Phylogeny of Dinosaurs

Kevin Padian. Encyclopedia of Dinosaurs. Editor: Philip J Currie & Kevin Padian. Amsterdam: Academic Press, 1997.

Since the early 1970s, the acceptance and use of phylogenetic systematics, or cladistics, has grown to a virtual hegemony. Vertebrate paleontology was one of the earliest disciplines to incorporate cladistic analyses of phylogenetic relationships, and analyses of major dinosaurian subgroups were well in place by the mid-1980s, with phylogenies of subgroups proliferating to the present.

There are still some unanswered questions about the monophyly and relationships of some major dinosaur lineages, as well as of individual taxa and small groups. The following summary indicates consensus as well as continuing controversy, which is discussed further in entries of individual taxa so noted. More extensive discussions of diagnoses and synapomorphies of individual taxa may also be found under their separate entries.


The limits of Dinosauria are determined by their definition in the phylogenetic system. The two stem groups of Dinosauria are Ornithischia and Saurischia; any taxon that does not belong to one of these two groups is not a dinosaur. Over the years, considerable controversy has surrounded the question of whether herrerasaurs are dinosaurs. The answer is yes if they share all the synapomorphies of Ornithischia + Saurischia. However, workers have yet to agree on a definitive list of synapomorphies for Dinosauria (e.g., Gauthier, 1986; Benton, 1990; Novas, 1991; Sereno, 1992, 1993; Sereno et al., 1993; Sereno and Novas, 1992; Holtz and Padian, MS.). Many characters are ambiguous or subject to interpretation, and some are not known in the immediate outgroups of dinosaurs, such as Pseudolagosuchus, Lewisuchus, Lagosuchus (Marasuchus), Lagerpeton, and pterosaurs, or they may be autoapomorphically transformed in certain taxa (e.g., the forelimbs of pterosaurs or the apomorphic reduction of the phalangeal formula in Eoraptor and Herrerasaurus). Clearly a great number of characters indicate that Eoraptor and Herrerasaurus are the closest known taxa to true dinosaurs, if they are not dinosaurs themselves.

Ornithischia + Saurischia appear to share the following characters. (Note that these hypotheses of synapomorphy do not indicate what all members of these taxa share, but what their most recent common ancestor had; obviously all characters are subject to modification as evolution proceeds—by definition.) The manus is asymmetrical, with its fourth and fifth manual digits reduced but still retaining a phalangeal formula of 2-3-4-3-2. Claws are present only on the first three digits. There are three fully incorporated sacral vertebrae, and there is a true U-shaped brevis fossa on the ventral surface of the ilium behind the acetabulum. Eoraptor and Herrerasaurus do not share these features, but they share many others with Dinosauria, including a semiperforate acetabulum with supra-acetabular buttress, the loss of the postfrontal (which participates in the supratemporal opening), reduced fourth and fifth digits on the manus, and a well-developed ascending process of the astragalus situated on the anterior face of the tibia.


This group is united by many features, including the toothless premaxillary tip, the predentary bone, reduced antorbital fenestra, at least slight buccal emargination of the jaw bones, the leaf-shaped teeth, jaw joint lower than the tooth row, five sacral vertebrae, and the opisthopubic pubis (Sereno, 1986). The basal members of this group include the poorly known Pisanosaurus (Late Triassic, Argentina) and Technosaurus (Late Triassic, Texas), as well as the better knownLesothosaurus (Early Jurassic, South Africa) (Weishampel and Witmer, 1990). The other ornithischians are united into a node-based group called Genasauria (Sereno, 1986) based on their possession of fully emarginated jaw bones, which suggest the presence of “cheeks” or muscular pouches lateral to the mouth for storing and manipulating food while it is being chewed, as well as a spout-shaped symphysis on the lower jaw and several other features. Genasauria is in turn divided into two node-based groups: Thyreophora, the stegosaurs and ankylosaurs and their relatives, and Cerapoda, the remaining ornithischians (Sereno, 1986).


This group is distinguished by keeled scutes of dermal armor in parasagittal rows along the body. The most basal well-known member of the group appears to be Scutellosaurus (Early Jurassic, Arizona), a small (skull length about 5 cm), apparently bipedal form with gracile forelimbs and hindlimbs, in which the tibia is slightly longer than the femur. Small scutes of various sizes and shapes appear to have covered the body and proximal tail. Scelidosaurus (Early Jurassic, England and Arizona) was considerably larger (skull length >20 cm) and presumably quadrupedal, though its tibia was also slightly longer than the femur, usually associated with a tradition of cursoriality and often bipedality. Scelidosaurusis considered closer to the remaining thyreophorans than is Scutellosaurus because it is larger and its jaws have more of the typical S-shaped curvature seen in stegosaurs and ankylosaurs, among other features (Sereno, 1986; Coombs et al., 1990).

The Stegosauria and Ankylosauria are the two largest clades of Thyreophora, both node defined (Sereno, 1986). Together, these two groups share two lateral supraorbital bones at the dorsal margin of the orbit, a rectangular scapula, long preacetabular and

short postacetabular processes of the ilium, short metapodials, the loss of the first pedal digit, and other features (Sereno, 1986). Stegosaurs reduced the dermal armor to a single row of plates or spikes, often considerably enlarged, arranged in either an opposite or alternating pattern along the backbone, depending on the genus (Galton, 1990a); pairs of thick spikes appear to have been the basal condition, and a single, long shoulder spike is also found in most forms. Stegosaurs also have small heads and short, stocky forelimbs, with long straight hindlimbs in which the femur is considerably longer than the tibia. The unguals are hoofed. Stegosaurs are known from about a dozen genera, ranging from the Middle Jurassic of China and Europe through the Late Jurassic of China, North America, Europe, and Africa and the Early Cretaceous of China and Europe. Two Late Cretaceous forms from India have been questioned but otherwise are the only known post-Albian records.

Ankylosaurs are generally divided into two node-based clades, the Nodosauridae and the Ankylosauridae (Sereno, 1986; Coombs and Maryañska, 1990). Both have short broad skulls covered with bony osteoderms that obscure the antorbital and temporal openings. The skull is very flat and low, and there is a complete bony palate separating the processes of respiration and alimentation. The body is broad and squat, and the limbs are short. Ankylosaurs, unlike stegosaurs, retained a full complement of dermal armor and elaborated it into rows of several, sometimes alternating shapes, including keeled or spined plates, polygonal pustules, spikes, and symmetrical ovals or rounded rectangles. Ankylosaurids additionally have a “tail club” formed of coalesced vertebrae and dermal scutes, absent in nodosaurids, and a pronounced fore-aft arch to the skull, which is broader than it is long and bears posterolateral “horns” formed of dermal scutes. Ankylosaurs are mostly Cretaceous forms: ankylosaurids are only known from the Late Cretaceous (North America and Asia), whereas nodosaurids range throughout the Cretaceous (Europe, North America, Australia, possibly South America, and Antarctica), with one form (Dracopelta) assigned from the Late Jurassic and another (Sarcolestes) from the Middle Jurassic of Europe. Nodosaurids are generally less specialized than ankylosaurids, which could create the impression that the latter may have evolved from the former sometime in the Cretaceous. Ankylosaurs and stegosaurs thus have approximately equally ancient first known appearances in the fossil record, suggesting an approximate divergence time by the Middle Jurassic.


The remaining ornithischians are assigned to this node-based group, which is distinguished by a broad diastema between the premaxillary and maxillary teeth, asymmetrical enamel in the cheek teeth (thicker on the buccal side of the upper teeth and thicker on the lingual side of the lower teeth), no more than five premaxillary teeth, and a fully open acetabulum, among other features (Sereno, 1986). Two principal node-defined taxa comprise the Cerapoda: the Ornithopoda [Euornithopoda of Sereno (1986); generally termed Ornithopoda by Weishampel and others in Dodsonet al. (1990) as well as in most other treatments] and the Marginocephalia (Sereno, 1986). The Ornithopoda as a group appear to show a distinct set of trends toward larger size, loss of premaxillary teeth, robusticity of cheek teeth, closure of antorbital and mandibular fenestrae, elaboration of the forward prong and reduction of the backward prong of the pubis, graviportal posture, and even quadrupedal stance, including hoof-like unguals on both fore and hind digits. This set of trends seems to follow a gradation from Heterodontosaurs through Hypsilophodonts through Iguanodonts to the noncrested and crested Hadrosaurs. However, it is also possible to find characters that unite several known forms into apparently monophyletic clusters for which the gradational names can be retained (Sereno, 1986; Norman, 1984; Sues and Norman, 1990; Norman and Weishampel, 1990; Weishampel and Horner, 1990; Horner, 1990), although experts disagree on this. Heterodontosaurs are mostly known from the Early Jurassic, though some persist later in that period. Hypsilophodonts are mostly known from the Late Jurassic and Early Cretaceous, with a couple of forms persisting to the latest Cretaceous. Iguanodonts, which few experts regard as monophyletic, first appear in the Late Jurassic and the forms closer to hadrosaurs appear in the Early Cretaceous; some members of the group appear to persist to the latest Cretaceous. Finally, the hadrosaurs are strictly Late Cretaceous. Thus, the known stratigraphic relationships of ornithopods appear to parallel closely their phylogenetic relationships.

The Marginocephalia comprise the node-based Pachycephalosauria and the Ceratopsia, united by the posterior skull shelf formed by the parietals and squamosals. Some authors (Sereno, 1986) regard this group as a separate stem from basal cerapodan stock, and others (Norman, 1984) regard some characters (e.g., the form of the pubis and the maxillary diastema) as indicating a closer relationship within Ornithopoda, perhaps branching off among the hypsilophodont nodes. Again, discovery of more complete basal taxa may inform this question further. Because marginocephalians are strictly Cretaceous—in fact, nearly all but the apparent basal form Stenopelix are restricted to the Late Cretaceous—there would be a considerable stratigraphic disjunction in the separation between the ornithopod (earliest Jurassic) and marginocephalian (late Early Cretaceous) lineages were they to be regarded as sister taxa (Dodson, 1990).

Pachycephalosaurs are distinguished by their flat, thickened skull roofs, ornamented with small tubercles continuing along the posterior sides and back of the skull. Through the history of the group the skull roofs thicken further, forming a patellar dome with clusters of spikes and nodules around its sides. In some forms these spikes become quite long, directed posteriorly, and the dome can develop a sagittal crest; the antorbital and temporal openings become closed (Sereno, 1986; Maryańska, 1990). Ceratopsians parallel ornithopods in many evolutionary ways: They tend to become larger, with a reduction of front teeth and the development of great dental batteries in the cheek teeth; the cranial ornamentation becomes more elaborate and separates two of the most derived subtaxa; and postcranially, as size increases, quadrupedality replaces bipedality and the unguals become more hoof like. The basal forms, the psittacosaurs, have a parrot-like beak and the distinctive rostral bone that characterizes all ceratopsians. These were apparently bipedal forms for the most part, but the protoceratopsids, which do not appear to be a monophyletic group (Sereno, 1986), grew to 3 m or more in length and were mostly quadrupedal; they first show the characteristic posterior skull frill of the rest of the ceratopsians, which is centrally open in all forms except Triceratops. The ceratopsids proper can be divided into the centrosaurs, which sported large nasal horns and small orbital horns, and the chasmosaurs, which had small nasal horns and large orbital horns (Dodson and Currie, 1990).


This second great branch of Dinosauria is distinguished by long posterior cervical vertebrae, axial postzygapophyses lateral to prezygapophyses, and fully developed hyposphene-hypantrum articulations in the trunk vertebrae. Digit II is the longest in the manus, both metacarpals IV and V are palmar to digits III and IV, and the “saurischian pollex” (Gauthier, 1986) is more robust than the other digits, bearing an enlarged ungual, with the first metacarpal no more than half the length of the second and the first with offset distal condyles that direct the thumb medially. There is ambiguity about many other proposed synapomorphies (Gauthier, 1986; Sereno and Novas, 1992; Novas, 1993; Sereno et al., 1993), but the monophyly of Saurischia is still well supported (Holtz and Padian, MS.).

Saurischia is separated into two stem-based groups, Sauropodomorpha and Theropoda (Gauthier, 1986; Padian and May, 1993), both ranging from the Late Triassic to the Late Cretaceous (except birds, which are members of Theropoda and still survive). Sauropodomorphs have small, high-snouted skulls, long necks, and stout, strong forelimbs with robust first toe and claw. The tibia is shorter than the femur, and the trunk between the shoulder and hip girdles is at least as long as the forelimb (Gauthier, 1986). Sauropodomorpha has been separated into Prosauropodaand Sauropoda, but the monophyly of the former (e.g., Galton, 1990b) has been strongly questioned (e.g., Gauthier, 1986), and most (though not all) of its hypothesized synapomorphies can be equally viewed as transitional characters to the larger, longer necked, more ponderous sauropods. Under this view, the “thecodontosaurids,” plateosaurids, and melanorosaurids can be seen as successively closer to sauropods, but more work is clearly needed. All “prosauropod” forms are restricted to the Late Triassic and Early Jurassic.


The large size of this group is accompanied by even longer necks and relatively smaller skulls (compared to the bulk of the animal) than in “prosauropods,” and the tails are also elongated. The nares are large and dorsally placed, the vertebrae are lightened by pleurocoels and their spines and processes become arch-like and laminar, and the number of phalanges is reduced in the manus and pes. There is, however, a fifth metatarsal, which is severely reduced in all “prosauropods.” Some major groups within Sauropoda have been traditionally identified; they include vulcanodontids, cetiosaurs, camarasaurs, diplodocids, brachiosaurs, and titanosaurs, but the interrelationships of these groups remain problematic. It is not clear if any of them (except titanosaurs) are monophyletic, and McIntosh (1990) feels that there is too much homoplasy and not enough associated skulls and skeletons known for most taxa to make cladistic analysis meaningful. The Early Jurassic vulcanodontids and the Middle to Late Jurassic cetiosaurs appear to be outside the remaining taxa; sauropods persisted in some parts of the world until the latest Cretaceous, but most lineages had become extinct before then.


The only carnivorous dinosaurs, theropods seem to have retained this ancestral habit among dinosaurs from their ornithodiran predecessors, as witnessed by the distinctive shape of the claws and teeth. Among the synapomorphies of theropods, the lacrimal is broadly exposed on the skull roof, the vomers are fused anteriorly, the ectopterygoid is expanded with a ventral fossa, the presacral vertebrae are pleurocoelous, there are five sacral vertebrae, a “transition point” (Gauthier, 1986) appears approximately halfway down the tail, distal carpal 1 overlaps metacarpals I and II, the penultimate manual phalanges are longer than the others, the preacetabular portion of the ilium is enlarged, there is a pronounced brevis fossa on the ilium, the femur is convexly bowed and not sigmoid, and the fibula is straplike and attached to a crest on the lateral side of the tibia.

Gauthier (1986) separated Theropoda into an initial dichotomy of Ceratosauria (Rowe and Gauthier, 1990), which includes Coelophysis, Syntarsus, and virtually all Late Triassic and Early Jurassic theropods, and Tetanurae, a group that bears more derived characters such as the reduction of the hand to three fingers, a large, posteriorly placed maxillary fenestra, a more anterior caudal transition point, a straplike scapula, a manus at least two-thirds as long as the humerus plus forearm, and other characters (Gauthier, 1986). These animals are virtually all post-Middle Jurassic forms. Tetanurae was in turn divided into redefined Carnosauria and Coelurosauria. Many characters uniting Carnosauria were suspected by Gauthier (1986) of being correlates of large size, and this was borne out by later analyses by Novas (1992) and Holtz (1994), who showed that tyrannosaurs were actually coelurosaurs, a group thought to comprise only relatively smaller forms. The Carnosauria still has an objective definition as theropods closer to Allosaurus than to birds, and the Coelurosauria are closer to birds than to allosaurs (Holtz and Padian, 1997). The systematics of both groups are largely problematic, and the nomenclature has been complicated by conflicting uses of the same terms (reviews in Holtz, 1994, 1996; Holtz and Padian, MS.). Within Coelurosauria, the Maniraptora includes birds and all coelurosaurs closer to birds than to ornithomimids, and Arctometatarsalia are all coelurosaurs closer to ornithomimids than to birds. The Dromaeosaurids appear to be the closest taxon to Archaeopteryx and the birds (AVES). Particularly in regard to Saurischia, the use of nomenclature, definitions, and group memberships needs to be standardized and synapomorphies hypothesized for these reorganized taxa before their phylogenetic relationships can be fully clarified. Nevertheless, the advances briefly summarized here may be regarded as the signal achievement of dinosaurian study within the past two decades.