GAIA N'15, LlSBOAlLISBON, DEZEMBROIDECEMBER 1998, pp. 219-226 (ISSN: 0871-5424) SKULL AND TOOTH MORPHOLOGY AS INDICATORS OF NICHE PARTITIONING IN SYMPATRIC MORRISON FORMATION THEROPODS Donald M. HENDERSON Department of Earth Sciences, University of Bristol. Wills Memorial Building, Queens Road, Clifton. BRISTOL SS8 1RJ, ENGLAND. UK Department of Cell Biology and Anatomy, School of Medicine, Jonh Hopkins University. 725 North Wolfe Street, BALTIMORE MD 21205-2196. USA E-mail: dhende@jhmLedu ABSTRACT: The Upper Jurassic Morrison Formation of the western United States has pro­ duced a diverse assemblage of large, carnivorous dinosaurs. Analysis of the skulls and teeth assigned to two of the best preserved of these sympatric Morrison Formation genera, Allosaurus MARSH and Ceratosaurus MARSH, reveals three distinct forms. These three skull and tooth morphologies are interpreted as evidence of feeding and behavioural niche par­ tioning among the top Morrison carnivores. One form, represented by some members ofthe genus Allosaurus, has a shortened face, and a tall, wide skull, with short, posteriorly di­ rected teeth. Other Allosaurus specimens have a long face, a low skull profile, and longer, more vertically oriented teeth. Ceratosaurus also has a long face, but is distinguished from the latter allosaurid by exaggeration of the depth of the skull and by having even longer and broader teeth. Distinct lacrimals and nasal ornaments seen in the three forms would aid in interspecific recognition, and reduce incidences of direct competitive interaction. The gen­ eral rarity of Ceratosaurus in the Morrison suggests competitive displacement of Cera to­ saurus by the long-toothed, long-skulled allosaurs. RESUME: La Formation Morrison du Jurassique superieur de l'Ouest des Etats Unis a livre un ensemble varie de grands dinosaures carnivores. L'etude des cranes et des dents des espe­ ces les mieux preserves, Allosaurus MARSH et Ceratosaurus MARSH, revele trois ensembles distincts. Ces trois ensembles indiquent une division entre differentes niches alimentaires et comportementales. Quelques specimens du genre Allosaurus ont un museau tres court, un crane haut et large, et des dents courtes et dirigees en arriere. D'autre specimens d'Allo­ saurus ont un museau allonge, un crane bas et etroit, et les dents sont longues et verticales. Ceratosaurus a egalement un museau allonge, mais Ie crane est plus haut, et les dents de Ceratosaurus sont plus longues et plus larges que chez Ie deuxieme type d'Allosaurus. Les lacrimaux et les ornements nasaux, qui sont distincts chez chaque forme, auraient pu aider a la reconnaissance interspecifique, et reduire ainsi la frequence des interactions directes. La rarete de Ceratosaurus dans la Morrison suggere une exclusion competitive par les allo­ saures avec des dents et un crane allonges. INTRODUCTION rus, Allosaurus, Marshosaurus, and Stokesosaurus (BRITT, 1991), and Como Bluff quarries have pro­ Many distinct genera of large, carnivorous dino­ duced associated Allosaurus, Torvosaurus, and saurs are known from quarries in the Upper Jurassic other large carnivorous forms (BAKKER, 1996). Morrison Formation. Some individual quarries have . produced multiple genera of carnivorous forms: A review of the literature and of published skele­ Cleveland-Lloyd Quarry in Utah has produced Cera­ tal illustrations and reconstructions suggests that if tosaurus MARSH, Allosaurus MARSH, Marshosaurus there are not two species of Allosaurus from the Mor­ MADSEN and Stokesosaurus MADSEN (MADSEN, rison Formation, there is at least evidence for two 1976), Dry Mesa Quarry Marsh, in Colorado has pro­ 'morphs'. PAUL (1988) argued for two, possibly duced Torvosaurus GALTON & JENSEN, Ceratosau- three, species. BRITT (1991) and CHURE (pers. 219 artigos/papers D.M. HENDERSON comm., 1997) also claim at least two species of Allo­ study (DAYAN et al., 1990), small felids were also saurus. MADSEN (1976: pI. 2) produced a composite shown to fall into regularly, spaced categories with skull for Allosaurus based on material from the respect to mean canine diameter. PI MM & GITTLE­ Cleveland-Lloyd quarry. This restored skull shows MAN (1990), commenting on the work of DAYAN et al. marked differences between right and left examples (1989,1990), noted thatthese small mustelids and of the maxillae, lacrimals, pterygoids, and nasals felids form a natural guild with members of the guild suggesting that this Allosaurus is a chimera. MAD­ sharing the same prey. SEN (1976), commenting on the variation in the There are also examples of character displace­ number of tooth alveoli shown by his extensive sam­ ment in fossil carnivores. WERDELIN (1996) found a ple of maxillae, states that the variation could not be regular partitioning of carassial total lengths and explained by different growth stages of the individu­ blade lengths in late Miocene and early Pliocene als who possessed the elements. Gilmore's illustra­ hyaenids from Eurasia and Africa. The similarity of tions (GILMORE, 1920) of the reconstructed skull of the hyaenid partitioning to that seen in extant, sym~ "Antrodemus" LEIDY (currently regarded as a junior patric canids led WERDELIN (1996) to infer similar synonym for Allosaurus) is very different from the re­ ecological roles for these early hyaenids who filled constructed skull presented by MADSEN (1976). the niches now filled by the true can ids. MASSARE The dinosaur remains at the Cleveland-Lloyd, (1987), in a study of tooth variation in Mesozoic ma­ Dry Mesa, and Como Bluff quarries were all depos­ rine diapsids, found a range of tooth forms similar to ited within 3 million years of each other (SMITH, that seen in modern carnivorous marine mammals. 1996). With several different types of large, bipedal The material studied by MASSARE (1987) did not carnivore in the Morrison environment there would come from a single deposit or time plane, but her have been extensive niche overlaps with resultant study indicates that it is possible to identify feeding interspecific competition if they were living together. specializations and niche partitioning among car­ The competition between these sympatric carni­ nivorous diapsids. vores would have been heightened by the fact that The large theropods of the Morrison Formation they were all of roughly the same body length of 6-8 form a guild, one specialized for locating, killing, dis­ m. The morphological differences within the genus membering, and ingesting large prey. The skulls and Allosaurus cou ld be interpreted as showing the ef­ teeth of large theropods are their most variable fea­ fects of natural selection reducing intraspecific com­ tures, and the best indicators of how predatory petition by producing two different forms of niches were divided up. In their review of mammal­ Allosaurus. The present paper aims to interpret cra­ ian carnivore teeth, VAN VALKENBURGH & RUFF nial and dental morphology, as it relates to niche par­ (1987: 380) state that " ... canine teeth oflarge preda­ titioning and feeding strategy, for three, well known, tors are used to kill and dismember prey and to sympatric theropods - Ceratosaurus and two forms wound or threaten other individuals." and " ... differ­ of Allosaurus. For the purposes of this paper, one ences in bite power and behaviour should be re­ form of Allosaurus, based on the MADSEN (1976) re­ fiected by differences in canine shape if teeth are construction, wil l be referred to as Allosaurus. The designed to resist expected loads." It seems reason­ other Allosaurus, the one illustrated by GILMORE able to extend these observations to other taxa such (1920), will be refered to as "Antrodemus". as dinosaurs to infer how their teeth, and the associ­ ated skull bones and muscles, wou ld have func­ CARNIVORE ECOLOGY tioned. It is a common observation in ecology that the in­ visible hand of natural selection acts to reduce the METHODS physiological, morphological, and behavioural simi­ Published illustrations of Ceratosaurus and "An­ larities between competing organisms, thus lessen­ trodemus" (GILMORE, 1920) and Allosaurus (MAD­ ing the depression of fitness due to niche overlap. SEN, 1976) were used as the primary data sources. These reductions in similarities were termed "char­ Outline drawings of the three skulls are presented in acterdisplacements" by BROWN & WILSON (1956). Figure 1. Evidence for character displacement in sympat­ Skull lengths were measured from the posterior ric carnivores can be found in extant faunas. DAYAN end of the occipital condyle to the anterior margin of et al. (1989, 1990) presented separate instances the premaxillae from ventral views of the three taxa. that show niche partitioning among modern carnivo­ The postorbital heights were taken along lines that rous mammals. In one study (DAYAN et al., 1989), ran from the skull table, through the approximate three sympatric species of weasel (Muste/a sp.), center of the lower temporal fenestra, and inter­ from eight different localities across North America cepted perpendicularly with the ventral margins of were shown to have consistent, regularly spaced the juga Is. The antorbital heights were taken from groupings of mean canine diameter. In another 220 SKULL AND TOOTH MORPHOLOGY AS INDICA TORS OF NICHE PARTITIONING IN THEROPODS Fig. 1 - Lateral and dorsal views of skulls and mandibles. Top - "Antrodemus va/ens" (LEIDY), redrawn from GILMORE (1920). Middle -Allosaurus (ragilis MARSH, redrawn from MADSEN (1976). Bottom - Ceratosaurus nasicomis MARSH, lat­ eral view redrawn from