct link between climate change and eitherthe origin or early radiation of angiosperms seems premature.However, it is interesting to note that the initial radiation ofangiosperms occurred during Aptian transition betweenicehouse and greenhouse conditions (Jahren et al., 2001). In theearly Aptian, carbon and oxygen isotope data show relativelylow temperatures and high marine productivity, suggesting

humid conditions (Hochuli et al., 1999), although regionalvariability in rainfall would be predicted then as today. Warm-ing began in the latest Aptian and continued through the

Albian. Oxygen isotope values in terrestrial calcite cementshow an increase of approximately 10C in the average tem-perature of rainwater from the mid Aptian to late Albianat high southern latitudes (Ferguson et al., 1999). Coolingbegan following the AlbianCenomanian boundary, withdramatic warming beginning at CenomanianTuronian boundarytime (Frakes, 1999).

Clearly, we can no longer assume a stable climate duringthe interval of angiosperm origin and early radiation. Thus,

we cannot dismiss climate as a player in the angiosperm eco-logical incursion. However, a mechanistic hypothesis linking

warming and/or fluctuating climate in the mid Cretaceous toangiosperms infiltration of pre-existing plant communitieshas not emerged. Changing climate may have disrupted exist-ing plant communities, perhaps causing extinctions, andallowing the new lineage to gain a roothold. We know thatplant species associations become flexible and are prone tochange during times of climatic flux (e.g. Coleman et al.,2003; Mueller et al., 2003; Schonswetter et al., 2003;McKinnon et al., 2004). However, a tightly coupled record ofmid Cretaceous climate and floristic change would be necessary

to test such a hypothesis.To complicate the picture further, climate change in themid Cretaceous appears to be a consequence of changesin global tectonics, which had other effects as well. TheCretaceous warming events described above have been linkedto increased atmospheric CO2(Arthur et al., 1985; Larson,1991) produced by elevated rates of seafloor spreading(Tarduno et al., 1991; Bralower et al., 2002a; Jenkyns, 2003;Poulsen et al., 2003) beginning in the Aptian. Beyond influ-encing climate, CO2flux may have had direct effects on thebiosphere. For example, the Jurassic and Cretaceous arecharacterized by at least seven episodes of increased organiccarbon burial in the worlds oceans and widespread dysoxia in

the deep ocean ( Jenkyns, 1980; Bralower et al., 2002a). SomeOAEs have been linked to the release and oxidation ofmethane hydrate trapped on continental shelves (Beerlinget al., 2002; Jahren et al., 2001 submitted). Others have beenlinked directly to volcanic CO2(Jenkyns, 2003). Whateverthe cause, these events produce rapid and significant (up to600 ppmv) increases in atmospheric CO2(Jahren et al., 2001;Beerling et al., 2002).

In addition to an impact on global climate, changes inatmospheric composition may have had direct effects on plant


19/26

Tansley review


Research 401

communities. Nonetheless, there is considerable debate overthe effects such changes might have had. Data from modernplant communities show that significant changes in theabundance of CO2alter competitive interactions among co-occurring species and can lead to changes in species com-petition and relative abundance within communities (Bazzaz &Miao, 1993). Some experimental results (Bazzaz et al., 1990) andmodeling (Bolker et al., 1995) suggest that angiosperms mayrespond more vigorously to increasing CO2than do conifers,resulting in changes in forest succession and communitycomposition over time. If such conclusions can be appliedto the Early Cretaceous, rapid increases in CO2such asthose observed in association with methane hydrate release( Jahren et al., 2000) might have facilitated penetration ofangiosperms into established Mesozoic plant communities.However, all of the experimental studies are based on thehighly derived physiology of modern angiosperms (i.e. exclu-sively temperate eudicots) and conifers (Pinaceae). More workon modern nonangiosperm seed plants and a better under-

standing of the physiology of Cretaceous angiosperms is neededto further test this hypothesis. Others have suggested that fall-ing atmospheric CO2levels during the Jurassic and Cretaceous(Ekart et al., 1999; Berner & Kothavala, 2001; Rothman, 2001)may have favored angiosperms (McElwain et al., 2005).McElwain et al. noted a correlation between changes in plantcommunities and Early Cretaceous declining CO2, predictedby global carbon models. They noted that increases inangiosperm diversity and relative abundance occurred duringfall of CO2concentrations, and argue that CO2 limitation

would favor more efficient angiosperm physiologies, throughevolution of xylem vessels, net-veined leaves and highly

responsive stomatal opening/closure kinetics, over lessadaptively responsive seed plant and fern lineages. Currentlythere is considerable debate in the geologic community overthe quantitative accuracy of the wide range of models andproxies used to deduce patterns in pCO2change through time(e.g. Boucot et al., 2004). Until there is greater agreement onthe patterns and magnitude of pCO2change during the criticalinterval, and until we understand more about the physiologicalresponses of basal angiosperms and nonangiosperm seedplants to changing CO2, mechanistic arguments about therole of atmospheric composition on the early evolution of thelineage remain unsatisfying

Both of these arguments suggest an underlying physiol-

ogical superiority of angiosperms compared to the otherMesozoic seed plant lineages with which they shared the EarlyCretaceous landscape. This assumption has haunted discus-sions about the early evolution of flowering plants for decades(Stebbins, 1974, 1981; Regal, 1977; Burger, 1981; Bond,1989). However, such assumptions are based on modern,derived physiologies for both angiosperms and conifers. Howrelevant were such distinctions in Early Cretaceous commu-nities? Until now, such questions have been unanswerable.However, the widespread use of carbon isotope discrimina-

tion to infer competitive ability in crop (Read et al., 1991;Condon et al., 1993) and wild (Toft et al., 1989) opens a newavenue test the hypothesis of angiosperm competitive superi-ority. Current agricultural applications use carbon isotopediscrimination to calculate water use efficiency (WUE), which,in turn, is a proxy for the efficiency of photosynthetic carbonfixation (Read et al., 1991; Condon et al., 1993). Once theconstraints of diagenetic alteration of the carbon isotopicsignature of fossil plant tissue are understood (Tu et al., 2004.),such approaches could be applied to seriate the members ofancient plant communities according to carbon fixationefficiency. This might allow us to finally determine whetherangiosperm dominance was an inevitable legacy of thelineages origin.

2. Small-scale biotic factors

Insects, particularly pollinators, have long been recognized asessential partners in angiosperm evolution (Crepet & Friis,

1987; Crepet, 1996), primarily because differentiation ofpollination syndromes has played such an important rolein the diversification of core angiosperm lineages. Manybasal angiosperms are pollinated primarily by insects (waterlilies,Austrobaileya, Chloranthus, Illicium, Sarcandra) or bya combination of insects and wind (Amborel la, Trimenia,Hedyosmum), suggesting that an insect or combinationsyndrome for ancestral angiosperms (Luo & Li, 1999; Tosakiet al., 2001; Bernhardt et al., 2003; Thien et al., 2003). Insectpollination has been inferred for most of the earlyangiosperms described to date on the basis of features such asabundant connective tissue, ethereal oil cells and valvate

anther dehiscence in stamens of Early Cretaceous flowers, aswell as the presence in the Portuguese floras of insectcoprolites consisting exclusively of angiosperm pollen (Friiset al., 1999, 2000).

Although the radiation of angiosperms and the establish-ment of discrete pollination syndromes in the early LateCretaceous coincides with the first appearance and/ordiversification of many pollinating clades (e.g. lepidopterans,dipterans and pollinating hymenopterans such as bees)(Grimaldi, 1999), there is little direct evidence to suggest thatinsect diversification played an important role in angiospermorigin or early evolution (Labandiera & Sepkoski, 1993).However, insect pollination in the earliest angiosperms may

have permitted outbreeding in isolated individuals or smallpopulations within Early Cretaceous forest understories.

Dinosaurs have also been promoted as evolutionary part-ners with angiosperms. Bakker (1978, 1986) suggested that atransition from canopy-feeding sauropodomorph dinosaurs(e.g. Diplodocus, Apatosaurus) in the Early Cretaceous toground-feeding ornithischian dinosaurs (e.g. Hyps

Documents

Fisiologia Extinçao Flores