A convincing hypothesis of angiosperm origins in the context of a secure understanding of seed plant phylogeny still remains to be achieved. A major impediment is the need for improved information on fossil seed plants, but there are also other challenges, particularly concerning how best to integrate the massive and increasing disparity between what is known about the genomes of living plants and what is known about the morphology and anatomy of extinct and extant taxa. In this chapter we briefly review the development of ideas concerning seed plant phylogeny and outline the status of research in this area. We also review the implications of ideas on relationships for understanding the origin and age of angiosperms.

Hypotheses of seed plant relationships

Ideas on relationships among seed plants that developed in the early twentieth century (e.g. Coulter and Chamberlain, 1917; Chamberlain, 1935) focused on gymnosperms (often excluding angiosperms completely) and recognised two groups: cycadopsids and coniferopsids. Cycadopsids included cycads, Bennettitales (Cycadeoidales) and the fossil plants grouped together at that time as seed ferns (e.g. lyginopterids, medullosans). Coniferopsids comprised conifers, cordaites and Ginkgo, and in many schemes also included Gnetales. This biphyletic interpretation of seed plant evolution gathered further support with the recognition of Devonian progymnosperms and the hypothesis that aneurophytalean progymnosperms might have given rise to the cycadopsid line, while archaeopteridalean progymnosperms might have given rise to the coniferopsid line (Beck, 1966, 1970, 1971, 1981). As developed by some authors (e.g. Doyle, 1977), this notion of two main lines of seed plant evolution interpreted angiosperms as the culmination of evolutionary elaboration in the cycadopsid line, while Gnetales were viewed in the equivalent position among coniferopsids (Figure 6.1).

Over the past two decades there has been rapid progress in developing a phylogenetic framework within which to interpret many aspects of angiosperm evolution. This progress has been brought about mainly through the application of molecular phylogenetics to an increasingly large sample of flowering plants. In this chapter we briefly review the development of ideas on angiosperm phylogeny and outline the current state of knowledge based on the most recent APGIII classification (2009). Phylogenetic patterns within particular groups are summarised in Chapters 8–15. In this chapter we also briefly discuss the issues that need to be considered in assigning fossils to extant groups. This is essential background to what follows in Chapters 8–15.

Early ideas on angiosperm phylogeny

The Euanthial (Anthostrobilus) Theory of the angiosperm flower, which interprets the flower as a uniaxial structure (Chapter 6), is at the core of most phylogenetic interpretations of angiosperm evolution developed in the second half of the twentieth century. In their emphasis on Magnoliaceae and their relatives as a starting point for angiosperm evolution, such phylogenetic schemes trace their origin to the pre-Darwinian classifications of De Candolle and Bentham and Hooker, but they have been broadly supported by comparative data from many sources, especially from wood anatomy (Bailey, 1944; Eames, 1961; Dickison, 1975) and palynology (Woodhouse, 1935; Walker, 1974a, b, 1976). The classifications of Hutchinson (1959), Cronquist (1968, 1981), Takhtajan (1969, 1980), Thorne (1976) and Dahlgren (1980) all represent different manifestations, each with their own idiosyncrasies, of the Euanthial paradigm.

Different applications of the magnoliid concept, and different conclusions as to which families should be included in the group, are a potential source of confusion in understanding early angiosperm evolution (Chapters 7, 8). However, recent phylogenetic studies have greatly clarified the situation and provide a firm basis for interpreting fossil material related to this important group of angiosperms. In this chapter we provide a brief outline of the current classification of eumagnoliids, and a review of those fossils, mainly from the Cretaceous, that can be assigned to the group. The fossil history of eumagnoliids is extensive, particularly from the Cenozoic, but there are also well-preserved and informative fossils from the Cretaceous. The early fossil record of eumagnoliids continues to increase rapidly as new palaeobotanical discoveries are made.

Classification of eumagnoliids

In the classification of Takhtajan (e.g. 1969) subclass Magnoliidae included six orders (Magnoliales, Laurales, Piperales, Aristolochiales, Rafflesiales, Nymphaeales), but it is now clear from phylogenetic analyses based on molecular data that this group is not monophyletic (e.g. Soltis et al., 2005; APGIII, 2009). Several families are now recognised to comprise the ANITA grade; others have been placed elsewhere in the angiosperm tree. Four orders, Magnoliales, Laurales, Piperales and Canellales (Zanis et al., 2003; APGIII, 2009), which were at the core of the former subclass Magnoliidae, comprise a monophyletic group, termed eumagnoliids by Soltis et al. (2000b) and magnoliids by APGIII (2009). Here we follow Soltis et al. (2000b) and refer to Magnoliales, Laurales, Piperales and Canellales as eumagnoliids. We use magnoliids informally in the broader conventional sense (Chapter 7). Magnoliales and Laurales form a clade that is sister to a clade comprising Piperales plus Canellales.

Preface

Developments in the study of fossil and living plants over the past few decades have greatly clarified many aspects of early angiosperm evolution. Explicit phylogenetic analyses, facilitated by the development of computer technology and based on both morphological and molecular data, have renewed interest in the relationships of angiosperms to other plants, the patterns of relationship among major groups of angiosperms, and the processes that have generated angiosperm diversity at both microevolutionary and macroevolutionary scales. At the same time, a rapid accumulation of new information on the structure and biology of many key groups of living angiosperms has catalysed comparative studies and brought to light many previously unrecognised features that provide new perspectives on angiosperm evolution.

Palaeobotanical studies have also been central in revitalising research on early angiosperm evolution and have advanced significantly our understanding of early angiosperm history. In particular, the discovery of diverse and exquisitely preserved fossil flowers and floral organs from the Cretaceous has yielded detailed information on the structural and systematic diversity of early angiosperms. These data complement the information available from living plants, and are also invaluable for testing evolutionary hypotheses based on extant taxa against palaeobotanical and stratigraphic evidence. The recognition of fossil pollen grains in situ within flowers has also provided new possibilities for interpreting the record of dispersed fossil pollen. Only a few decades ago the abundant occurrence of fossil angiosperm flowers in Cretaceous strata was unimagined, but today there is a rich floral record, much of which still remains to be analysed in detail. The key breakthrough was the recognition that numerous small fossil flowers, which are generally not visible to collectors in the field, can be extracted from Cretaceous sediments by using bulk-sieving techniques and studied with scanning electron microscopy (SEM), and now also with synchrotron X-ray microtomography (SXRTM). These techniques, modified from standard approaches to Cenozoic fossil floras in Europe, and pioneered in the Late Cretaceous of Scania, Sweden, have now yielded diverse angiosperm flowers from many new fossil floras (mesofossil floras) discovered in Lower and Upper Cretaceous strata in Europe, North America, Asia, New Zealand and Antarctica.

The phylogenetic diversification and ecological radiation of angiosperms (flowering plants) that took place in the Early Cretaceous, between about 135 and 65 million years ago, was one of the major biotic upheavals in the history of life. It had dramatic consequences for the composition and subsequent evolution of terrestrial ecosystems. Ancient Mesozoic vegetation, which was dominated by ferns, conifers, ginkgos and cycads, as well as Bennettitales and other groups of extinct seed plants, was eventually almost entirely replaced by more modern ecosystems dominated by angiosperms. Since the Early Cretaceous, high diversification rates have generated more than 350 000 extant angiosperm species. Today there are more living species of angiosperms than all other groups of land plants combined.

In their rise to ecological dominance angiosperms have exhibited extraordinary developmental and evolutionary plasticity. This has resulted in overwhelming morphological diversity and a great variety of adaptive types. Angiosperms are far more diverse in vegetative form and in the structure of their reproductive organs than any other group of land plants.

Rosids are a very large and heterogeneous group of eudicots that exhibit great diversity in both reproductive and vegetative morphology. The group is recognised based on molecular data, but support is currently weak and there are no unique morphological or anatomical features that define the rosids as a group. Among the shared characters noted by Soltis et al. (2005) is the presence of a reticulate pollen exine, but this is a common feature in many other groups of angiosperms. Other potential defining characters, such as obdiplostemony, mucilage cells in flowers and/or leaves, simple perforations of vessel end-walls, and alternating inter-vessel pitting, are also found occasionally in other groups and their value as synapomorphies of rosids remains to be determined.

Classification of rosids

Core rosids include Vitales, which are sister to two large clades: fabids sensu Judd and Olmstead (2004; Eurosids I sensu APGII, 2003) and malvids sensu Judd and Olmstead (2004; Eurosids II sensu APGII, 2003) (Figure 14.1). Fabids include Zygophyllales, which are sister to the nitrogen-fixing clade, comprising Cucurbitales, Fabales, Fagales and Rosales, and the COM clade, comprising Celastrales, Oxalidales and Malpighiales. Molecular support for these groupings is not strong (Endress and Matthews, 2006a). Malvids include Geraniales + Myrtales, which are sister to the remaining orders (Crossosomatales, Picramniales, Sapindales, Huerteales, Brassicales, Malvales) (APGIII, 2009).

Core eudicots include the bulk of extant angiosperm species diversity (Magallón et al., 1999) and can be divided into two large clades (rosids, asterids) together with a small number of lineages of uncertain phylogenetic position (Figure 13.1). In this chapter we deal with those lineages of core eudicots that fall outside the species-rich rosids and asterids, but that potentially occupy an important position in the early evolution of the group.

Classification of core eudicots

A four-gene phylogenetic analysis places Gunnerales (Gunneraceae and Myrothamnaceae) as the sister group to all other core eudicots (Soltis et al., 2003). In the APGII classification (2003) the precise position of such lineages as Berberidopsidales, Saxifragales, Santalales and Caryophyllales–Dilleniaceae was not yet securely determined, but in the latest APG classification (APGIII, 2009) the positions of some of those taxa have been clarified. Berberidopsidales, Santalales and Caryophyllales are resolved as a grade of successive sister lineages to the asterid clade, while the Saxifragales are placed as sister to the rosid clade. In this chapter we describe the fossil record of these lineages of core eudicots that appear to have diverged before the diversification of both rosids and asterids. Some of these lineages, such as the Saxifragales, have a well-documented fossil history extending back to the Late Cretaceous.

Overwhelming palaeobotanical evidence from many geographical areas indicates that angiosperms first attained ecological prominence during the mid-Cretaceous and that this led ultimately to profound changes in the composition of terrestrial plant communities and ecosystems. Knowledge of how these changes were manifested in changes in ecosystem structure and function, and how this affected, and was influenced by, larger-scale changes in the global environment is still at an early stage. A major difficulty is the relatively poor stratigraphic resolution for many Cretaceous terrestrial deposits as well as difficulties of correlating floras from different environmental settings, different latitudes and different continents. An exhaustive treatment of these issues is beyond the scope of this book, but here we provide an overview of vegetational change through the Cretaceous to place the evolutionary changes discussed in previous chapters in a broader context.

Transition to angiosperm-dominated vegetation

The transition to angiosperm-dominated vegetation did not occur simultaneously or uniformly in all parts of the world, and our understanding of how this great transition unfolded geographically is still rudimentary. Nevertheless it is possible to distinguish some broad patterns, and estimates of the timing and magnitude of angiosperm diversification through the Cretaceous have been made at both regional (e.g. Samylina, 1976; Lupia et al., 1999; Cantrill and Poole, 2005) and global scales (e.g. Niklas et al., 1980; Crane, 1987; Crane and Lidgard, 1990; Lidgard and Crane, 1990). Data on the transition to angiosperm-dominated vegetation are most complete from middle- and high-palaeolatitude regions of Laurasia.

The palaeontological information summarised in earlier chapters provides direct historical evidence of the pattern of angiosperm evolution through time. Such information is an important complement to contemporary approaches that use data from living plants to look back into evolutionary history. At the same time, living plants are the essential points of reference for any meaningful interpretation of the palaeontological record. Only from studies of the modern world can we understand the lives of plants and the roles they play in ecological systems. For these reasons the emphasis in this book has been on the integration of information from living and fossil plants, which we see as vital for a full understanding of plant evolution.

In this chapter we provide a brief integrated overview of the major patterns of angiosperm evolution as revealed by studies of living plants and the fossil record. We also provide more detailed consideration of selected groups of angiosperms with a particularly interesting or informative palaeontological history. Finally, we conclude with a brief consideration of angiosperm evolution through the Cenozoic to make the temporal connection between our primary focus on the Cretaceous fossil record, and the diversity of angiosperms that exists today.

Monocots (monocotyledons) are a major clade of angiosperms that have been recognised as a natural group since John Ray in the early eighteenth century (Ray, 1703). The name refers to the single cotyledon, a feature found in all members of the group. Monocots also lack the ability to produce secondary xylem and phloem in the same way as other angiosperms and most other seed plants. Nevertheless, they are hugely diverse in habit and form. Monocots include aquatic herbs and tall palms as well as Arctic grasses and tropical epiphytes. Taken together, monocots account for a little over a fifth of angiosperm species. A recent attempt to develop a global checklist of all monocots at the species level suggests that about 70 000 extant species of monocot are currently known (The Board of Trustees of the Royal Botanic Gardens, 2008). In this chapter we provide a brief overview of monocot classification and consider the fossil history of the group, focusing particularly on the earliest records and other evidence of monocots from the Cretaceous.

Classification of monocots

Rapid progress in the development of a phylogenetic classification of monocots over the past 25 years was stimulated in large part by the important synthetic work of Dahlgren et al. (1985). This was taken forward in a series of symposia (Rudall et al., 1995; Wilson and Morrison, 2000; Columbus et al., 2006) as well many individual research contributions. Sampling for phylogenetic analyses based on molecular data has been especially intensive and a broad consensus has emerged on the major clades of monocots (Figure 11.1), as well as many aspects of phylogenetic pattern within these groups (e.g. Chase, 2004). There is also agreement on some aspects of the relationships among these clades.

In addition to great diversity in floral structure and modes of pollination, extant angiosperms also exhibit astonishing variety in fruit and seed morphology. This variety reflects a great range of different modes of dispersal as well as other aspects of dispersal biology. Dispersal and establishment is a key phase in the plant life cycle. Dispersal allows new populations to be established and new habitats to be colonised. It therefore has an important influence on the structure of plant populations. The evolution of angiosperm dispersal modes may have had important consequences for large-scale patterns of angiosperm evolution.

As with pollination, dispersal in angiosperms and other seed plants may be abiotic or may involve interactions with animals. However, in contrast to pollination, where pollen is carried from flower to flower, in dispersal the destination for the propagule is generally much less tightly constrained (Wheelwright and Orians, 1982). The potential reward for the disperser is also available only at the outset, rather than also on completion. The evolutionary dynamic is therefore different. As a result, co-evolution between plants and their dispersers may be much less specific than is often the case for pollination (Wheelwright and Orians, 1982). Co-evolution is more likely to be ‘loose’ rather than ‘tight’ (Herrera, 1985; Fleming, 1991).

Eudicots are an important and well-supported monophyletic group of angiosperms. The clade includes almost all dicotyledons; the only groups excluded are eumagnoliids and those dicots at the ANITA grade plus Chloranthaceae and Ceratophyllum. Eudicots are therefore broadly equivalent to all dicot lineages except the Magnoliidae sensu Takhtajan (1980) (Chapter 7). The term eudicotyledons, or eudicots, was introduced to recognise the monophyletic status of this major group (Doyle and Hotton, 1991), and has been widely accepted in subsequent works. Earlier studies referred to eudicots as non-magnoliid dicots (Walker and Doyle, 1975; Crane, 1989) or tricolpates (Donoghue and Doyle, 1989b; Judd and Olmstead, 2004). The term tricolpates refers to the tricolpate aperture configuration, which is characteristic of the pollen of many early-diverging eudicots. While many eudicots have pollen with other aperture configurations, almost all are based on the triaperturate ground plan (Doyle and Hotton, 1991).

Eudicots are extremely diverse, and exhibit an almost bewildering breadth of morphological and ecological variation. The group contains about three-quarters of all extant angiosperm species (Magallón et al., 1999). The fossil record of eudicots is extensive and informative about their evolutionary history. In this chapter we provide a brief outline of eudicot classification and the early fossil record of the group. We then focus on those lineages that diverged at an early stage from the line that gave rise to the bulk of eudicot species. We emphasise particularly the Cretaceous fossil record, and those taxa that can be recognised based on fossil flowers or other reproductive organs.

The global environment during the earliest phases of angiosperm diversification was radically different from that of today. The geography and distribution of continents was unlike that of our modern world and the Cretaceous was also a time of generally higher sea levels and higher global temperatures. There were low thermal gradients from the equator to the poles (DeConto et al., 2000a; Gale, 2000; MacLeod et al., 2000), patterns of rainfall were quite different (Parrish, 1987) and there is no unequivocal evidence of polar ice caps during the Cretaceous and Early Cenozoic (Gale, 2000). The Cretaceous is often considered the classic example of ‘greenhouse’ or ‘supergreenhouse’ conditions in Earth history. These unusual conditions compared with today profoundly influenced the ecology and evolution of life on land as angiosperms underwent their initial radiation and then diversified to become the dominant primary producers in most terrestrial ecosystems. Our modern world provides a poor analogue for the environmental backdrop against which more than half of the evolutionary history of angiosperms unfolded. In this chapter we provide a brief overview of changes in palaeogeography and palaeoclimate through the Cretaceous period, a time interval of 80 million years (Figure 3.1). More detailed accounts can be found elsewhere (e.g. Skelton, 2003b). The aim here is to provide a short summary that places patterns of early angiosperm evolution in their environmental context.

Palaeogeography

During the later Palaeozoic and Triassic a single supercontinent, Pangaea, was formed through the coalescence of all pre-existing major continental masses. The southern part, Gondwana, consisted of South America, Africa, Apulia (present-day Italy), Arabia, Australia, New Zealand, Antarctica, India, and Madagascar. The northern part, Laurasia, consisted of many smaller and larger continents with North America, Greenland, Europe, Iberia, Siberia, Kazakhstan, Kolyma and China as its main components. Pangaea persisted into the early Mesozoic, but began to break up during the Triassic and Jurassic through a series of processes that continued into the Cretaceous (Figures 3.2–3.5) (Smith et al., 1981; Scotese et al., 1988).

Pollination, the successful transfer of pollen from the pollen sacs (microsporangia) into proximity with the ovule, is the essential precursor to fertilisation and therefore to sexual reproduction in seed plants. Pollination has been studied most intensively in angiosperms, although few species have been examined in detail compared with the great variety of flowers within the group (e.g. Proctor et al., 1996; Thien et al., 2009). Pollination in extant non-angiosperm seed plants has received less attention, but studies over the past few decades now provide a more complete context within which pollination in angiosperms can be evaluated and studied (e.g. Owens et al., 1998).

Interpretation of pollination in extinct plants faces significant difficulties. Only rarely is there relatively direct evidence of flower–pollinator interactions (e.g. insect gut contents, coprolites, insects preserved within flowers, insects carrying pollen) and interpretations of pollination in extinct plants therefore depend heavily on extrapolations to extant taxa based on structural similarities. This often leads to plausible interpretations, but it may also be constraining. There is no a priori reason why the spectrum of plant–pollinator interactions existing today should also include all of those that existed in the past, and inferring floral function from floral structure, even in extant plants, can sometimes be difficult. It is therefore especially challenging to infer pollination in extinct seed plants (e.g. Caytonia, Bennettitales) that have no clear close living relatives.

In this chapter we describe plant fossils, mainly from the Early Cretaceous, that for various reasons cannot be assigned reliably to any extant group of angiosperms. In a few cases the relationship to angiosperms themselves is also uncertain (sections 9.1, 9.7). Some of these fossils are well-preserved floral structures, and the available information is reasonably extensive, but assignment to an extant order or family is precluded because the fossil has characters, or suites of characters, that are difficult to relate to a particular extant angiosperm group. In some instances the lack of relevant comparative data for extant taxa is also a problem. Other fossils considered in this chapter are dispersed floral parts, such as pollen, fruits and seeds. These provide further documentation of diversity among early angiosperms, but are difficult to assign to an extant group because they lack sufficient diagnostic characters.

In addition to the material discussed in this chapter there are many other fossils, especially from the Portugal and Potomac Group mesofossil floras, that remain to be described and for which relationships are not yet evaluated. A comprehensive account of these fossils is beyond the scope of this book. However, many of them probably represent early-diverging lineages of angiosperms of uncertain relationship that would probably also be included here.

The current high islands of the Hawaiian archipelago are among the most remote land masses in the world. They lie 3500 km from California, the nearest continental source, and approximately 2300 km from the Marquesas, the nearest islands (Fig. 4.1). They are the southernmost islands in the Hawaiian Ridge, formed successively over a ‘hot spot’ that has allowed magma to penetrate the Pacific Plate. The plate has moved gradually north and northwestwards over the past 85 Ma, leaving the previously formed islands to gradually erode and subside (Clague, 1996). The current high islands (Fig. 4.1, inset) range in age from Kauai/Niihau (5.1–4.9 Ma), to Oahu (3.7–2.6 Ma), to Maui Nui (2.2–1.2 Ma), during the Pleistocene comprising several islands – West Maui (1.3 Ma), East Maui (0.75 Ma), Molokai (1.76–1.90 Ma), Lanai (1.28 Ma) and Kaho’olawe (1.03 Ma) – and Hawaii (0.5 Ma to present) (Price & Clague, 2002).