- The Paleontological Society
A new centrosaurine ceratopsid, Albertaceratops nesmoi, is described from the lower Oldman Formation (Upper Cretaceous) of southern Alberta, and is based on a single, almost complete skull. Referred material is described from equivalent beds in the Judith River Formation of north-central Montana. A limited phylogenetic analysis of the Ceratopsidae places the new taxon as the basal member of the Centrosaurinae and indicates that robust, elongate postorbital horncores that form a synapomorphy of (Ceratopsidae + Zuniceratops) are also present in Centrosaurinae.
The ceratopsidae includes 13 named genera disposed in two subfamilies, the Centrosaurinae and the Chasmosaurinae (Ryan and Russell, 2001). All valid taxa are from the Late Cretaceous sediments of the Western Interior Seaway of North America. All of the ceratopsids are diagnosed almost exclusively on cranial characteristics that are postulated to have undergone rapid modification in the Late Cretaceous as a result of sexual selection (Horner et al., 1992; Sampson, 1995). The Centrosaurinae includes Achelousaurus Sampson, 1995, Avaceratops Dodson, 1986, Centrosaurus Lambe, 1904, Einiosaurus Sampson, 1995, Pachyrhinosaurus Sternberg, 1950, and Styracosaurus Lambe, 1913. Sampson et al. (1997) reexamined the status of Brachyceratops Gilmore, 1917 and Monoclonius Cope, 1876 and declared each a nomen nudum due to their juvenile characters. The centrosaurines include the first named horned dinosaur Monoclonius (=Centrosaurus); next to the hadrosaurids, centrosaurines are the most commonly encountered dinosaur fossils in Canada (Brinkman et al., 1998).
This paper describes a new species of centrosaurine ceratopsid, discovered and collected in the summer of 2001 from the lower Oldman Formation (Fig. 1) of southern Alberta (Fig. 2), based on a single, almost complete, skull, TMP 2001.26.1 (Figs. 3, 4, 6– 8). Additionally, referred material comes from a bone bed in time-equivalent sediments of the Judith River Formation of Montana, approximately 16 km south of the type locality (Fig. 2). Some specimens from this bone bed are in the collections of the Royal Tyrrell Museum and the Wyoming Dinosaur Center. Additionally, disarticulated material from this bone bed was used to assemble a composite ‘chasmosaur’ skeleton now on display at the Fukui Prefectural Dinosaur Museum. This skeleton consists of mostly original postcranial material and a skull incorporating both original and sculpted elements.
The new taxon is distinguished by cranial characters, including the morphology of its nasal, parietal, and postorbital ornamentation. To determine the phylogenetic relationship of this new taxon to other ceratopsids, a restricted phylogenetic analysis of the Ceratopsidae is presented.
American Museum of Natural History (AMNH), Fukui Prefectural Dinosaur Museum (FPDM), Royal Tyrrell Museum of Palaeontology (TMP), Royal Saskatchewan Museum (RSM), and Wyoming Dinosaur Center (WDCB).
Order Ornithischia Seeley, 1888
Suborder Ceratopsia Marsh, 1890
Neoceratopsia Sereno, 1986
Family Ceratopsidae Marsh, 1888
Subfamily Centrosaurinae Lambe, 1915
Genus Albertaceratops new genus
Albertaceratops nesmoi n. sp.
Centrosaurine ceratopsid with low, thick, elongate nasal ornamentation; unmodified adult-sized postorbital horns wide-based, elongated, massive, and oriented rostrolaterally; parietal with pair of wide-based, large, pachyostotic, dorsoventally depressed, curved processes originating from the caudolateral margin of the parietals.
The unfused, adult-sized nasal horncores of Albertaceratops (broken in the holotype) closely resemble the small, juvenile, unfused, nasal horncores of Pachyrhinosaurus but they are tenfold larger. The large parietal processes at the caudolateral margins of the parietal differ from those of Pachyrhinosaurus in being shorter but with longer bases, and having highly rugose surficial texture.
Alberta (name of province where the holotype skull was discovered) + ceratops (horned-face, Latinized Greek).
Albertaceratops nesmoi new species Figures 3–11
As for genus.
A patronym in honor of Cecil Nesmo, a rancher from southern Alberta, whose assistance and hospitality has facilitated the collection of many important paleontological specimens, including the holotype of Albertaceratops n. gen.
Holotype TMP 2001.26.1. An almost complete skull (Figs. 3, 4, 6–8). Type locality, Milk River badlands, south side of Milk River on the Pinhorn Grazing Reserve, approximately 15 km south of Manyberries, southeastern Alberta [UTM; 12 U 510019E 5440905N (datum = WGS 84), Fig. 2]. Lower Oldman Formation, 914 m above sea level (asl), approximately 9 m above the contact with the Foremost Formation.
Other material examined
Referred specimens (Figs. 5, 9–11) come from a bone bed in the lower Judith River Formation near Havre, Montana, along the west side of Kennedy Coulee bordering the Milk River (Fig. 2) (exact locality data are on file with the Royal Tyrrell Museum of Palaeontology). The bone bed occurs at roughly the same elevation as the locality of the holotype skull (899 m asl), and appears to be stratigraphically, chronologically, and lithologically equivalent to the locality of the holotype. The material consists of numerous disarticulated elements (except for at least one partial skull), primarily of adult size, representing almost the entire skeleton. Most of this material is privately owned, but some original material (TMP 2002.69.1–10) and casts (TMP 2002.28–38) are in the collections of the Royal Tyrrell Museum (Alberta), and several elements, including a partial skull (WDCB-MC-001), in the Wyoming Dinosaur Center. Original material from this bone bed was also assembled with fabricated elements into the composite skeleton (FDMJ-V-10) of a “Chasmosaurus sp.” and is now in the collections of the Fukui Prefectural Dinosaur Museum, Japan.
description holotype TMP 2001.26.1
The skull (Fig. 3) is almost complete, with portions of almost all elements preserved; however, many are damaged and/or incomplete. No postcranial elements were recovered. The skull was preserved in a friable, light grey, muddy siltstone. Numerous bones are penetrated or overlain with fossilized Equisetum root traces (up to 10 mm in diameter), some of which have formed deeply inscribed channels. This suggests that the skull lay almost exposed in a wet-to-marshy region, possibly along the quiet margin of a channel, long enough for these plants to take root.
The skull was buried on its left side, which is the best preserved aspect. Portions of skull (including the predentary, both premaxillae, the right maxilla, jugal, squamosal, and partial right dentary and surangular) were displaced up to 0.5 m northeast of the skull. Additionally, the skull has been flattened mediolaterally, and some elements have been somewhat displaced. Although the frill is still in contact with the back of the skull as in life, the lateral parietal rami have been crushed toward the midline. The right side exposes the braincase and some elements of the palate, but these are too poorly preserved to provide significant information. The well-developed parietal and postorbital ornamentation and the highly fused nature of the cranial elements suggest that this specimen is from a large adult-sized animal that still retained some subadult features at the time of death (Sampson et al., 1997; Ryan et al., 2001). Significantly, the preserved nasal (Figs. 3, 4.1) is unfused. Large adult-sized, unfused nasals are also known from the Montanan bone bed (Fig. 5), suggesting that, atypically for other centrosaurines, the nasals of this taxon did not fuse and develop the characteristic adult morphology until the animal had obtained full adult size (Sampson et al., 1997; Ryan et al., 2001). Diagnostic postorbital (Fig. 3) and parietal ornamentation (Figs. 3, 6) had been developed prior to the time of death.
The specimen displays several characters, most significantly the ‘short,’ stepped squamosal (sensu stricto Dodson, 1986; Fig. 7), that refer it to Centrosaurinae, albeit as the basal member of this subfamily.
Much of the rostral and caudal margins of the partial nasal (Fig. 4.1), and the lateral surface, have been lost to erosion. The unmodified smooth bone texture on the unfused flat medial surface suggests that in this taxon fusion of the nasals did not occur until late in ontogeny. The dorsal margin of the nasal is almost complete and has a long-based, low, bladelike ornamentation with a slightly raised rostral section when viewed in profile. This closely resembles the profile of the large, unfused nasals from the Montanan bone bed (Fig. 5), as well as the small, unfused, juvenile-sized nasals of Pachyrhinosaurus. Medially and laterally, the preserved bone textures are consistent with those seen on other subadult-sized centrosaurine nasals (Sampson et al., 1997; Ryan et al., 2001). The medial shelf from which the nasal ornamentation arises is crushed. The rostrally directed narial process at the rostroventral margin of the nasals, diagnostic of centrosaurines, is not preserved on this specimen, but is preserved on some specimens from the Montanan bone bed (e.g., Fig. 5.3, 5.4).
Portions of both premaxillae (Fig. 4.2) were found associated with the skull. Although they appear to be preserved in their life position, they are crushed and flattened together. The rostral margin of the left premaxilla has a distinctive, irregularly pinched surface marking the contact with the caudally projecting ventral flange of the rostral. A small pit on this lateral surface is typical of the rostral contact surface of ceratopsid premaxillae. Caudal to the slightly thickened anterior rostral articulation, the thin narial walls are broken, so it is not possible to determine the presence or absence of either a small secondary opening or a narial strut [present in chasmosaurines but absent in centrosaurines (Lehman, 1990)]. A portion of the right medial narial surface is preserved, with a short portion of the premaxillary shelf for the maxilla, but the element is too crushed to allow accurate comparison with either ceratopsid subfamily. Neither premaxilla has the diagnostically important ascending caudodorsal process preserved.
The shape of the premaxilla suggests that the nares were large and that the rostral was probably elongate and narrow, as it is in other centrosaurines. An isolated caudoventral premaxillary fragment was recovered from the Montanan bone bed and was incorporated into the Fukui specimen, suggesting that the ventral margin of the premaxillae was of the unique centrosaurine shape (App., char.12).
Both maxillae are present and were found displaced several centimeters from their life positions. The left maxilla (Fig. 4.3, 4.4) is the better preserved and appears to be almost completely undistorted, exhibiting typical ceratopsid morphology. Externally the maxilla is roughly triangular in shape and is divided caudally into a dorsal ascending branch and a ventral horizontal branch, the latter of which includes the caudal continuation of the tooth row. It cannot be determined if the rostrodorsal margin of the ascending branch contacts the premaxilla as it does in most other centrosaurines, or if it contacts the nasals, as it does in chasmosaurines and some individual centrosaurine specimens. The tip of the ascending process has a large contact facet for the jugal, and a large antorbital fossa was enclosed by a contribution from the lacrimal. Internally, a medial shelf arches above the tooth row and ends rostrally as the short, rostromedially projecting premaxillary shelf. Caudally the medial shelf narrows and flattens to meet the palatine (not preserved). The caudal margin of the medial shelf has a thickened, convoluted margin that would have supported the ventral margin of the palatine. Externally the maxilla has a modest caudal buccal excavation, less defined than on some other ceratopsid specimens (possibly due to the effects of crushing), which becomes shallower towards the rostral margin. Typically on the caudolateral surface of ceratopsid maxillae the ectopterygoid is evident as either a raised swelling or by the presence of an oval articular facet, neither of which are clearly present on this Albertaceratops n. gen. The lingual surfaces of the exposed teeth exhibit high-angle wear surfaces, typical of centrosaurines (Dodson and Currie, 1990). Both maxillae have 27 tooth positions, typical of other adult centrosaurines of this size, and most of the replacement teeth are present.
Individual ceratopsid taxa can show variation as to the depth, width, and length of the maxilla and the contact its margins make with adjacent elements. In general, however, the form of the maxilla is highly conservative across the Ceratopsidae, and provides no diagnostic information.
Both jugals and squamosals are present, with the left jugal and squamosal (Fig. 3) being complete and connected with the skull, while the right elements are isolated and less complete (Figs. 4.5 and 6, respectively). The squamosal is discussed with the remainder of the frill.
The rostral margin of the left jugal is broken away from the orbital margin at its point of contact with the lacrimal. The margin of its scarf suture with the postorbital is visible on the caudal margin of the orbit. This straight rostral border of the jugal gives it the impression of having a chasmosaurine appearance (i.e., being longer and narrower than the equivalent condition in centrosaurines) but, in reality, the length/width ratio of this element is approximately the same in both subfamilies (Ryan, personal data).
The right jugal (Fig. 4.5) lacks most of its rostral margin (including most of the contact surface for its epijugal) and its proximocaudal area, including the squamosal flange and the contact with the postorbital.
The rostral and caudal margins of the infratemporal fenestra (Fig. 3) are formed, as they are in all ceratopsids, by the jugal and squamosal, respectively. The fenestra is closed ventrally in most centrosaurines and some chasmosaurines by a thin, short projection of the jugal that may (Chasmosaurus belli Lambe, 1914, C. russelli Sternberg, 1940, and Pentaceratops Osborn, 1923) or may not (other chasmosaurines) contact the jugal process of the squamosal. Although both jugals are broken at this point, the breakage point on the left jugal suggests the presence of a short, robust process that could have contacted the squamosal. The evidence for which bone forms the dorsal margin of the infratemporal fenestra is equivocal, but the preserved suture on the right squamosal suggests that, as in the centrosaurines, a slim extension of the jugal overlapped the rostroventral margin of the squamosal to form the dorsolateral margin of the fenestra.
The distolateral apex of the left jugal is thickened and preserves a deep (13 mm), wide (30 mm × 30 mm) pit for the unfused epijugal. Such a pronounced contact for the epijugal is a typically chasmosaurine feature, and is rarely seen on centrosaurines. One isolated skull fragment (Fig. 4.6) might be an epijugal. It has a short (55 mm in height), compressed, conical shape with sharp margins. One surface has a series of wide, convoluted channels similar to those seen on the proximal surface of large epijugals. The element's asymmetrical shape argues against it being an epinasal. It is also unlikely to be an isolated epoccipital, given that the centrosaurine-like shape of the Albertaceratops holotype epoccipitals are well documented from the left lateral parietal ramus.
The quadrate/quadratojugal complex is partially visible on the right side, adjacent to a portion of the left basioccipital (Fig. 3). These elements have been displaced caudally by the crushing of the skull, and are more exposed than they would have been in life. The quadrate and quadratojugal are of typical ceratopsid shape.
The arrangement of the supraorbital (formed by the fusion of the frontal, lacrimal, palpebral, postorbital, and prefrontal in adult-sized centrosaurs) and jugal (the dorsal margin of which forms the ventral and caudal margin of the orbit) is highly conservative in ceratopsids, with the postorbital forming most of the dorsal margin of the orbit, and the lacrimal and palpebral of the supraorbital contributing to the rostrodorsal and rostral margins of the orbit, respectively. Most of the left circumorbital region is preserved on TMP 2001.26.1, with the participating elements remaining intact (Fig. 3).
The most striking feature of the specimen is its long, massive postorbital horncores (Fig. 3), a feature shared with the basal neoceratopsid Zuniceratops Wolfe and Kirkland, 1998 and all chasmosaurines other than Chasmosaurus belli, C. irvinensis Holmes et al., 2001, and C. russelli (Holmes et al., 2001). Although both horncores have been pressed together and displaced caudodorsally, they are not significantly crushed, probably due to their solid construction. Each horncore is complete, lacking only approximately 20 mm from their apices due to erosion, and each has an expanded oval base and tapers towards a blunt tip. The shafts are oval in cross section. Near the base of each horncore, approximately 20 mm above the inner orbit margin, is a distinctly rugose lip, below which each base is slightly constricted. This lip is most pronounced on the lateral and caudal surface of each horncore, and may represent the basal point of attachment of an overlying keratinous sheath during life. At the midpoint on the lateral surface of the right horncore is a shallow (10 mm) oval depression (110 mm × 40 mm), probably resulting from postmortem crushing.
The horncores curve gently rostrally along their distal halves, typical of most long-horned chasmosaurines other than Chasmosaurus mariscalensis Lehman, 1989. In life these horncores would have been inclined rostrolaterally from the skull roof, with the dorsal horncore surfaces approaching the horizontal. A reconstruction of these horncores, and those of the referred material, suggests that the lateral orientation of the horncores appears to exceed that noted for any other ceratopsid with large postorbital horncores.
Both horncores exhibit typical ceratopsid horncore texturing (i.e., shallow to deep longitudinal grooving running for significant distances along their surfaces). Unlike most chasmosaurines with large orbital horncores, the exposed base of the horncore of Albertaceratops (visible on the right side) does not appear to be excavated as an extension of the supraorbital cavity.
The supracranial cavity and the braincase (visible in right lateral view) are poorly preserved and taxonomically uninformative. Exceptionally, the postorbital contact of the right laterosphenoid is well preserved and measures 65 mm × 25 mm. This contact surface is not significantly larger than the largest laterosphenoid contacts measured on numerous Centrosaurus apertus postorbitals in the TMP collections (Ryan, 1992).
The palpebral is a long, narrow bone that forms most of the rostral margin of the orbit. Only the dorsal margin of the right orbit is preserved on the skull, but a portion of its rostral margin is preserved on the isolated right jugal. Sutures on the margins of the left palpebral (Fig. 3) are distinct except medially where it is obscured. Unlike in some other ceratopsids (e.g., Achelousaurus and Einiosaurus Sampson, 1995), the palpebral does not form a rostral shelf or buttress.
Two small bones (not illustrated), tentatively identified as prefrontals, were recovered as isolated elements. These are flattened (approximately 9 mm at their thickest points), with one margin turned up at a right angle to the rest of the element. Each bone is the mirror image of the other and is similar in shape to the element(s) of the frontal region on the articulated Montanan skulls (Fig. 9) and the composite skull of FDMJ-V-10. The presence of a frontal fontanelle on TMP 2001.26.1 cannot be determined, due to the crushing of the skull.
On the left side of the skull, a small portion of the lacrimal, which forms the rostoventral margin of the orbit, is in place, but the remainder has broken away. A portion of the caudal articular facet with the lacrimal is preserved on the partial right jugal. This appears to be wider and deeper than that seen on other centrosaurine jugals, but its actual shape is difficult to determine, given the broken nature of the specimen.
In ceratopsids the frill is composed of the paired squamosals and the parietal whose paired components appear to have fused early in ontogeny. Although crushed, the frill (Figs. 3, 6, 7) is nearly complete, with both squamosals preserved and most of the parietal, except for the proximal portion of the right lateral parietal ramus. The lateral parietal rami are both broken near the caudal corners and have been pushed up against midline ramus, deforming the parietal fenestrae.
Chasmosaurines and centrosaurines can be easily distinguished by the form of the squamosals. All known chasmosaurines have an elongated squamosal in which the caudal portion has an elongated triangular shape in which the squamosal reaches, or almost reaches, the caudolateral corner of the parietal. The lateral chasmosaurine frill margin is formed almost exclusively by the squamosals. All centrosaurines have relatively short squamosals, with an approximately square caudal portion (caudal to the quadrate groove). In centrosaurines the squamosal forms approximately the rostral one-half of the lateral frill margin. Additionally, these squamosals have a distinct caudal portion that is “stepped-up” (s.s. Dodson, 1986) relative to the rostral portion along the medial margin. The angle formed by the step between the rostral and caudal portions also contains a continuation of the quadrate groove from the ventral to the dorsal surface—a feature not currently known for any chasmosaurines. Avaceratops is the only centrosaurine lacking this stepped up medial margin and the continuation of the quadrate groove seen on other centrosaurines.
Both squamosals of TMP 2001.26.1 (Figs. 3, 6) are of typical centrosaurine shape, having rostral and caudal portions that are subequal in length, a stepped-up medial margin with a dorsal continuation of the quadrate groove, meeting the rostral margin of the lateral parietal rami at a thick, concave-shaped half of a butt suture, and forming approximately one-half of the lateral frill margin.
Although the caudomedial margin is partially broken away on the right squamosal (Fig. 6), the width of its parietal contact has the typical ovoid centrosaurine shape, with a length of 190 mm and a maximum width of 22 mm adjacent to its caudal margin. Each squamosal had at least four pronounced marginal scallops, but, unfortunately, each is incomplete (the caudolateral margins of all ceratopsid squamosals are at least slightly scalloped and may have an associated epoccipital. As for the parietal, the squamosal scallops are sequentially numbered, progressing rostrally from the caudomedial corner). On the left the second and third scallops bore fused epoccipitals (Fig. 3), although the complete shape of these cannot be determined. Each squamosal clearly shows the open suture for the epoccipital that straddled the parietal/squamosal contact, as it does in all centrosaurines and the chasmosaurine Triceratops Marsh, 1889. On the right squamosal two scallops are lateral to this suture, but the caudolateral corner is broken away. The second scallop has a small fragment of bone attached in the position of an epoccipital, but this could be an unrelated fragment of bone. The surface of the third scallop, although partially eroded, appears to have been capped with a low, long-based, crescent-shaped epoccipital. Maximum thicknesses of scallops 1–3 are 21 mm, 17 mm, and 17 mm, respectively.
Each squamosal has a saddle-shaped caudal portion, with the marginal scallops being rostrally inflected. As on most ceratopsid squamosals, three raised bumps sit, respectively, at the midpoint of the rostral portion, over the paroccipital process/quadrate groove, and rostrodorsal to the last scallop. Together these three bumps and the one on the caudal postorbital form a low, raised arc on the surface of the squamosal. Contra Penkowski and Dodson (1999), these bumps are variably developed on both centrosaurine and chasmosaurine squamosals, and are not an autapomorphy of Avaceratops (personal observation).
Rostrally, as in other centrosaurines, the squamosal ends in a thin, squared-off margin that overlaps the caudal portion of the postorbital, and contacts the jugal along its ventral margin. The rostroventral margin of the squamosal, including its contribution to the infratemporal foramen, is complete on the left squamosal. The jugal flange is 115 mm in length and tapers to a blunt point, as it does in all ceratopsids. It has a basal width of approximately 75 mm. Of note, the dorsal surface of the squamosal immediately caudal to the jugal process is perforated by a small, ovoid foramen which may have been closed across the ventral edge. This resembles the lesions described for other ceratopsid squamosals (Forster, 1996) and appears to have a small drainage channel opening into it from its dorsocaudal margin. The feature does not appear to be a rehealed horn-induced injury (e.g., no indication of subsequent bone growth is visible), but its origin and function are unknown.
The parietal has a complex shape, with a medial ramus, two caudal rami which run lateral from the midline, and lateral rami which pass rostrally on centrosaurs to meet the squamosal and form the frill margins. The parietal is reconstructed in Figure 7.2. The left lateral parietal ramus (Figs. 3, 7.1), almost completely preserved, has been crushed against the midline ramus. The right ramus (Fig. 7.1) has broken from its point of attachment with the parietal ramus, rotated laterally, and been pushed several centimeters over the caudal midline ramus so as to obscure the latter's dorsal margin.
The midline parietal ramus has the triangular cross-sectional shape and broad dorsal profile common to all centrosaurines, and to chasmosaurines with small parietal fenestrae. The dorsal midline has a distinct midline ridge with several small bumps that are not as prominent as they are on some of the most mature centrosaurines (their expression is highly variable between individuals). Each side of the frill would have had a large, probably oval, fenestra but its actual size and shape cannot be determined from this specimen. The caudal midline ramus margin is U-shaped (Figs. 7, 9.3) as it is in most other centrosaurines with large, laterally curling epoccipitals forming the P3 processes emanating from close to the midline caudal margin.
The left parietal ramus is complete to its squamosal contact on the left (Fig. 3), but the right parietal is only partially complete (Fig. 7.1). No evidence of parietal processes at locus 1 or 2 is visible on the parietal, and they are presumed to have been absent from this taxon (see Sampson et al., 1997 and Ryan et al., 2001, for a complete description of the loci for parietal processes on centrosaurs). The large processes covering most of the caudal and the caudolateral margins of the parietal are partially preserved on both sides on the holotype and are interpreted here as the P3 processes (Figs. 1, 7.1) and have been reconstructed (Fig. 7.2) based on the ornamentation of the parietals recovered from the Montanan bone bed (Fig. 11).
The left P3 process is eroded along the distal and ventral margins, but its overall shape is preserved. Reconstructed in its orientation in life (Fig. 7.2), this process formed a laterally oriented, thickened, slightly rostrally curving hook as seen on the Montanan bone bed specimens (Fig. 11). The base of the process would have covered a significant portion of the caudal parietal ramus. The process is thickest along its dorsal margin and thins towards the ventral and distal margins. The right P3 process shows the same general shape, but has undergone significant flattening via postmortem distortion, with approximately its distal half lost to erosion. Both processes are heavily textured on both surfaces, similar to the Montanan bone bed material, and have deeply inscribed grooves that follow the arc of their midline axes.
The left lateral ramus of the parietal has, in addition to the large pachyostotic P3 process, four elongated, flattened, tab-shaped epoccipitals at loci 4–7 (the process at locus 7 is broken away above the base) and the caudal portion of the squamosal contact (Fig. 7.1). Each of processes P4–7 has a thickened base and tapers to its tip. As seen on other centrosaurine lateral processes, the basal fusion lines of the epoccipitals with the underlying parietal are distinct. These fusion lines appear to be asymmetrical, with the medial margin lying higher up on the underlying bone than the lateral margin. Each epoccipital is separated from its neighbors by a short, saddle-shaped notch similar to those seen between well-developed epoccipitals on other centrosaurines (e.g., Centrosaurus and Styracosaurus). This notch is broken between P3 and P4 on the right side, but clearly would have been present here as well. P4 closely resembles P5–6, but has the distinctive surface texture of P3, suggesting that it was in the process of developing into the modified P4 hook seen on adult specimens from the Montana bone bed. Although the P4–6 processes have been somewhat distorted by crushing, their bases clearly show the same imbrication pattern evident on mature centrosaurines with marginal epoccipitals, with the medial edge being rostrally inflected relative to the lateral.
The right lateral parietal ramus has the P4 and 5 processes partially preserved, both being crushed dorsoventrally and eroded, so that only their general, tablike shape is preserved.
The lower jaw is represented by the ramus of the right dentary, the complete predentary, and a fragmentary left surangular.
The right dentary (Fig. 8.1, 8.2) is missing its rostral portion and coronoid process. The lateral surface is badly eroded but the medial surface is relatively well preserved. The large contact groove for the splenial is almost complete. Although the preserved teeth are badly shattered due to erosion, they are preserved in situ and all have deeply worn, high-angle wear surfaces. The actual number of teeth present cannot be determined.
The predentary (Fig. 8.3) is almost complete, missing only the caudal margin of the caudoventral process. Like the skull, the predentary has been mediolaterally compressed, but its overall shape is preserved. The cutting surfaces of the predentary are steeply inclined, as seen in all centrosaurines, although it is possible that this has been altered by crushing. Of note, the rostolateral margins of the cutting margins each bear a thin, dorsally projecting flange of bone, giving the rostral cutting surfaces a more pronounced groove shape than typically seen on ceratopsids.
other referred material
All other material referred to Albertaceratops n. gen. is from a bone bed on private land in northern Montana, approximately 16.1 km southeast of the holotype locality. Until recently all of this material was in private hands (principally Canada Fossils Ltd., of Calgary, Alberta). Some of this material (including casts of selected original material) now resides in the collections of the Royal Tyrrell Museum and the Wyoming Dinosaur Center, and is available for examination. Additionally, at least one composite skeleton (FDMJ-V-10) was assembled by Canada Fossils from the disarticulated bone bed material as a chasmosaurine ceratopsid of uncertain affinities (most of the diagnostic parietal material was not used in this composite) and is on display in the Fukui Prefectural Dinosaur Museum in Japan. Due to the nature of the construction, the individual original skull elements have been fiberglassed into place and are, for the most part, not easily distinguishable from the adjacent reconstructed portions.
The material collected from the Montanan bone bed includes multiple examples of most elements, including postcrania. However, other than the mount of FDMJ-V-10 assembled from elements from the bone bed, no articulated postcranial material is available for study. A study of the postcrania of Albertaceratops is in progress. The present description is limited to original or cast material that has been catalogued in a recognized, publicly accessible institution.
The most complete specimen of Albertaceratops from the Montanan bone bed is a partial skull lacking the frill (WDCB-MC-001; Fig. 9) in the collections of the Wyoming Dinosaur Center. The skull (total preserved length 670 mm) is from an adult-sized, mature individual, based on the fusion of the nasals, the supraorbital pitting on the remodeled right orbital horncore (see Sampson et al., 1997 for a full discussion of this age related centrosaurine feature), and the lack of any visible sutures anywhere on the skull. The specimen comprises the dorsal portion of the nasals and the dorsal skull roof, including the right, and the base of the left, orbital horncore.
The most complete rostral is now incorporated into FDMJ-V-10. Fragments of three specimens were used to form this rostral, including a partial rostral lacking the dorsal and ventral margins, a partial ventrocaudal flange fused to a partial left premaxilla, and a partial dorsal rostral margin fused to the dorsal margin of a partial premaxilla. Taken together, these pieces suggest a large, elongate rostral of chasmosaurine affinities, but this shape is equivocal.
All premaxillae from the bone bed are fragmentary. The most complete example preserving a ventral angle diagnostic of centrosaurines has been incorporated into FDMJ-V-10. TMP 2002.69.9 is a fragmentary specimen preserving a Centrosaurus-like ascending process with an expanded rostrodorsal margin that underlies the diagnostic centrosaurine narial process of the nasal.
The fused nasal on WDCB-MC-002 and WDCB-MC-001 (Fig. 5.1, 5.2) preserves the complete adult-sized fused nasal ornamentation morphology, with neither showing any evidence of a medial suture. The ornamentation takes the form of a long-based, low, rounded boss. The rostral surface of the ornamentation is somewhat flattened and ascends sharply from the rostral base, thickens above the caudodorsal margin of the nares, and gently tapers to the contact with the frontal/prefrontal. The dorsal surface of the fused nasals on WDCB-MC-001 is rounded and each nasal bears two low humps when viewed in lateral profile that roughly separate the rostral two-thirds from the caudal one-third; these humps are not present on the isolated WDCB-MC-002. When viewed dorsally, the perimeter of the base of the ornamentation is an elongated oval (Fig. 5.2), clearly showing that as the animal matured the nasal ornamentation fused and inflated mediolaterally (as it does in all centrosaurines). The lateral nasal walls below the ornamentation are pinched together, forming a distinct proximal base.
Portions of the dorsal narial margins are preserved on WDCB-MC-002 and on the nasal incorporated into FDMJ-V-10. In both cases the ornamentation sits over the caudal half of the narial margin and extends caudally back to the level of the prefrontals, as do the enlarged nasal bosses on Achelousaurus and Pachyrhinosaurus.
The remaining isolated nasals (TMP 2002.03.37, Fig. 5.3, 5.4; TMP 2002.3.32, Fig. 5.5–5.7; and TMP 2002.3.31, Fig. 5.8, 5.9) are of adult size, but unfused, indicating that, unlike the situation in other centrosaurines, and as seen on the holotype skull, these paired elements did not fuse until late in life. Each of these nasals has ornamentation that takes the shape of a long-based, fin-shaped blade with a sharply angled rostral margin and a more gently sloping caudal margin. TMP 2002.3.32 is complete enough to preserve the bilobed profile seen also on the fused nasal of WDCB-MC-001. The almost complete right nasal TMP 2002.3.37 (total length = 292 mm) has a blade of 235 mm in basal length and is 72 mm at its maximum height, giving it a basal length to height ratio of approximately 3:1. In adult-sized, fused nasals this ratio increases and varies from 4–5:1, giving Albertaceratops the longest basal length to height ratio for any centrosaurine nasal ornamentation. Maximum height is located along the rostral one-third of the blade, just over the caudal margin of the external nares. Each of the unfused nasals has blades ranging in maximum thickness (through their bases) between 14 to 24 mm. The dimensions of the unfused nasals indicate that the diameter across the snout on the living animal was not wide.
The ornamentation of each nasal is heavily eroded medially, but appears to have a flat, slightly textured surface typical of that seen on the pre-fusion nasals of other centrosaurines. Below the base of the ornamentation, the unfused Albertaceratops nasal is of typical centrosaurine morphology, both laterally and medially. TMP 2002.03.37 (Fig. 5.3, 5.4) is complete enough ventrally to preserve a significant portion of the diagnostic centrosaurine narial process.
TMP 2002.3.32 (Fig. 5.5–5.7) preserves a significant portion of the diagnostic nasal ornamentation. On the medial side of this region is a 50 mm suture, presumably for contact with the other nasal. Between this and the lateral surface of the nasal is a large (36 mm × 52 mm), shallow cavity. The ventral margin is formed by the caudal edge of the midline shelf, and the dorsal margin has a smooth, rounded margin like that seen on the rostolateral margins of centrosaurine frontal fontanelles. It is possible that this structure may represent a portion of the frontal and/or prefrontal, and that the cavity represents the rostralmost margin of the frontal fontanelle. Unfortunately no sutures are visible in this region to clarify this point.
The jugal is also poorly represented from the bone bed. The best-preserved isolated specimen is TMP 2002.3.33 (Fig. 10.1). It has the typical triangular shape of a ceratopsid jugal, but is otherwise undiagnostic. Of note is what appears to be the broken base of the lower infratemporal ramus on the lower caudal margin, suggesting that this jugal contributed to the lower infratemporal margin, as it does in all centrosaurines (Avaceratops excluded) and Chasmosaurus Lambe, 1914.
The circumorbital region and the dorsal skull roof are best preserved on FDMJ-V-10, WDCB-MC-001 (Fig. 9), the subadult-sized postorbital horncore TMP 2002.69.10 (Fig. 10.4, 10.5), TMP 2002.3.35, and the isolated postorbital horncore TMP 2002.3.36 (Fig. 10.6). The postorbital horncores incorporated into FDMJ-V-10 and TMP 2002.3.36 closely resemble the size and shape of those of the holotype. The subadult, right postorbital horncore, TMP 2002.69.10, is large and robust for a specimen of this size (total height of 100 mm, estimated basal length and thickness of 50 mm and 60 mm, respectively) and more closely resembles the postorbital of subadult Triceratops (e.g. RSM P2299.1, Tokaryk, 1997: fig. 1; RSM P2623.1) than that of any other known centrosaurine. The horncore has an oval base and rises from above the rostrodorsal surface of the orbit. It projects dorsorostrally and has a slight medial arch reminiscent of adult postorbital horncores. The frontal suture is visible along the medial surface as a dorsally arched concavity with a highly convoluted contact surface.
The skull WDCB-MC-001 is notable for having a short, massive right horncore (Fig. 9; only the base of the left is preserved) that projects dorsolaterally from the skull roof, with a modest rostral inflection. The medial base of the right horncore merges smoothly with the skull roof, while the lateral base forms the dorsal margin of the orbit and is confluent with it. The apex of the horncore has four distinct supraorbital pits (diameters = 16 mm × 12 mm; 28 mm × 17 mm; 14 mm × 10 mm; and 13 mm × 9 mm) (Fig. 9.1, 9.2), but the base has a similar circumference to that of the holotype skull, suggesting that this horncore has been reduced by more than half of the original length.
This region comprises the dorsal skull elements (principally the frontal, prefrontal, rostral parietal, and dorsal braincase) that surround the supracranial cavity. WDCB-MC-001 preserves a small portion of the U-shaped (centrosaurine) margin of the frontal fontanelle (Figure 9.1–9.3), but it is impossible to distinguish the separate contributions of the frontal or prefrontal to it (or to the dorsal skull roof). Due to the massive vaulting of the frontal, the region between the caudal nasal ornamentation and the rostral margin of the frontal fontanelle rises abruptly.
The vaulted frontals are massive in WDCB-MC-001 with the height of the supracranial cavity being at least 100 mm. The cavity has the typical ceratopsid arrangement of being subdivided into three smaller chambers within each frontal, each separated by thin vertical walls of bone that buttress the dorsal frontal roof and the overlying postorbital horncore. As seen on ceratopsids possessing large postorbital horns, the supracranial cavity of WDCB-MC-001 extends into the lower 100 mm of the base of the orbital horncore (this region is not preserved on any other Albertaceratops specimen). This is unusual when compared to all other centrosaurines which show no excavation of the base of the horncore (but have relatively modest horn development), but the excavation is quite small when compared to those of chasmosaurines like Triceratops that can have up to one-half of their lengths (several hundred millimeters) hollowed out.
The frill is represented by numerous uncatalogued small fragmentary pieces of parietal and squamosal, including one complete right lateral parietal ramus (TMP 2002.3.28; Fig. 11.1, 11.2). No complete caudal parietal rami are known, nor have any portions of the midline parietal ramus been retrieved from the Montanan bone bed.
Squamosals are poorly represented from the Montanan bone bed. The most complete is a large centrosaurine-like specimen lacking most of the rostral process that was incorporated into FDMJ-V-10 and modified to resemble a chasmosaurine squamosal. A slightly smaller specimen (TMP 2002.69.10, estimated total length 400 mm; Fig. 10.2, 10.3) shows the same overall structure. Although it only bears two well-developed, long-based caudal epoccipitals it does possess the elongate, concave suture for the parietal that covers the caudomedial surface of the squamosal. Other fragments from this locality (e.g., TMP 2002.69.2, 2002.69.3, 2002.69.4) closely resemble the squamosals incorporated into FDMJ-V-10.
The most complete parietal, TMP 2002.3.28 (Fig. 11.1, 11.2), is an adult-sized right lateral ramus and, unlike that of the holotype, is uncrushed. It has a large, dorsoventrally depressed, laterally directed, and rostrally curved, hooklike process at putative locus 3 (P3) that closely resembles the left P3 process of the holotype. This process on both specimens resembles a shorter, pachyostotic version of the parietal locus 3 (P3) spike on the unpublished Pachyrhinosaurus-like material from Grande Prairie, Alberta (e.g., TMP 86.55.239, 87.55.164, and TMP 89.55.1033). The P3 process originates dorsally from a thickened point of fusion with the underlying parietal that extends medially onto the caudal margin. Ventrally, no distinct point of fusion for the process is visible, which instead grades progressively into the smooth adult bone texture. The P3 process is thickened along the caudal margin and becomes progressively thinner along the rostral and lateral margins towards the terminus. Portions of the lateral margin have been broken or eroded away, but, if complete, would have rounded out this side to form the stout hook (as on Fig. 7.2). The P3 hook is separated from the P4 process by a rounded, saddle-shaped surface. The P4 process is a smaller version of the P3 hook and, as on the P3 process, the lateral margin is broken away. The point of basal fusion of the epoccipital is not clearly defined on either surface. The P5 process is a small, compressed, triangular epoccipital sitting on the margin of the parietal. It is distinctly offset from the P4 process by a small, rounded notch and sits next to a large, rounded marginal scallop that is confluent with the squamosal contact. The midpoint of the base is penetrated by a small, 4 mm wide foramen.
A 150 mm long squamosal suture along the rostral margin of the ramus is similar in shape to that of the holotype. In both, the squamosal contact suture extends approximately 50 mm onto the lateral surface, indicating that an epoccipital fit into the marginal gap formed between the parietal and the squamosal.
TMP 2002.3.28 (Fig. 11.1, 11.2) is the only bone bed parietal to preserve any of the caudal margin medial to the P3 process. A 110 mm section of the caudal ramus extends medially from the base of the P3 process. The caudal surface of this section is rounded, and the dorsal and ventral surfaces taper rostrally. No indication of the U-shaped caudal margin of the parietal midline seen on the holotype is visible, and this short section of caudal parietal more closely resembles the straplike caudal parietal rami of Chasmosaurus belli and C. russelli. However, given the large size of the specimen, it is likely that it only preserves a section of the caudal parietal ramus adjacent to the midline parietal ramus.
The remaining catalogued parietal specimens (TMP 2002.03.38, Fig. 11.3, 11.4; TMP 2002.69.5, TMP 2002.69.6, TMP 2002.69.7, and TMP 2002.3.29) are all from adult-sized specimens and preserve some portion of the putative P3 or P4 processes. A portion of a lateral parietal ramus similar to that of TMP 2001.26.1, preserving P5–7, was incorporated into the caudal parietal ramus of FDMJ-V-10. This region closely resembles the same section of the holotype in size, shape, and degree of imbrication of the parietal processes.
Dentaries (e.g., TMP 2002.69.8; Fig. 10.7) are one of the most common cranial elements collected from the Montanan bone bed, and include many uncatalogued specimens, as well as the pair incorporated into FDMJ-V-10. The elements are unremarkable except that the largest specimens (e.g., FDMJ-V-10) have estimated total lengths of approximately 500 mm, exceeding the length of the largest dentaries of other Campanian centrosaurines (e.g., Centrosaurus, TMP 79.11.14, 410 mm [Ryan, 1992]; Pachyrhinosaurus, TMP 89.55.57, 415 mm). The largest dentaries have rugose texturing along their ventral margin that can extend up to 20 mm onto the lateral and medial surfaces. A similar texture is evident on the largest Centrosaurus dentaries, but is only comparatively lightly inscribed on these. This texture appears to be related to ontogenetic development, being deeper and more pronounced on larger specimens, just as similar texturing is more pronounced on larger ceratopsid parietals and squamosals. This is most notable in the development of deep vascular texturing seen on the surfaces of large ceratopsids such as Triceratops (Forster, 1996), that also becomes more pronounced during ontogeny.
To resolve the relationship of Albertaceratops n. gen. within Ceratopsidae, a phylogenetic analysis was undertaken that includes all recognized large-bodied ceratopsid genera (with excluded taxa outlined below). The entire Ceratopsidae has been the subject of few cladistic analyses (Sereno, 1986; Pisani et al., 2002—genus-level supertree analysis; Dodson et al., 2004); most analyses have been limited to some subset of the group [(basal ceratopsians—Chinnery and Weishampel, 1998; Sereno, 2000; Xu et al., 2002); Centrosaurinae—Sampson, 1995; Penkowski and Dodson, 1999; Chasmosaurinae or portions thereof—Forster et al., 1993; Forster, 1996; Lehman, 1996; Holmes et al., 2001]. Dodson and Currie (1990) and Wolfe and Kirkland (1998) presented cladistic analyses of the Neoceratopsia, but without published data sets.
Excluded from this study are the centrosaurine genera Monoclonius and Avaceratops, and the chasmosaurine Diceratops Lull, 1905. Monoclonius was designated a nomen nudum by Sampson et al. (1997), based on the unequivocal juvenile status and lack of diagnostic adult characters, and thus not considered further. Avaceratops has been problematic since its description by Dodson in 1986. Although it is a valid centrosaurine taxon, based at least on one apomorphy of the squamosal, it is either missing key diagnostic cranial characters (postorbital and nasal), or lacking diagnostic characters (adult parietal ornamentation) due, at least in part, to its putative subadult size. Until more complete, adult-sized material can be unequivocally referred to this taxon I follow the example of Sampson (1995) and set this taxon aside from analysis. The taxon Diceratops is based on a single skull (USNM 2414) lacking lower jaws or postcrania, and shares characters with both Torosaurus Marsh, 1891 and Triceratops. It has been briefly described by Forster (1990, 1996), who asserted its validity, yet formal diagnosis remains. It is set aside until a more complete study can be made of it.
The taxon Chasmosaurus has four recognized species, C. belli, C. irvinensis, C. mariscalensis, and C. russelli, and is dimorphic for postorbital horncore length (short in C. belli and C. russelli; long in C. mariscalensis). In this analysis, Chasmosaurus has been coded as a combination of C. belli, C. russelli, and C. mariscalensis. C. irvinensis has been excluded due to its autapomorphic parietal ornamentation and its equivocal lack of postorbital horns. The taxa Torosaurus and Triceratops have two recognized species, but they cannot be differentiated using the characters included in this analysis.
To construct a hypothesis of the phylogenetic relationships within the Ceratopsidae, 36 cranial characters of 14 ingroup taxa, plus two outgroups (Protoceratops Granger and Gregory, 1923 and Zuniceratops) (Table 1) were scored for cladistical analysis. In addition to Albertaceratops these ingroup taxa include the previously recognized centrosaurines Achelousaurus, Centrosaurus, Einiosaurus, Styracosaurus, Pachyrhinosaurus, and the previously recognized chasmosaurines Anchiceratops Brown, 1914, Arrhinoceratops Parks, 1925, Chasmosaurus, Pentaceratops, Torosaurus, and Triceratops. Postcranial characters are excluded because 1) the two ceratopsid subfamilies are diagnosed almost exclusively on cranial characters; 2) complete postcranial material is even less well known for some taxa than is the cranial material; and 3) postcranial characters have proven to be of limited utility in higher level ceratopsid diagnoses (Chinnery, 2002). Characters used in this study include those taken or adapted from previously published works (i.e., Sereno, 1986; Forster et al., 1993; Sampson, 1995; Forster, 1996; Holmes et al., 2001), as well as newly defined characters.
In this analysis 13 characters (5, 9, 10, 14–16, 24, 27–32) are multistate. Traditionally, such characters are avoided because they are often difficult to polarize and order (Sampson, 1995). However, as variation among ceratopsids is principally contained within the cranial ornamentation, these apomorphies are usually difficult to subdivide into valid separate characters in such a way that they are either completely independent of each other or so that they do not result in the addition of a large amount of missing data to the data set if they are excluded.
All characters were optimized using both the delayed and accelerated transformation options, and all characters were run unordered. The data matrix was analyzed using the Branch-and-Bound option in Phylogenetic Analysis Using Parsimony (PAUP, version 4.0b10 for Windows). The data set was subjected to a bootstrap analysis of 1,000 iterations using all characters subjected to 50% resampling to assess the support for the generated clades, and a Bremer decay analysis to determine the number of additional steps required to collapse each node.
The analysis produced a single most parsimonious tree (Fig. 12) of 66 steps [consistency index (CI) = 0.818; retention index (RI) = 0.885; rescaled consistency index (RCI) = 0.724]. The analysis supports monophyly of Ceratopsidae, Centrosaurinae, and Chasmosaurinae. The basal neoceratopsid Zuniceratops is the sister taxon of the Ceratopsidae. Albertaceratops n. gen. is the least derived centrosaur and is defined by four ambiguous characters (91, 300, 312, and 322) and one unambiguous character (102) using DELTRAN optimization. The sister group to Albertaceratops [Centrosaurus + (Styracosaurus + (Einiosaurus + (Achelousaurus + Pachyrhinosaurus)))] is supported by four unambiguous characters (141, 161, 171, and 292).
Albertaceratops is unequivocally placed within the Centrosaurinae, requiring eight steps to remove it from this clade and a further nine steps to place it within the Chasmosaurinae. Bremer support (three steps) for the Chasmosaurinae node is the largest recovered from this analysis.
Albertaceratops nesmoi n. sp. (Fig. 3) represents a new basal centrosaurine ceratopsid from the lowermost Oldman Formation of Alberta, Canada, diagnosed by its cranial ornamentation. Referred material comes from a time- and lithologically equivalent horizon 16 km to the southeast of the holotype locality in the Judith River Formation of Montana. The holotype skull, TMP 2001.26.1, is referable to the Centrosaurinae based on the shape of the squamosal, which is shared by all centrosaurines (except Avaceratops), the ornamentation of the lateral parietal ramus, and the shape of the cutting margin of the predentary. The inferred morphology for incomplete elements such as the nasal, premaxilla, and jugal also supports this hypothesis, which can only be tested by the discovery of additional complete skulls.
With the basal neoceratopsian Zuniceratops (Wolfe and Kirkland, 1998), and all chasmosaurines other than three of the four species of Chasmosaurus (C. belli, C. mariscalensis, and C. russelli), this new taxon shares the plesiomorphic character state of large, massive postorbital horns, making this the first occurrence of this character state in the Centrosaurinae.
The new Montanan Albertaceratops bone bed material was first noted by Sweeney and Boyden (1993), who suggested that the material represented the southernmost occurrence of Styracosaurus albertensis Lambe, 1913 based on the misidentification of the postorbital horns as parietal spikes. When the postorbital horncores were later correctly identified, Trexler and Sweeney (1995) noted the similarity of the cranial material to that of type material of “Ceratops montanus” Marsh, 1888. Marsh (1888) described the new genus and species “Ceratops montanus” based on an occipital condyle and a pair of large chasmosaurine-like postorbital horncores (USNM 2411). Although the exact locality of this specimen is not known, Marsh (1888) reported it to be from the top of the Judith River beds in the Cow Creek Valley, about 16 km upstream from the confluence of this stream with the Missouri River. Stratigraphically it is approximately equivalent to both the Albertaceratops type locality in the Oldman Formation of Alberta and the Albertaceratops bone bed in the equivalent Judith River Formation of northern Montana.
Both orbital horncores of Ceratops montanus are long (each being approximately 240 mm long) and robust, and, when oriented in a life posture, bear some resemblance to those of Albertaceratops n. gen. in being more laterally than rostrally directed from the skull. Unfortunately isolated horncores have limited utility for diagnosing taxa. All Chasmosaurinae, with the exception of Chasmosaurus belli, C. irvinensis, and C. russelli, have robust orbital horncores, and individuals of some taxa can show a pronounced lateral inflection of the horns. Given that the holotype material of Ceratops montanus lacks diagnostic features it must remain a nomen dubium.
Penkowski and Dodson (1999) also considered the relevance of Ceratops montanus to their cf. Avaceratops skull (MOR 692); however, they limited their discussion to referred Ceratops squamosal material and did not discuss the horncores. The referred Ceratops squamosals (Hatcher et al., 1907, pl. a, fig. 1) closely resemble those of Avaceratops, and are probably referable to this taxon, but their association with the holotype of Ceratops montanus is tenuous as they were collected “many miles from the locality that furnished the type” (Hatcher et al., 1907, p. 102). MOR 692 does share the presence of large orbital horncores with Albertaceratops, but the thickened margins of the frill lacks well-developed epoccipitals, indicating that this skull is not referable to Albertaceratops.
Albertaceratops material from the Montanan bone bed reveals a number of interesting features of key elements. The unfused gracile and fused robust, massive nasals each appear to represent different stages of an ontogenetic growth series. The smallest, unfused nasals are not significantly smaller than the largest fused nasals. This suggests that, as is seen on TMP 2001.26.1, they did not fuse until late in life, after the parietal ornamentation had developed. The ornamentation of the unfused nasals is low and long-based, and almost identical in morphology to that of the smallest unfused Pachyrhinosaurus nasals. Both share a slight bilobed dorsal margin that is mirrored in the largest fused Montanan nasals. Similar to the condition in Pachyrhinosaurus, the fused nasals of adult Albertaceratops do not form an erect horn but rather take on the unique shape of a long, low boss. Although this ornamentation differs markedly from that of Pachyrhinosaurus, it more closely resembles that taxon than any other centrosaur.
Unfortunately the complete nares are not preserved on any of the specimens, and on only one specimen (TMP 2002.3.37, a large unfused nasal) is a portion of the diagnostic centrosaurine narial process preserved. Based on this latter specimen the remainder of the nasals has been inferred as being centrosaurine.
Both TMP 2001.26.1 and the Montanan bone bed material share a large, robust horncore morphology. Of note is one subadult postorbital (TMP 2002.69.1) known from the bone bed. This is of typical chasmosaurine shape and has been so scored in the cladistic analysis. If it had been recovered as an isolated element it would no doubt have been catalogued as chasmosaurine.
The Montanan parietal material is problematic as no complete specimens are known. Frustratingly there are also no complete caudal parietal rami and no recognized fragments of the midline parietal ramus. Consequently it is impossible to determine the shape of the midline ramus or the presence or absence of any medially positioned caudal parietal processes (P2). The presence of a thin, straplike caudal and/or midline ramus would necessitate the coding of this material as having chasmosaurine character states, making the centrosaurine referral of this material more questionable. Notably, each of the catalogued parietals has lateral rami that have a total of only three lateral processes (P3–P5) rather than the at least five (P3–P7) present on the Albertaceratops holotype skull. The lateral processes of TMP 2001.26.1 are all of typical centrosaurine shape, being (with the exception of the hooklike P3 process) laterally compressed, tab-shaped epoccipitals. The P4–7 processes also show the typical centrosaurine imbrication of the epoccipitals. The most complete parietal from the Montanan bone bed, TMP 2002.3.28, clearly shows only three lateral processes: 1) the large pachyostotic hook at the caudolateral corner of the parietal; 2) a smaller hooklike process rostral to it; and 3) a small, triangular chasmosaurine-like epoccipital adjacent to the squamosal contact. All chasmosaurines with well-developed parietal processes (Anchiceratops ornatus Brown, 1914, Chasmosaurus belli, C. russelli, C. mariscalensis, and Pentaceratops sternbergi Osborn, 1923) have only three parietal processes, suggesting that the Montanan bone bed parietals should be referred to the Chasmosaurinae. The exception to the chasmosaurine rule is the anomalous Chasmosaurus irvinensis, which has at least five fused epoccipitals that line the dorsal margin of the caudal rami, suggesting that the number of parietal processes within the Chasmosaurinae is variable. However, arguing against the chasmosaurine nature of TMP 2002.3.28 is the shape of the rostral margin of the lateral parietal ramus that forms the squamosal suture. This centrosaurine-like suture runs almost parallel to the caudal margin and gives the specimen a typical centrosaurine profile. Additionally, the angle of the squamosal contact indicates that the squamosal did not reach the caudolateral margin of the parietal as it does in all chasmosaurines. Indeed, all the squamosal material from the bone bed appears to have the caudally short, stepped medial margin seen on centrosaurines, and would be referable to this subfamily based on this feature alone.
Several of the smaller, possibly subadult-sized, uncatalogued, parietal specimens owned by Canada Fossils, Ltd., and fragments used to construct the Fukui display mount, appear to preserve the parietal processes P4–P7, supporting a centrosaurine referral. Thus, unless the ceratopsid material in the Montanan bone bed comes from more than one taxon, the number of lateral parietal processes appears to be variable, a feature that is rarely observed in ceratopsid taxa. Given the close proximity of the holotype to the Montanan bone bed, it is unlikely that this difference is a result of regional variation. It is possible that this variation represents sexual dimorphism; however, this is considered unlikely given that such dimorphism has not been conclusively shown for any large-bodied ceratopsid.
Finally, the bone texture on the largest specimens from Montana have what has been previously described as chasmosaurine (i.e., highly vascularized) bone texture, a feature used as a characteristic to diagnose Triceratops (Forster, 1996). This texture is clearly absent from TMP 2001.26.1 and all known centrosaurines. However, based on the material present in the Montanan bone bed, this texturing appears to be a size-related feature which, although present on smaller specimens, becomes prominent on elements larger than those associated with the largest centrosaurines. All the adult material from this bone bed are also massive, and it is probable that the bone texture seen on the Montanan skull elements, in addition to being related to absolute size, is related to the robustness of the specimen.
As the Montanan bone bed was not excavated using rigorous scientific methodology, field notes, and excavation maps, the associated taphonomic data are not available to assist in assessing the recovered material. It is also unclear whether all excavated material was kept or whether some elements were selectively excluded from collection. It is possible that the bone bed preserves the remains of at least two different ceratopsids, and that the fragmentary material presented here only has a superficial resemblance to the holotype skull, TMP 2001.26.1. The unequivocal assignment of ceratopsid material from the Montanan bone bed will only be possible with the collection of more complete specimens in the future.
The analysis presented here supports the monophyly of the Ceratopsidae (node 2) based on two unambiguous characters—351 and 361 (superscripts indicate character states [0–4] as indicated in the character description), but the monophyly of the clade is also well supported by numerous characters excluded from this analysis (e.g., Sereno, 2000). The Centrosaurinae (node 4) is supported by eight characters (Fig. 12), only one of which is ambiguous (131), and the Chasmosaurinae (node 10) is supported by nine characters (Fig. 12), of which three (111, 190, and 280) are ambiguous.
The basal neoceratopsid Zuniceratops is the sister taxon of the Ceratopsidae, supporting the assertion of Wolfe and Kirkland (1998) that was published without phylogenetic analysis.
The Centrosaurinae was first proposed by Lambe (1915) and was rediagnosed by Lehman (1990). Of the eight characters supporting this clade (Fig. 12) only one (presence of the jugal infratemporal flange) is shared with the Chasmosaurine. Albertaceratops is determined to be the least derived centrosaur and is defined by four ambiguous characters (91, 300, 312, and 322), and one unambiguous character (102), the shape of the adult nasal ornamentation. The length (but not the shape) of the nasal ornamentation is shared with Achelousaurus, and orientation of the spike at locus 3 is shared with Pachyrhinosaurus.
The sister group to Albertaceratops is [(Centrosaurus + (Styracosaurus + (Einiosaurus + (Achelousaurus + Pachyrhinosaurus))))] (node 5), which is supported by three or four unambiguous characters (ACCTRAN and DELTRAN, respectively) (Fig. 12). Node 6 is supported by only one ambiguous character (311), while node 7 is supported by one (DELTRAN) or two (ACCTRAN) unambiguous characters (the shape of the subadult postorbital horncore, 142, in the former, and additionally 162, the shape of the unmodified adult postorbital horncore in the latter). Node 8 (Achelousaurus + Pachyrhinosaurus) is supported unambiguously by the shape of the nasal (102) and the postorbital ornamentation (300), and one (DELTRAN) or two (ACCTRAN) ambiguous characters [the basal length of the subadult nasal ornamentation (91) which these taxa share with Albertaceratops, and the shape and length of the parietal process at locus 2, 302].
Within the Chasmosaurinae node 10 is supported by one unambiguous character (the absence of an epoccipital straddling the squamosal-parietal border). Node 11 is supported by one unambiguous character under DELTRAN analysis (the orientation of the parietal epoccipitals), and a further three ambiguous characters under ACCTRAN analysis. The clade (Anchiceratops + Arrhinoceratops) are united at node 13 by a single unambiguous character, the orientation of the narial strut (52). The sister-group relationship of Chasmosaurus and Pentaceratops (node 12) is supported by seven (DELTRAN) or eight (ACCTRAN) characters, of which two (70 and 181, DELTRAN) or three (additionally 320, ACCTRAN) are ambiguous, and confirms the previous findings of Lehman (1996) and Holmes et al. (2001).
Bootstrap support for most clades within this analysis is relatively poor. Only the Centrosaurinae has greater than 95% support over the 2,000 replicates. Bremer support for the recovered tree is not robust, but the Centrosaurine clade (node 4) has a support value of 3, the Ceratopsidae (node 3), the Chasmosaurinae (node 9), (Chasmosaurus + Pentaceratops) (node 12), centrosaurs more derived than Albertaceratops (node 5), and (Achelousaurus + Pachyrhinosaurus) (node 8) each have a support value of 2, while the remaining clades are poorly supported with each having a value of 1, indicating that one additional tree step (tree length = 67 steps) would cause the node to collapse. This is reflective, in part, of the large number of character states of the key cranial characters shared between some chasmosaurines and centrosaurines that alternatively yield polychotomous outcomes due to multiple potential homoplasies. These results support the observation that the ceratopsids are a tight-knit group in which the pattern of relationships between the members can only be teased apart by using a relatively small number of polymorphic cranial characters, primarily those relating to ornamentation of the nasal, parietal, and supraorbital regions. These characters are postulated to have been under a high degree of sexual selection (Sampson, 1997), allowing a large number of taxa to have quickly evolved a wide array of different ornamentations (e.g., the centrosaurines in this analysis span approximately seven million years). Elimination of any of the cranial ornamentation characters quickly reduces the ability to resolve patterns of relationship within the Ceratopsidae (e.g., elimination of the features of the caudal parietal of Centrosaurus and Styracosaurus would make adult specimens of these taxa indistinguishable). While both subfamilies can be distinguished exclusive of their parietal ornamentation, rerunning the analysis without these characters (28–32) yielded a polytomy among the centrosaurines (excluding Albertaceratops) arranged as (Centrosaurus + Styracosaurus + Einiosaurus + (Achelousaurus + Pachyrhinosaurus)), with the latter two taxa being united primarily by the shared presence of nasal and orbital bosses. Given the variation shown within some of these characters (e.g., the extensive remodeling that occurs on the postorbital horncores through ontogeny), and the variability in the development of some parietal ornamentation (e.g., parietal processes 5–7 can be either expressed as short spikes or be unmodified on Styracosaurus), it is not surprising that distinguishing between these taxa has proven difficult. In previous analyses (Sampson, 1995; Penkowski and Dodson, 1999), the coding of cranial characters was relatively straightforward, with little overlap between character states. The addition of Albertaceratops has brought new character states to the analyses, effectively blurring boundaries between certain taxa. The character states of some key characters, such as the shape of the postorbital horncore and various parietal processes, become increasingly more difficult to describe as discrete states.
As new ceratopsids continue to be discovered, it can be predicted that resolution of cladistic patterns will become poorer as new forms fill in the morphological gaps between previously described taxa. Albertaceratops lessens the distinction between the two subfamilies, while other recent discoveries, such as Einiosaurus, bridge the parietal and postorbital morphologies between Pachyrhinosaurus and Styracosaurus. The repeated pattern seen in new ceratopsid discoveries is an increase in variation within the key cranial characters without significant changes within other skeletal features. Compounding this problem is that some taxa, such as Arrhinoceratops, are only known from limited material that will require repreparation to clarify key characteristics.
1. The Ceratopsidae, the Centrosaurinae, and the Chasmosaurinae are all supported as monophyletic clades. Zuniceratops is supported as the sister taxon of the Ceratopsidae. Bootstrap support for most clades is relatively weak, with all clades but the Centrosaurinae being supported well below a 95% confidence level. This is probably reflective of the difficulty in establishing discrete and distinctive character states for most taxa. Homoplasy appears to be extensive. Most sister clades show variation among only a small set of cranial characters that display conspicuous homoplasy and reversal, even within localized areas of the tree.
2. Within the Centrosaurinae, Albertaceratops n. gen. is the basalmost centrosaurine. Its large postorbital horncores make this character state a synapomorphy of the Ceratopsidae.
Characters included in analysis
Comments are included for characters new to, or modified for, this analysis.
1. Rostral, size and shape (Sereno, 1986)
(0) triangular in lateral view, with short dorsal and ventral processes
(1) elongate, with deeply concave caudal margin and hypertrophied dorsal and ventral processes
2. Premaxillary septum shape
(1) rostrally elongate
5. Narial strut of premaxilla (after Holmes et al., 2001)
(1) present, rostrally inclined
(2) present, caudally inclined
The thickened caudal portion of the premaxillary septum is found in all chasmosaurines but is absent in outgroup taxa. The strut is rostrally inclined in Anchiceratops and Arrhinoceratops.
(0) absent, caudoventral margin of premaxilla unexpanded and level with alveolar margin of maxilla
(1) present, expanded ventrally to extend well below alveolar margin of the maxilla
9. Nasal ornamentation, basal length (subadult; Sampson et al., 1997)
(0) short-based, restricted in length rostrocaudally
(1) long-based, ornamentation covers almost the entire length of the nasal
10. Nasal ornamentation type (adult; after Sampson, 1995)
(0) absent or poorly developed
(1) elongated horn
(2) long-based, low thickened ridge
All basal ceratopsians previously classified as protoceratopsids lack nasal ornamentation, except for some specimens of Protoceratops that show a bilateral dorsal swelling on the dorsal surface of the nasals (Holmes et al., 2001). All centrosaurines develop ornamentation after the paired nasals fuse during ontogeny. All adult chasmosaurines carry a short nasal horn, formed by a separate element (epinasal) in at least some taxa.
(0) centered caudal or caudodorsal to caudal margin of the endonares
(1) centered dorsal or rostrodorsal to the caudal margin of the endonares
12. Narial spine of nasal, a pronounced tablike process projecting rostrally into the nasal vestibule from the caudal narial margin (Sereno, 1986)
This process is formed primarily by the nasal but it can include a contribution ventrally from the premaxilla, forming the caudal margin of the nares.
The jugal forms the lower part of the infratemporal opening and may or may not contact the squamosal. This is modified from Lehman (1996) and Forster (1996), who defined the character as the jugal flange contacting/not contacting the squamosal. Using this definition the character is variable within taxa, but the presence or absence of the flange is not. The character is defined here to reflect the latter condition.
14. Form of postorbital ornamentation (subadult; Sampson, 1995)
(0) conical horncore with rounded base and pointed apex; height at least three times as long as rostrocaudal basal length
(1) pyramidal horncore with approximately a 1:1 ratio of height to rostrocaudal basal length
(2) horncore longer rostrocaudally than high, with rounded apex
15. Supraorbital ornamentation type (adult; Sampson, 1995)
(1) present, horn
(2) present, boss
16. Postorbital horncore shape (unmodified adult; after Sampson, 1995)
(0) elongate with pointed apex and round to oval base
(1) pyramidal with rounded apex, at least as tall as base is long
(2) rounded apex, base longer than horn is tall
17. Postorbital ornamentation height (unmodified adult; after Holmes et al., 2001)
(1) short, less than 40% length of face
(0) long, greater than 60% length of face
The postorbital horncores of both centrosaurines and chasmosaurines can exhibit reduction in length as a result of the development of supraorbital pitting (e.g., in Centrosaurus this pitting can completely eliminate the horn). All chasmosaurines except Chasmosaurus belli, C. irvinensis, and C. russelli have long postorbital horncores. C. belli and C. russelli are the chasmosaurine taxa that most frequently exhibit supraorbital pitting, but even unmodified specimens have horncores that are significantly shorter than other chasmosaurines. The lack of postorbital horns on C. irvinensis is equivocal, as they may have been lost as the result of supraorbital pitting. This character has been scored as dimorphic for Chasmosaurus.
(0) caudal to orbit
(1) over or rostral to orbit
19. Postorbital horncore curvature (adult; after Holmes et al., 2001)
(0) straight, dorsally, rostrally or rostolaterally procurved (not recurved)
20. Length of squamosal relative to parietal (Sereno, 1986)
(0) equal or subequal in length
(1) squamosal less than 60% total parietal length
(1) caudal portion stepped-up relative to rostral portion (after Dodson, 1986)
All centrosaurines (excluding Avaceratops) have relatively short squamosals on which the caudal portion of the element (caudal to the paraoccipital groove) is “stepped-up” relative to the rostral margin. All chasmosaurines have relatively long squamosals with straight margins.
22. Epoccipital crossing the squamosal-parietal contact (Penkowski and Dodson, 1999)
23. Profile shape of epoccipitals on squamosal (Holmes et al., 2001)
(0) crescentic to lozenge-shaped
24. Parietal fenestra (after Holmes et al., 2001)
(1) present, occupies less than 40% of the total parietal length
(2) present, occupies 45%, or greater, of the total parietal length
25. Rostrocaudal width of caudal parietal ramus behind parietal fenestrae (after Holmes et al., 2001)
(0) relatively wide, ≥20% of total parietal length
(1) narrow and straplike, width ≤10% of parietal length
26. Mediolateral width of median parietal ramus (after Holmes et al., 2001)
(0) relatively wide, ≥15% of total parietal length
(1) narrow and straplike, ≤10% of parietal length
27. Number of loci for epoccipitals on parietal rami lateral to the midline margin (this study)
The caudal and lateral parietal margins have points of potential fusion (loci) for overlying epoccipitals which may be elaborated into distinctive processes. All chasmosaurines have three loci, except for Chasmosaurus irvinensis, which possesses five on either side of the parietal midline, as an autapomorphy for the clade. All centrosaurines typically possess seven or eight loci. Albertaceratops n. gen. appears to have parietals with either three (the Montana bone bed material) or six (the holotype) well-developed processes, and the taxon has been scored as dimorphic for this character.
28. Process at locus 1 (after Sampson, 1995)
(1) unelaborated epoccipital on caudal margin
(2) short procurving hook on dorsal margin; length of hook ≤ diameter of base
(3) long procurving hook on dorsal margin; length of hook ≥ twice diameter of base
(4) triangular epoccipital on dorsal margin
29. Orientation of epoccipital at locus 1 (after Sampson, 1995)
(0) caudally directed
(1) dorsally directed
(2) rostrally directed (pronounced rostral curl)
30. Process at locus 2 (after Sampson, 1995)
(1) unelaborated epoccipital on caudal margin
(2) small, medially curled hook; length of hook ≤ length of base
(3) large, medially curled hook; length of hook ≥ twice length of base
(4) large, triangular profile
Most centrosaurines have medially directed process at loci 2. This can take the shape of a large curled hook in Centrosaurus and Pachyrhinosaurus, or a small unmodified process in Achelousaurus and Einiosaurus. Although most specimens of Styracosaurus have a small medially curled hook, the holotype has a process similar in size and shape to that of Achelousaurus and Einiosaurus, and has been scored as dimorphic for this character.
31. Process at locus 3 (after Sampson, 1995)
(0) small, unelaborated epoccipital on caudal margin
(1) narrow-based hook; length between 1 and 3 times basal diameter
(2) narrow-based long spike; spike greater than 4 times basal diameter
(3) broad-based short pachyostotic spike
(4) large, triangular profile
32. Orientation of spikelike epoccipital at locus 3 (after Sampson, 1995)
(0) caudally directed
(1) caudolaterally directed
(2) laterally or rostrolaterally directed
(3) dorsolaterally directed
33. Imbrication of lateral marginal undulations of the parietal (s. Sampson et al., 1997)
(0) nearly horizontal
(1) steeply inclined laterally
35. Rooting of teeth (Brown and Schlaikjer, 1940)
36. Vertical replacement series of teeth (Brown and Schlaikjer, 1940)
(0) one or two replacement teeth
(1) more than two replacement teeth
This work was part of a Ph.D. dissertation at the University of Calgary, Department of Biological Sciences, supervised by A. P. Russell, and I am very grateful for his guidance and assistance during the course of this work. Thanks to the Royal Tyrrell Museum of Palaeontology and its Field Experience program for supporting this project, especially D. Brinkman, P. Currie, J. Gardner, and B. Naylor; the staff of Dinosaur Provincial Park; Canada Fossils, Ltd. (especially A. Dznic and R. Vandervelde); B. Pohl and the staff of the Wyoming Dinosaur Center for graciously allowing me access to the Montanan material; S. Godfrey and D. Tanke for helpful discussions; W. Sloboda for her expert preparation of the holotype and to A. Dzinc for his equally good preparation of material from the Montanan bone bed; D. Evans for help with Bremer support analysis; R. Holmes, A. Murray, and A. P. Russell for comments on an early draft of this manuscript; D. Sloan for drawing Figure 3.2; C. Kerychuk (Digital Dream Machine, Edmonton) for assistance with photography; J. Anderson, B. Chinnery-Allgeier, A. Fiorillo, and T. Tokaryk for careful reviews; and the Phaeton Group. Special thanks to C. Nesmo for his generous support of paleontological research in southern Alberta. The Dinosaur Research Institute provided funding to support, in part, the casting of selected elements from the Montanan bone bed. This work was supported, in part, by a NSERC Discovery grant (9747-03) to A. P. Russell and monies from the Dinosaur Research Institute, Calgary, Alberta.
Current address: Department of Vertebrate Paleontology, Cleveland Museum of Natural History, <>
- Accepted 25 October 2005.