Extending the scope of Darwin’s ‘abominable mystery’: integrative approaches to understanding angiosperm origins and species richness

Abstract

Background and aims

Angiosperms are the most species-rich group of land plants, but their origins and fast and intense diversification still require an explanation.

Scope

Extending research scopes can broaden theoretical frameworks and lines of evidence that can lead to solving this ‘abominable mystery’. Solutions lie in understanding evolutionary trends across taxa and throughout the Phanerozoic, and integration between hypotheses and ideas that are derived from multiple disciplines.

Key Findings

Descriptions of evolutionary chronologies should integrate between molecular phylogenies, descriptive palaeontology and palaeoecology. New molecular chronologies open new avenues of research of possible Palaeozoic angiosperm ancestors and how they evolved during as many as 200Myr until the emergence of true angiosperms. The idea that ‘biodiversity creates biodiversity’ requires evidence from past and present ecologies, with changes in herbivory and resource availability throughout the Phanerozoic appearing to be particularly promising.

Conclusions

Promoting our understanding of angiosperm origins and diversification in particular, and the evolution of biodiversity in general, requires more profound understanding of the ecological past through integrating taxonomic, temporal and ecological scopes.

INTRODUCTION

Angiosperms constitute approx. 90 % of all land plant species and over 95 % of vascular plant species (Crepet and Niklas, 2009), and are undoubtedly the most successful major land plant clade in terms of species number and diversity, proliferation, prevalence and biomass. This stands in contrast to their being relatively young, first appearing towards the end of the Jurassic (for a geological time scale, see Fig. 1) and becoming prominent in the fossil record only during the Cretaceous (Crane et al., 1995; Qiu et al., 1999; Crepet, 2000; Davies et al., 2004; Crepet and Niklas, 2009; Magallon and Castillo, 2009; Bell et al., 2010; Krassilov and Silantieva, 2013; Augusto et al., 2014; Magallon et al., 2015), compared to pteridophytes and gymnosperms that have dominated many ecosystems since at least the Carboniferous (e.g. Niklas et al., 1983; DiMichele et al., 2001). Remarkably, some of the most species-rich angiosperm clades (e.g. Asterales, Fabaceae, Lamiales, Poales and Orchidaceae) emerged only during the Late Cretaceous (Wikström et al., 2001; Janssen and Bremer, 2004; Magallon and Castillo, 2009; Magallon et al., 2015), and some of the most species-rich families (e.g. Asteraceae and Poaceae) emerged and proliferated only during the Palaeogene (e.g. Magallon et al., 2015).

Fig. 1.

(A) Number of plant species (redrawn from Niklas et al., 1983) and (B) number of tetrapod families (redrawn from Sahney et al., 2010a) belonging to major clades throughout the Phanerozoic. Dots represent maximal species/family number ever recorded at a single point in the palaeontological record against the estimated time of initial evolutionary radiation and success. Note that ‘gymnosperms’ in this figure also includes angiosperm ancestors (between the gene divergence time of extant gymnosperms and angiosperm approximately 350Mya and the first emergence of true angiosperms), as described in the section ‘New Chronology Opens New Horizons’. Key to geological periods: O, Ordovician; S, Silurian; D, Devonian; C, Carboniferous; P, Permian; T, Triassic; J, Jurassic; K, Cretaceous; Pg, Palaeogene; Ng, Neogene.

The puzzles of angiosperm origins and how this young clade achieved such great species richness and ecological dominance over a relatively short time are commonly known as Darwin’s first and second ‘abominable mysteries’, respectively, although this term deviates from what Darwin originally meant (Friedman, 2009). The second ‘abominable mystery’ is even more striking because angiosperm species number today is much greater than the species number of any other vascular plant group ever, even if one accounts for representation biases in geological time (Fig. 1A) (Niklas et al., 1983; Crepet and Niklas, 2009). For example, the gymnosperm Ginkgophyta has probably never consisted of more than a few dozens or few hundreds of species (Niklas et al., 1983), a species number that is considerably less than the species number in several angiosperm families today, and considerably less than the over 10000 species in the angiosperm family Orchidaceae alone (Gravendeel et al., 2004).

Despite recent discoveries and considerable methodological and theoretical advances in recent years, angiosperm origins and early diversification remain elusive (Frohlich and Chase, 2007; Crepet and Niklas, 2009; Augusto et al., 2014). Suggested explanations for the fast origination and early diversification and success of angiosperms include explanations that rely on novelties in angiosperm biology, such as physiology and leaf habits (Bond, 1989; Feild and Arens, 2007; Brodribb and Feild, 2010; Feild et al., 2011), reproduction through double fertilization within animal-pollinated flowers (Crane et al., 1995; Grimaldi, 1999; Friis et al., 2006; Hu et al., 2008) and polyploidization and subsequent diploidization (De Bodt et al., 2005; Soltis et al., 2008; Fawcett and Van de Peer, 2010; Dodsworth et al., 2016; Alix et al., 2017; Paule et al., 2017). Other hypotheses put more emphasis on angiosperms’ interactions with herbivores and competitors (Bakker, 1978; Barrett and Willis, 2001; Butler et al., 2010) or on climate change and plate tectonics (Heimhofer et al., 2005; Chaboureau et al., 2008; Fawcett and Van de Peer, 2010) as possible stimuli for this diversification.

As Crepet and Niklas (2009) and Augusto et al. (2014) noted, the truth possibly lies in a synthesis of many of the existing hypotheses, but such a synthesis is not easy to achieve. In this viewpoint, I demonstrate how this synthesis can be done and bring us closer to resolving both the question of angiosperm origins and the question of angiosperm diversity, by putting more emphasis on how a synthesis of palaeontology, physiology and ecology can be of special significance. Most previous studies of these matters, and consequently the hypotheses they produced, have attained a relatively limited temporal scope, focusing on the timing of angiosperm origins and early diversification. Such an approach may have hindered our advancement towards a more complete understanding of angiosperm origins and evolution. Instead, I suggest a more holistic and integrative approach that puts more emphasis on long-term patterns throughout Life’s history. I demonstrate how new discoveries and realizations about the evolution of terrestrial biodiversity and the complex dating of evolutionary events offer us new perspectives on the issues in hand. I also discuss what questions these new discoveries raise, and what future methodological advancements can lead to solving these questions.

PARALLELS AND GLOBAL INCREASE IN BIODIVERSITY

That younger major clades are more species-rich than older ones have ever been is not unique to land plants (Cascales-Miñana et al., 2016). The younger tetrapod clades of birds and mammals are also more species-rich compared to the older amphibians and reptiles (Fig. 1B) (Sahney et al., 2010a); among insects the most species-rich orders (Hymenoptera, Lepidoptera, Diptera, but probably not Coleoptera) have also diversified relatively late (Mayhew, 2007; Misof et al., 2014); and a comparable pattern can also be seen in Sepkoski’s (1997) three marine faunas, although these do not represent taxonomic units. Thus, angiosperm diversification is part of a general increase in biodiversity over geological time, which is often suggested to arise from positive feedbacks caused by escalation and specialization (Sepkoski, 1997; Gravendeel et al., 2004; Erwin, 2008; McNamara, 2008; Benton, 2010; Sahney et al., 2010a; Butterfield, 2011). When angiosperms are discussed, examples of such feedbacks include hypotheses about coevolution of flowers with pollinators (Gravendeel et al., 2004; Hu et al., 2008), of grasses with mammalian herbivores (Osborne, 2008; Edwards et al., 2010), of fruits with frugivores (Fleming and Kress, 2011), and perhaps of early angiosperms or angiosperm orders with certain dinosaur clades (Barrett and Willis, 2001; Butler et al., 2010; Katz, 2015).

Therefore, there are many parallels to angiosperm diversification from which insights can be drawn. Putting angiosperm emergence and evolution in this broader context, and more generally putting more effort into studying the evolution of biodiversity and Earth–life systems at large temporal scales (Benton and Emerson, 2007; Erwin, 2008; Beerling, 2012), is likely to contribute to our understanding of angiosperms’ early evolution and current species richness. The following sections will demonstrate this by reviewing recent advancements in methodology, evidence and theory, and I highlight some possible avenues for future research.

NEW CHRONOLOGY OPENS NEW HORIZONS

Looking deeper into geological time and describing long-term processes requires a more elaborate view of the tempo of evolutionary changes and differences between timings of various evolutionary events. It should be borne in mind that gene ages, clade ages and trait ages are not the same. Molecular (gene) ages, either those of stem groups or those of crown groups, represent the time in which two genetic lineages diverged, but does not mean that taxa or traits have also diverged at that time (Edwards and Beerli, 2000; Nichols, 2001; Pulquerio and Nichols, 2006). Many plant phylogenies, including recent angiosperm phylogenies (Wikström et al., 2001; Janssen and Bremer, 2004; Magallon and Castillo, 2009; Bell et al., 2010; Magallon et al., 2015), are based on mitochondrial, plastidial or ribosomal genes that do not encode morphological and anatomical traits such as vascular structure, leaf morphology or reproductive system. However, morphology and anatomy are the key traits that define clades (as evident in practically any analytical flora), whose evolution is traced in the fossil record (e.g. Friis et al., 2006; Feild et al., 2011; Krassilov and Silantieva, 2013), and on which natural selection mostly acts. Therefore, when tracing the origins, evolution and diversification of angiosperms, or any other taxon for that matter, we need to distinguish between the gene, trait and clade ages. Molecular phylogenies mark the divergence of genetic lineages, and remain the more appropriate method for constructing multiple-trait phylogenies and for molecular dating. However, fossils are more appropriate to identify and date the emergence of characteristic traits, diversification or ecological success. Integrating molecular and fossil evidence sometimes yields unexpected mismatches, which raise intriguing questions that may reveal hidden chapters in evolutionary history, and angiosperms are no exception.

Recent molecular phylogenies suggest that extant gymnosperms and angiosperms are monophyletic sister clades that have diverged approx. 300–50Mya (Qiu et al., 1999; Bowe et al., 2000; Magallon and Castillo, 2009; Magallon et al., 2015). A recent discovery of angiosperm-like pollen from approx. 243Mya (Hochuli and Feist-Burkhardt, 2013) fits this molecular dating. This gene divergence date is nearly 200Myr earlier than other major events in angiosperm evolution: the emergence of the first unequivocal angiosperm traits during the Early Cretaceous or at best the very Late Jurassic (Niklas et al., 1983; Crane et al., 1995; Sun et al., 1998; Friis et al., 2006; Feild and Arens, 2007; Bell et al., 2010; Brodribb and Feild, 2010; Feild et al., 2011; Herendeen et al., 2017), the early diversification of the angiosperm crown group during the Late Cretaceous and into the Palaeogene and Neogene (Davies et al., 2004; Janssen and Bremer, 2004; Crepet and Niklas, 2009; Magallon and Castillo, 2009; Bell et al., 2010; Magallon et al., 2015), and their ecological success that probably did not pre-date the Late Cretaceous (Krassilov and Silantieva, 2013) and became prominent only during the Palaeogene. This disparity has been noted by Herendeen et al. (2017), who argued correctly that there is no pre-Cretaceous fossil that can be confidently assigned to the angiosperms. However, the lack of fossil evidence thus far does not negate the new molecular chronology. If anything, it shows that more attention should be put into this possible 200Myr gap.

During those nearly 200Myr, angiosperm ancestors probably did not differ morphologically and anatomically from gymnosperms, and their fossils that did not differ from those of other land plant clades were classified as pteridosperms or gymnosperms. This suggests that at least some of the gymnosperms referred to in Fig. 1A during this time period are probably angiosperm ancestors that do not belong to the lineages leading to extant gymnosperms. Thus, Fig. 1A may be biased towards overrepresentation of gymnosperms on the one hand and underrepresentation of the lineage leading to extant angiosperms on the other, and underestimates the extent of the recent angiosperm species richness over the historical peak of true gymnosperm species richness.

This chronological gap and the difficulty in identifying angiosperm ancestors raise two questions that, if answered, may shed new light on angiosperm origins and possible late success. The first question is whether fossils from the two clades can be distinguished prior to the emergence of key angiosperm traits. If such a distinction is made, it may provide new insights on the biology and ecology of angiosperm ancestors. To date, no clear criteria or strong analytical tools for making such a distinction have been developed for either macrofossils (e.g. wood, leaves and reproductive systems) or microfossils (e.g. pollen and phytoliths). Advancements in phytolith morphological analysis improve taxonomic resolution, and the strong taxonomic signal in phytolith morphological variations (e.g. Prychid et al., 2004) suggests that the evolution of phytolith morphologies may be traced and provide a method to distinguish between the two clades. Nevertheless, identifying the evolution of gymnosperm and angiosperm phytolith morphologies remains undeveloped at this point in time, despite its great potential to answer some questions in plant evolutionary history. This potential still waits to be realized. A Late Cretaceous phytolith assemblage with confidently identified grass phytoliths (Prasad et al., 2005), the procurement of Devonian phytoliths by Carter (1999) and the identification of phytoliths within petrified wood macrofossil of the Permo-Triassic boundary (Looy et al., 2012) are encouraging finds that suggest phytoliths of early angiosperms and their ancestors can be preserved and identified.

Other possible solutions may exist if organic molecules are preserved within fossils. In recent years we have witnessed some exciting revelations on vertebrate biology from such remains, such as paravian theropod feather colour deciphered from preserved melanosomes (Li et al., 2010) and successful attempts to extract DNA from Cretaceous bones (Woodward et al., 1994). If comparable remains can be found in plant fossils, then it may be possible to use them to identify angiosperm ancestors and perhaps even learn something about their traits. Admittedly, informative organic compounds are rarely preserved in deep geological time, and methods to extract DNA from plant fossils are still in their infancy (Gugerli et al., 2005), and only rarely has DNA been extracted from pre-Quaternary plant fossils (Golenberg et al., 1990, 1991; Soltis et al., 1992; Parducci and Petit, 2004). However, this potential source of data should not be overlooked.

Alternatively to this first question, one may argue that pre-Cretaceous angiosperm ancestors were actually angiosperm-like, i.e. possessing all or most of the key traits that we currently associate with angiosperms, but were just too rare or geographically limited to be found in the fossil record. While this possibility cannot be refuted and fits some ecophysiological studies (Feild et al., 2004), it appears unlikely (Hochuli and Feist-Burkhardt, 2013). Let us assume that angiosperm key traits, which the fossil evidence suggests emerged only during the Cretaceous (e.g. Herendeen et al., 2017), have indeed existed within the angiosperm clade since the Carboniferous. If the early and fast diversification and success of angiosperms are the result of their many novel key traits (e.g. Bond, 1989; Crane et al., 1995; Grimaldi, 1999; Friis et al., 2006; Feild and Arens, 2007; Brodribb and Feild, 2010; Feild et al., 2011), then there should be a most convincing and revolutionary argumentation to why they did not rise to dominate the Earth very shortly after this suite of traits had emerged. After all, many of the existing hypotheses suggest that this was the case in the Cretaceous. Hence, it is more reasonable (although still requires proof) that diversification follows innovation (i.e. trait emergence) rather than gene divergence.

Therefore, the second question raised by the possible new chronology is whether there is an explanation for the gap itself, i.e. why angiosperm ancestors did not evolve angiosperm-like traits for 200Myr or why a putative Palaeozoic emergence of such traits did not lead to rapid diversification and success. If some differences between the two clades can be observed, they may explain why the two clades did not differ morphologically for nearly 200Myr, or inform us of very early stages of the transition from angiosperm ancestors that resemble pteridosperms and gymnosperms to true angiosperms. If key evolutionary and ecological changes in the gymnosperm clade, angiosperm ancestors and other events in geological time (e.g. environmental change or events in animal evolution) appear to temporally coalesce, new explanations for the evolutionary changes that led to angiosperm emergence, diversification and success may arise.

Geologists and palaeobiologists have discussed such cases for over a century. One of the best examples is the Cretaceous Terrestrial Revolution (Lloyd et al., 2008). The Jurassic/Cretaceous transition, which marks the early rise of true angiosperms, also marks the early rise of paraves (Sereno, 1999), the early diversification of Hymenoptera, Lepidoptera and Diptera (Misof et al., 2014), and the replacement of herbivorous dinosaur sauropods by ornithischians (Bakker, 1978; but see Barrett and Willis, 2001; Butler et al., 2010). Early angiosperm diversification during the Mid/Late Cretaceous coincides with the early diversification of placental mammals (Archibald, 2003; Springer et al., 2003; Wible et al., 2007) and insect pollinators (Hu et al., 2008). Another phenomenon associated with this period was the evolution of novel herbivorous dinosaur groups (Lloyd et al., 2008), including grit-adapted dentition in hadrosaurs and ceratopsians (Williams et al., 2009; Prieto-Marquez, 2010), and gondwanatherians (Verma et al., 2012), which may be related to the evolution of some silicon-rich angiosperm clades (Katz, 2015).

Other examples include the Mid/Late Devonian emergence of megaphyllous leaves coinciding with a drop in atmospheric CO₂ concentrations (Beerling et al., 2001; Beerling, 2005; Beerling and Berner, 2005), and the rise of silicon-rich C₄ grasslands coinciding with decreasing atmospheric CO₂ concentrations and the rise of hypsodonty and grazing (Stebbins, 1981; Christin et al., 2008; Edward et al., 2010; but see Strömberg et al., 2013, 2016).

Until now, coincidences of key events of angiosperm evolution with other environmental and evolutionary changes have focused on the Cretaceous or later, assuming that there were no earlier non-gymnosperm angiosperm ancestors or early angiosperm forms. The new chronology suggests that there was as long as 200Myr in which such ancestors existed. Hypotheses of temporal coalescence in angiosperm evolution can now be sought at broader temporal scales, and comparisons with angiosperm ancestry in deeper geological time become possible.

GLOBAL INCREASE IN BIODIVERSITY AND COMPLEXITY

Key innovations in plant evolution have long been suggested to culminate from changes in atmosphere and climate, such as decreases in atmospheric CO₂ concentrations that are thought to be associated with the evolution of megaphyllous leaves during the Devonian (Beerling et al., 2001; Beerling, 2005; Beerling and Berner, 2005) and rise of C₄ photosynthesis during the early Neogene (Christin et al., 2008; Edwards et al., 2010; but see Osborne, 2008). The same probably applies for the effects of ecology on early diversification of major clades. The general increase in biodiversity over geological time (e.g. Niklas et al., 1983; Sepkoski, 1997; Sahney et al., 2010a; Fig. 1) is often associated with the idea of ‘biodiversity creating biodiversity’. New species create new niches, form more complex ecological nexuses and provide more opportunities and stimuli for specialization and the evolution of even more species, in a series of positive feedback loops that continues for hundreds of millions of years (Sepkoski, 1997; Gravendeel et al., 2004; Erwin, 2008; McNamara, 2008; Forbes et al., 2009; Benton, 2010; Sahney et al., 2010a; Butterfield, 2011).

In view of the new suggested chronology of the gymnosperm–angiosperm divergence, the possible similar gene ages of gymnosperms and angiosperms cannot explain the difference in species richness seen in Fig. 1A. It is therefore more reasonable that if geological time affects biodiversity, the time of early diversification is more important than the time of early emergence. This can also be seen in Fig. 1B, where mammals and birds, which have both started to rise to success around the Cretaceous/Palaeogene boundary, have reached comparable species diversities despite mammals emerging approx. 50Myr earlier. A combination of external abiotic changes that exert new stresses or release previous ones with increased biodiversity and ecological complexity may be a major driver of early clade diversification. An example for this is the high CO₂ constraints that have hindered the emergence of megaphyllous leaves and their removal once CO₂ concentrations decreased (Beerling et al., 2001; Beerling, 2005; Beerling and Berner, 2005). Comparing the ecological realities during the Carboniferous rise of seed plants and the Cretaceous rise of angiosperms suggests that such effects can last for tens of millions of years, and that the search for what drove angiosperm diversification requires comparison with early seed plant diversification during the Carboniferous.

HERBIVORY AND DIVERSITY

Herbivory, in the broader sense of plant tissue consumption by animals, can have profound effects on plant fitness. At the individual level, herbivory can have negative effects because of the destruction of tissues that are essential for the plant to complete its life cycle (e.g. leaves, roots and reproductive systems), or positive if it assists in seed dispersal and their excretion into deposits rich with organics, nutrients and moisture. At the evolutionary scale, negative and positive effects can promote defensive or mutualistic specialization (respectively) and thus diversification. Feeding by different herbivores may further affect competition among plants, and thus affect plant community composition and natural selection, and further affect diversification by promoting niche differentiation.

Bakker (1978) suggested that the shift from long-necked treetop-feeding sauropods to short-necked ornithischians during the Jurassic/Cretaceous transition resulted in more severe damage to the slow-growing and less defended gymnosperm saplings, thus releasing the fast-growing and better defended early angiosperm seedlings from competition and facilitating their ecological success. This explanation is now disputed because of temporal mismatches between the shift in dinosaur herbivory and early angiosperm emergence, and may only apply for the Late Cretaceous angiosperm evolutionary radiation (Barrett and Willis, 2001; Butler et al., 2010). This idea is also criticised because it assumes that dinosaurs consumed early angiosperms rarely or in very small amounts, which has been challenged by some recent evidence (Prasad et al., 2005).

Bond (1989) offered a somewhat more relaxed version of Bakker’s (1978) hypothesis, suggesting that frequent disturbances (e.g. herbivory) favour angiosperms over gymnosperms because of the former’s faster growth rates in early developmental stages. When seed plants emerged and began to diversify during the Carboniferous, terrestrial arthropods and tetrapods were mostly carnivorous or detritivorous, whereas terrestrial herbivory first appeared relatively late in the Carboniferous (Edwards et al., 1995; Sues and Reisz, 1998; Labandeira, 2006, 2007; Sahney et al., 2010b; Pearson et al., 2013). With herbivores being a minor threat, early seed plants were not driven to evolve intricate and diverse anti-herbivory defences or to grow fast, and therefore did not evolve into many species. If this is the case, then the increase in pteridosperm species number during the Late Triassic and Early Jurassic (Fig. 1A) may be related to the onset of dinosaur herbivores stimulating the evolution of anti-herbivory defences among seed plants, possibly including both true gymnosperms and angiosperm ancestors. Frugivory is also unlikely to have played a significant role in plant diversification prior to the Cainozoic. Most specialized frugivorous mammal and bird clades evolved and diversified only toward the end of the Cretaceous, probably shortly after the first fleshy-fruited angiosperms evolved (Crane et al., 1995; Fleming and Kress, 2011), while there is no evidence for specialized frugivorous dinosaurs.

Extending our knowledge of the evolution of herbivory in deep time can contribute to assessing whether and how herbivory may have been involved in the early evolution of seed plants and angiosperms. While evidence for herbivory by small invertebrates is relatively abundant, consisting of piercing, mining, boring and galling signs in fossil leaves (e.g. Labandeira, 2006), plant fossils rarely preserve evidence of vertebrate herbivory because of the frequent consumption of whole plant parts. Nevertheless, coprolites can be used to identify herbivory and herbivore diets (e.g. Rodriguez-de la Rosa et al., 1998; Tiffney, 2004; Prasad et al., 2005; Chin, 2007; Wood et al., 2013). For example, if seeds are preserved in coprolites, they may indicate frugivory and seed dispersal, but such evidence is rare (Tiffney, 2004). Likewise, leaf cuticles in coprolites can serve as evidence for leaf consumption by herbivores (Rodriguez-de la Rosa et al., 1998). A further question is whether herbivory by invertebrates and vertebrates was equally significant in the short and long term and at different points in geological time. Devising new methods to identify herbivory, and more importantly its effects on evolution of plant traits, may provide the needed perspective of how herbivory and plant–animal coevolution may have been driving the evolution of plants, animals and biodiversity in general.

RESOURCE AVAILABILITY, UTILIZATION AND CYCLING

Resource availability, utilization and cycling may also play an appreciable role in driving evolution. Plants require certain resources (most notably sun irradiance, water, carbon and nutrients) to complete their life cycles and reproduce. A relative scarcity of such resources can limit plant success and favour slow-growing species that do not need to acquire substantial amounts of resources to support fast growth rates, whereas in more resource-rich environments, more efficient utilization of resources and faster growth rates make species better competitors (e.g. Grime, 1977; Reich, 2014). Therefore, increasing resource availability along environmental gradients can be associated with the decrease in requirement for a narrow range of specific adaptations, increase in competition that induces niche partitioning, and therefore an increase in functional diversity and divergence (Callaway et al., 2002; Brooker et al., 2008; Yu et al., 2017; for a discussion over the validity of this model in varying scales, see Pugnaire and Luque, 2001; Maestre et al., 2005, 2009; Gross et al., 2013).

Extant angiosperms dominate most of Earth’s terrestrial ecosystems. However, they more frequently dominate resource-rich or disturbed habitats that favour their fast growth rates, whereas extant gymnosperms dominate mostly resource-poor or low-temperature habitats in which growth rates are limited (Bond, 1989). This pattern may provide a further explanation of angiosperms’ rise to success. Recent evidence suggests that Cretaceous early angiosperms were more abundant in disturbed habitats such as forest gaps and riparian habitats (Feild et al., 2004) and in aquatic or coastal habitats that are affected by sea or lake levels (Krassilov and Silantieva, 2013). These early angiosperms also had more efficient vascular systems compared to contemporaneous or extant gymnosperms (Brodribb and Feild, 2010), which supplied them with ample amounts of water and supported more efficient photosynthesis compared to gymnosperms (Feild and Arens, 2007; Brodribb and Feild, 2010; Feild et al., 2011). These supported fast growth rates that allowed angiosperms to outcompete gymnosperms in resource-rich and disturbed habitats and eventually restricted the latter to resource-poor or low-temperature habitats, where growth rates are limited (Bond, 1989; Augusto et al., 2014).

Berendse and Scheffer (2009) further suggested that angiosperms could have gained dominance in resource-limited habitats, continually proliferating over the world and pushing gymnosperms to more marginal habitats, not only by their ability to acquire and utilize resources but also by habitat facilitation. Angiosperms can partially overcome resource scarcity because their litter decomposes faster than gymnosperm litter (Cornwell et al., 2008), thus releasing nutrients back to the soil more quickly and gradually enriching adjacent habitats and appropriating them for angiosperm expansion.

Berendse and Scheffer’s (2009) hypothesis offers a reasonable and testable explanation for angiosperm prominence in ecosystems and biomes, but has made no attempt to explain their diversification. However, clues can be found if again we broaden our perspective and the scope of this hypothesis. A comparative look at the possible involvement of terrestrial animals in resource cycling during the Devonian/Carboniferous and the Cretaceous offers an extension to this hypothesis. Herbivores and detritivores play appreciable roles in resource cycling through their participation in plant tissue removal, digestion and decomposition. Detritivores directly decompose dead plant tissues and reincorporate water, nutrients and simple organic compounds into the soil, whereas herbivores consume plant tissues before they wilt and die. Therefore, compared with detritivores, herbivores accelerate resource turnover rates (Kitchell et al., 1979; Detling, 1998; Belovsky and Slade, 2000; Metcalfe et al., 2014), and may further facilitate seed germination and seedling survival by excreting seeds and propagules in dung, which is rich in organics, nutrients and moisture (Miller, 1995; Traveset, 1998; Sanchez de la Vega and Godinez-Alvarez, 2010; and see also a hypothesis regarding silicon by Katz, 2015).

During the Carboniferous, the scarcity of large terrestrial herbivorous tetrapods (Sues and Reisz, 1998; Labandeira, 2006, 2007; Sahney et al., 2010b; Pearson et al., 2013) meant that resources in plant tissues were probably not recycled before plant death. Therefore, resource cycling was controlled by detritivores, although even lignin-decomposing fungi had just started to evolve (Floudas et al., 2012). Hence, large quantities of organic matter may have not been decomposed but buried in sediments for millions of years (Robinson, 1990; Berner, 1998, 2003), depleting the atmosphere of carbon and the soils of nutrients. Ecosystem engineering by animals was also minor compared with later periods (Erwin, 2008; Butterfield, 2011). All these point to slow resource cycling and gradual resource depletion, which inhibited facilitation as a driver of evolutionary diversification during the Carboniferous.

The Mesozoic/Cainozoic combination of herbivory as well as fast and efficient resource utilization and cycling offered early angiosperms conditions that Carboniferous early seed plants could not have been exposed to. This meant that early angiosperms could not only utilize resources better than their forerunners, but also had access to more resources and evolved in a more complex ecological setting that provided more room and stimuli to specialize and diversify. Pteridophytes and gymnosperms were left behind, perhaps also because they are more phylogenetically conservative than angiosperms (Crisp and Cook, 2011), and were pushed mostly to habitats where angiosperms cannot outcompete them (Bond, 1989).

Extending Berendse and Scheffer’s (2009) hypothesis to explain diversification is feasible, as shown above, but requires firmer theoretical and evidential foundations. The effects of biotic conditions and resource availability on extant plant community composition and diversity are inconclusive (Maestre et al., 2005, 2009). The effects of large-scale resource availability and cycling on terrestrial biodiversity over geological time scales also lag behind evidence from marine biomes, and require further supporting evidence (Martin, 2003). Differences in effect sizes of herbivores and detritivores on resource cycling and availability and how they evolved over geological time are also not coherently known. Further studies of long term effects of resource availability and cycling on plant community composition and functional diversity, and especially explaining the mechanisms that underlie these effects (e.g. increasing habitat heterogeneity, competition, niche partitioning and specialization), can provide answers to many of these questions.

CONCLUSIONS

Broadening the taxonomic and temporal scopes in which angiosperm origins and early diversification is studied provides us with new insights. The greater species richness of angiosperm compared to other major land plant clades is not unique (Fig. 1), so there are parallels to which we can compare and from which we can gain insight. The rise of angiosperms should be studied as part of the global increase in biodiversity. The 200Myr starting with the gene divergence of the clades leading to extant gymnosperms and angiosperms and ending with the first emergence of angiosperm traits and of angiosperms as a clade is a newly discovered knowledge gap. These 200Myr are likely to hold key clues as to early origins of angiosperms and the traits that define them, as well as to the evolutionary leap from gymnosperm-like angiosperm ancestors to extant angiosperms as we know them. We need to have a more profound understanding of ancient species interactions, functional diversity and ecosystem functioning and how these evolved throughout the Phanerozoic, and especially from one key evolutionary point in time to another. Achieving these goals depends on integration between the ecologies of extant and past plants, ecosystems and biomes.