The ecophysiology of early angiosperms


Taylor S. Feild. E-mail:


Angiosperms first appeared during the Early Cretaceous, and within 30 million years they reigned over many floras worldwide. Associated with this rise to prominence, angiosperms produced a spectrum of reproductive and vegetative innovations, which produced a cascade of ecological consequences that altered the ecology and biogeochemistry of the planet. The pace, pattern and phylogenetic systematics of the Cretaceous angiosperm diversification are broadly sketched out. However, the ecophysiology and environmental interactions that energized the early angiosperm radiation remain unresolved. This constrains our ability to diagnose the selective pressures and habitat contexts responsible for the evolution of fundamental angiosperm features, such as flowers, rapid growth, xylem vessels and net-veined leaves, which in association with environmental opportunities, drove waves of phylogenetic and ecological diversification. Here, we consider our current understanding of early angiosperm ecophysiology. We focus on comparative patterns of ecophysiological evolution, emphasizing carbon- and water-use traits, by merging recent molecular phylogenetic studies with physiological studies focused on extant basal angiosperms. In doing so, we discuss how early angiosperms established a roothold in pre-existing Mesozoic plant communities, and how these events canalized subsequent bursts of angiosperm diversification during the Aptian–Albian.


The flowering plants, or angiosperms, are the most species-dense and ecologically diverse branch on the green plant tree of life (Soltis et al. 2005). In addition to impressive diversity (∼ 250 000 species), angiosperms have evolved an unparalleled breadth of ecophysiological performance, underpinned by high-capacity and flexible metabolisms and extreme structural diversity (Bond 1989; Sperry 2003). In metabolic performance, angiosperms range from deep-shade rain forest herbs growing at the limits of land plant autotrophy to sun-loving weeds bounding the other extreme with the highest CO2 assimilation, O2 evolution and water loss rates known. Associated with broad metabolic performance, angiosperms exhibit corresponding diverse whole-plant vascular plumbing designs and ecomorphological variation. Beyond impressive diversity, angiosperms are the modern vegetational dominants. Angiosperm ecophysiologies dictate present-day terrestrial water, nutrient and carbon cycles, as well as feedback on community disturbance and regeneration dynamics (Knoll & James 1987; Volk 1989; Berner & Kothavala 2001). Finally, angiosperm ecophysiological and structural diversity is foundational in the generation and maintenance of terrestrial biodiversity by providing diverse resources for pollinators, dispersers and herbivores (Farrell 1998; Grimaldi 1999).

Angiosperm diversity and ecological dominance, however, arose relatively recently. Unlike other major land plant lineages (i.e. ‘bryophytes’, lycophytes, ferns and other seed plants), which extend into the Paleozoic, the first unequivocal angiosperm fossils appear nearly 100 Ma later in low paleolatitudes of southern Laurasia and northern Gondwana (∼ 135 Ma; Knoll, 1986; Crane, Friis & Pedersen 1995; Brenner 1996; Lupia, Lidgard & Crane 1999). Within ∼ 15 Ma, angiosperms radiated into a riot of unprecedented variation in foliar, reproductive and pollen morphological diversity (Hickey & Doyle 1977; Upchurch 1984; Lidgard & Crane 1990; Friis, Pedersen & Crane 1999, 2000; Lupia et al. 1999; Crepet, Nixon & Gandolfo 2004; Hochuli, Heimhofer & Weissert 2006). The interpretation of these fossils and molecular clock analyses of divergence times within the clade demonstrate that most of the major angiosperm lineages appeared during a relatively short (10–15 Ma) geological window (Crane et al. 1995; Friis et al. 1999, 2000; Bremer 2000; Magallón & Sanderson 2001, 2005; Crepet et al. 2004; Davies et al. 2004; Bell, Soltis & Soltis 2005).

Angiosperm diversification produced a cascade of consequences for Mesozoic plant communities. Angiosperms advanced to the poles of both hemispheres and began dominating a few low-latitude communities by the Aptian times (∼ 112 Ma; Romero & Archangelsky 1986; Wing, Hickey & Swisher 1993; Lupia et al. 1999; Hochuli et al. 2006). Although they accounted for by far the largest proportion of species, angiosperms did not dominate the majority of ecosystems, even in the Latest Cretaceous (Wing et al. 1993). However, by the Late Cretaceous, many plant communities were transitioning into angiosperm-replete ones as ferns, lycophytes and other seed plant lineages declined in abundance. Several ancient groups such as theBennettitales, Caytoniales, Chierolepidaceae conifers and Pentoxylon dwindled to extinction (Knoll 1986; Lupia et al. 1999). In contrast, some non-angiosperm clades (e.g. pleurocarp mosses, epiphytic ferns, epiphytic lycophytes and some podocarp conifers) diversified during the Mid- to Late Cretaceous, possibly spurred by new opportunities created by angiosperms (Brodribb & Hill 1997; Wikström & Kenrick 2001; Schneider et al. 2004). Angiosperms also energized diversification within many insect clades (Farrell 1998; Grimaldi 1999; Moreau et al. 2006). Examples of these novel ecological interfaces created by angiosperms include changed understorey light and disturbance regimes created by multi-layered canopies, diverse floral and bark morphologies, and innovations in leaf form and chemistry (Brodribb & Hill 1997; Wikström & Kenrick 2001; Shaw et al. 2003; Schneider et al. 2004). Moreover, some have argued that once angiosperms reached dominance their physiologies changed atmospheric composition. The generally greater leaf photosynthetic capacities and root-weathering properties of angiosperms may have ended the extremely high-CO2 (i.e. 1500–3500 μLL−1CO2) world of the Mesozoic by drawing down atmospheric pCO2 during the Cretaceous (Knoll & James 1987; Volk 1989).

The general pattern of Cretaceous angiosperm diversification is now broadly sketched out, and new fossil discoveries continue to refine our understanding of its systematic scope and the rates of vegetation change (Friis et al. 1999, 2000; Lupia et al. 1999; Nagalingum et al. 2002; Heimhofer et al. 2005; Hochuli et al. 2006). However, the ecophysiology and environmental interactions stimulating the early angiosperm radiation remain unresolved. This constrains our ability to diagnose the selective pressures and habitat contexts responsible for the evolution of fundamental angiosperm features, such as flowers, rapid growth, xylem vessels and net-veined leaves, which drove phylogenetic and ecological diversification.

As a starting point for attaching a comprehensive ecophysiological dimension – one merging biotic and physical processes – to the early angiosperm radiation, a picture of what the first flowering plants looked like and how they functioned in their environment is needed. From these vantage points, we can polarize the evolutionary directions of ecophysiological mechanisms that ultimately birthed the high-capacity, stream-lined and plastic photosynthetic/hydraulic functions that are the basis of an angiosperm overprint on energy flow and evolutionary diversification in the modern biosphere (Stebbins 1974; Bond 1989). The aim here was to examine how recent phylogenetic, ecophysiological and geological discoveries are bringing new perspectives to our understanding of early angiosperm ecology and functional biology. We begin by briefly characterizing what we have learned recently about the fundamental phylogenetic context of angiosperms. We then discuss how the identification of extant ‘basal’ angiosperm lineages and understanding their ecophysiology has revised our image of how early angiosperms functioned in their environments and the directions of their early ecological evolution.


Angiosperm out-groups

It is not hyperbole to claim ‘all bets are off’ on the question of angiosperm sister groups. Beginning in the 1980s, many believed that the living and extinct gymnosperm out-groups to angiosperms had been identified (Crane 1985; Doyle & Donoghue 1986). The ‘anthophyte’ hypothesis emerged as the consensus of several morphological phylogenetic analyses (Crane 1985; Doyle & Donoghue 1986). Under this hypothesis, the Gnetales are the closest living relatives of angiosperms and two extinct groups, the Pentoxylon and the Bennettitales, emerged as anthophyte stem lineages (Doyle & Donoghue 1986). Resolution of the anthophyte clade suggested that Gnetales, Bennettitales, Pentoxylon and angiosperms had homologous flower-like reproductive organs (Crane 1985; Doyle & Donoghue 1986).

However, molecular phylogenetic data, which bear only on the relations of living taxa, have not supported the anthophyte topology (Bowe, Coat & dePamphilis 2000; Burleigh & Mathews 2004). The Gnetales appear to be unrelated to angiosperms, being sister group to conifers or perhaps derived from within conifers (Bowe et al. 2000; Burleigh & Mathews 2004). These data further suggest that angiosperms have a much deeper root, diverging before the separation of extant gymnosperms. If the purported homologies between living Gnetales and angiosperms (Crane 1985; Doyle & Donoghue 1986; Friedman 1998) are rejected as suggested by recent molecular phylogenies, similar interpretations of fossil taxa such as Bennettitales and Pentoxylon may also be on shaky ground. Currently, a plethora of sister group hypotheses are being explored (Bateman, Hilton & Rudall 2006; Doyle 2006; Hilton & Bateman 2006; Taylor et al. 2006). Still, addressing the question of angiosperm out-groups is axiomatic for polarizing ecological evolution within angiosperms (Bateman et al. 2006). A new approach for inferring phylogenetic affinities of extinct taxa is the biochemical analysis of fossilized organic matter. For example, the presence of oleanane, a biomolecule common to crown-group angiosperms, has been used to discriminate taxa putatively on the angiosperm stem lineage (Taylor et al. 2006). Interestingly, oleanane has been detected in Bennettitales, but not Gnetales. However, oleanane has been reported from ferns (e.g. Adiantum; Nakane et al., 2002), suggesting that the evolutionary dynamics of oleanane are not confined to angiosperms and their close relatives.

Basal angiosperm lineages

Molecular data have greatly clarified our view of the angiosperm phylogenetic tree. After decades of conflicting hypotheses (Soltis et al. 2005), a diversity of data has converged on a largely consistent topology near the root of the angiosperm phylogeny (Mathews & Donoghue 1999, 2000; Qui et al. 1999; Soltis et al. 2005;Saarela et al. 2006). Here, we refer to lineages resolved near the root of the angiosperm phylogenetic tree as ‘basal’. The term ‘basal’ has been criticized as a synonym for ‘primitive’ (Crisp & Cook 2005). Basal groups, however, can be apomorphic (containing derived characters; Crisp & Cook 2005; Feild & Arens 2005). Similarly, calling these lineages ‘early diverging’ is unsatisfying because the sister groups to basal lineages diverged at the same time (Crisp & Cook 2005). Finally, only nodes on the tree are truly ‘basal’. The extant plants (terminals on the tree) that provide characters to populate reconstructed ancestral nodes are not. While recognizing these points, we prefer ‘basal lineage’ as a shorthand for ‘a clade whose stem lineage is attached to a basal or near-basal node’. More importantly, when several lineages branch off successively below a major clade, as is the case for angiosperms, any shared character states can be inferred to be ancestral to angiosperms as a whole, even if each of these lineages is apomorphic in other characters.

Most molecular analyses have the New Caledonian shrub Amborella trichopoda diverging closest to the angiosperm root (Mathews & Donoghue 1999, 2000; Qiu et al. 1999; Soltis et al. 2005). The next branch includes the aquatic herbs Nymphaeales (‘the water lilies’; two families, Cabombaceae: Brasenia, Cabomba and Nymphaeaceae: Barclaya, Euryale, Ondinea, Nuphar, Nymphaea, Victoria; Mathews & Donoghue, 1999; Qiu et al. 1999), although some studies unite water lilies with Amborella to form the sister group to all other angiosperms (Barkman et al. 2000; Zanis et al. 2001). An astonishing recent result placed the enigmatic monocot, Hydatellaceae (Hydatella and Trithuria; a small group of highly reduced ‘isoetid’ freshwater aquatic herbs; Hamann 1976), sister to Nymphaeales (Saarela et al. 2006). Morphological characters of Hydatellaceae, a former Poales member, also suggest affinities with Nymphaeales and the basal angiosperm position: monosulcate pollen, anomocytic stomata, ascidiate carpels, four-cell nucleate embryo and a palisade exotesta in the seed (Saarela et al. 2006). Members of the third basal lineage, Austrobaileyales, include Austrobaileya (a liana) at the base of the clade, and Trimenia (small trees to lianas) forms a clade with Illicium (shrubs to small trees) and the Schisandraceae (consisting of the lianous genera Kadsura and Schisandra).

Following the Amborella-Nymphaeales plus Hydatellaceae-Austrobaileyales grade, combined morphological and molecular analyses (Doyle & Endress 2000; Eklund, Doyle & Herendeen 2004; Duvall et al. 2006) suggest that Chloranthaceae (Ascarina, Chloranthus, Hedyosmum and Sarcandra) or Ceratophyllum, a genus of highly reduced freshwater aquatics followed by the Chloranthaceae (Duvall et al. 2006) form the next divergence(s). Other phylogenies have variously placed both of the taxa as sister groups to monocots or eudicots (Mathews & Donoghue 2000; Soltis et al. 2005). The remaining angiosperms form three, well-supported monophyletic lineages: eudicots (tricolpates), eumagnoliids (Laurales, Magnoliales Piperales and Winterales) and monocots. Unlike the widespread agreement on the Amborella-Nymphaeales-Austrobaileyales basal grade, relations among these lineages are still in flux with a wide range of alternative topologies (Soltis et al. 2005; Duvall et al. 2006).


Because out-group relations remain so uncertain, many workers have used the recent overhaul of extant ‘basal’ angiosperm relationships to infer Early Cretaceous angiosperm biology (so-called ‘top-down’ approach; Bateman et al. 2006). Here, the form and function of basal lineages is surveyed and specific traits are mapped onto hypotheses of angiosperm phylogeny to reconstruct the traits of the angiosperm common ancestor and early patterns of character evolution (Doyle & Endress 2000; Doyle 2001, 2006; Williams & Friedman 2002; Feild et al. 2004; Feild & Arens 2005; Soltis et al. 2005).

An analysis of ecofunctional character evolution among extant basal angiosperms indicated two contrasting options for the ancestral ecology of angiosperms. One option is represented by Amborella, Austrobaileyales and some of the Chloranthaceae as woody plants residing in damp, dark and disturbed habitats (Feild et al. 2004). Another possibility is represented by Nymphaeales (and now also Hydatellaceae; Saarela et al. 2006) as aquatic herbs (Feild et al. 2004). For the purpose of this review, we consider the suite of ecological and physiological characteristics predicted for early angiosperms under the first hypothesis. Because an aquatic origin for angiosperms is less supported phylogenetically and unlikely for several reasons (see further discussion), the ecophysiology of Nymphaeales is not reviewed here, but has been recently reviewed elsewhere (Feild & Arens 2005).

Growth forms, habitats and regeneration ecologies

Feild et al. (2004) hypothesized that early flowering plants grew as woody shrubs to small trees that passed through developmentally flexible juvenile growth forms. Seedlings of Amborella and many Austrobaileyales grow with a soil-hugging ‘creeping’ habit and develop thickened, ray tissue-rich lignotubers (Feild & Arens 2005). Seedlings and saplings generally consist of several scandent, leaning and/or gangly shoots arising from buds in the lignotuber. Gradually, saplings adopt more typical multi-stemmed shrub or small tree forms with erect branches, but growth habits in other taxa remain lax, with scandent cane-like branching patterns. In response to wounding, many arborescent taxa revert back to the juvenile pattern and produce a viny phalanx of basal sprouts that explore the forest floor for suitable establishment sites.

The climates of extant basal lineages suggested that early angiosperms occurred in wet, upland tropical regions. These habitats vary from 1500 to 7000 mm rainfall per year, with frequent cloud immersion and occult precipitation. In these sorts of habitats, basal lineages regenerated in forest understorey microsites characterized by low light and disturbance (Feild & Arens 2005). Examples of these ‘damp, dark and disturbed’ habitats include well-drained, sandy and brittle clay soils along small understorey water courses, on washout-prone slopes and on or in between rotting ‘nurse’ logs (Fig. 1a,d; Feild et al. 2004). Multiple sapling shoots and early commitment to resource storage seems consistent with dual tolerances of disturbance and shade. By presenting a sparse target for understorey disturbance and rooting over a broad soil surface area, viny saplings can colonize and persist on unstable substrates. Furthermore, a lateral spread across the forest floor may enable saplings to ‘forage’ for patchy understorey resources, such as sunflecks and soil nutrients.

Figure 1.

Seeding regeneration habitats and high-irradiance responses of basal angiosperms. (a) Regeneration of Amborella trichopoda (Amborellaceae) seedlings on exposed mineral soils underneath the uproot of a fallen subcanopy tree in New Caledonia. Amborella seedlings recruit alongside a diversity of ferns and mosses in these shady disturbed patches. (b) A plagiotropic cane-branch from a mature A. trichopoda shrub growing into a small light gap. Along a gradient of increasing light, from base to tip, leaves along the branch gradually fold up along the mid-vein. (c) Sun acclimation of Illicium floridanum leaves when plants are forced to grow under full sunlight in a common garden (Auburn, AL, USA). Leaves of I. floridanum become bronzed and chlorotic (on the right) with localized spotting resulting from small clusters of dead cells on the leaf upper surface. This contrasts with the healthy appearance of understorey and subcanopy expanding I. floridanum leaves (on the left). (d) A wet, dark and disturbed habitat, exemplified by I. floridanum from a mesophytic subtropical forest in northern Florida, USA. Here, I. floridanum seedlings recruit along small perennial spring-fed creeks. The sandy soils that plants are rooted often wash out during flooding. Plants are also frequently impacted by falling debris from dense vegetation that overhangs these streams.

Amborella, Austrobaileyales Chloranthaceae seeds are small, which is consistent with disturbance adaptation. An exception is Austrobaileya scandens, which produces the largest fruits (up to 7 cm long and 3 cm thick) and seeds (73 mm3, whereas other basal angiosperm seeds ranged from 0.24 to 18 mm3) among basal angiosperms (Feild et al. 2004). Austrobaileya also appears to be the only basal angiosperm whose seeds are dispersed by living mega-fauna (i.e. by cassowary birds; Williams, unpublished data), which may have been selected for increased seed size. Although large seeds may be viewed as the best strategy in shady environments (Walters & Reich 2000; Moles & Westoby 2004), recent work suggests that small seeds can succeed in the understorey (Metcalfe, Grubb & Turner 1998; Lusk & Kelly 2003). Small seeds may in fact be well suited to steep patches of exposed soil because they do not roll off or washout easily. In addition, soils cleared of leaf litter accommodate the limited amount of etiolated growth (no basal angiosperms germinate green in the dark; Mathews, Burleigh & Donoghue 2003) that is possible for small-seeded basal angiosperms.

Leaf ecophysiological function

Leaf cross-sectional anatomy

The comparative leaf anatomy of Amborella, Austrobaileyales and Chloranthaceae reveals features linked to high performance in wet, shady habitats. Here, basal angiosperms exhibit anatomies with increased durability and geared for efficient photosynthesis in light-limited habitats. For example, a mesophyll dominated by spongy parenchyma tissue (i.e. no palisade layers) is ancestral among angiosperms (Feild et al. 2004). Spongy parenchyma creates numerous light reflecting air–water interfaces that increase light scattering within the leaf (Terashima & Saeki 1983; DeLucia et al. 1996). In combination with a horizontal leaf orientation, the spherical cells of spongy parenchyma are very effective in collecting the diffuse light characteristic of dense understorey habitats under frequent cloudy conditions (Smith et al. 1997). By extending the path lengths of photons inside leaves, spongy mesophyll increases light absorbance per unit of chlorophyll invested (Terashima & Saeki 1983; DeLucia et al. 1996; Smith et al. 1997). High intercellular reflectance may maintain light intensities within leaves above the photosynthetic light compensation point during potentially long idling times between understorey sunflecks, thus priming the light induction state of photosynthetic enzymes (DeLucia et al. 1996).

Although not ancestral characters, ethereal oil cells and mucilage ducts developed early in angiosperm evolution, as represented by occurrences in Austrobaileyales and Chloranthaceae leaves (Upchurch 1984; Metcalfe 1987; Carpenter 2006). Such structures may favour understorey success by bolstering leaf durability. Leaf durability matters in deep-shade habitats because leaves must live long to recoup the carbon invested in their production (Chazdon et al. 1996). Ethereal oil cells consist of a large lacuna filled with oil that connects to the leaf surface through an epidermal gland (i.e. radiostriate cells; Upchurch 1984; Carpenter 2005). In Trimenia, Chloranthaceae, Illicium and Schisandraceae (Metcalfe 1987; Todzia 1988), mucilage ducts ramify within the leaf mesophyll. Oil volatilized by radiostriate cells may deter herbivore browsing. Furthermore, extrusion of mucilage when insects chew leaves could seal breached tissues and block continued attack by gumming up insect mouth parts (i.e. analogous to latex canals; Zakucki et al. 2001). Considering the probably low food quality of most basal angiosperm leaves (i.e. low investment in photosynthetic enzymes, hence the high carbon-to-nitrogen ratio), oil cells and mucilage canals would seem to further dissuade herbivore attack. Field studies, however, are needed to test these hypotheses because no basal angiosperm species have been sampled for herbivore community structure or ecology.

Epidermal anatomy

The leaf epidermal anatomy of Amborella, Austrobaileyales and Chloranthaceae is consistent with ecophysiological performance under high humidity and dimly lit conditions. Nearly all basal angiosperm leaves develop hypostomy (stomata confined to the abaxial surface; Metcalfe 1987), low stomatal densities and large stomatal sizes (Upchurch 1984; Metcalfe 1987; Feild et al. 2001; Kong 2001; Feild, Franks & Sage 2003b; Oh, Denk & Friis 2003; Denk & Oh 2005). Amphistomy, stomata on both leaf surfaces, is weakly developed in several Kadsura and Schisandra chinensis vines (Yang & Lin 2005; Denk & Oh 2005). Stomatal densities in basal angiosperms range from 13 to 88 stomata mm−2(average = 36 stomata mm−2, n = 35 species) and stomatal sizes range from 35 to up to 100 µm (average = 50 µm, n = 35 species) in some species (Jalan 1962; Metcalfe 1987; Kong 2001; Oh et al. 2003; Denk & Oh 2005; Yang & Lin 2005; Feild, unpublished data). These species averages are nested in the low range of stomatal densities and in the upper range of stomatal sizes for angiosperms as a whole (Bongers & Popma 1988; Hetherington & Woodward 2003).

An advantage of hypostomy for understorey leaves is that during sunflecks, stomatal photoinhibition (guard cells are more exposed and thus vulnerable to excess light than internal mesophyll cells; Lawson et al. 2003) is avoided because stomata occur on the more-shaded leaf undersurface (Smith et al. 1997). Hypostomy, in conjunction with large stomatal size and sparse stomatal density, may reduce leaf construction and maintenance costs. Stomata can be highly metabolically active and are associated with a network of physiologically specialized subsidiary cells. Moreover, stomatal movements entail additional energetic costs to run ionic pumps through a local synthesis of hormones (although some of these costs may be offset by guard cell photosynthesis; Roelfsema & Hedrich 2005). Nonetheless, stomatal complexes are probably more expensive to develop as compared with ‘normal’ epidermal cells (Roelfsema & Hedrich 2005). Thus, low stomatal investments may be more cost efficient in light-limited habitats. Cheaper epidermises, however, are likely to be traded off against a fine control of within-leaf gas exchange that would be enabled by a denser, smaller-sized stomatal network. In humid, dense forest understories, however, highly responsive stomatal control may be disadvantageous, because even the fastest stomatal movements will lag behind flickering sunflecks (Knapp & Smith 1990; Chazdon et al. 1996; Kaiser & Kappen 2000).

Finally, the abaxial epidermis of basal angiosperm leaves is more ornamented than the adaxial epidermis. Braided and sinuous cuticular striations, ridges, epicuticular crystals and/or flanges encircle and radiate from the guard cells of Amborella, Austrobaileyales and Chloranthaceae (Upchurch 1984; Metcalfe 1987; Kong 2001; Feild et al. 2003b; Yang & Lin 2005; Carpenter 2006). Highly sculpted cuticles on the abaxial surface have been suggested to contribute to the strong bicolouration noted in Amborella and Austrobaileya leaves (Feild et al. 2001, 2003b). Although air spaces near the abaxial surface of the leaf also contribute to the high reflectance of bicoloured leaves (Woolley 1971; Smith et al. 1997), rugged abaxial cuticle textures could enhance light trapping in the spongy mesophyll, thus increasing leaf carbon-use efficiency, by providing a reflective surface on the internal side of the abaxial epidermis to reduce the amount of light leaking from the leaf bottom (Smith et al. 1997).

Photosynthetic performance

Consistent with their anatomy, Amborella, most Austrobaileyales and many Chloranthaceae share low, steady-state maximum photosynthetic electron transport rates (measured with chlorophyll a fluorescence) and low light-saturation points (∼ 20% full sunlight; Feild et al. 2004; Griffin, Ranney & Pharr 2004). Light-saturated leaf CO2 uptake rates under steady-state and physiologically optimum conditions (i.e. under low leaf water potentials, high humidity and non-photoinhibitory conditions) vary from low to moderately high (i.e. from 3 to 12 µmol CO2 m−2 s−1; Feild et al. 2003b; Griffin et al. 2004). However, some large-gap-establishing Chloranthaceae, particularly Ascarina and Hedyosmum species, evolved higher photosynthetic rates similar to those of colonizing sun-tolerant tropical angiosperm trees and shrubs (Feild et al. 2004). In light-limited understories, evolution of low photosynthetic capacity is functionally advantageous because such leaves are less costly to build and maintain as well as less attractive to herbivores (Chazdon et al. 1996; Smith et al. 1997).

Low photosynthetic capacity, however, can limit the ability to take advantage of greater light availability (Smith et al. 1997). For example, A. trichopoda, A. scandens and several Illicium taxa displayed only small increases in leaf photosynthetic capacity and ability to dissipate excess light (i.e. non-photochemical quenching) in response to artificial gaps or when forced to grow under high-light conditions (Feild et al. 2001, 2003b; Griffin et al. 2004). Instead, leaves of the species examined folded up along the mid-vein to reduce light interception and/or bronzed in association with chlorosis and sustained photoinhibition (Fig. 1b,c; Feild et al. 2001; Griffin et al. 2004). Under common garden conditions with high water and nutrient availability, most Illicium taxa did not survive more than a year under full sun (Griffin et al. 2004).

Below dense forest canopies, light regimes are often characterized by stochastic, intense (near full sun intensity) and short duration (< 2 min) sunflecks that are juxtaposed on a background of dim, diffuse light (i.e. 2–10% full sunlight; Chazdon & Fetcher 1984). To take advantage of these high-light pulses that can dominate the daily photon receipt of understorey leaves, many plants have evolved a host of leaf-level physiological processes (Mooney et al. 1980; Knapp & Smith 1990; Chazdon et al. 1996; Kaiser & Kappen 2000; Naumberg & Ellsworth 2000; Leakey, Scholes & Press 2005). Among these features, desensitizing stomata to environmental perturbations in order to time-average rapid fluctuations in leaf microclimate in the understorey as a result of rogue winds, bumping of shoots by animals or during sunflecks themselves improves carbon gain during sunfleck use. By fixing stomatal conductance, photosynthetic enzymes can rapidly process quick changes in light intensity because internal CO2 supply is not limited by much slower stomatal aperture changes (Mooney et al. 1980; Chazdon et al. 1996; Kaiser & Kappen 2000). Alternatively, sluggish stomatal movements prevent rapid temperature excursions during sunflecks by transpirational cooling (Leakey et al. 2005). Consistent with these predictions, steady-state leaf stomatal conductances of A. scandens responded slowly (∼ 3 h to reach a new steady state; compared with 15–40 min of sampled eudicot angiosperms; Robinson 1994) when challenged with dry air conditions or transitioned from high to low CO2 (Feild et al. 2003b; Feild & Franks, unpublished data).

The anatomical features of Austrobaileya, which are also frequent in other basal angiosperms, may to contribute to slow stomatal response kinetics. Austrobaileya stomata are large, and large-sized stomata change shape slowly (Aasamaa, Sober & Rahi 2001; Hetherington & Woodward 2003). Compounding the movement inertia of large guard cells, cuticular vestibules over the guard cells may slow stomatal responses to dry air because the headspace above the guard cells must be dehumidified before guard cells can respond to atmospheric conditions (Feild et al. 2003b). In addition, stomata-subsidiary cell patterning is complex and highly variable in Austrobaileya. One or two rosettes of smaller subsidiary cells (up to 10 cells per stomatal complex; Carpenter 2006) encircle each guard cell pair. Such a pattern contrasts with stomatal development in more derived angiosperms, where guard cells are generally situated between two to four relatively larger subsidiary cells.

Because liquid-phase signalling within the guard-subsidiary cell network regulates epidermal turgor and stomatal aperture (Buckley 2005), hydraulic communication may be tortuous because of the numerous cell walls that separate subsidiary cell rosettes from guard cells. Other basal angiosperms share large stomata, cuticular vestibules and variable and heavily ‘cellularized’ stomata-subsidiary cell systems (Upchurch 1984; Metcalfe 1987; Kong 2001; Oh et al. 2003; Carpenter 2005; Denk & Oh, 2005; Yang & Lin 2005). This suggests slow stomatal responses may be widespread, but much more experimental work is needed. In many other basal angiosperms, the presence of a water-rich cell layer below the epidermis (i.e. a hypodermis, Amborella; Metcalfe, 1987) and hydrophilic mucilages in ducts running alongside veins, may increase leaf water storage (Morse 1990; Balsam & Thomson 1995). Water storage in the leaf may provide a short-term buffer against rapid drops in epidermal turgor, and thus stomatal closure during sunflecks as leaf temperature ramps up and drives transient, but rapid, transpiration (Leakey et al. 2005).

Leaf teeth

Placements of Amborella, Austrobaileyales and Chloranthaceae at the base of angiosperm phylogeny suggest that early angiosperm leaves bore ‘chloranthoid’ teeth along the leaf margin (Doyle & Endress 2000; Doyle 2001). A typical chloranthoid tooth consists of an acuminate tip with a glandular (epithem-containing) region and three veins (two from along the margin) and one larger secondary vein enters the tooth apex (Fig. 2a,f). Epithem is a poorly studied, but curious, tissue type. Structurally, the epithem of basal angiosperms is densely packed with organelles, has an extensive endocytotic system and is often pigmented by anthocyanins (Fig. 2c,f; Feild et al. 2005). Epithem cell walls are also thin, contain abundant pectins and consist of elongate cell shapes that form a high surface area for material exchange with the tracheary elements emptying into epithem tissue (Bürkle et al. 2003; Feild et al. 2005). The epithem of leaves resembles transmitting tissue cells of basal angiosperm flowers, a region of intense metabolic and developmental gene expression (Bernhardt et al. 2003; Hristova et al. 2005). Phloem appears to be lacking in chloranthoid teeth (Feild et al. 2005). While Amborella and Chloranthaceae taxa are always toothed (Todzia 1988; Todzia & Keating 1991), the leaf margins of Austrobaileyales are variable in morphology. For example, Illicium, Austrobaileya and three species of Trimenia are entire (Fig. 2e; Feild, Arens & Dawson 2003a).

Figure 2.

Leaf marginal and chloranthoid tooth functional diversity among extant basal angiosperms. (a) Transmitted light profile of a living leaf of Chloranthus japonicus (Chloranthaceae). Chloranthoid teeth are plumbed by veins running along the leaf margin as well as by a larger secondary vein looping off the mid-vein. (b) Guttation from the leaf adaxial surface in C. japonicus. (c) Leaf marginal teeth with prominent glandular (epithem-bearing) regions of Trimenia papuana (Trimeniaceae). (d) Guttating inflorescence of C. japonicus. Droplets emerge from the upper surfaces of bracts, with hydathodes and epithem tissue, that clasp the base of each flower. (e) example of an entire-margined basal angiosperm, Trimenia neocaledonica (Trimeniaceae). In this species, teeth and epithem are absent, but secondary veins merge into a intramarginal vein lined with anthocyanin pigments. (f) Marginal chloranthoid teeth of Ascarina rubricaulis (Chloranthaceae). Teeth in this species are heavily pigmented with anthocyanins. Small dark spots at the tooth tips result from build up crusty salts following evaporation of guttation sap from the leaf teeth tips. Scale bars: (a, e, f) = 2 mm; (b) = 8 mm; (c) = 1.5 cm.

Chloranthoid teeth appear to be functionally linked to wet habitats. In Chloranthus japonicus, a perennial herb from the temperate deciduous forests of East Asia, leaf teeth release water (via specialized stomata called ‘water pores’ or hydathodes) during root pressure-driven guttation when transpiration ceases under high humidity and soil saturation (Fig. 2b; Feild et al. 2005). Experimental blocking of leaf teeth caused guttation sap to rapidly fill the leaf intercellular air spaces. Mesophyll flooding inhibited photosynthetic activity (∼ 40%) during experimental lightflecks because of impeded CO2 diffusion (Feild et al. 2005). Thus, chloranthoid teeth seem important in enabling guttation and root pressure to operate without drowning the leaves. Guttation in the field has also been observed in Amborella, Trimenia (Trimenia papuana, Trimenia weinmanniifolia), Schisandra (Schisandra henryi, Schisandra glabra, Schisandra chinensis), Kadsura (Kadsura longipedunculata, Kadsura japonica) and all genera of Chloranthaceae, whereas entire-margined leaves species (Austrobaileya, four Illicium taxa) appear to lack guttation (Feild et al. 2003a).

Root pressure is an important whole-plant regulatory mechanism in wet understorey habitats. Firstly, root pressure maintains the flow of xylem-mobile nutrients and hormones (cytokinin and abscisic acid) coming from roots that would otherwise be limited by suppressed transpiration in the damp and dark habitats frequented by most basal angiosperms. Running root pressure and the release of sap from leaves sidesteps this transport bottleneck by maintaining hydraulic flow from root to shoot (Pedersen & Sand-Jensen 1997; Tanner & Beevers 2001). The epithem in leaf teeth appears to complete the transport circuit. Nutrients and hormones are removed from the xylem sap by the epithem before it is expelled through the hydathodes (Wilson, Canny & McCully 1991; Bürkle et al. 2003; Aloni et al. 2003, 2005). In Arabidopsis, the epithem functions as a hub for diverse ion, amino acid and hormone transport (Largarde et al. 1996; Shibagaki et al. 2002; Bürkle et al. 2003; Aloni et al. 2003, 2005; Pilot et al. 2004; Wang et al. 2004). The epithem can also add substances to the guttation fluid, such as excess calcium and proteins that protect vulnerable hydathodes from microbial and fungal attack (Fukui, Fukui & Alvarez 1999). Although gene expression in the epithem of basal angiosperms invites future study, calcium secretion and presence of an extensive epithem endocytotic system in C. japoncius suggests that nutrient and hormonal cycling is active (Feild et al. 2005). White crusts, resembling calcium salt deposits found on tooth tips in other guttators (Ivanoff 1963), are frequently observed in chloranthoid leaf teeth. This suggests ionic release by teeth (Feild, unpublished data).

Secondly, root pressure can regulate xylem water relations by refilling cavitated conduits (Sperry et al. 1987; Vogt 2001). Although xylem cavitation may seem unlikely in the understorey, excessive transpiration during opportunistic sunfleck use or air sucked into the xylem as stems break from the impact of falling canopy debris, may both cause xylem embolism (Schultz & Mathews 1997; Leakey et al. 2005). Reversing embolism is critically important in understorey plants because deep shade drives a low allocation of xylem sapwood area relative to the supported leaf area (Brodribb, Holbrook & Hill 2005). This means that any prolonged loss of conducting area by air blockage will strongly curtail shoot gas exchange because the conducting area is low to minimize stem carbon construction cost under shade. Furthermore, recent data indicate that basal angiosperms are especially vulnerable to drought-induced cavitation (Hacke et al. 2006).

Finally, root pressure may regulate shoot growth by pressurizing expanding cells (Stahlberg & Cosgrove 1997; Munns et al. 2000). Rates of cell and tissue expansion reflect a complex interplay of turgor as well as biochemical and biomechanical processes (i.e. cell wall-loosening enzymes, pH, ionic gradients, cell wall biomechanics; Munns et al. 2000). When transpiration stops and root pressure propagates to the shoot (note that guttation droplets appear on Chloranthus, Sarcandra, Kadsura and Amborella leaves ∼ 30 min; Feild, unpublished data), increased shoot turgor from water pushed up from the roots will ensure that cell wall loosening events driving expansive growth are not turgor limited (Munns et al. 2000). It is interesting to note that the tips of other growing shoot organs, for example, stipules and bracts sheathing flowers on inflorescences of C. japonicus, guttate and possess water pores and epithem (Fig. 2d). Inflorescence hydathodes of Chloranthaceae do not appear to be extrafloral nectaries since insects do not feed on the guttation droplets, and rewards such as sugars are not present in the guttation fluid (von Balthazar & Endress 1999). Amborella also has chloranthoid gland-like protuberances in its flowers (Buzgo, Soltis & Soltis 2004). Such hydathodes could be easily co-opted for extrafloral nectary function, however, if phloem developed in thevasculature of the epithem, as in carbohydrate and terpene secreting foliar hydathodes of Populus (Curtis & Lersten 1974).

Long-distance xylem water transport

Xylem vessels have long been viewed as important features promoting diversification in early angiosperms (Frost 1930; Bailey 1944; Carlquist 1975; Bond 1989; Pammenter, Midgley & Bond 2004; Brodribb et al. 2005; Feild 2005). Because of lower lumen and end wall resistances on a conduit basis, vessels can dramatically cut the costs of transpiration. For example, vessels enable greater rates of water transport and photosynthesis, while maintaining a lower pressure gradient between root to shoot (Brodribb & Feild 2000; Sperry 2003). Although some tracheid-bearing conifers can produce similar or even higher xylem hydraulic capacities on a leaf area basis as vesselled angiosperms (Brodribb & Feild 2000; Pammenter et al. 2004; Brodribb et al. 2005), vessels allow these conductivities with lower investment in wood conduit area (Sperry 2003; Feild 2005).

However, these high-efficiency vessels do not characterize basal angiosperm lineages. Vessels of Austrobaileyales and Chloranthaceae (note that Amborella is vessel-less; Feild et al. 2000) are relatively small in diameter, angular in cross-section and bear long perforation plates with up to 150 scalariform pits (Carlquist 1982, 1984, 1987, 1990, 1992a,b, 2001a,b; Carlquist & Schneider 2002a,b). Vessels of Austrobaileyales and Chloranthaceae also occasionally retain intact pit membranes (Carlquist 1992b; Carlquist & Schneider, 2002a). These pit membrane remnants are diverse in morphology, ranging from highly porous skeins (holes 300–500 nm) draped across the pit aperture to encrustations that partially occlude pits (Carlquist, 1987, 1990, 1992a,b, 1999, 2001a,b; Carlquist & Schneider 2002a,b). There are some exceptions. For example, some lianescent taxa develop wider diameter vessels and more streamlined perforation plate anatomies (Carlquist 1984, 1999, 2001b). Consistent with anatomy, recent studies indicate that stem wood hydraulic capacities of most Austrobaileyales and Chloranthaceae are similar to vessel-less angiosperms (Amborella, Winteraceae, Trochodendraceae) and lower than many eudicot vascular systems (Feild et al. 2000; Hacke et al. 2006; Hacke et al., unpublished data). The relatively small, tracheid-like vessels typical of basal angiosperms suggest that the earliest vessels did not confer a quantum leap in hydraulic performance. Rather, vessel functions were likely honed in incremental steps within or close to the ancestral habitat before achieving high conductivities necessary for functional advantage in more open or drier environments.

Opening the pit membranes is an early step in vessel origin (Frost 1930; Bailey 1944; Carlquist 1975). All else being equal, increased end wall porosity will lower the pressure necessary to pull air through the pit membranes of one conduit into the next (Sperry 2003) and increase the risk that embolisms will spread. Indeed, the stem xylem of Austrobaileyales and Chloranthaceae species is very vulnerable to tension-induced cavitation, exhibiting a 50% percent loss in xylem hydraulic conductivity (PLC) on average at −1.5 MPa (n = 15 species; Hacke et al. 2006). Importantly, basal vessel-bearing angiosperms are more vulnerable to drought than vessel-less angiosperms (mean = −4 MPa at PLC50%; Hacke et al. 2006). This suggests that the transition to vessels is associated with a loss of drought tolerance. A ‘vulnerability bottleneck’ to vessel evolution may explain why wet, dark, understorey habitats with low transpirational demand are also where early vessel experimentation occurred in other seed plant lineages (Feild 2005).

With no immediate gain in hydraulic capacity and a decrease in xylem drought tolerance, what did early vessels offer their bearers? Vessels may have opened up new developmental options for wood function and specialization (Hacke et al. 2006). Early angiosperm vessels enable roughly the same water transport capacity as ancestral tracheids, but this wood design requires less space devoted to axial water-conducting tissue (Hacke et al. 2006). Thus, for the same stem cross-sectional area, different types of cells can be developed and packed in for resource storage,biomechanical support, radial vascular transport and possibly embolism refilling under tension. This outcome of more-for-less may have been particularly advantageous in the understorey, where carbon gain is limited and the environment is heterogeneous.

Vessel-bearing wood also allows hydraulic and mechanical functions to be decoupled and specialized independently. Tracheids must multitask mechanical and hydraulic functions, which pulls anatomy in opposite directions. Optimal form for mechanical strength stresses low lumen volume and thick walls. This contrasts with a design for hydraulic efficiency, which involves high lumen volume and open conduits between conducting elements (Carlquist 1975, 2001a). In contrast, vessels are free to become hydraulically optimized while the evolution of fibres compensates for the loss in mechanical strength (Carlquist 1975, 2001a). Interestingly, one interpretation of the widespread distribution of scandent architectures among basal angiosperm lineages is that mechanical function lags behind hydraulics. Fibres in many basal groups tend to be thin walled, sometimes living at maturity and retaining normal-sized bordered pits (Carlquist 1982, 1984, 1990, 1992a,b, 2001b). All of these features limit mechanical support, and at least of the taxa that bear them are transitional between a liana and tree (e.g. Trimenia neocaledonica; Feild, unpublished data). Concomitant mechanical and hydraulic measurements would be informative in deconvoluting the interplay between mechanics and hydraulics associated with early vessel evolution.


Before the debut of angiosperms, a diverse assemblage of seed plants (Bennettitales, cycads, numerous conifers and the enigmatic group Czekanowskiales) and ferns filledterrestrial communities (Rees, Ziegler & Valdes 2000). Thus, it is clear that angiosperms did not begin their rise by simply shifting into the ecological space emptied out by a mass extinction event – a ‘victors by default’ process (Lupia et al. 1999). Instead, we know that the processes determining the ways in which fundamental resources were processed by light-, water- and carbon-use mechanisms played important roles in enabling angiosperms to intercalate themselves into intact communities and radiate. We are left with three key questions. What functional features contributed to angiosperms' initial invasiveness? What role did the environment play? And how did early evolutionary events canalize later directions of functional innovation?

A major hypothesis to account for the rise of angiosperm dominance is the competitive superiority hypothesis. This hypothesis views the earliest angiosperms as highly competitive, functioning with well-honed ecophysiological performances and therefore growing much faster, cheaper and therefore better than co-occurring Cretaceous gymnosperms and ferns (Stebbins 1965, 1974; Hickey & Doyle 1977, Doyle & Donoghue 1986; Bond 1989). Suggested key ecophysiological enablers of early angiosperm invasion include innovations in leaves and wood, such as xylem vessels and reticulate-veined leaves with freely ending veinlets and herbaceousness – all traits appearing with the origin of the clade and predicted to immediately imbue the clade with high opportunistic performance and rapid speciation potential (Stebbins 1974; Doyle & Donoghue 1986, 1993; Bond 1989; Taylor & Hickey 1996). Previous whole-plant images of early angiosperms cultivated this concept. Early angiosperms were imaged as ruderal, drought-tolerant and sun-loving shrubs (Stebbins 1974), aquatic herbs (Burger 1981; Sun et al. 2002; Ji et al. 2004) and high-light-tolerant floodplain/riparian herbs or shrubs with high photosynthetic and transpiration rates (Hickey & Doyle 1977; Taylor & Hickey 1996). These views motivated numerous studies aimed at discovering the floral and vegetative ‘key innovations’– intrinsic organismic traits catalysing ecological, morphological and/or species diversification – of early angiosperm success (Stebbins 1974, 1981; Carlquist 1975; Burger 1981; Doyle & Donoghue 1986, 1993; Friedman 1998; Gorelick 2001).

However, the notion that angiosperms sprinted out of the evolutionary blocks propelled by functionally superior traits is due for reconsideration (Sanderson & Donoghue 1994; Magallón & Sanderson 2001). Firstly, the bulk of angiosperm diversity (represented by orchids, grasses, rosiids and asterids; Stebbins 1974, 1981) did not arise near the base of the angiosperm tree. Instead, basal angiosperm lineages are relatively species poor (Magallón & Sanderson 2001). Secondly, many of the oft-mentioned angiosperm innovations, such as simple pitted xylem vessels, flowers, out-crossing mechanisms, insect associations and double fertilization, assembled within angiosperm lineages that have low diversity today (Carlquist & Schneider 2002a; Williams & Friedman 2002; Friedman & Williams 2004; Bateman et al. 2006; Friedman 2006). Thirdly, many of the so-called innovations themselves appear to have evolved through a series of intermediates with modest functional capabilities (e.g. vessels; Hacke et al. 2006). While many angiosperm stem lineages have been lost to extinction (Friis et al. 2000, Friis, Pedersen & Crane 2001), it seems clear that the overwhelming modern diversity of angiosperms cannot be explained by the evolution of a single character or even a discrete complex of characters acquired at the origin of the clade (Magallón & Sanderson 2001).

Moreover, an exclusive focus on the origin of key innovations obscures the co-ordinate role of environment – the key opportunity – under which new traits found utility and were refined by natural selection (de Queiroz 2002; Donoghue 2005). We suggest that the damp, dark and disturbed ancestral habitat of angiosperms allowed the origin and refinement of ecophysiological traits – such as vessels and broad, net-veined leaves – that permitted the new lineage to gain a roothold in the Mesozoic understorey (Boyce 2005; Feild & Arens 2005). These traits were later co-opted or modified as the lineage broke out into more demanding environments and began its meteoric rise (Lupia et al. 1999) to vegetation dominance.

Ecophysiology of the initial roothold

By analogy with modern basal angiosperms, early vessels were relatively small in diameter for a given length, present in low density and probably retained pit membrane remnants (Carlquist & Schneider 2002a; Hacke et al. 2006). Consistent with their anatomy, these early vessels probably did not confer a major advantage in hydraulic capacity to their bearers (Hacke et al. 2006). In fact, one might argue that early vessels were initially disadvantageous to the plants that produced them because of an increased risk of drought embolism. However, improved hydraulic capacity to support high transpiration and greater drought resilience were not necessary for success in the damp, shady understorey. Here, high humidity and wet soils minimized transpirational demand and enabled high leaf water potentials – conditions that lower the risk of embolism. Frequent and reliable root pressure generation in the wet understorey also ensured that hydraulic embolism, when it occurred, did not last long (Feild et al. 2005). Instead, vessels may have offered relatively similar hydraulic capacity as tracheids but with lower carbon investment, an advantage in a light-limited habitat and where frequent disturbance by understorey foragers damaged or destroyed shoots (Feild 2005).

Similarly, the ancestral leaf of angiosperms – broad and net-veined, with large stomata, complex subsidiary cell network, low photosynthetic rates per unit area and high optical efficiency (via uniform ‘spongy’ mesophyll cells and an internally reflective lower epidermis) – offers little functional advantage outside heavily shaded and humid microsites (DeLucia et al. 1991, 1996; Smith et al. 1997). Yet under the canopy, where light is a limiting resource, leaves that function in these ways are advantageous in carbon-use efficiency. Thus, the competitive advantages conferred by early vessels and shade-adapted leaves were not of globally superior function or drivers of rapid speciation, but more modest accommodation to the rigours of a specific ancestral habitat – the damp, Mesozoic forest understorey. What habitats did early angiosperm shift to next, and how did an ancestral ecophysiology as wet- and shade-adapted canalize these shifts?

Early breakout habitats and ecophysiologies

Angiosperms explored sunny, aquatic zones early in their evolution (Feild et al. 2004). Beyond the all-aquatic basal lineages Nymphaeales and Hydatellaceae, other early aquatic experiments include several basal monocots (Acorus, Alismatales; Les & Schneider 1995), basal eudicot lineages (i.e. Nelumbo, some Ranunculus species) and the phylogenetic vagabond lineage Ceratophyllum (see Soltis et al. 2005). Traits that evolved in the damp, shady understorey seem to have primed early angiosperms for aquatic exploration. For example, root pressure and hydathodes present in the common ancestor of extant angiosperms (Feild et al. 2003a) could by co-opted for the long-distance transport of ions and water for submersed ‘transpiration’ (Sculthorpe 1967; Pedersen & Sand-Jensen 1997). Early angiosperms also exploited the open water by losing the vascular cambium and up-regulating leaf photosynthesis via large increases in leaf thickness (i.e. multiple-palisade layers), increases in stomatal densities and loss of stomatal control of leaf gas exchange (Brewer & Smith 1995; Feild et al. 2004). However, aquatic shifts and the suite of physiological modifications wrapped around them did not set the stage for later diversification on land (see further discussion).

To break into brighter and more evaporative terrestrial habitats, several ecophysiological modifications to whole-plant function are involved as lineages transitioned from light-limited to more water-limited environments. In addition to metabolic up-regulation and allocation, leaves must evolve palisade mesophyll and amphistomy to process collimated light and increase photosynthetic rate per unit area to take advantage of greater sunlight (Smith et al. 1997). Furthermore, evolution of smaller stomata, finer guard cell regulation and better control of gas exchange across the leaf may increase water-use efficiency (Sack et al. 2003). Changes in leaf gas exchange function were probably coupled with increased density and regularization of the leaf vein network, and perhaps the evolution of xylem vessels in the venation to minimize pressure drops within the leaf under high transpiration (Zwieniecki et al. 2002; Sack et al. 2003; Arens 2006).

Modifications to secondary xylem vessels in the wood were also probably associated with movement out of the shade. To accommodate greater transpirational demand, evolution of greater sapwood hydraulic efficiencies can cut the hydraulic costs of transpiration (Sperry 2003). Here, vessels changed in a number of ways, including the evolution of simple perforation plates, increase in vessel densities, as well as lengthening and widening of individual vessels (Hacke et al. 2006). Finally, dependence on root pressure for growth and embolism refilling would be expected to decline as angiosperms branched out into brighter and drier habitats or as plant size increased. This is because opportunities to generate root pressure drop off as rainfall and humidity decline. The unyoking of root pressure ‘addiction’ likely involved the increases in xylem cavitation resistance, rapid stomatal control to humidity and leaf water potential, as well as evolution of physiological mechanisms of osmotic adjustment to maintain turgor and embolism refilling under tension (Stahlberg & Cosgrove 1997; Sperry 2003).

A preliminary look at early-derived ecologies among the extant basal flora indicates that angiosperms gradually entered more exposed and increasingly disturbed habitats, but these zones were very wet (i.e. wet, montane cloud forests; Feild et al. 2004). Specific habitats include as colonizers of steep landslips, along forest margins and in large forest-light gaps (Sugden, Tanner & Kapos 1985; Todzia 1988; Wagner & Lorence 1999; Feild et al. 2004; Martin & Ogden 2005). Ascarina, Hedyosmum and Trimenia represent independent lines that evolved greater leaf photosynthetic performances and abilities to recruit seedlings in brighter, more large-scale disturbed habitats. Consistent with greater ecological amplitude, Ascarina- and Hedyosmum-like chloranths, represented by fossil pollen and flowers, appear to lead the initial waves of angiosperm invasion into high-latitude Northern and Southern Hemisphere communities during the Early Cretaceous (Romero & Archangelsky 1986; Eklund et al. 2004; Feild et al. 2004). Trimenia, however, lack a Mesozoic fossil record. Among basal angiosperms, the initial transitions to higher-light environments are characterized by a high degree of lineage-dependent, functional experimentation, in which fine-tuned performances were assembled piece-by-piece. Gap-establishing Chloranthaceae and Trimenia species, for example, evolved high leaf photosynthetic rates, but maintained a lack of palisade cells, leaf teeth and, therefore, reliance on root pressure and large stomata at low density. Similarly, other taxa that stay in the understorey, such as several Illicium species, evolved a single palisade layer and lost leaf teeth, but leaf photosynthetic rates remain low and leaf epidermises stayed tuned to low evaporative demand (Feild et al. 2004).

Hydraulically, stem xylem vessels of Ascarina, Hedyosmum and Trimenia seem to remain functionally undifferentiated from understorey-dwelling basal groups, with no major divergences in hydraulic resistance or vulnerability to cavitation (Hacke et al. 2006). Xylem venation of patterns of angiosperm leaves exhibit a similar pattern. When the architectural variability of basal angiosperm venation was quantified and compared with that of magnoliids and eudicots, basal angiosperms emerged as significantly more variable than later diverging clades (Arens 2006). This suggests that early in angiosperm history, the lineage honed its leaf vein architecture towards greater efficiency, perhaps minimizing pressure drops through the leaf system (Zwieniecki et al. 2002; Sack et al. 2003) in supplying water to the lamina's photosynthetic cells. These functional patterns in basal angiosperms suggest that functional experimentation for higher-light use occurs first in the optical- and CO2-processing traits of leaves, while xylem modifications for intrinsic hydraulic efficiency are slower to evolve. Such experimentation was only possible under the relatively low evaporative demands (as compared with a lowland tropical rain forest) and high water availabilities characterizing the sorts of large-scale disturbed habitats that Chloranthaceae and Trimenia recruit in.

The evolution of lianous habits represents another track for exploring different environments in early angiosperms. Basal angiosperm lianas typically establish in the damp, dark understorey, but later climb into more well-lit and drier environments as they climb up their hosts into the forest subcanopy. Associated with the evolution of climbing, the traits related to vessel hydraulic efficiencies ramp up considerably (Carlquist 1984, 1999, 2001b), but leaf photosynthetic capacities and the ability to process collimated light remain low (Feild et al. 2003b, 2004).

Much more detailed work is needed to piece together the nature of ecophysiological transitions promoting breakthroughs in higher light tolerance and emerging out of the clouds by increasing tolerance to greater evaporative demand. Interestingly, the observation that independent forays out of the shade as wet, gap-colonists and subcanopy vines did not entrain considerable species diversification (Magallón & Sanderson 2001) suggests that co-ordinated shifts in hydraulic and photosynthetic performances of hydraulic efficiency are necessary to make a substantial movement out of wet habitats. In the later phases of early angiosperm history, we predict that the forms and functions refined in the understorey and along the wet forest edge are co-opted for evolution of high productivity in unstable habitats with fluctuating resource availabilities – an ability so emblematic of highly successful eudicot and monocot lineages (Stebbins 1974, 1981; Bond 1989; Taylor & Hickey 1996; Brodribb et al. 2005). From this point, the uniquely invasive features and high physiological performance of angiosperms promotes further biogeographic, ecological and taxonomic expansion with significant consequences for the lineages with whom they share the biosphere (Schneider et al. 2004).


Some have proposed that angiosperms originated as aquatic herbs. This argument is supported by the placement of an aquatic line (Nymphaeales plus Hydatellaceae) near the base of extant angiosperm phylogeny and the discovery of several Cretaceous fossils of aquatic angiosperms (Barkman et al. 2000; Sun et al. 2002; Ji et al. 2004). Fossils representing early angiosperm aquatics include a variety of nymphaeid (long-petioled, oval floating leaves) leaf imprints from Aptian to Albian lake deposits in Brazil and North America (Mohr & Friis 2000), as well as of water lily fossil flowers (Friis et al. 2001), with one placed in the Victoria lineage (Turonian age, 93.5–85.8 Ma; Gandolfo, Nixon & Crepet 2004).

Whole-plant fossils of Archaefructus (∼ 30 cm tall) are additional early angiosperm aquatics from lake deposits in China (Sun et al. 1998, 2002; Friis et al. 2003; Zhou, Barrett & Hilton 2003; Crepet et al. 2004; Ji et al. 2004). Archaefructus was heralded as the oldest flowering plant, based on a Late Jurassic age (Sun et al. 1998). However, radioisotopic evidence revised the age of Archaefructus as Early Cretaceous, Barremian ∼ 124 Ma (Swisher et al. 1999; Friis et al. 2003; Zhou et al. 2003). Consequently, Archaefructus fossils are too young to represent the earliest flowering plants because the angiosperm radiation was well underway by that time (Friis et al. 1999, 2000). Sun et al. (2002) phylogenetically placed Archaefructus below the common ancestor of all living angiosperms and suggested that angiosperms originated underwater. Subsequent cladistic studies moved Archaefructus to various positions away from the angiosperm root (Friis et al. 2003, but see Ji et al. 2004), either as a Cabomba relative or a basal eudicot. These nested evolutionary positions significantly weakened the phylogenetic claim that angiosperms started underwater. The discovery of Hydatellaceae as the sister group to water lilies represents yet another contender for an extant Archaefructus relative because these taxa bear unicarpelate inflorescences like Archaefructus (Saarela et al. 2006). Until its relationships are resolved it will be difficult to evaluate the ecological significance of the Archaefructus.

Advocates for an aquatic origin argue that the lineage leading to extant angiosperms entered submerged habitats to escape competition on land and diversified in ecologically open freshwater habitats (Sun et al. 2002). Adaptation to drying pools, characteristic of most living Nymphaeales (Sculthorpe 1967; Schneider 1983) and Hydatellaceae (Hamann 1976), may have sparked the evolution of rapid growth (via herbaceousness and rapid photosynthesis), accelerated reproduction and condensation of the reproductive axes – a functional trinity thought to underlie much of the ecological ‘success’ of angiosperms (Stebbins 1974, 1981; Burger 1981; Bond 1989; Taylor & Hickey 1996). While this scenario is likely valid for diversification within the water lily clade, it does not hold water for the larger angiosperm radiation.

Shifting from an aerial to a submerged lifestyle requires radically different biophysical and biomechanical function. Fluid density of water is a thousand times greater than that of air (Niklas 1997). Consequently, aquatic plants operate under dramatically increased diffusion resistance for CO2. Aquatic plants also experience predominantly tensile mechanical forces, from waves and currents, rather than from the compressive effects of gravity (Niklas 1997). These fundamental differences between terrestrial and submerged conditions have pushed aquatic lineages into novel vegetative morphologies occurring along a limited set of highly replicated themes (Sculthorpe 1967; Cook 1999; Keeley 1999). Examples of these vegetative motifs found in aquatic basal angiosperms include thread-like filliform leaves (i.e. Cabomba, Archaefructus, Ceratophyllum) that reduce boundary layer resistance, shield- and heart-shaped (nymphaeid) leaves with palmate venation for efficient light capture and stability on the water's surface (most Nymphaeales) and tufted linear leaf display atop a dense, fibrous root system (Hydatellaceae; Hamann, 1976). The latter converges on the aquatic lycophyte Isoetes and other angiosperms (the ‘isoetid’ habit; Keeley 1999; Robe & Griffiths 2000), which may be related to harvesting of CO2 from aquatic sediments.

While highly functional in the aquatic context, these specializations lead the lineages that bear them into morphological cul-de-sac from which they cannot easily escape. This conclusion is supported by the lack of phylogenetic evidence for aquatic lineages reinvading land. Of approximately 200 terrestrial-to-aquatic transitions in flowering plants (Cook 1999), monocots may be the only example of a land-to-aquatic-to-land transition (Les & Schneider 1995; Tomlinson 1995). The roadblock to re-invading the land may be explained by the difficulty in re-acquiring the key traits for life on land. For instance, re-invasion of land would require re-invention of the simple bifacial cambium observed in most lignophytes (i.e. seed plant out-groups) and many angiosperms (Tomlinson 1995; Donoghue 2005; Feild & Arens 2005). The vascular cambium seems easy to lose in the shift to an aquatic habit (Carlquist 1975), but once lost much harder to (if not impossible) to re-evolve without being marked by an aquatic legacy. For example, arborescent monocots have evolved a unique type of unifacial cambium (Tomlinson 1995). Importantly, Nymphaeales and Hydatellaceae do not retain a vestigial vascular cambium (Hamann 1976; Schneider & Carlquist 1995). Thus, a bifacial cambium was not minimized under an aquatic bottleneck and was later aggrandized on land. It seems unlikely that an ancestral angiosperm aquatic could have escaped its pond to spawn the Cretaceous angiosperm radiation.


Consensus on the root of the flowering plant phylogeny and detailed studies of developmental, physiological and genomic patterns of extant basal angiosperms has recast our imagery of the earliest flowering plants (Feild et al. 2004; Soltis et al. 2005; Friedman 2006; Hacke et al., unpublished data). Despite this conclusion, an important issue looms over the comparative approach: how relevant are living basal lineages for reconstructing the botany of Early Cretaceous angiosperms (Endress 2001; Crisp & Cook 2005; Feild & Arens 2005)? Three caveats warn against a literal reading of living basal lineages as ecological look-alikes to early angiosperms.

Firstly, living basal angiosperm may not represent the root of the flowering plant tree. Such pattern could arise from the extinction of stem lineages branching from nodes below the extant basal grade. Such extinction would produce a pattern of later-derived features among the surviving, thus extant, basal flora. The fossil record of early angiosperm flowers offers evidence of extinction and turnover during the initial angiosperm diversification. Surveys of five Barremian through Aptian assemblages from Portugal identified 150 angiosperm taxa, represented by entire flowers, dispersed stamens, fruits and seeds (Friis et al. 1999, 2000). Most (85%) of these fossils preserve features familiar from the basal angiosperm grade plus eumagnoliids and monocots. However, these characters were combined in ways not found in modern flora (Friis et al. 1999, 2000), suggesting a variety of stem lineages that did not survive to the present day. At present, there has been little success in incorporating these fossil angiosperms into the phylogeny of living taxa (Friis et al. 1999, 2000). Without a detailed assessment of the phylogenetic position of purported stem lineages, we are unable to assess whether this is a significant problem for the reconstruction of ancestral ecology.

Secondly, ancestral traits reconstructed from living basal angiosperm lineages may actually be derived. Even if extant basal lineages truly represent nodes near the beginning of the angiosperm radiation, their traits may have changed in the ∼ 130 Ma from their divergence to the present. Still, other traits present in the earliest angiosperms may have been lost to extinction. The Earth has changed significantly since angiosperms arose. Global climate warmed and cooled, pCO2 dropped, major herbivore lineages became extinct and others diversified, pollinators and dispersers have come and gone, the continents shifted and sea levels rose and fell (Grimaldi 1999; McElwain, Willis & Lupia 2005). In fact, the genera of extant basal lineages, such as Illicium, Schisandraceae, Chloranthaceae and Nymphaeales, are products of a much later (early- to mid-Tertiary) diversification – diversification events far removed from the early angiosperm radiation (Zhang & Renner 2003; Denk & Oh 2005; Yoo et al. 2005; Morris et al. in press). Species diversity in some groups may reflect very recent bursts of speciation. For example, half of the extant Hedyosmum species likely arose with the uplift of the northern Andes (∼ 20 species out of 40 in ∼ 2 Ma; Todzia, 1988, see Hughes & Eastwood 2006 for an example radiant plant lineage) and a mini-radiation of three Trimenia species occurred since the birth of the Marquesas archipelago (1.3–5.8 Ma; Wagner & Lorence, 1999; Clouard & Bonneville 2001). Yet Hedyosmum, Trimenia and other basal angiosperm radiations appear to be characterized by ecological stasis. For example, no major growth form changes or shifts from wet to dry habitat characterize the Andean sections of Hedyosmum or the Marquesan Trimenia (Todzia 1988; Todzia & Keating 1991; Wagner & Lorence 1999), despite the fact that such shifts commonly typify adaptive radiations (Hughes & Eastwood 2006).

Further support for ecological stasis can be tentatively seen in the ecomorphology of Early Cretaceous fossils. A preliminary look at the anatomy and morphology of early angiosperm leaf fossils reveals vegetative features characteristic of those features occurring in damp, dark and disturbed basal taxa. Some early angiosperm cuticles possess large and sparse stomata that are concealed by cuticle vestibules, highly variable subsidiary cell arrangements and rugose and thick abaxial cuticles (Upchurch 1984; Carpenter 2005). Some of the oldest compression fossils of early angiosperm leaves also reveal the presence of leaf teeth and irregular venation (Hickey & Doyle 1977; Upchurch 1984; Coifford et al. 2006; Mohr et al. 2006). Based on an analogy with modern basal lineages, all of these traits may be the hallmarks of a wet, disturbed understorey ecology (Feild et al. 2004; Arens 2006).

The angiosperm phylogeny itself sounds a final note of caution. While current topologies enjoy broad consensus, recent and unexpected discoveries emphasize that the basal angiosperm phylogeny is not definitive. For instance, the Hydatellaceae recently moved from the Poales to a water lily sister group. While this phylogenetic placement does not change the distribution of ecological traits associated with common ancestor of extant angiosperms (Saarela et al. 2006), the result highlights how conventional wisdom can be overturned by a run of the PCR machine. Moreover, the Amborella mitochondrial genome has experienced more horizontal gene transfer (e.g. from eudicot and moss donors) than any other eukaryote yet examined (Bergthorsson et al. 2004). Such extreme genome fluidity may turn out to be key to the angiosperms prodigious ability to adapt and diversify, but it could also cast doubt on favoured topologies if the same degree of lateral transfer is discovered in the other genomes of Amborella and other currently basal taxa. Including fossils could also change the picture, as illustrated by the debate surrounding the phylogenetic placement of the Cretaceous aquatic Archaefructus (Sun et al. 2002; Friis et al. 2003; Ji et al. 2004; Saarela et al. 2006).

Despite these uncertainties, living basal angiosperms have provided fertile ground for posing testable hypotheses on the form and function of early angiosperms. To the degree that the comparative approach has suggested specific ecomorphological traits to look for in the fossil record, it has already been successful. In this context, there is good paleobotanical evidence to suggest that we are on the right track in studying the form and function of living plants. It seems unlikely that such consistency between Early Cretaceous angiosperm cuticles and extant basal angiosperms would emerge simply by chance (Feild et al. 2004). However, the definitive test for ancestral ecology will come from ecomorphic analyses of early angiosperm fossils.


The fusion of ecophysiology and extant phylogeny offers new perspectives on early angiosperm form and function as well as the routes of ecophysiological evolution taken by later evolving lines. While this approach has allowed has crystallized a hypothesis of early angiosperm ecology, the fossil record will provide definitive testing of patterns emerging from living plants. Such future paleobotanical analyses, focusing on anatomical traits and biochemical markers (carbon isotopes) of paleo-ecophysiological carbon/water use, can test the image of early angiosperms damp, dark and disturbed (Feild & Arens 2005). Ultimately, analyses of this sort will address the long-standing issue of how relevant extant taxa are to questions of angiosperm ecology, morphology and development in deep time. But an accurate interpretation of fossil paleo-ecophysiology will depend on continued work in the living basal angiosperm flora. We need concrete knowledge on how anatomy (i.e. stomatal traits, leaf teeth characters, which can be observed directly in fossils) translates into ecophysiological performance (i.e. gas exchange, xylem hydraulic fluxes, which cannot be assayed directly from fossils; Feild et al. 2003a). It will also be important to consider the sedimentological context of early angiosperm fossils to obtain for clues on local habitat that are independent of anatomy.

Some of the next research steps also need to include biogeographic work to understand how the spread of angiosperms is associated with tectonic and climate reconstructions to test hypotheses about the biomes exploited by Early Cretaceous diversifying lineages. We also need to look for the experimentation, canalization and breakout phases of evolution during angiosperm diversification using ecomorphic indicator traits suggested by basal angiosperms. Finally, while the ecophysiological understanding of angiosperms continues to progress at a steady place, now is the time to explore the functional biology of long-neglected phylogenetic groups, such as lycopods, ferns, cycads, Gnetales and tropical conifers (Brodribb & Hill 1997; Brodribb et al. 2005). Understanding the functional biology of these groups is essential because many of them went leaf-to-leaf and root-to-root with early angiosperms in Mesozoic. Such comparisons are needed to critically address: How do living basal angiosperms and other non-angiosperm seed plants respond to the sorts of environmental changes proposed for the Early Cretaceous? How do changes in atmospheric CO2 influence the competitive interactions of these plants? Much in the same way that angiosperms have been viewed as uniformly highly competitive, a statement which now appears not to characterize the base of the angiosperms, these groups have been monolithically viewed as non-competitive and functionally inferior (Bond 1989). Is such a view well founded? Integrating our understandings of community, ecophysiological, environmental and phylogenetic perspectives during the Cretaceous will allow holistic analysis of the ecological mechanisms underlying early angiosperms' evolutionary success.


We thank Keith Mott for the kind invitation to compose this review. We also appreciate the insightful comments and discussions on the manuscript from the anonymous reviewer, Michael Donoghue, Erika Edwards, Sarah Mathews, Lawren Sack, William K. Smith, John S. Sperry and Joe Williams. We finally thank Sean Graham for the discussions on the phylogenetics of Hydatellaceae, and Terry McFarlane for information on the Hydatella/Trithuria ecology.