ONTOGENY OF CRANIAL EPI-OSSIFICATIONS IN TRICERATOPS

Journal of Vertebrate Paleontology 28(1):134 144, March 2008 2008 by the Society of Vertebrate Paleontology ARTICLE ONTOGENY OF CRANIAL EPI-OSSIFICATIONS IN TRICERATOPS JOHN R. HORNER *,1 AND MARK B. GOODWIN 2 1 Museum of the Rockies, Montana State University, Bozeman, Montana 59717-0040, U.S.A. jhorner@montana.edu 2 Museum of Paleontology, University of California, Berkeley, CA 94720-4780, U.S.A. mark@berkeley.edu ABSTRACT Historically, the scarcity of non-adult Triceratops fossils collected from Upper Cretaceous sediments of North America limited our understanding and promoted controversy with regard to the morphology, and presence or absence of cranial epi-ossifications in this widely known horned dinosaur. The recent discovery of several exceptionally well preserved juvenile and subadult Triceratops skulls and numerous juvenile, subadult, and adult cranial elements, from the Hell Creek Formation of eastern Montana, confirms the ontogeny and morphology of epi-ossifications in this study. We propose to standardize the terminology for these four cranial epi-ossifications: epinasal, epijugal, epiparietal, and episquamosal. We describe the ontogeny and timing of the fusion of each of these epi-ossifications and the rostral from a cranial growth series. Although the timing is variable, the epinasal fuses first, followed by the rostal, the epijugals, the episquamosals, and lastly by the epiparietals. Co-ossification of the epinasal, rostral and epijugals unites several of the anterior (rostral-nasal-premaxillae) and lateral (jugal-quadratojugal) skull elements. In combination with forward directed postorbital horns and a massive fan-shaped frill, cranial epi-ossifications may have enhanced visual display and species communication in Triceratops. INTRODUCTION Triceratops was one of the largest and most abundant horned dinosaurs from the Late Cretaceous of western North America. Historically, the majority of skulls collected and described were adults (Marsh, 1889; Hatcher et al., 1907; Lull, 1933). A few subadult skulls and isolated juvenile horns were also collected but remained undescribed or not recognized as such (Horner and Goodwin, 2006). A massive bony frill (parietal + squamosal), a pair of prominent postorbital horns, and a nasal horn characterize the adult Triceratops skull (Dodson, 1996). The epinasal, epijugal, epiparietal, episquamosal, and rostral are distinct accessory ossifications that attach to or overlie cranial elements early in ontogeny. Their articulation, morphology, and eventual fusion are variable. As a result, a degree of uncertainty regarding the timing, morphology, and presence or absence of epiossifications persists in the literature (Marsh, 1891; Hatcher et al., 1907; Sternberg, 1949; Ostrom and Wellnhofer, 1986; Forster, 1996a; Dodson et al., 2004). In this study, we document the morphology and ontogeny of the epinasal, epijugal, epiparietal, episquamosal, and the rostral from a cranial growth series of Triceratops (see Horner and Goodwin, 2006) and from previously undescribed specimens catalogued in the collections of the Museum of the Rockies and the University of California Museum of Paleontology. We recommend standardizing the names of these epi-ossifications based on their anatomical relationship to underlying bones or contacts in the skull. We concur with Hatcher et al. (1907:34) that the term, epoccipital, is not ideal since a connection with any of the occipital bones is lacking. At the same time, we do not advocate introducing an additional term for this ossification. If this bone can be identified independently as we confirm in this study, or if it is fused onto the parietal or squamosal, the terms epiparietal and episquamosal are recommended following this early suggestion by Hatcher et al. (1907) and more recent use by Forster and Sampson (2002). If the element is isolated and cannot be identified beyond epiparietal or episquamosal, it is still diagnostic as one element or the other and these terms should suffice. Many of the epi-ossifications described and figured in this study are from relatively complete, disarticulated Triceratops skulls allowing confirmation of its ontogenetic stage (see Table 1). We presume that the isolated epi-ossifications (Table 2) found in microvertebrate sites, channel lags, and overbank deposits within the Hell Creek Formation, Garfield and McCone Counties, eastern Montana, are also derived from Triceratops, though we cannot absolutely rule out they may represent the closely related sympatric taxon, Torosaurus. It is impossible to distinguish the two taxa without the presence of a fenestrated parietal and we lack a juvenile or subadult Torosaurus in our sample. For purposes of this study, we assume that the cranial ontogeny and morphology of the epiossifications in Triceratops and Torosaurus are nearly identical. More material is needed in order to determine if patterns and rates of growth in Triceratops are applicable to other ceratopsians. In this study of cranial epi-ossification and ontogeny, our focus is on the description of patterns of development in Triceratops. We use the term epi-ossification rather than dermal or epidermal bone to categorize these bones that overlie dermal elements of the cranium. We cannot be absolutely certain that these elements are similar to osteoderms, scutes, or ossicles found in a variety of reptiles, anurans, and edentates (Hill, 2006; Vickaryous and Hall, 2006) and lacked a cartilage precursor or developed intramembranously within the dermis as dermal bone (Kardong, 1998). Analysis of the interactions of processes related to dermal and epidermal development in the skull of Triceratops is beyond the scope of this paper and will be investigated in a following study. Institutional Abbreviations MOR, Museum of the Rockies, Bozeman, MT; UCMP, University of California Museum of Paleontology, Berkeley. 134

HORNER AND GOODWIN EPI-OSSIFICATIONS IN TRICERATOPS 135 TABLE 1. Fourteen skulls and partial skulls of Triceratops represent three growth stages. Triceratops Locality name Ontogenetic stage Epinasal Nasal length (cm) Rostral Epijugal Epiparietal or episquamosal # Epiparietal # Episquamosal # MOR 1110 SG-5 Juvenile x 33 x x MOR 1199 Sierra Skull Juvenile 18 x 6 UCMP 128562 Harley s Puzzle Juvenile x 6 UCMP 136306 Jack s Bay 5 Juvenile x x 1 UCMP 159233 Juvenile Trike Juvenile 1 MOR 1120 Getaway Trike Subadult x 1 38 x x 3 2 5 6L,7R MOR 2574 Quitin Time Subadult x 44 x UCMP 137263 High Ceratopsian Subadult x 47 UCMP 137266 Bob s Bonebed Subadult x MOR 1604 Baker Trike Adult x 1 50 (est) x 2 x 3 MOR 1625 Haxby Trike Adult x 1 50 (est) x 2 x 3 MOR 2570 Scavenged Skull Adult x 1 x 3 UCMP 144297 Bay Stud Coulee Adult 50 (est) x 2 UCMP 174838 Coyote Basin Adult x 1 1 Epiossification fusion is highly variable in subadults. Note that the relatively smaller nasals in MOR 1120 have a fused epinasal, while the larger nasals in MOR 2574 and UCMP 137263 possess an unfused epinasal. Blank cell indicates element not preserved. Abbreviations: est, estimate; L, left; R, right; x 1, epinasal fused to nasals; x 2, rostral fused to premaxillae; x 3, epijugal fused to jugal. Epinasal EPI-OSSIFICATIONS Marsh (1891:169) recognized that the nasal horn core ossifies from a separate centre, an ontogenetic pattern noted and illustrated by Hatcher et al. (1907:33, fig. 29). Our sample of juvenile and subadult Triceratops skulls confirms this. The over-growth of the epinasal onto the underlying nasals and premaxillae often obscures the sutural relationships between these bones in adults. Ostrom and Wellnhofer (1986:121) questioned whether the nasal horn core was an outgrowth of the nasals. Forster (1996a: 248) noted that a separate epinasal cannot be distinguished in most Triceratops specimens but suggested in this study that the occurrence of an epinasal element was evident. Seven epinasals are identified in the MOR and UCMP collections (Tables 1, 2), five of which are illustrated in Figures 1 and 2. Three of the epinasals (UCMP 136306, MOR 1110 and 2574) are associated with skulls. The smallest epinasal (Fig. 1A, MOR 1167) is 4 cm high, laterally compressed, conical with a rounded apex. Nearly all the dorsal surface is abraded by transport. Unabraded surfaces reveal a porous texture (Fig. 1B) resulting from traversing vascular canals of the fibrolamellar bone tissues. This texture is characteristic of ontogenetically young bone (see Johnson, 1977; Bennett, 1993; Sampson et al., 1997; Tumarkin- Deratzian et al., 2006). The ventral surface of MOR 1167 is convex with a very porous texture. A few large channels perforate the ventral surface, and a couple of these channels exit on the dorsal surface close to the line demarking the dorsal-ventral border. The two epinasals associated with juvenile skulls (Fig. 1C, UCMP 136306; Fig. 1D, MOR 1110) are similar with rounded anterior and straight-sided posterior surfaces. Both specimens have a porous surface similar to the smallest epinasal (Fig. 1A, MOR 1167). Unlike MOR 1167, the ventral surface of these two epinasals is flat with a small posteroventral convex bump of bone (Fig. 1D, arrow) that inserts into an open space between the paired nasals (Fig. 2 A, arrow) where it meets the paired posterodorsal processes of the premaxillae. The ventral surface of these two epinasals is oval. The paired nasals and epinasal simply rest against one another in articulation (Fig. 2B). The opposing surfaces are smooth. A slightly larger epinasal, MOR 989 (Fig. 1E) is nearly the same height as the epinasals UCMP 136306 and MOR 1110 but has a relatively broader base with a near circular cross-section. The dorsal surface of MOR 989 is smooth, indicative of a more mature ontogenetic stage (see Johnson, 1977; Bennett, 1993; Tumarkin-Deratzian et al., 2006). The largest of the epinasals, MOR 2574 (Figs. 1F; 2C, D) is convex on the anterior side, concave on the posterior side, with a pointed apex. The lateroventral sides of MOR 2574 have a process (Fig. 1F, arrow) that overlaps a portion of the nasals anterolaterally. The dorsal surface of MOR 2574 is sculptured by indented vessel grooves. Ventrally, the surface is oval, much more rugose than the smaller epinasals, and has distinct grooves for the union of the nasals and premaxillae (Fig. 2C, D). The fusion of the nasal bones occurred at approximately the same time as fusion of the epinasal to the nasals, although the timing of this fusion appears to be variable in our sample of skulls and associated elements (see Table 1). Figures 2C and 3B D illustrate the variable timing of the epinasal-nasal-premaxilla union in Triceratops. The dorsoposterior ascending processes of the paired premaxillae fit snugly between the anteroventral processes of the nasals (Fig. 2C, lateral view; Fig. 3D, ventral view), but this complex apparently does not fuse at the same time the epinasal fuses with the nasals. In MOR 2574, the epinasal-nasal-premaxilla complex is unfused (Fig. 2C). In MOR 1120, the epinasal is fused to the nasals (Fig. 3A, C), the paired nasals coalesce dorsally (Fig 3A) but remain open ventrally (Fig. 3B), and the ascending process of the premaxilla is not fused anteroventrally at the point of insertion into the nasal (Fig. 3D at white arrow). The premaxillary fusion with the nasals and epinasal may occur in conjunction with the fusion of the rostral described in more detail below. Throughout the juvenile early subadult ontogenetic stages, the disarticulated nasals and premaxillae demonstrate little evi- TABLE 2. Isolated Triceratops cranial epi-ossifications and the ontogenetic stage represented. Triceratops Element Ontogenetic stage MOR 1167 Epinasal Juvenile UCMP 186583 Epinasal Juvenile MOR 989 Epinasal, rostral Juvenile MOR 2582 Epiparietal or episquamosal Juvenile MOR 2588 Epiparietal or episquamosal Subadult UCMP 172487 Epiparietal or episquamosal Subadult MOR 2583 Epiparietal or episquamosal Subadult MOR 2586 Epiparietal?adult MOR 2584 Epijugal Subadult MOR 2585 Epijugal Subadult

136 JOURNAL OF VERTEBRATE PALEONTOLOGY, VOL. 28, NO. 1, 2008 FIGURE 1. Triceratops epinasals in right (A, C E) and left (F) lateral view. A, juvenile epinasal, MOR 1167; B, magnified image of the surface of MOR 1167 illustrates the porous texture characteristic of bone in young individuals; C, juvenile epinasal, UCMP 136306; D, juvenile epinasal from a Triceratops skull, MOR 1110; the arrow points out the convex bump of bone that inserts between the anteroventral processes of the paired nasals and meets the dorsoposterior processes of the premaxillae; E, juvenile epinasal, MOR 989; F, subadult epinasal, MOR 2574; the arrow points out the lateral flange of bone that overlaps the anterolateral surface of the anteroventral process of the left nasal. Abbreviation: ant, anterior. Scale bar equals 5 cm for A, C F and 5 mm for B. dence of their eventual fusion. The flat, medial surface where the paired nasals contact remains nearly smooth throughout these ontogenetic stages. Only the anterodorsal surface where the epinasal eventually attaches to the nasals becomes rugose. Rostral The skull of Triceratops tapers anteriorly to a narrow, keeled edentulous beak formed by the rugose and indented vessel grooved rostral and opposing predentary. Hatcher et al. (1907: 33) acknowledged the rostral as a dermal or epidermal ossification and grouped it with the nasal horn core, epijugals, epoccipitals (epiparietal + episquamosal), and the predentary, and we concur. We do not discuss the predentary in this study as it appears phylogenetically earlier in the basal ornithischian clade (Norman et al., 2004). The rostral is disarticulated in our juvenile and subadult skulls (Table 1). Based on the subadult skull MOR 2574, the anterior edges of the premaxillae inserts into the rostral posterolaterally on either side of an inverted Y shaped midline septum (Fig. 4A, B). The long dorsal process of the rostral extends posteriorly between the disc-like premaxillae anterodorsally, both separating and obscuring the junction with the premaxillae. A bony spur off each premaxilla projects dorsally into the overlying nasal and epinasal to complete a union of the rostral-premaxillae-nasals-epinasal. This union expands ontogenetically into a tightly fused complex where sutural contacts become completely concealed in dorsal and lateral view (Fig. 4C). The posterolateral extent of the rostral unto the premaxillae is trace- FIGURE 2. Triceratops epinasal-nasal-premaxillary complex. A & B, nasals and epinasal of showing the three elements disarticulated (A) and articulated (B) in the juvenile Triceratops skull, MOR 1110, in dorsal view. Arrow in A points to the convex bump that fits between the anteroventral processes of the nasals but is no longer visible upon articulation in B; C, left nasal and left premaxilla (white arrow marks the apex of the premaxilla inserted into the nasal) and the attachment position of the epinasal remain unfused in the subadult ontogenetic stage shown by MOR 2574, in medial view. D, ventral surface of the epinasal showing the paired premaxillary grooves that receive the anterior face of the ascending anterodorsal processes of each premaxilla. Abbreviations: ant, anterior; dor, dorsal; en, epinasal; n, nasal; ng, nasal groove; pmg, premaxillary groove. Scale bars equal 5 cm.

HORNER AND GOODWIN EPI-OSSIFICATIONS IN TRICERATOPS 137

138 JOURNAL OF VERTEBRATE PALEONTOLOGY, VOL. 28, NO. 1, 2008 FIGURE 3. Nasal complex of a subadult Triceratops skull, MOR 1120. A, anterior end of the nasal complex showing the extent of the nasal horn (epinasal) indicated by bracket relative to the fused nasals in dorsal view. Note that the dorsal sutural contacts between these elements are obscured by fusion dorsally. B, the nasals remain unfused ventrally; arrows point to the open internasal suture and the nasal horn formed by the epinasal in ventral view; C, the nasal horn (epinasal) and nasal in left lateral view; D, close-up of the anteroventral surface of the nasal horn (epinasal) nasal union with the left premaxilla inserted into the notch between the nasal horn (epinasal) and nasal above the white arrow; the white arrow delineates the posterior extent of the ascending nasal process of the premaxilla into the nasal. Abbreviations: n, nasal; nh, nasal horn (epinasal); ns, internasal suture; pm, premaxillary. Scale bar equals 5 cm for A, C, D and 10 cm for B.

HORNER AND GOODWIN EPI-OSSIFICATIONS IN TRICERATOPS 139 able only by the contrast between the highly porous, vesselchanneled morphology of the rostral with the relatively smoother premaxillae (Fig. 4C, white arrows). The position of the posterodorsal contact of the rostral with the epinasal is obscured because of the considerable bone growth that develops at this union (Fig. 4C, black arrow). Surprisingly, even in large adult Triceratops skulls, sutures remain open between the nasals ventrally and for the premaxillae posteriorly (Fig. 4D). Based on FIGURE 4. Triceratops rostral and rostral complex. A, a subadult rostral, MOR 2574, in left lateral and B, posterior view. The thin midline septum noted by the black arrows inserts between the anterior margins of the paired premaxillae. C, the edentulous beak and snout shows the posterolateral extent of the rostral (white arrows) in the adult skull, MOR 1625. The black arrow points to a region where the bony overgrowth may be an extension of the nasal, rostral, or both. D, the nasal/premaxillary junction showing the open internasal suture between the paired nasals and premaxillae (horizontal white arrows), MOR 1625, in posteroventral view. Vertical white arrow shows the position of the descending posterior nasal process into the left premaxilla. The same view in a juvenile skull is shown in Fig. 3D. Abbreviations: n, nasals; pm, premaxillary. Scale bar equals 5 cm in A, B, D and 10 cm in C.

140 JOURNAL OF VERTEBRATE PALEONTOLOGY, VOL. 28, NO. 1, 2008 FIGURE 5. Triceratops epijugals. A, left and B, right epijugal from the juvenile skull, MOR 1110, in lateral view; C, right subadult epijugal, MOR 2585, in lateral view; D, left subadult epijugal, MOR 2584, in lateral view; E, articulated subadult jugal and quadratojugal, MOR 1120 in left lateral view; white arrows indicate the position of the union with the epijugal; F, internal face of the subadult epijugal, MOR 2585, showing the rugose surfaces that contact and unite the jugal and quadratojugal; G, adult jugal and epijugal, MOR 2570, in dorsal view; the white line traces the intersection between the jugal and epijugal; H, anteroposterior oriented vessel groove noted by the white arrow on the epijugal, UCMP 174838. Abbreviations: ant, anterior; ej, epijugal; j, jugal; qj, quadratojugal. Scale bars equal 5 cm.

HORNER AND GOODWIN EPI-OSSIFICATIONS IN TRICERATOPS 141

142 JOURNAL OF VERTEBRATE PALEONTOLOGY, VOL. 28, NO. 1, 2008 MOR 1625 (Fig.4D) and confirmed by other specimens, complete fusion of the ventral nasal suture occurs prior to fusion of the posterior premaxillary suture in the largest of adult skulls. Epijugal A pair of unfused epijugals (Fig. 5A, B) preserved with a juvenile skull (MOR 1110) confirms their emergence early in ontogeny. Larger, disarticulated epijugals (Fig. 5C, D) indicate fusion to the skull occurs relatively late in ontogeny, likely during the subadult adult stages. The epijugal ontogenetically fuses onto the lateroventral surfaces of the jugal and quadratojugal (Fig. 5E, F). Juvenile epijugals are hoof -shaped with a rugose lateral surface and are slightly asymmetrical. Their concave anterodorsal surface becomes planer in older individuals, contributing to a more horn-like appearance of the epijugal (Fig. 5G). The lateral surface increases in rugosity with numerous vascular pits, deeper indented vessel grooves, and presumably supported a keratin structure (Hatcher et al., 1907). A shallow, longitudinal indented vessel groove on the lateroventral surface (Fig. 5H) deepens ontogenetically and is present in all Triceratops and Torosaurus epijugals and jugals examined. Similar to the condition observed in the nasals prior to fusion of the epinasal, the articular surfaces of the jugal and quadratojugal are smooth, and it is the fusion of the epijugal to these two bones that appears to bind them together. Epiparietal and Episquamosal Epiparietals and episquamosals (Fig. 6A I) are commonly found associated with juvenile and subadult Triceratops skulls. An epiparietal or episquamosal (Fig. 6A) found with a partial juvenile skull (UCMP 159223) isa2cmequilateral delta-shaped triangle, flat ventrally, with a rugose dorsal surface marked by vascular pits. Numerous vascular pits are also present on the ventral surface. The very spongy, striated surface is highly vascularized. An equilateral triangular shape is retained throughout the juvenile stage of ontogeny (Fig. 6A G), and begins to flatten in subadults (Fig. 6 H J) against the edge of the frill (Horner and Goodwin, 2006). As the epiparietals and episquamosals grow along the edge of the frill, they become more rugose and concave ventrally (Fig. 6K). The posteroventral surface of the epiparietals and episquamosals increases greater than the anterodorsal, signaling a closer association with, and eventually fusion to the posteroventral side of the frill. This sequence may be restricted to the epiparietals, as the episquamosals appear to have a different morphology. The apex of the episquamosals migrates from a lateral orientation to a more anterior facing direction during ontogeny (Fig. 6L). The apices of the epiparietals and episquamosals compress dorsoventrally and the ventral surfaces of these ossifications become narrower and wider (Horner and Goodwin, 2006). Eventually, the lateral edges of the epiparietals and episquamosals nearly join or contact each other in subadults as fusion onto the frill gradually proceeds. This transformation continues into adulthood where it is often difficult to trace any sutural contact between the epiparietals and episquamosals with the underlying frill (Fig. 6M). Generally, seven episquamosals ornament the posterior margin of each squamosal, plus an additional epiparietal/episquamosal spans the contact between both parietals and squamosal in subadult and adult Triceratops; five to seven epiparietals ornament the posterior parietal. The number of epiparietals and episquamosals can vary from one specimen to another, as observed by Hatcher et al. (1907), and occasionally between left and right squamosals on a single individual (see Table 1). DISCUSSION From a structural point of view, it is clear that the fusion of the epinasal, rostral, and epijugals aid in binding various cranial elements together, thus providing a more rigid skull. Interestingly, the epiparietals and episquamosals do not provide this type of bond, except at the posterior margin of the frill where they appear to overlap the parietal-squamosal contact. The parietal and squamosals remain separate throughout ontogeny (Horner and Goodwin, 2006). Postmortem taphonomic events, distortion and crushing may result in the appearance of parietal-squamosal fusion in adult skulls, but closer inspection indicates otherwise. In one subadult skull (MOR 1120), the episquamosals are fused onto the frill, but the epiparietals are unattached. The number of episquamosals may vary within and between individuals in our growth series, indicating some epiparietals and episquamosals did not always fuse onto the frill in ontogeny. We observe this uneven fusion in the epinasal and epijugals. Forster (1990b) also recognized that epinasals are occasionally missing due to the lack of fusion. Loss of an epi-ossification may occur in a variety of ways: in vivo, postmortem, taphonomically, or evolutionarily. It is unlikely, however, that the absence of an epinasal in Triceratops serratus, T. obtusus, and T. hatcheri ( Diceratops hatcheri), is an autapomorphy. We hypothesize that loss of the epinasal in vivo would most likely result in a rugose, irregular nasal boss as bone remodeling occurs and the site heals. This type of pathology of the nasal boss is observed in Torosaurus (MOR 1122), and Ugrosaurus olsoni (UCMP 128561) Cobabe and Fastovsky 1987 ( Triceratops sp. (Forster, 1993)). Although the timing of fusion of these epi-ossifications is variable, from our sample it appears that the epinasal probably fuses first, followed by the rostral, epijugals, and episquamosals. The epiparietals apparently fuse last. The extensive vascularization indicated by the indented vessel grooves and vascular pits of all the epi-ossifications supports the hypothesis that each of these elements supported a thick keratin covering from their earliest appearance, suggested by Hatcher et al. (1907) and reaffirmed by Horner and Marshall (2002). With the exception of the rostral, we interpret these epiossifications as cranial ornaments. We hypothesize that the ontogenetic development of the epinasal, epijugals, epiparietals, and episquamosals (combined with the solid fan-shaped frill and >1 m long forward facing postorbital horns in adults) function primarily in visual communication to enable the identity of juveniles and signal their attainment of sexual maturity (Horner and Goodwin, 2006). The epi-ossifications, except for the epiparietals and episquamosals, bind the dermal cranium in adults. FIGURE 6. Triceratops epiparietals and episquamosals. A, juvenile epi-parietal or squamosal, UCMP 159223; B E, juvenile epi-parietals or squamosals, MOR 1199; F and G, juvenile epiparietals, MOR 2582; H and I, epiparietals from the subadult Triceratops skull, MOR 1120; J, posterior view of MOR 1120, episquamosals attached to the Triceratops squamosal with a distinct demarcation (white arrows) between the squamosal and episquamosals. The black arrows point to the apex of each episquamosal; K, disarticulated?adult epiparietal in cross-section; arrow points to the apex, MOR 2586. Note the long bony extension that drapes over the dorsoventral edge of the parietal; L, posterior squamosal fragment with the attached episquamosal, MOR 2570, in lateral view. The white arrow points to the anterior oriented apex of the episquamosal; M, posterior squamosal showing raised area that represent the episquamosals but no demarcation between the episquamosal and squamosal is present in this partial adult skull, MOR 1625. Black arrows point to the apex of these relatively low relief episquamosals. Abbreviations: ant, anterior; esq, episquamosal; sq, squamosal. Scale bar equals 10 cm for A J, L, M and 5 cm for K.

HORNER AND GOODWIN EPI-OSSIFICATIONS IN TRICERATOPS 143 visually dominating frill and postorbital horns, skin color may have enhanced intraspecific communication and sexual display. CONCLUSION Prior to this study, cranial epi-ossifications in Triceratops were poorly documented and often misinterpreted. We confirm the expression, morphology, and ontogeny of epi-ossifications in Triceratops using a cranial growth series supplemented by recently collected juvenile adult skulls and cranial elements from the Late Cretaceous Hell Creek Formation, eastern Montana. Timing of epi-ossification fusion to the cranium is variable. We confirm that the Triceratops epinasal is derived from a separate center of ossification, indicated very early by Marsh (1891) and illustrated by Hatcher et al. (1907), but still a subject of discussion in the literature ever since (Brown and Schlaikjer, 1940; Sternberg, 1949; Ostrom and Wellnhoffer, 1986; Forster, 1996a). The rostral functioned with the predentary to acquire food and bind the premaxillae-nasal-epinasal complex as it fuses with the skull ontogenetically. The epijugal binds the jugal-quadratojugal-quadrate union in adult skulls. We hypothesize that epiossifications in Triceratops are cranial ornaments that enhanced visual display, played a role in species communication, and signaled the attainment of sexual maturity in combination with the forward directed postorbital horns. FIGURE 7. Restoration of an adult Triceratops with epiparietals, episquamosals, epinasal, and forward-directed postorbital horns. This is observed in the expansion of the nasal horn above the rostral-premaxillae-nasal complex and in the epijugal with the underlying jugal and quadratojugal. As a major cranial ornament in adults, the prominent and morphologically diverse epinasal enlarges the profile of the narrow rostral-premaxillae-nasal complex as a nasal horn. The epijugal capped by a keratin spike accentuates the side of the face. A series of epiparietals and episquamosals adorns the posterior frill margin, exaggerating the size and dimensions of the skull. The rostral possesses a thin, beveled ventral cutting edge. The opposing predentary has a thick, flat cutting edge and terminates in a sharp, upcurved point and rests inside the rostral (Dodson et al., 2004). This edentulous beak provides an efficient mechanism for nipping and acquiring plant material. The unfused juvenile rostral grows allometrically into a robust elongate beak, tightly fused onto the premaxillae and nasals, and capped by a substantial epinasal that forms a nasal horn. This arrangement and variable fusion could reflect different feeding strategies, behaviors, or food sources depending on the age of the individual. Sexual maturity in large herbivorous mammals is often signaled by some manner of morphological change in the juvenile and subadult states (Jarman, 1983). The final shape and size of these keratin-covered epi-ossifications remains unknown, but based on the adult morphology, a head-on view of an adult Triceratops (Fig. 7) offers an array of cranial ornamentation that we hypothesize is a signal of sexual maturity, for attracting a mate, or to intimidate an adversary in competition with rivals. Bright integumental colors around the face and head occur in a diverse assortment of extant birds (see Prum et al. 1994, 1999). Following this line of evidence and the apparent lack of morphological sexual dimorphism in our assemblage and in ceratopsids overall (Padian et al., 2005), we propose that dimorphism may have been expressed with structural color arising from the scattering of ultraviolet light by collagen fibers in the hard keratin of the face, head, horns, and epi-ossifications in Triceratops. Paired with a ACKNOWLEDGMENTS We thank Carrie Ancell for her exceptional fossil preparation; Bob Harmon, Nels Peterson and the many MOR and UCMP field crew members who participated in the Hell Creek Project. David Smith illustrated the restoration of the adult Triceratops skull. We thank William A. Clemens and Patricia Holroyd for their comments on an earlier version of this manuscript. Andrew Farke, Thomas Lehman, and Frank Varriale provided thoughtful reviews from their collective knowledge of ceratopsians that improved this study. We thank David Weishampel for his attentive editorial guidance. Nathan Myhrvold generously provided funding for the Hell Creek Project and is gratefully acknowledged. UCMP provided funding to M.B.G. The assistance of the Bureau of Land Management, the United States Fish and Wildlife Service, and the Charles M. Russell Wildlife Refuge is sincerely appreciated. This is UCMP contribution no. 1937. LITERATURE CITED Bennett, S. C. 1993. The ontogeny of Pteranodon and other pterosaurs. Paleobiology 19:92 106. Brown, B. and E. M. Schlaikjer. 1940. The origin of ceratopsian horncores. American Museum Novitates 1065:1 7. CoBabe, E. A. and D. E. Fastovsky. 1987. Ugrosaurus olsoni, a new ceratopsian (Reptilia: Ornithischia) from the Hell Creek Formation of eastern Montana. Journal of Paleontology 61:148 154. Dodson, P. 1996. The Horned Dinosaurs. A Natural History. Princeton University Press, Princeton, New Jersey, 30 pp. Dodson, P., C. A. Forster, and S. D. Sampson. 2004. Ceratopsidae; pp. 494 513 in D. B. Weishampel, P. Dodson, and H. Osmólska (eds.), The Dinosauria. University of California Press, Berkeley. Forster, C. A. 1993. Taxonomic validity of the ceratopsid dinosaur Ugrosaurus olsoni (Cobabe and Fastovsky). Journal of Paleontology 67: 316 318. Forster, C. A. 1996a. New information on the skull of Triceratops. Journal of Vertebrate Paleontology 16:246 258. Forster, C. A. 1996b. Species resolution in Triceratops: cladistic and morphometric approaches. Journal of Vertebrate Paleontology 16: 259 270. Forster, C. A. and S. D. Sampson. 2002. Phylogeny of the horned dinosaurs (Ornithischia, Ceratopsidae). Journal of Vertebrate Paleontology Supplement 22(3):54A. Hatcher, J. B., O. C. Marsh, and R. S. Lull. 1907. The Ceratopsia. U.S. Geological Survey, Monograph XLIX:1 300.

144 JOURNAL OF VERTEBRATE PALEONTOLOGY, VOL. 28, NO. 1, 2008 Hill, R. V. 2006. Comparative anatomy and histology of xenarthran osteoderms. Journal of Morphology 267:1441 1460. Horner, J. R., and C. Marshall. 2002. Keratinous covered dinosaur skulls. Journal of Vertebrate Paleontology 22(3, Supplement):67A. Horner, J. R., and M. B. Goodwin. 2006. Major cranial changes during Triceratops ontogeny. Proceedings of the Royal Society, Biological Sciences 273:2757 2761. Jarman, P. 1983. Mating system and sexual dimorphism in large, terrestrial, mammalian herbivores. Biological Review 58:485 520. Johnson, R. 1977. Size independent criteria for estimating relative age and the relationships among growth parameters in a group of fossil reptiles (Reptilia: Ichthyosauria). Canadian Journal of Earth Sciences 14:1916 1924. Kardong, K. V. 1998. Vertebrates: Comparative Anatomy, Function, Evolution. WCB McGraw Hill. Boston, 747 pp. Lull, R. S. 1933. A revision of the Ceratopsia or horned dinosaurs. Yale Peabody Museum Memoir 3, 175 pp. Marsh, O. C. 1889. Notice of gigantic horned Dinosauria from the Cretaceous. American Journal of Science, series 3, 38:501 506. Marsh, O. C. 1891. The gigantic ceratopsidae, or horned dinosaurs, of North America. American Journal of Science 41:167 178. Norman, D. B., L. M. Witmer, and D. B. Weishampel. 2004. Basal ornithischia; pp. 325 334 in D. B. Weishampel, P. Dodson, and H. Osmólska (eds.), The Dinosauria. University of California Press, Berkeley. Ostrom, J. H. and P. Wellnhofer. 1986. The Munich specimen of Triceratops with a revision of the genus. Zitteliana 14:111 158. Padian, K., J. Horner, and A. Lee. 2005. Sexual dimorphism in dinosaurs?: a review of the evidence. Journal of Vertebrate Paleontology 25 (2, Supplement):98A. Prum, R. O., R. L. Morrison, and G. R. Ten Eyck. 1994. Structural color production by constructive reflection from ordered collagen arrays in a bird (Phillepitta castanea: Eurylaimidae). Journal of Morphology 222:61 72. Prum, R. O., R. Torres, C. Kovach, S. Williamson, and S. M. Goodman. 1999. Coherent light scattering by nanostructured collagen arrays in the caruncles of the Malagasy asities (Eurylaimidae: Aves). Journal of Experimental Biology 202:3507 3522. Sampson, S. D., M. J. Ryan, and D. H. Tanke. 1997. Craniofacial ontogeny in centrosaurine dinosaurs (Ornithischia: Ceratopsidae): taxonomic and behavioral implications. Zoological Journal of the Linnean Society 121:293 337. Sternberg, C. M. 1949. The Edmonton fauna and description of a new Triceratops from the Upper Edmonton Member; phylogeny of the Ceratopsidae. Annual Report of the National Museum of Canada, Bulletin 113:33 46. Tumarkin-Deratzian, A. R., D. R. Vann, and P. Dodson. 2006. Bone surface texture as an ontogenetic indicator in long bones of the Canada goose Branta Canadensis (Anseriformes: Anatidae). Zoological Journal of the Linnean Society 148:133 168. Vickaryous, M. K. and B. K. Hall. 2006. Osteoderm morphology and development in the nine-banded armadillo, Dasypus novemcinctus (Mammalia, Xenarthra, Cingulata). Journal of Morphology 267: 1273 1283. Submitted January 8, 2007; accepted July 30, 2007.