1.1. Harry Govier Seeley and the division of Dinosauria

Many traditional sources recount that Harry Govier Seeley was the first to subdivide dinosaurs into Saurischia and Ornithischia. However, as he summarised in his 1887 paper, Cope, Huxley and Marsh had already subdivided them into various schemes of orders and suborders (e.g. Huxley Reference Huxley1870). The difference was that Seeley thought that the distinctions among them implied that the animals called dinosaurs did not form a natural group. That is, they had some general features in common, but features of the pelvis, pneumaticity of the vertebrae, and braincase implied that they could not have had a monophyletic origin. This is why Seeley erected Ornithischia and Saurischia and denied the monophyly of dinosaurs.

Seeley, who often commented on presentations at scientific meetings and was consistently a taxonomic splitter, was something of a maverick. He had studied law and, like many of his contemporaries, came to palaeontology as an avocation after hearing Richard Owen's lectures (Seeley Reference Seeley1901). He became an expert on pterosaurs (which he called ornithosaurs) and dicynodonts, but he was generally interested in issues of vertebrate paleontology and the philosophy behind natural science. Critical to the analysis here is that, although Seeley accepted evolution, when it came to classification he was what I have called a casual Quinarian (Padian Reference Padian and Sarjeant1995). Quinarianism was the practice of classifying organisms in groups of five, arranged in a circle of adjacent groups that shared certain features with their neighbors. This system was developed by the English (later Australian émigré) entomologist William Sharpe MacLeay, who popularised it in his tome Horae Entomologicae (Reference MacLeay1821). It was widely popular in England, largely among entomologists, until Darwin's explanation of “affinities” through common descent (Thomson Reference Thomson2009). The system mostly failed in the end because Darwin's (Reference Darwin1859) recasting of classification based on ancestry explained so much (Padian Reference Padian1999), and there was no conceivable empirical reason why the Quinarian system should work. So it is strange to find Seeley (Reference Seeley1892, p. 368, and later in his great popular book on pterosaurs Dragons of the Air, 1901), using Quinarian diagrams to postulate relationships among his “ornithosaurs” and other reptile groups (Fig. 1, from Seeley Reference Seeley1901, p. 190).

Figure 1

Seeley's (Reference Seeley1901) Quinarian diagram of the relationships of Archosauria. “Ornithosauria” was what Seeley called Pterosauria. “Aristosuchia” was an old term of Cope's for what we would loosely call basal Archosauromorpha and other things. Note that Seeley has not only separated Ornithischia and Saurischia; they are not even “osculating” (connecting) groups.

The importance of Seeley's use of Quinarianism speaks to why he was so ready to do away with the concept of Dinosauria. In his Reference Seeley1887 paper (p. 170, my italics), he writes: “the Dinosauria has no existence as a natural group of animals, but includes two distinct types of animal structure with technical characters in common which show their descent from a common ancestry rather than their close affinity.”

What could Seeley have meant by having common ancestry but not close affinity? The term “affinity” was used in the 19th Century in a non-evolutionary (non-transmutationist) sense, that is to say, not implying a genealogical (evolutionary) relationship (e.g. Owen Reference Owen1870). Seeley was saying that what he called Ornithischia and Saurischia shared common ancestry, but were not necessarily each other's closest relatives. This can be seen in the Quinarian diagram reproduced in Figure 1: the two dinosaurian orders that he erected are not placed next to each other. Seeley (Reference Seeley1887) supported this reasoning by analysing characters that he thought gave more weight to relationships than others. For him, the differences in pelvic (pubic) structure, presence or absence of pneumatisation, and configuration of the braincase were more reliable as characters than other structures of the “teeth, mandible, ilium, femur, and the absence or presence of dermal armour” (Seeley Reference Seeley1887, p. 166).

Seeley (Reference Seeley1887, p. 167) felt that, “The characters on which these animals should be classified are, I submit, those which pervade the several parts of the skeleton, and exhibit some diversity among the associated animal types” So his reasons for dividing dinosaurs were based on two major convictions: a Quinarian perspective on classification that was not tree-like but built on a non-evolutionary scheme of groupings into five, and a presumption (common to most systematists until the advent of numerical taxonomy in the 1950s and cladistics in the 1970s) that some characters should be weighted more heavily than others in establishing relationships.

Seeley was silent on the functional and ecological circumstances involved in the rise of dinosaurs and other groups, as indeed were most later 19th Century authors, because there was very little evidence of possible precursors.

1.2. Dinosaur origins: archosaurs, thecodontians, and pseudosuchians in the 19th–20th centuries

As a result of Seeley's (Reference Seeley1887) analysis, the monophyly of Dinosauria was placed in considerable doubt. Although “Dinosauria” was used by scientists and commonly by the public afterward, it was largely disregarded as a formal category for another century (see e.g. Huene Reference Huene1914, Reference Huene1956; Romer Reference Romer1945, Reference Romer1968). This taxonomic limbo is instantiated in illustrations that accompanied some of the most influential works on dinosaurs; namely, the “bubblegrams” typical of the 1940s through comparatively recent times (e.g. Romer Reference Romer1945; Carroll Reference Carroll1990; Fig. 2). Bubblegrams allowed “stem groups,” which were explicitly paraphyletic, to “give rise” to a series of different, more derived groups, without specifying their relationships to each other. So, for example, a bubble of “thecodonts” (sic: “thecodont” is a type of socketed dentition; a “thecodontian” was a type of archosaur [sensu lato] that had this dental configuration but was not a dinosaur, pterosaur, or crocodilian) was broadly viewed as “ancestral” to other archosaurian groups.

Figure 2

Sample bubblegrams to show postulated relationships of some archosaurian groups in the middle decades of the 20th Century: (a) from Romer (Reference Romer1945); (b) from Charig (Reference Charig and Kuhn1976a)

– reproduced by permission of Verlag Dr. Friedrich Pfeil. 

But where did the terms “archosaur” and “thecodont” and even “pseudosuchian” come from? Scholars today are brought up with late 20th-Century versions of these terms, which are mostly based on phylogenetic systematics and, therefore, rooted in diagnostic synapomorphies. But it is not possible to understand the literature of the earlier 20th century on dinosaur origins without knowing something of the historical genesis of these terms, and remembering that synapomorphies were not explicitly recognised or required for diagnosing taxa.

2.1. Alan Charig and the mid-century problem of dinosaur origins

The work of Alan Charig and his colleagues was considerably influential in the thinking of the early 1960s through the 1970s on the problems of dinosaur origins. Charig was Curator of Fossil Amphibians, Reptiles and Birds at the British Museum (Natural History), now the Natural History Museum, in London, from 1961 to 1987 (see Moody & Naish Reference Moody, Naish, Moody, Buffetaut, Naish and Martill2010). Although he published relatively few descriptions of taxa, Charig was a stimulating and critical thinker, and he had decisive views on the origins of dinosaurs, based largely on some field work and anatomical analyses that he had performed on Triassic archosauromorphs from eastern and South Africa in the course of his dissertation work.

However, before citing Charig's views here, several particulars should be recalled about palaeontological consensus views at mid-century: (1) Dinosauria was not regarded as monophyletic; (2) almost no Triassic ornithischians were recognised, even though, at the time, many formations now considered Early Jurassic were considered Late Triassic (Olsen & Galton Reference Olsen and Galton1977); (3) as a result, and because ornithischians were considered so derived morphologically, they were not usually considered part of the “origin of dinosaurs” problem; (4) saurischians were usually divided into Sauropoda, Theropoda and Prosauropoda, the last of which were often considered part of Theropoda (but also as having “given rise” to Sauropoda or classified within Sauropodomorpha: e.g. Romer Reference Romer1945, Colbert Reference Colbert1955); (5) so the problem of the “origin of dinosaurs” usually reduced to the problem of the “origin of Saurischia,” because theropods (which were seen as the least derived saurischian group) were regarded as the most primitive saurischians. To be precise, the origin of dinosaurs was focused on what was then understood to be Coelurosauria: specifically, in the sense of von Huene (Reference Huene1932), who had essentially regarded the large ones as Carnosauria and the small ones as Coelurosauria (in contrast to how Gauthier parsed them cladistically in 1986, although von Huene recognised tyrannosaurs as coelurosaurs). It must be said, however, that this description simplifies a complex picture. Some “prosauropods” were classified as theropods or Sauropodomorpha by some anothers. It was not out of the question to classify the same animal within theropods but to accept that it could have been close to the origin of sauropods, for example.

Clearly, the methods by which phylogenetic relationships were assessed were less prescriptive than those of today. Nevertheless, it seems clear that for most mid-century paleontologists, the origin of dinosaurs was essentially a problem of getting from a generalised “pseudosuchian” (read basal archosaur) to a generalised, small “coelurosaur” (read basal theropod). At the time, Triassic theropods were known from forms such as Coelophysis, which Colbert had described briefly in Reference Colbert1947; Podokeosaurus, known only by that time from a cast of an impression of a partial skeleton on a rock of Triassic (now considered Jurassic) sandstone (the original had been destroyed in a fire; Talbot Reference Talbot1911); some fragmentary taxa from Europe that von Huene (Reference Huene1932) had summarised; and Segisaurus, which Charles Camp (Reference Camp1936) had described from the Navajo Sandstone of Arizona (then thought to be Triassic, but now considered Jurassic), which was often overlooked. The difficulty, again, was that the “pseudosuchians” were generally larger than the “coelurosaurs,” which meant that they were usually quadrupedal, often armoured, and tended to approach crocodilian habits more than dinosaurian ones.

Ornithosuchus, from the Late Triassic of Scotland, was an exception because it seemed to have bipedal tendencies and had short arms, which partly accounted for inferences about its bipedal tendencies. Alick Walker (Reference Walker1964), whose main research focus was the study of the fossil reptiles from the Triassic Lossiemouth Sandstone Formation of northeastern Scotland (often called the “Elgin Sandstones,” which included both Permian and Triassic rocks), produced a painstaking and anatomically thorough monograph on this animal which T.H. Huxley had originally described. Walker, however, thought that Ornithosuchus prefigured not the ancestry of all the dinosaurs, but specifically the Carnosauria, which (in von Huene's (Reference Huene1932) sense) were known only from the Late Jurassic (Allosaurus) through the Cretaceous (Tyrannosaurus). This created a huge complication: how could Carnosauria and Coelurosauria have separate origins if they are all theropods, and how could a Late Triassic animal like Ornithosuchus be the ancestor of a group of large theropods that did not show up until 70 million years later (and where could carnosaurs have hidden for that long)?

Alfred S. Romer followed Walker's (Reference Walker1964) assignment of Ornithosuchus to Saurischia in his third edition of Vertebrate Paleontology (1966), but not without misgivings (Romer Reference Romer1968). This situation requires some context. Romer was the dean of vertebrate paleontology until his sudden death in late 1973. His was the only American textbook on the market except E.C. Olson's Vertebrate Paleozoology (Reference Olson1971), influential in a different way but without Romer's broad success. As a result, Romer was constantly besieged by colleagues to include their findings in his next printing, and he was often berated for not adequately considering their discoveries or points of view (his legendary reply was “we'll get it in the next edition”). It was important for him to be diplomatic, especially in the absence of decisive evidence. In Notes and Comments (1968, pp. 131–132, 136–137), he says that he accepted Ornithosuchus as a saurischian rather than a pseudosuchian “with considerable qualms,” and that it might rather be “an advanced, near carnosaur thecodont.” The flavour of phylogenetic thinking of those days might be had from Romer's statement that “Very probably the saurischians arose in mildly polyphyletic fashion from two or several pseudosuchian forms.” (Reference Romer1968, p. 132) So perhaps it was not out of the question in those days for Carnosauria and Coelurosauria to have had separate origins but to be included in Theropoda.

However, not everyone agreed, and one strong voice to the contrary was Alan Charig's. Charig did not think that the first dinosaurs evolved from small, lightly built, bipedal “pseudosuchians.” He thought that the first dinosaurs were quadrupedal, not bipedal, and he based this on the kinds of animals that he and his colleagues found in the earlier Triassic localities of eastern and South Africa. The South African animals were not well known. Sometimes, there were isolated bones, and sometimes footprints. The bones in eastern Africa were not of dinosaurs, but of animals that shared some features with dinosaurs (although, as currently understood, not to the exclusion of other Triassic archosaurian groups). For his thesis at Cambridge (1956), Charig described some Middle Triassic archosauromorphs from Tanzania, which he informally named “Mandasuchus,” “Teleocrater,” and “Nyasasaurus” (see Appleby et al. Reference Appleby, Charig, Cox, Kermack, Tarlo, Harland, Holland, House, Hughes, Reynolds, Rudwick, Satterthwaite, Tarlo and Willey1967 for some details). Regrettably, he never actually published formal descriptions of these and other animals, although he referred to them in various abstracts and commentaries for some years to come. Charig frequently alluded to studies in preparation as manuscripts in press and this caused substantial confusion.

Moody & Naish (Reference Moody, Naish, Moody, Buffetaut, Naish and Martill2010) provided a nuanced picture of Alan Charig's Weltanschaaung and his contributions to the field over the years. The judgment of his colleagues suggests that Charig, for all his abilities and insights, was a better rhetorician than an empirical scientist, and intensely interested in the rhetoric of scientific arguments. More than one of his colleagues remarked that he should have been a barrister. As Moody and Naish noted, Charig often published papers that were styled a “commentary” or “critical review” or “reasoned approach” or “point of view,” and so on, of the works of others, which seldom introduced new empirical information. But Charig was more than this; in his capacity as Curator at the Natural History Museum and the only dinosaur worker in the UK, he commanded a substantial pulpit and had considerable influence in the field, as well as among prestigious journals for which he was a frequent reviewer.

Given that dinosaurian palaeontology at mid-century was something of a Wild West, it may be understood how Charig could have imputed to his African fossils a critical role in dinosaurian evolution, even though there was little evidence that they were particularly close to dinosaurs. Most of Charig's Middle Triassic Tanzanian archosauromorphs were indeed quadrupedal, but recent research has shown that most of them may not even be crown-group archosaurs (reviewed by Moody & Naish Reference Moody, Naish, Moody, Buffetaut, Naish and Martill2010). In the late 1950s, Charig collaborated with John Attridge and A.W. (“Fuzz”) Crompton in South Africa, who had come across some deposits (considered Triassic at the time) that preserved the bones and footprints of large and early quadrupedal dinosaurs. The three collaborated on an influential paper (Charig et al. Reference Charig, Attridge and Crompton1965), in which they argued that dinosaurs or effectively saurischians (because ornithischians were considered an afterthought, having almost no Triassic representatives) originated from large quadrupedal forms that only secondarily attained bipedal gait in some lineages. Complications arose, however, when Charig concluded that his eastern African Manda archosaurians were related to the South African prosauropods.

Charig et al. (Reference Charig, Attridge and Crompton1965) began with a re-assessment of melanorosaurid (“prosauropod”) material in the South African Museum. They observed that most inferences of previous authors had been based on isolated bones or on collections of bones that had been mistakenly put together and so were useless for taxonomic or proportional studies. They pointed instead to newer studies of material from southern Africa in the “Passage Beds” between the Molteno and “typical Red Beds” (now mostly considered of Jurassic age), saying: “This [material] comprises several incomplete skeletons of a large saurischian (the ‘Maphutseng dinosaur’), the articulated hind-limb of a second large saurischian (the ‘Blikana dinosaur’), footprints ascribed to both of these, and the footprints of yet a third large saurischian (the ‘Soebeng trackways’).” These remains largely comprised the basis of their hypothesis of dinosaurian origins.

Charig and his colleagues never actually described any of this material. The “Maphutseng dinosaur” was later informally given the name Thotobolosaurus by Ellenberger (Reference Ellenberger1970); it has never been described scientifically and appears to be based on a collection of poorly preserved bones found in association with a trash heap in Lesotho, probably pertaining to a prosauropod. The “Blikana dinosaur” was eventually named Blikanosaurus by Galton & Van Heerden (Reference Galton and Van Heerden1985), and it is a basal sauropod. The “Soebeng trackways” have never been described, although Thulborn (Reference Thulborn2006) and others have regarded them as prosauropod tracks. So the argument of Charig et al. (Reference Charig, Attridge and Crompton1965) reduces to the contention that, although these early sauropodomorphs are variously quadrupedal or bipedal (or both facultatively), their original condition was quadrupedal, based on the Manda archosaurians and other proterosuchians; yet there was no real evidence for this at the time, and it is generally understood today that the first sauropodomorphs were bipedal. Romer (Reference Romer1968, p. 137) remarked that this paper influenced him in writing his 1966 textbook, concluding that none of the large, mainly bipedal dinosaurs of the Late Triassic have any relationship to later theropods but are related to sauropod ancestry. In contrast, the more quadrupedal melanorosaurids of the later Triassic seem to be closest to the great quadrupedal sauropods. In this sense Charig et al. (Reference Charig, Attridge and Crompton1965) did the profession a service in removing Prosauropoda from Theropoda and reaffirming that it should be united with Sauropoda as von Huene (Reference Huene1932) had proposed. However, it was never broadly accepted that dinosaurs evolved from large quadrupeds, nor that bipedality was secondary in the group. To the contrary, the idea persisted that dinosaurs must have evolved from small, bipedal, lightly built “pseudosuchians.” But just how this must have happened remained a mystery.

Charig (Reference Charig, Joysey and Kemp1972), in a second very influential paper, complicated the issue further, based largely on his studies (that were never published) of the Middle Triassic archosauromorphs from eastern Africa that he studied for his dissertation (Charig Reference Charig1956). He was convinced that forms such as “Mandasuchus” were related to dinosaurs, but that they were neither sprawling in their posture nor erect and columnar. He essentially invented a posture intermediate between a sprawling and upright gait that he called “semi-improved.” Others had used the terms “semi-improved” or “semi-erect” (see e.g. Bakker Reference Bakker and Hallam1977, Reference Bakker, Thomas and Olson1980), but Charig's deliberate use of it as an intermediate stage in locomotor evolution was original. His choice of terms effectively assumed an evolutionary progression of locomotor function among grades of Triassic archosaurs.

In his 1972 paper, Charig maintained that “Thecodontia were certainly incapable of adopting a ‘vertical’ limb posture,” but that beyond the Chasmatosaurus (Proterosuchian) grade of organisation, the “thecodontians” (i.e. archosauriforms and basal archosaurs) were “semi-improved” and could modify their stance to a limited extent. In this respect, Charig regarded them as comparable to crocodylians. However, Charig interpreted crocodylian stance and gait in an unusual way. Rather than adopting a “semi-erect” stance, crocodylians today tend either to sprawl or to walk with a virtually erect stance and parasagittal gait in their hindlimbs (Charig quoted Bobb Schaeffer (Reference Schaeffer1941), accurately on this point), although the forelimbs and the hindlimbs have different arcs (see e.g. Brinkman Reference Brinkman1980; Padian & Olsen Reference Padian and Olsen1984). There was at the time (and still is) no evidence from living crocodiles of a stance and gait that is evolutionarily intermediate between an all-out “sprawling” condition of most lizards and a fully “upright” condition seen in today's birds: Bakker (Reference Bakker and Hallam1977, Reference Bakker, Thomas and Olson1980) appears to have followed Charig in this scenario. It is important to distinguish between terms used to describe stance and those used to describe gait, because conflicting combinations of stances and gaits are possible (Padian et al. Reference Padian, Li and Pchelnikova2010). What an animal is doing with its limbs (in a neuromuscular sense) to actuate its muscles to adopt a particular limb posture or gait is again a wholly different question, experimentally outside the realm of paleontology in almost all cases. Padian et al. (Reference Padian, Li and Pchelnikova2010) also showed that, if the Triassic trackway Apatopus is indeed attributable to a phytosaur as historically asserted, the ability to draw the limbs under the body and walk parasagittally was common to crown-group Archosauria at least, and perhaps even more basal, depending on whether Phytosauria are basal Pseudosuchia or outside Archosauria.

Charig based his view that Thecodontia could not adopt a vertical limb posture on the structure of the pelvis and hindlimb, compared to those of crocodiles and birds. This was a reasonable view at the time, but it was soon contested. Re-analyses of extinct archosaurian hindlimb morphology, plus the ichnological evidence of narrow-gauge trackways, suggested that “Thecodontia” could indeed adopt a “vertical” limb posture in the sense of an upright stance and parasagittal gait (see Parrish Reference Parrish1986). As Bonaparte (Reference Bonaparte1984) showed, they did this in two different ways: the “crocodile-normal” forms by producing a lateral overhang of the ilium beneath which the femur could swing laterally (abduct and adduct) to adopt different poses; and the “crocodile-reversed” forms by evolving a well-offset femoral head that fit snugly into the (usually perforated) acetabulum but could not adopt any other gait than parasagittal.

2.2. Ankles and ankles

The change in flexion of the ankle from crurotarsal to mesotarsal was critical to explaining the origin of dinosaurian stance and gait in the later 20th century, but the transition had long stymied palaeontologists. The problem was that, in the absence of an independently derived phylogeny for dinosaurs and other archosaurs, no pattern of ankle evolution was independently supported, and so process explanations of functional shifts were the only other available approach. The Swiss palaeontologist Bernard Krebs approached this problem in the early 1960s, with important new evidence. Krebs, a student of the influential European diapsid palaeontologist Emil Kuhn-Schnyder, was describing an animal called Ticinosuchus, from the Middle Triassic of the San Giorgio region of the foothills of the Alps (Krebs Reference Krebs1963, Reference Krebs, Kuhn-Schnyder and Peyer1965). He referred to it, as was customary in those times, as a “rauisuchian,” although the term was coming to be considered something of a waste basket. Krebs realised that his animal was crurotarsal; that is, it flexed its ankle between the astragalus and calcaneum, so that the astragalus was linked to the tibia and the calcaneum to the distal tarsals and metatarsus, whereas dinosaurs flexed their ankles mesotarsally: the tibia, astragalus, fibula, and calcaneum were bound in a functional unit, and the distal tarsals and metatarsus flexed in a straight plane against these (Fig. 3).

Figure 3

A latter 20th-Century view of archosaurian ankle structure, from Thulborn (Reference Thulborn1982). The views in (a–c) are anterior of the left ankle, with functional configurations according to Thulborn: (a) mesotarsal ankle, in which the proximal tarsals (AS+CA) flex with the tibia and fibula against the distal tarsals and foot (dinosaurs, pterosaur, and other ornithodirans); (b) the “duplex” ankle (generally thought primitive), in which the calcaneum can move with either the upper leg or foot (proterosuchians and perhaps other archosauromorphs); (c) a crurotarsal ankle, in which the calcaneum and the foot rotate against the astragalus and upper leg (crocodiles and other pseudosuchians); (d) diagram of the “crocodile-normal” ankle (crocodiles and other pseudosuchians), in which the “peg” of the archetypal “peg-and-socket” joint is on the astragalus and the “socket” is on the calcaneum; (e) the “crocodile-reversed” ankle (ornithodirans) in which the peg is on the calcaneum. Abbreviations: AS=astragalus; CA=calcaneum; F=fibula; T=tibia; 1–5=numbered metatarsals.

Reprinted by permission from Macmillan Publishers Ltd: Nature 299, 657 (14 October 1982) © 1982 www.nature.com. 

How could this transition from crurotarsal to mesotarsal ankle have been accomplished? Krebs (Reference Krebs, Kuhn-Schnyder and Peyer1965) thought that it was not possible. He reasoned that dinosaurs could not have evolved from any known crurotarsal “pseudosuchians.” Their ancestry would have to be sought farther back, in the earliest Triassic or even the latest Permian, among archosaurs with more generalised ankle types, such as proterosuchians or erythrosuchians (e.g. Euparkeria). This idea was generally accepted in those years for a variety of reasons: stratigraphically, dinosaurs and pseudosuchians were considered contemporaneous, so one could not have “given rise” to the other by the reasoning of those times; moreover, each types was specialised. In this sense, the perceived inability for a functional change to occur between two configurations found in two different groups was allowed to force back common ancestry to a more remote time.

Ironically, it may have been Charig (Reference Charig, Joysey and Kemp1972, pp. 150–151) who first suggested an “alternative solution” to the evolution of the dinosaurian ankle other than his favoured idea of a direct evolution of large quadrupedal saurischian dinosaurs from large quadrupedal “thecodontians.” Charig envisioned (almost as a thought experiment) the evolution of digitigrady in the context of erect posture and parasagittal gait, as a way to make locomotion more efficient. Digitigrady would obviate the need for the calcaneal tuber, which actuated the lateral flexion of the crocodile-normal ankle through the astragalo–calcaneal joint; the tuber could therefore be reduced. The elongation of the metatarsus could have followed, to extend the stride. Charig effectively advanced the possibility (which he did not accept) that, to evolve the dinosaurian condition, first the ankle joint would have had to be “stiffened” – that is, the flexion between the astragalus and calcaneum would have constricted, and parasagittal motion would have become largely limited to the interphalangeal and metatarso–phalangeal joints. Then, after the digitigrade stance and parasagittal gait were fully evolved, flexion could have returned between the proximal and distal tarsals, thus creating a mesotarsal ankle.

Charig's “alternative scenario” was not picked up as the most likely hypothesis to solve the problem of dinosaurian origins in the functional-ecological sense, although elements of what he described were already commonly accepted, such as the fact that dinosaurs had upright posture and parasagittal gait, and that their ankle type must have evolved from some kind of more basal ankle type.

However, at nearly the same time that Charig's paper appeared, the dinosaurian world was rocked by the publication of Romer's preliminary descriptions (Reference Romer1971, Reference Romer1972) of the South American Triassic “pseudosuchians” Lagosuchus and Lagerpeton. These small, lightly built animals were not dinosaurs, but were very close to them, as José Bonaparte showed in his Reference Bonaparte1975 monograph on Lagosuchus. And they had mesotarsal ankles. These discoveries showed, as Bonaparte realised, that dinosaurs did not evolve their stance and gait at large size, but at very small size. The problem of the origin of the dinosaurian ankle, with its erect stance and parasagittal gait, was effectively solved. Bernard Krebs appeared to be vindicated. The importance of the discovery of these animals for the understanding of dinosaurian origins cannot be overestimated.

And yet, controversy continued over the details. Because “pseudosuchians” were such a mixed group of animals with no clear phylogenetic lines, arguments about ankle evolution persisted. It was still not clear from which “pseudosuchians” dinosaurs and lagosuchids had evolved, and so the origin of the mesotarsal ankle was still in doubt. Arthur Cruickshank (Reference Cruickshank1979) first laid out the contrast between the “crocodile-normal” and “crocodile-reversed” ankle, the first common to most “pseudosuchians” and the latter common to most dinosaurs, ornithosuchids, and other forms such as Euparkeria. In the former configuration the “peg-and-socket” joint had the peg on the astragalus and the socket on the calcaneum, such that the calcaneum, its heel, the distal tarsals, and the foot rotated as a functional unit through this joint against this astragalus and the upper leg. In the latter configuration, the peg was on the calcaneum but there was relatively little rotation in the ankle. Cruickshank's view was that both ankle types arose from a more generalised (and more mesotarsal) proterosuchian-type ankle, probably in the earliest Triassic or earlier. For Cruickshank, the “prosauropod” ankle type was derived (to the exclusion of the ankles of other dinosaurs) from a crocodile-normal configuration, a view that never gained much currency. He also accepted (following Walker Reference Walker1964) that “carnosaurs” evolved from ornithosuchians, and so he saw a polyphyletic origin of the “advanced mesotarsal” ankle of various dinosaurian groups. Most people agreed with Krebs (Reference Krebs1963) that the mesotarsal ankle must have evolved from something much older and more generalised, such as a proterosuchian or erythrosuchian ankle (e.g. Euparkeria).

Chatterjee (Reference Chatterjee1982) followed Cruickshank in some respects of deriving various archosaurian ankle types from others, including suggesting a phylogenetic arrangement of groups based on ankle structure. He recognised that all dinosaurs shared the same type of “advanced mesotarsal” joint (as compared to the “primitive mesotarsal” joint of proterosuchians, which is too generalised to be so named). He also recognised the importance of Lagosuchus, a “pseudosuchian,” sharing the same dinosaurian configuration and to acknowledge this properly, he proposed the erection of a suborder Lagosuchia within the Order Thecodontia. Although it was still not clear how these types could have transmuted functionally, the pattern of evolution was becoming clearer. There were still quibbles about “advanced mesotarsal-normal” and “advanced mesotarsal-reversed” dinosaurian configurations, and about just which configurations were present in certain ambiguous fossils, but the general outlines of ankle evolution were established. As far as they went, they showed that the mesotarsal ankle was correlated with longer and more gracile hindlimbs, an offset femoral head, a tibia longer than the fibula, elongated metatarsals, upright stance, and parasagittal gait (e.g. Charig Reference Charig, Joysey and Kemp1972).

Oddly enough, even long after descriptions by Romer (Reference Romer1971, Reference Romer1972) and Bonaparte (Reference Bonaparte1975) of the “rabbit thecodonts” Lagosuchus and Lagerpeton, and Bonaparte's demonstration of their importance to dinosaurian origins, Charig (Reference Charig1980) continued to prefer scenarios based on only large quadrupedal forms, with only secondary reversions to bipedality, and he persisted in his progressivist view of “sprawling” to “semi-improved” to “fully improved” archosaurs. But it was generally felt that there was little obstacle to deriving a “crocodile-normal” ankle from any other kind; after all, it had clearly happened. This vindicated what the late Don Baird used to call “The Harvard Law of the Improbable: it happened, therefore it was possible.”

3. The “Dinosaur Renaissance” of the 1970s and the problem of dinosaur origins

4.1. Stratigraphic realignment and the discovery of an Early Jurassic terrestrial record

The role of biostratigraphy in assessing the timing of faunal succession and the origin of dinosaurs has a long and important history. “Romer's Gap” is the name commonly given to the absence of terrestrial vertebrate fossils between the advent onto land in the Devonian and the radiation of tetrapods in the Carboniferous. But Romer (Reference Romer1968) bemoaned a second gap, the absence of fossils between the first dinosaurs of the Late Triassic and the fully developed dinosaurian ecosystems of the Late Jurassic (complete with allosaurs, coelurosaurs, stegosaurs, huge sauropods, and various ornithopods). Briefly, a contributing factor was that the geological timescale was established on European deposits. In the Early Jurassic, epeiric seas covered most of Europe, so there was little idea of what an Early Jurassic European terrestrial vertebrate fauna should look like. Triassic vertebrate faunas of phytosaurs, aetosaurs, prosauropods, and some theropod remains were known from Europe, and these plus associated footprint faunas created a diverse picture of life in the Triassic. These faunas were correlated worldwide with formations in North America, South America, Africa, China and India.

However, studies of European deposits suggested that “thecodontians” did not survive the Triassic. As a result, horizons worldwide were correlated by a simple metric: if “thecodontians” were present, it was Triassic; if they were absent, it was Jurassic. Absence of “thecodontians” was not necessarily definitive, and it was possible, for example, for Welles (Reference Welles1954) to make the argument that the Kayenta Formation of Arizona was of Jurassic age because the hindlimb of the theropod Dilophosaurus was of “Jurassic grade.” To this argument, Colbert (Reference Colbert1981) countered that aetosaurs were known from the Kayenta, and therefore it was of Triassic age. Padian (Reference Padian1989) showed that the scutes on which the aetosaur identification was based actually belonged to Scelidosaurus, a thyreophoran dinosaur known from the Sinemurian–Pliensbachian (Early Jurassic) of England. This removed evidence that the Kayenta Formation was of Triassic age, and tended to lend support to a Jurassic age. But this was only a minor piece of the whole puzzle.

In the 1970s, a group of researchers independently studying the palaeontology of the Newark Group of the eastern seaboard of North America came across a conundrum. Exposures of the Newark Group extended from Nova Scotia to North Carolina and represented a series of rift lake basins formed by the split of North America and Africa beginning in the Middle Triassic or even earlier. Among many others, Nick McDonald and Bobb Schaeffer were studying the fishes; Bruce Cornet tackled the palynology; Paul Olsen, Don Baird, and Peter Galton worked on the skeletal and ichnological fossil tetrapods. After most of a decade of work, the results were clear. Although the Newark Group had been traditionally considered entirely Triassic, approximately half of it was situated firmly in the Jurassic. The basis of this work was lithostratigraphic correlation, but the fossils told an impressive story. Not all basins contained the same fossil floras and faunas; and stratigraphically, within basins, the associations of plants and animals correlated well with each other. According to Cornet's studies, the Jurassic was marked by: the loss of striate Classipollis pollen; the loss of “thecodontians”; the evolution of larger theropods and sauropodomorphs; and the advent of ornithischians.

Through a series of papers, Paul Olsen and his colleagues (e.g. Olsen & Galton Reference Olsen and Galton1977; Olsen et al. Reference Olsen, Shubin and Anders1987; Olsen & Sues Reference Olsen, Sues and Padian1986) characterised a distinct set of terrestrial faunas and floras across the Triassic-Jurassic boundary in the Newark Group (now renamed the Newark Supergroup), and revised the dating of geological formations worldwide to reflect these biostratigraphic correlations. Following this revision, about half of the horizons previously considered Late Triassic were placed in the Early Jurassic. Romer's second gap was filled – at least through the Early Jurassic. To some observers, the loss of “thecodontians” in the latest Triassic was both close to the Triassic–Jurassic boundary and relatively abrupt (e.g. Olsen et al. Reference Olsen, Shubin and Anders1987), even though the faunal evidence was limited both in specimens and in geographic extent.

What did these advances mean for the question of dinosaur origins? At the very least, the stratigraphic realignment appeared to both order and constrain the first and last appearances of dinosaurs and their relatives during the Late Triassic and Early Jurassic. “Thecodontians” did not persist into the Jurassic; almost no ornithischians were known from the Triassic (with almost the sole exception of the poorly preserved and questionable Pisanosaurus from Argentina, and a couple of fragments from elsewhere); and Triassic theropods and basal sauropodomorphs were unevenly represented among the continents (“prosauropods” were absent in North America, theropods rare in Europe and southern Africa, for example). By the Early Jurassic, all three major lineages of dinosaurs were found worldwide.

The Triassic picture of faunal succession became clearer. Research by Romer's Harvard field crews and later by José Bonaparte and his colleagues in Argentina had discovered faunas of terrestrial tetrapods ranging from what was reckoned the Middle Triassic up through the latest Triassic (Bonaparte Reference Bonaparte1982). The earliest of these, from the Middle Triassic, were dominated by various therapsids but (among other reptiles) Lagosuchus and Lagerpeton were also present (Chañares Formation). The last two dinosaur relatives were absent from the early Late Triassic assemblage (Ischigualasto Formation), but the basal dinosaurs (or very close thereto) Herrerasaurus and Ischisaurus were present, as was Pisanosaurus, and Staurikosaurus, usually considered close to Herrerasaurus, is present in the roughly coeval deposits of Brazil. By the latest Triassic various basal sauropodomorphs were found, but no ornithischians and no theropods. Because South America was the only continent on which dinosaur relatives such as Lagosuchus (later named Marasuchus by Sereno & Arcucci [Reference Sereno and Arcucci1994]) and Lagerpeton (and herrerasaurids) had been found, it was presumed that these animals were quickly replaced by true dinosaurs in the latest Triassic. However, the South American fossil record, did not exactly say that because true dinosaurs were not well represented in the latest Triassic, apart from sauropodomorphs.

4.2. Recent discoveries and yet another paradigm shift

Since the 1970s and 1980s, when the paradigm shift just described provided us with our Early Jurassic record, palaeontological exploration of Triassic–Jurassic sediments has not been idle (Fraser & Sues Reference Fraser and Sues1994; Fraser Reference Fraser2006; Sues & Fraser Reference Sues and Fraser2010). And some very important new works have re-analysed and synthesised our knowledge of the Triassic–Jurassic record of dinosaurs and their relatives (e.g. Upchurch et al. Reference Upchurch, Hunn and Norman2002; Irmis et al. Reference Irmis, Nesbitt, Padian, Smith, Turner, Woody and Downs2007a, Reference Irmis, Parker and Nesbittb; Nesbitt et al. Reference Nesbitt, Irmis and Parker2007; Brusatte et al. Reference Brusatte, Nesbitt, Irmis, Butler, Benton and Norell2010; Langer et al. Reference Langer, Ezcurra, Bittencourt and Novas2010).

From South America came a variety of new dinosaurs, including the basal sauropodomorphs Saturnalia, Panphagia, Chromogisaurus, and the theropods Guibasaurus and Zupaysaurus (reviewed in Irmis Reference Irmis2011), but no ornithischians except a possible heterodontosaurid jaw fragment from Patagonia (Báez & Marsicano Reference Báez and Marsicano2001). More sauropodomorphs have been found in the South African Late Triassic, but no theropods, and the ornithischian Eocursor (Butler et al. Reference Butler, Smith and Norman2007) comes from what is recognised as the latest Triassic beds (although there are suggestions that it could equally be from the Early Jurassic). Ornithischians and theropods begin to radiate in the Early Jurassic. In western North America, there are no ornithischians or sauropodomorphs in the Late Triassic, but the pattern differs in eastern North America (where footprint records suggest their presence, somewhat questionably) and Europe (Nesbitt et al. Reference Nesbitt, Irmis and Parker2007). These patterns appear in the time of Pangaea, when all the continents are together and physical barriers to dispersal are more difficult to establish. As Irmis (Reference Irmis2011) noted, no single hypothesis appears to explain the distribution of the various dinosaur groups worldwide in the Late Triassic; sometimes they are there, and sometimes they aren't. But as a general pattern, ornithischians are virtually absent from the Late Triassic record. Revueltosaurus, known only from teeth, was widely considered a Triassic ornithischian for many years, but Parker et al. (Reference Parker, Irmis, Nesbitt, Martz and Browne2005) showed that these teeth actually belonged to a pseudosuchian. And, given the poor preservation and incompleteness of Pisanosaurus and the heterodontosaurid jaw described by Báez and Marsicano (Reference Báez and Marsicano2001), and the absence of certified Triassic ornithischian footprints, Eocursor may be the earliest valid ornithischian.

So, stratigraphically, we are left with almost no ornithischians worldwide in the Triassic, and a disjunct distribution of sauropodomorphs or theropods (occasionally both) among various continents at the same time. This stratigraphic disjunction would be odd if ornithischians and saurischians are sister groups, because we would expect them to be equally ancient in their representation in the fossil record, absent circumstances of preservation such as the often-invoked but untested “upland vs. lowland” differences among taxa. But theropods and sauropodomorphs are also sister taxa, and they do not often co-occur in the Late Triassic.

In a recent paper, Rowe et al. (Reference Rowe, Sues and Reisz2011) proposed that Early Jurassic “prosauropods” of North America have been inappropriately lumped taxonomically, and that the three currently known North American forms evidently do not form a monophyletic group to the exclusion of other basal sauropodomorphs on other continents. These authors conclude that the North American “prosauropod” fauna evolved from three or four independent sources on other continents. If this pattern is confirmed (and it obtains even using two quite different phylogenetic analyses of other authors), it will be interesting to see whether other taxa follow a similar pattern.

6.1. Abiotic factors in extinction

The loss of taxa at or near the Triassic–Jurassic boundary (Olsen & Sues Reference Olsen, Sues and Padian1986) has been considered relatively rapid (Olsen et al. Reference Olsen, Kent, Sues, Koeberl, Huber, Montanari, Rainforth, Fowell, Szajna and Hartline2002, Reference Olsen, Kent and Whiteside2011), and perhaps associated with both volcanic activity and the Manicouagan impact in eastern Canada, although the latter hypothesis has been generally discounted because the events are not temporally coincident. Moreover the hypothesis that the groups that became extinct by the end of the Triassic persisted until very close to the Triassic–Jurassic boundary needs further substantiation. The Cretaceous–Tertiary boundary is the only major extinction event known to be associated with a bolide impact, but it is also associated with large-scale volcanism; in fact three of the five largest extinctions are (Archibald et al. Reference Archibald, Clemens, Padian, Rowe, Macleod, Barrett, Gale, Holroyd, Sues, Arens, Horner, Wilson, Goodwin, Brochu, Lofgren, Hurlbert, Hartman, Eberth, Wignall, Currie, Weil, Prasad, Dingus, Courtillot, Milner, Milner, Baipai, Ward and Sahni2010), but there are many instances of large impacts with no evident biotic effect (Shoemaker Reference Shoemaker1983). Furthermore, it is difficult to explain why the effects of a bolide impact would have favoured dinosaurs at the Triassic–Jurassic boundary but disfavoured them at the Cretaceous–Tertiary boundary (Archibald Reference Archibald2011).

6.2. Competition

Competition among amniote groups has long been a popular hypothesis to explain the Triassic–Jurassic land vertebrate extinctions. But who is competing with whom: is it (1) archosaurs with therapsids throughout the Triassic (e.g. Bakker Reference Bakker and Hallam1977); (2) dinosaurs with therapsids in the Middle to Late Triassic (e.g. Bonaparte Reference Bonaparte1982; Charig Reference Charig and Ferguson1984); or (3) dinosaurs and their closest relatives with pseudosuchians and other reptiles in the Late Triassic (e.g. Charig Reference Charig and Ferguson1984)? Only the latter two questions involve dinosaurs. The second question is problematic because, compared to their Late Permian and Early Triassic diversity, therapsids had waned considerably before the dinosaurs diversified extensively (Bonaparte Reference Bonaparte1982; Charig Reference Charig and Ferguson1984). The third question was initially answered by Bonaparte's (Reference Bonaparte1982) stratigraphic scheme showing that dinosaurs (notably prosauropods and the dinosauriform herrerasaurids) replaced lagosuchids and related forms in South America without taxonomic overlap; however, as noted above, this pattern did not hold in lower-latitude faunas (Irmis et al. Reference Irmis, Nesbitt, Padian, Smith, Turner, Woody and Downs2007a).

In any event, it is difficult to substantiate the hypothesis that extinct groups were directly competing. Benton (Reference Benton, Jablonski, Erwin and Lipps1996) surveyed several hundred cases in which competition in the fossil record had been hypothesised or presumed, and could find plausibility or even possibility in only a handful of them. It is difficult to establish whether hypotheses of competition are even testable in the fossil record.

6.3. Felicity

The “lucky break” hypothesis – the idea that other organisms vacated ecological space in becoming extinct, and the dinosaurs eventually just moved in – is based on an observation of non-overlapping geological ranges and the presumption that dinosaurs occupied the same resource zones or adaptive zones (e.g. Brusatte et al. Reference Brusatte, Benton, Ruta and Lloyd2008a, Reference Brusatte, Benton, Ruta and Lloydb). Ideas about resource zones or adaptive zones can only be examined very generally, because trophic webs are difficult to establish from direct evidence. And, the observation that one group of organisms became extinct and another radiated in its wake is not really a hypothesis. If the idea is that there is “nothing to explain” in terms of competition or differential success under environmental stress, then it is a classic “null hypothesis” in historical biology (e.g. Raup Reference Raup1991). The hypothesis that there is “nothing to explain” because there is nothing unusual, connected, non-random, or requiring causal explanation in the macroevolutionary pattern is viable, and it can only be rejected by the demonstration of a correlative pattern of differences among the groups affected, which suggests a causal explanation that then can be tested independently.

Because early dinosaurs co-existed with their close relatives for so long at lower latitudes, it is difficult to accept that they got a “lucky break” when other taxa became extinct, particularly because dinosaurs did not demonstrably radiate into many adaptive zones of their erstwhile contemporaries. During the Late Triassic dinosaurs were a minor component of the ecosystem, functioning as small carnivores (theropods) and omnivores to generalised herbivores (sauropodomorphs). Early Jurassic theropods and sauropods included larger forms (Dilophosaurus, Vulcanodon) that do not show departures from the habits of their Late Triassic relatives. Early Jurassic ornithischians (lesothosaurids and thyreophorans such as Scutellosaurus and Scelidosaurus), virtually unknown in the Late Triassic, show leaf-shaped teeth quite similar to those of basal sauropodomorphs (and the Triassic crocodile-line revueltosaurs); the only new invention comes in the molariform battery of cheek teeth in heterodontosaurids. However, the specialised feeding structures of trilophosaurs, rhynchosaurs, phytosaurs, and many other groups were not duplicated by Early Jurassic dinosaurs, so it is difficult to argue that it was necessary for other groups to vacate the territory in order that dinosaurs could move in.

6.4. Superiority

In a non-competitive sense, this means that dinosaurs evolved adaptations that are not found in their contemporaries and that enabled them to thrive differentially in the world of the Late Triassic and beyond. Romer (Reference Romer1968) noted many of these, such as elongated hindlimb elements, light build, erect stance, parasagittal gait, offset femur, “cursorial” proportions (e.g. tibia longer than femur, and elongated metatarsals), mesotarsal ankle, and so on. Bakker (Reference Bakker and Hallam1977, Reference Bakker, Thomas and Olson1980) provided a litany of arguments to support the superiority of dinosaurs. The features that Romer and Bakker listed indeed differentiate dinosaurs (and other ornithodirans) from their contemporaries. Bakker, perhaps more than Ostrom, was convinced that dinosaurs were endothermic, although terms such as endothermic, “warm-blooded,” homeothermic, and tachymetabolic were often conflated (Bennett & Ruben Reference Bennett and Ruben1979).

Eventually, comprehensive surveys of the ontogeny and phylogeny of dinosaur long bone histology established that dinosaurs and pterosaurs grew considerably more quickly than other reptiles (Horner et al. Reference Horner, de Ricqlès and Padian1999, Reference Horner, de Ricqlès and Padian2000; Sander Reference Sander1999; Ricqlès et al. Reference Ricqlès, Padian, Horner and Francillon-Viellot2000, Reference Ricqlès, Padian and Horner2003, Reference Ricqlès, Padian, Knoll and Horner2008; Padian et al. Reference Padian, de Ricqlès and Horner2001, Reference Padian, Horner and de Ricqlès2004). Their bones rarely showed the periodic “annuli” of poorly vascularised bone that characterised many other reptiles; in fact, Mesozoic dinosaur bone is as well vascularised as the bones of most large birds and mammals, suggesting comparable growth rates (Köhler et al. Reference Köhler, Marín-Moratalla, Jordana and Aanes2012; Padian Reference Padian2012). It is difficult to explain these sustained growth rates without underlying high metabolic rates, and so it is difficult to escape the conclusion that dinosaurs and pterosaurs were relatively tachymetabolic (Padian & Horner Reference Padian and Horner2002). High growth and metabolic rates imply a more active feeding style, faster growth to a size that escapes smaller predators, and likely the earlier onset of reproductive maturity, all of which are related to size. Unlike virtually all other Triassic reptiles, ornithodirans (dinosaurs, pterosaurs, and their common relatives) shared a suite of characters related to habitual upright posture, parasagittal gait, bipedality, and cursorial proportions, as well as high growth and metabolic rates (Romer Reference Romer1968; Padian et al. Reference Padian, de Ricqlès and Horner2001; Padian Reference Padian2008). The two groups of tetrapods today that share these features are birds and mammals (although most mammals are not bipedal), and they are also the most numerically diverse groups of tetrapods.

Superiority can only be inferred as a post hoc metric of success. In that sense, post- Triassic dinosaurs did better than their Triassic contemporaries, and they were successful in a variety of forms and roles for another 135 million years (and longer in the case of birds). In the post-Triassic world they radiated into a variety of forms that matched or exceeded previous reptilian groups in diversity and disparity. However, in the Late Triassic they were less conspicuous. In any event, although the features of growth, posture, and physiology that distinguish dinosaurs and their ornithodiran relatives from their Late Triassic contemporaries, and perhaps contributed substantially to their later success, became much more obvious after the Triassic, those features evolved during the Triassic.

7.1. Filling the evolutionary gaps

It is clear, then, that we have to re-think our conception of what really comprises the “beginning of the Age of Dinosaurs.” Several other discoveries of the past decades have changed our view of this.

(a) The first piece of evidence to shift thinking was the discovery of Lagosuchus and Lagerpeton in the Middle Triassic of Argentina (Romer Reference Romer1971, Reference Romer1972; Bonaparte Reference Bonaparte1975). In his monograph on Lagosuchus, Bonaparte took pains to show what parts of the skull and skeleton were “dinosaurian,” “thecodontian,” or a mixture of the two. The distinction was beginning to blur and it was because the characters were truly intermediate. The iliac blade was short as in “thecodontians” but the pubis had a curved, flattened blade that looked dinosaurian; the acetabulum might have been open (or incipiently open, or just damaged). The femoral head was offset but not as much as in most dinosaurs. The astragalus had an ascending process but not exactly like those in dinosaurs. And so on.

(b) Another realisation was that pterosaurs had many of the characters that appeared to distinguish dinosaurs from other reptiles (Gauthier Reference Gauthier1984; Padian Reference Padian, Reif and Westphal1984), particularly in the hindlimbs, although much of the rest of the skeleton was highly derived even in basal forms (Padian Reference Padian1983a, Reference Padianb). This nexus of pterosaur-dinosaur forms, which turned out to include Lagosuchus, Lagerpeton, and some of their relatives, comprised a group that Gauthier (Reference Gauthier1984, Reference Gauthier1986) called Ornithodira, and they were very distinct from other archosaurs. It seemed that the features of posture and locomotion that were thought to characterise dinosaurs were shared by a larger group on the “bird-line archosaur” side of the ledger. These anatomical features also conveyed functional changes in the whole lineage, including upright posture, parasagittal gait, and bipedality (Padian Reference Padian1983b, Reference Padian2008).

(c) New discoveries of basal dinosaurs and dinosauriforms such as Panphagia, Saturnalia, Pantydraco, and the silesaurs further blurred the distinction between traditionally known non-dinosaurs (pterosaurs, pseudosuchians) and traditionally known basal dinosaurs such as Coelophysis and Plateosaurus, both saurischians (Irmis Reference Irmis2011).

(d) Studies of the bone histology of basal dinosaurs and their archosaurian relatives (e.g. de Ricqlès et al. Reference Ricqlès, Padian, Horner and Francillon-Viellot2000, Reference Ricqlès, Padian and Horner2003, Reference Ricqlès, Padian, Knoll and Horner2008; Padian et al. Reference Padian, de Ricqlès and Horner2001; Knoll et al. Reference Knoll, Padian and de Ricqlès2010; Werning et al. Reference Werning, Irmis, Smith, Turner and Padian2011) revealed that many microanatomical features linked to rapid growth and elevated metabolic rates evolved before dinosaurs, were shared by pterosaurs, and may have first appeared much earlier in archosaurs.

7.2. The Triassic–Jurassic boundary marks the evolutionary inflection point

One insight that emerges from all the recent work on Triassic–Jurassic terrestrial vertebrate paleontology is that the “Age of Dinosaurs” really does not begin until the Early Jurassic.

(a) Regardless of whether Pisanosaurus is an ornithischian or something else, there are virtually no ornithischians known from the Triassic; nearly all their records have been reassigned for one reason or another (Parker et al. Reference Parker, Irmis, Nesbitt, Martz and Browne2005; Irmis et al. Reference Irmis, Parker and Nesbitt2007b; Nesbitt et al. Reference Nesbitt, Irmis and Parker2007). In the Early Jurassic the ornithischians quickly radiated into fabrosaurids, heterodontosaurids, thyreophorans, and other forms and by the Sinemurian they reached lengths of two meters or more (e.g. Scelidosaurus).

(b) Saurischian dinosaurs were smaller before the Triassic-Jurassic boundary than soon afterwards, which is to say that, notably in theropods, their maximum known size increased from one or two metres (e.g. Coelophysis) to four metres (Dilophosaurus). Less clear is the maximum size difference between Late Triassic and Early Jurassic sauropodomorphs. The largest basal sauropodomorphs (“prosauropods”) of the Late Triassic are the size of Plateosaurus (about 6 m on average, although one morph apparently reached 10 m: Sander & Klein Reference Sander and Klein2005). Late Triassic sauropods such as Isanosaurus were similarly about 6·5 m long, and most Early Jurassic sauropods are also in the range of 6·5 m (e.g. Vulcanodon).

(c) All non-crocodylomorph pseudosuchians and all other “indigenous” Triassic forms (Padian Reference Padian and Padian1986) became extinct by the end of the Triassic. Most of these forms had growth rates and attendant metabolic rates that were lower than in ornithodirans. Exceptions that survived include crocodiles, turtles, lepidosaurs, and champsosaurs; it was a great winnowing, and the causes are still not well understood.

7.3. The need for a new paradigm to open the “Age of Dinosaurs”

As for the previous section, we have to reconsider what we have previously considered “the beginning of the Age of Dinosaurs” – namely, the Late Triassic.

(a) The origin of Dinosauria per se was not much of an evolutionary event, now that we have filled in so many of the gaps in the Late Triassic that bridge dinosaurs with their closest relatives. The “Age of Dinosaurs” does not properly begin in the Late Triassic, although dinosaurs are present and reasonably diverse (but not much more so than several other groups). The “Age of Dinosaurs” properly begins in the Early Jurassic, with the diversification of ornithischians and the radiation of saurischians (which soon attain larger mean size than in the Triassic). Also, by the Early Jurassic a variety of large and diverse pseudosuchians and other terrestrial reptiles were out of the way.

(b) The event of note in the Triassic in this regard is the origin of Ornithodira, with their upright posture, bipedality, parasagittal gait and elevated growth rates and metabolic regimes. This had certainly happened by the early Middle Triassic. The “Age of Ornithodira,” though perhaps not as catchy as “The Age of Dinosaurs,” is the more important macroevolutionary event.