Tylocephalonyx Coombs, 1979

HUNT, ROBERT M., 2005, An Early Miocene Dome-Skulled Chalicothere from the ‘‘ Arikaree’ ’ Conglomerates of Darton: Calibrating the Ages of High Plains Paleovalleys Against Rocky Mountain Tectonism, American Museum Novitates 3486, pp. 1-46 : 12-28

Tylocephalonyx Coombs, 1979

TYPE SPECIES: Tylocephalonyx skinneri Coombs, 1979: 10 .

INCLUDED SPECIES: The type species only .

KNOWN DISTRIBUTION: Early Hemingfordian to early Barstovian of western North America.

DIAGNOSIS: As given in Coombs (1979: 7).

REMARKS: This report extends the temporal range of the genus reported by Coombs (1979) to include the early Hemingfordian chalicothere hypodigm from the Carpenter Ranch Formation, Goshen County, Wyoming. Material listed below generally conforms in anatomical detail to the descriptions given by Coombs (1979), but the species is considered indeterminate due to the fragmentary nature of much of the skeleton.

REFERRED MATERIAL: Included in this indeterminate species from the Carpenter Ranch Formation are a right mandible with m2–3, a partial cranial dome, and the diaphysis of the left femur, all found together at the same stratigraphic level within a 2­m interval (fig. 9); also a proximal right ulna; a distal metapodial; a proximal left metacarpal 4 and the dorsal process of an ungual phalanx; a distal right femur (the first evidence of this chalicothere found in the caprock sandstones of the southern buttes); a distal left femur in two parts (scavenged by an entelodont) found within 5 m of a proximal left humerus at the same stratigraphic level; a distal right tibia; a glenoid­bearing fragment of the scapula; and a fragment of the ectoloph of a right M3.

Localities and dates of collection of these elements are listed in table 3.

HORIZON AND AGE: All referred material is from the early Hemingfordian Carpenter Ranch Formation, Goshen County, Wyoming. The associated mandible (UNSM 44800), cranial dome (UNSM 44801), and partial femur (UNSM 44802) were found in place in a fluvial lens of sand and gravel exposed in a south­facing cliff at the western end of Deahl Butte (fig. 9). Here a channel of the Carpenter Ranch paleovalley had incised, and mostly removed, the gray Arikaree sands, cutting to the level of the local paleosurface of the White River beds at the chalicothere locality. The dome, mandible, and femur were found within a single gravel lens at the same stratigraphic level, situated; 25 ft (7.6 m) above the base of the channel.

DESCRIPTIVE OSTEOLOGY

MANDIBLE AND DENTITION: The mandible (UNSM 44800) preserves; 24 cm of the horizontal ramus from the angle forward to below m1 (figs. 10, 11). Height of the mandible below the center of m3 is 67.6 mm, and below the center of m2 is 60.6 mm. Greatest mandibular width below m3 is 37.4 mm. The horizontal ramus appears to rapidly diminish in height from m3 forward, suggesting that the length of the premolar row was somewhat reduced.

TABLE 3

Referred Material of the Dome­Skulled Chalicothere Tylocephalonyx from the Carpenter Ranch Formation, Goshen County, Wyoming

The m2–3 are larger and more robust than these molars in Moropus elatus , including those of large males. Maximum antero­posterior length of m3 is 55.5 mm and of m2 is 52.1 mm. Greatest width of the m3 trigonid is 29.1 mm and of the m3 talonid, 27.9 mm. The combined length of m2–3 is 108.8 mm. The m2 is moderately worn, the m3 fully erupted and essentially unworn. The crown of m3 is intact and shows an anterior lophid with a low yet distinct small paraconid, a higher protoconid, and an even taller metaconid, which is the tallest cusp. The metaconid and metastylid are distinctly separated cusps. The metastylid is the tallest cusp of the posterior lophid, whereas the somewhat lower hypoconid and entoconid are of about equal height.

Although only part of an upper molar (UNSM 44808, presumably M3) represents the upper dentition, several characters indicate its referral to the family and to Schizotheriinae . Coombs (1989: 438) noted that chalicotheres are distinguished by upper molars with a complete metaloph with no metaconule and an incomplete protoloph in which the paraconule is retained. UNSM 44808 preserves part of the ectoloph from the mesostyle forward to the paracone. If the ectoloph is restored, based on the M3 of Moropus , its length is estimated at; 52 mm. A small protoloph 12 mm in length, typical of schizotheriine chalicotheres, extends linguad from the paracone, and includes the paraconule. The ectoloph height measured at the unworn paracone is; 34.7 mm, indicating a relatively tall­crowned molar typical of schizotheriines, and similar to molar crown heights (e.g., M3, 38.2 mm) known in Moropus elatus . The enamel rib found on the labial face of the ectoloph in schizotheriines such as Moropus elatus is absent in UNSM 44808, and thus is similar to Tylocephalonyx , which also lacks these ribs.

CRANIAL DOME: Found at the same level; 2 m from the mandible, the cranial remnant appears to represent only half of the dome, broken to reveal an apparent sagittal section that exposes a massive boxwork of compartmentalized sinuses reinforcing the interior of the dome (figs. 12, 13). At the time of its discovery, the boxwork was packed with coarse sand and gravel. It likely survived due to rapid burial, presumably after the carcass was scavenged, because the bone shows no evidence of prolonged exposure to subaerial weathering. It is larger in its dimensions than any of the previously reported Tylocephalonyx specimens: height of the dome from base to summit is; 247 mm; anteroposterior length at its base is; 278 mm, gradually tapering upward so that 10 cm above the base this length is 178 mm, and 4.5 cm below the apex it is 114 mm.

The orientation of the partial dome is somewhat uncertain: the presumed posterior face tapers downward more gradually than the anterior surface, which descends more abruptly (fig. 12). The external surface is moderately rugose, more so toward the posterior margin. Small nutrient vascular foramina are clustered toward the posterior base, similar to parietal­squamosal vascular foramina found in some mammalian skulls. Com­ pact bone 4 cm thick occurs at the posterior base of the dome, interpreted as a remnant of the skull roof. Thick bone may have continued forward under the dome but this part of the skull has been lost. At the anterior base of the dome the bone transforms from an open network of thick­walled sinuses to a more normal, finely cancellous structure that parallels the skull surface, suggesting that one is observing the transition between the interior of the dome and the surface of the skull in the fronto­parietal region.

FORELIMB

SCAPULA: A scapular fragment (UNSM 44811) about 14 cm in length that includes the glenoid fossa appears to belong to a chalicothere. It compares with the glenoid of Moropus elatus and is similar in size to some individuals from the Agate bonebed. Only; 7–8 cm of the glenoid rim are preserved, heavily reinforced by thick bone with abundant nutrient vascular foramina.

HUMERUS: A proximal left humerus (UNSM 44812), damaged by preburial scavenging, was found within 5 m of a distal femur at Childers Butte. The distal femur displays a percussion fracture produced by a large entelodont (fig. 14), and it seems likely that the damaged humerus and femur came from the same scavenged carcass. The form of the proximal humerus does not differ in any respect from that of M. elatus . The antero­posterior length of the humeral head in this case is 16.2 cm whereas the same dimension in a large M. elatus from University Quarry is 19.2 cm.

Coombs (1979: 23) observed that none of the humeri of Tylocephalonyx skinneri retains the proximal end, so a comparison with UNSM 44812 is not possible. Scavengers often damage the proximal humerus when focusing on the concentration of muscle available in the shoulder region. The proximal humerus was also damaged or destroyed in individuals of M. elatus from University Quarry, where evidence of scavenging is commonplace in the Agate waterhole bonebed ( Hunt, 1990).

ULNA: A proximal ulna (UNSM 44810) preserves the olecranon process, semilunar notch, and coronoid process, demonstrating the nature of the articulation with the radius. The proximal ulna was similar in size and form to the ulna of M. elatus . Features of M. elatus mentioned by Coombs (1978) are present: a prominent anconeal process, deep semilunar notch with strong medial expan­ sion, radius facet distal and at an acute angle to the semilunar notch and forming a deep articular fossa. Ulnar width at the level of the coronoid process is; 11 cm. The shape of the semilunar notch for articulation with the humerus and the reduced coronoid process demonstrate that the radial head, which fits tightly against the radial notch, was immobile and did not rotate. Only extension and flexion occurred at the elbow; there was no rotation of the radius around the ulna, an aspect of the anatomy of M. elatus previously noted by Coombs (1978: 25). The ulnar fragment exhibits considerable preburial flaking and cracking of the cortical bone—it must have been scavenged and exposed to weathering for some time prior to burial.

Coombs (1979: 23) regarded the radius/ ulna of Tylocephalonyx skinneri as little different from that of M. elatus . She did note that the lateral extent of the proximal facet on the ulna for the radial head is slightly smaller in T. skinneri ; despite the damaged state of UNSM 44810, it seems to show this more restricted condition of the proximal facet. In both species the radius and ulna were incapable of independent movement— locked by fusion of the distal shafts, the form of the proximal articulation, and by strong interosseous ligaments.

METACARPAL: A massive, robust proximal metacarpal 4 (fig. 15, UNSM 44804) was found in talus of the Carpenter Ranch Formation at a small satellite butte; 0.6 km southeast of the site at Deahl Butte where the dome and mandible occurred. It was found above the base of the formation and can only have been derived from that rock unit.

The articular surfaces and facets for the unciform and for metacarpals 3 and 5 are diagnostic for various chalicotheres, and it is worthwhile to describe these in some detail for UNSM 44804.

The radial side of metacarpal 4 (UNSM 44804) bears two articular facets for metacarpal 3: a dorsal facet and a volar facet (fig. 15A). In M. elatus the two facets are in contact, the volar facet placed somewhat more proximally on the side of the metacarpal, and only slightly smaller in area than the dorsal facet. In UNSM 44804 the volar facet appears reduced in area relative to the dorsal facet, and is separated from it by a; 1 cm gap. In addition, the volar facet does not extend proximad above the dorsal facet as it does in M. elatus but rather is situated directly posterior to it.

The greater part of the dorsal surface of metacarpal 4 (fig. 15B) is for articulation with the unciform. This surface is much wid­ er and appears to be flatter than those of either males or females of M. elatus from University Quarry at Agate National Monument. However, this configuration may be influenced by the robust size of UNSM 44804, probably a large male. In larger individuals of M. elatus the articular surface for the unciform becomes broader and more planar than seen in small individuals, but not to the extent evident in UNSM 44804.

UNSM 44804 demonstrates that the Carpenter Ranch chalicothere had retained a reduced metacarpal 5 in the manus as does M. elatus . On the ulno­volar side of proximal metacarpal 4 in M. elatus, Coombs (1978: 31) described a facet for the articulation of metacarpal 5, located on the ulno­volar angle of metacarpal 4 and separated from the proximal edge by paired tubercles. This facet was ‘‘on the distal volar surface of the ulnar tubercle’’. On two metacarpal 4s of M. elatus in the UNSM collection from University Quarry, this facet is present, as well as a second articular facet behind it, the two facets together creating a modest concavity for the head of metacarpal 5, which clearly was a reduced element. UNSM 44804 also shows a facet for a reduced metacarpal 5, slightly concave, and similarly situated on the distal volar surface of the ulnar tubercle. However, the second facet found in M. elatus is questionable in UNSM 44804, in which a very small, smooth articular(?) surface does occur directly behind the principal facet, separated from it by a deep channel; 7 mm in width. This could represent a secondary area of contact for metacarpal 5.

Coombs (1979: 11) listed three metacarpal 4s belonging to Tylocephalonyx skinneri , one of which was complete (F:AM 54887b). Of the two facets (dorsal and volar) for metacarpal 3 present on the proximal metacarpal 4 of M. elatus , T. skinneri entirely lacked the volar facet, which Coombs considered an important distinction. However, UNSM 44804 displays both a volar and dorsal facet, which are of a different shape and placement than those seen in M. elatus and T. skinneri . Other subtle differences from the metacarpal 4 of UNSM 44804 can be seen in photographs of the T. skinneri metacarpal 4 ( Coombs, 1979: 33, figs. 16B, C, D), and these suggest that the manus of the early Hemingfordian Carpenter Ranch chalicothere was somehow differently configured relative to the manus of the late Hemingfordian T. skinneri . Because the presence of both a dorsal and volar facet for metacarpal 3 represents a plesiomorphic trait, it is possible that early Hemingfordian Tylocephalonyx possessed both facets, with only the dorsal facet retained in late Hemingfordian T. skinneri .

Coombs (1978: 53, figs. 25C,D) also illustrated the metacarpal 4 of Moropus hollandi (FMNH P13000) from near Jay­Em, Wyoming, presumably from the Upper Harrison beds. The placement of the facets for articulation of metacarpal 3 shows a closer correspondence to that of UNSM 44804 than to M. elatus or T. skinneri ; the dorsovolar length measured across these facets on the radial side of metacarpal 4 is; 48 mm in M. hollandi and 63.5 mm in UNSM 44804, the latter a much larger individual, probably male. Coombs (1978: 50), on good evidence, believed P13000 to be a female. One can entertain the possibility that the Carpenter Ranch metacarpal 4 might represent the earlier M. hollandi and not the dome­skulled Tylocephalonyx . However, the only known skull of M. hollandi (FMNH P12094) lacks any indication of doming, but whether a dome was an exclusively male trait is not known because of the scarcity of intact skulls ( Munthe and Coombs, 1979: 85).

UNGUAL PHALANX: A dorsal process of the ungual phalanx of a chalicothere (UNSM 44805) was found at the same locality as UNSM 44804 (proximal metacarpal 4). The remnant of the process is 4.6 cm in length and; 2.5 cm in width; it includes the apical termination of the process and much of the articulation with the intermediate phalanx. Its robust size and rugosity suggest it is the dorsal process of the ungual phalanx of digit II of the manus.

HINDLIMB

FEMUR: Remains of several femora, including a diaphysis of a left femur (UNSM 44802) found in proximity to the dome and mandible described above, come from sites at Deahl Butte and adjacent Childers Butte. A distal right femur (UNSM 44807) found at Childers Butte in November 1977 provid­ ed the first evidence of the presence of chalicotheres in the Carpenter Ranch Formation. These femora differ in no evident features from those of M. elatus , showing the same placement and development of the lesser trochanter and conspicuous third trochanter. Such bones often display green bone fractures and appear to be scavenged (fig. 16).

Another distal left femur and accompanying diaphysis (UNSM 44806) were found in outcrop in 1979, together with a proximal humerus (UNSM 44812), over a distance of; 5 m at the same stratigraphic level on Childers Butte. Despite damage due to preburial scavenging, the distal femur and diaphysis (UNSM 44806) can be identified confidently as chalicothere. These pieces show green­bone fractures, and the distal femur carries a depressed circular percussion fracture produced by the premolar of a large entelodont (fig. 14). This femur had an estimated length of; 60 cm, thus a size similar to the femur of M. elatus . Lengths of the femur in male and female M. elatus have been reported as 65 and 43 cm, respectively ( Holland and Peterson, 1914:362), and two intact femora from University Quarry in the University of Nebraska collections measure;60 and 64 cm in length.

TIBIA: A distal right tibia (UNSM 44809) from the Duncan Buttes is similar to that of M. elatus . Coombs (1978: 55) noted that both the tibia and femur of M. hollandi could not be distinguished from these same bones of M. elatus . The tibia of Tylocephalonyx skinneri was also considered by Coombs (1979: 34) to be ‘‘similar in proportions and morphology to tibiae of Moropus ’’. The few differences she cited do not involve the distal part of the bone. A fibula articulated with the distal tibia in these chalicotheres: the distal facet for attachment of the fibula is present in UNSM 44809 and suggests a fibula much like that of M. elatus ( Holland and Peterson, 1914: pl. 71).

Of some importance is that the distal tibia preserves the articular surface for the astragalus and demonstrates rather shallow grooves for the astragalar trochlea relative to those of M. elatus (particularly with regard to the medial trochlear depression). The width of the astragalar trochlea in M. elatus , based on specimens in the UNSM collection from the Agate waterhole bonebed, ranges from 58.6 to 79.6 mm (N 5 14; mean, 69.3 mm). Measurement across the trochlear depressions in the distal tibia of UNSM 44809, which serves as a proxy for the width of the astragalar trochlea, indicates that the Carpenter Ranch chalicothere had a broad astragalus with a trochlear width of; 92 mm, larger than that of M. elatus . However, in addition to this difference in size, it is the relative proportions of these astragalar depressions that suggest referral of the Carpenter Ranch chalicothere to Tylocephalonyx . Coombs (1979: 34, fig. 19A) observed that the tibial and fibular parts of the astragalar trochlea in T. skinneri are of about equal width, whereas in M. elatus the tibial part of the trochlea is narrower, occupying less than half the trochlear width ( Coombs, 1978: 33, figs. 14C,D). In UNSM 44809 the trochlear width of the tibial side is 45 mm, and the fibular side 47 mm, which is more similar to Tylocephalonyx . These same measurements taken from a tibia of M. elatus from University Quarry are 31 mm and 40 mm, respectively.

DISCUSSION

The tall­crowned molar fragment lacking an enamel rib on the ectoloph, the large and robust m2–3, a distal tibia, and the cranial dome all suggest identification of the material as the schizotheriine Tylocephalonyx . The partial femora, humerus, ulna, and other fragments do not in themselves provide a reliable generic identification for this chalicothere. Until discovery of more complete material proves otherwise, I have assumed that chalicothere elements from the Carpenter Ranch beds belong to a single species. It is improbable that any of the chalicothere fossils from the Carpenter Ranch Formation belong to a taxon such as Moropus elatus , since the presence of two such similar chalicotheres in the same environment seems unlikely, and because none of the remains shows any particular affinity to the Agate species. Specifically, the length and width of the m2–3 exceed almost all known M. elatus molars in size, the proportions of the astragalar trochlea differ from Moropus , and domed skulls are unknown in M. elatus males and females.

Among the postcranial elements referred here to Tylocephalonyx , it is the metacarpal 4 that is the most difficult element to include in that genus; it does not appear to correspond to that bone in T. skinneri but better matches metacarpal 4 of Moropus hollandi . Nonetheless, because the size and form of chalicothere postcranial elements from the southern buttes suggest the presence of only a single species, and the postcranial elements occur with dental and cranial material attributable to a dome­skulled animal, all chalicothere elements from the Carpenter Ranch beds are tentatively attributed here to Tylocephalonyx , while recognizing the possibility of the presence of a second species similar to or the same as M. hollandi .

At the time of the initial discovery of dome­skulled chalicotheres, Munthe and Coombs (1979) did not describe the internal architecture of the dome in detail due to the existence of only three skulls. Although radiographic examination of the two complete skulls provided no additional information, cranial fractures did allow construction of transverse sections of one of the skulls, including a schematic profile through the dome. They reported ‘‘bony partitions connecting the dome to the roof of the braincase which divided this cavity into a series of irregularly shaped sinuses’’ ( Munthe and Coombs, 1979: 82).

The partial dome of the Carpenter Ranch chalicothere (UNSM 44801) preserves a boxwork of compartmentalized sinuses extending from the summit to the base of the dome (fig. 13), which conforms to the preliminary observations of Munthe and Coombs (1979: 82, fig. 6). In domed skulls from the Split Rock Formation (Wyoming) and from Greenside Quarry (Nebraska), they found evidence of a similar internal osseous boxwork, a remnant of which can be seen in their photograph of the Greenside Quarry skull ( Munthe and Coombs, 1979: fig. 3; note bony septa immediately above the roof of the braincase). However, they were not able to determine the exact location and orientation of the cranial sinuses from the available skulls. Thus, UNSM 44801 makes possible for the first time an examination of the internal architecture of the dome.

UNSM 44801 shows that the interior of the dome was entirely supported by the sinusoidal boxwork from the base to the summit (figs. 13, 17). The sinuses are nearly intact in the upper half of the dome, an area of;17 by 14 cm seen in parasagittal section: about 40 sinus compartments occupy the interior of the dome as preserved, with individual diameters ranging from; 9mm to 3 cm (fig. 17A). The diameters of 20 of these compartments measured at random in the upper half of the dome averaged 16 mm. Similarly, the thickness of the bony partitions (septa) separating the compartments averaged 3.3 mm with a range from 1.6 to 6.0 mm. In the upper part of the dome the thickness of the bone forming the roof ranges from thin (3.2 mm) to very thick (24 mm). Of considerable interest is that the inner margins of almost all septa in the upper half of the dome are smooth, in life apparently covered by membrane, indicating that one is observing the intact, unbroken inner edges of individual compartments. However, compartments in the lower part of the dome exhibit broken margins, suggesting that they were continuous across the base of the dome, and were broken apart at the time the skull was damaged, presumably by scavengers.

Thus, the interior of the dome, at least in its upper part, must have been an open airfilled space, yet strongly reinforced along the sides by bone. This morphology provides a plausible explanation for the fact that only half of the dome was found, for if scavenged or otherwise damaged, the dome might tend to split apart into two halves along a mid­ sagittal plane through the hollow upper portion. One can conclude that the dome probably had a hollow dorsal cavity, bordered on its margins by strong sinusoidal bone reinforcing the walls throughout, and with a network of sinus compartments continuous across the basal part of the dome above the braincase.

In their evaluation of the functional role of the dome, Munthe and Coombs (1979) considered several plausible hypotheses. Although recognizing that information at their disposal was limited, they suggested that the dome may have been involved in acoustic signaling, in visual displays, or for low­impact butting (or some combination of these behaviors). UNSM 44801 fortunately supplies evidence, lacking in the Munthe and Coombs study, that establishes the thickness, orientation, and distribution of the dome’s internal bony partitions. The pronounced thickness of the walls of the sinus compartments and their distribution within the dome indicate a well­reinforced architecture. Robust bony struts bridge the area above the braincase, supporting the walls of the dome from base to apex. Only the upper part of the dome appears to have been hollow. One can conclude that the dome was reinforced, durable, and could have withstood fairly strong external impacts.

After carefully considering a variety of possible hypotheses, Munthe and Coombs (1979: 88) focused their attention on low­impact flank­butting behavior in the living giraffe ( Giraffa camelopardalis ). The skull of a male giraffe, which they examined in sagittal section, contained a cranial sinus dorsal to the braincase, partitioned by bony septa, that they judged similar in form and size to that of the dome­skulled chalicothere. Considering that the internal septa of the Carpenter Ranch dome are thicker and more robust than those of the giraffe, low­impact butting cannot be ruled out. However, other functions attributed to the dome seem equally plausible: as a male visual display or (if both sexes possessed the domed skull) in species recognition. In addition, the air sinuses within the dome also might have served an acoustic role similar to that attributable to the expanded cranial sinuses in the living African and Asian elephants (Loxo­ donta africana, Elephas maximus ). Elephants are known to employ low­frequency sound in communication over long distances ( Payne et al., 1986; Poole et al., 1988). Low­frequency sound perception is enhanced when middle ear volume is expanded so that the ossicular vibrations within the middle ear are not damped by an air space of small volume. The living African elephant possesses a large foramen in the exoccipital bone forming a passageway between its middle ear and the voluminous cranial sinuses in its skull. This communication between middle ear and cranial air sinuses would prevent ossicular damping and favor the perception of low frequency sound. Possibly a similar passageway existed in dome­skulled chalicotheres between their middle ear and the sinuses in the dome, but if so, domes would be expected in both sexes. Only dissection or noninvasive CT (computerized tomography) scanning of a complete skull can demonstrate the presence of such an anatomical arrangement in these large perissodactyls.

Another rationale for the dome lies in the enhancement of olfaction. Negus (1958) not­ ed that to obtain greater room for ethmoturbinals, hence improved olfaction, extension into surrounding bones and the development of spaces dorsal to the braincase were viable strategies. Extension of the ethmoturbinals covered with olfactory mucosa often occurs in mammals by invasion of the frontal and sphenoid paranasal sinuses. In some Carnivora , such as felids, the ethmoturbinals invade newly created spaces within the frontal and nasal bones, which Negus (1958: 300) considered as ‘‘the construction of a superimposed upper story’’ added to the dorsal surface of the skull. The extension of ethmoturbinals into paranasal sinuses clearly serves an olfactory role because olfactory nerves supply the invading turbinals.

Most such olfactory extensions occur in living Carnivora but it is noteworthy that among living perissodactyls the tapir has evolved olfactory penetration into both the nasal and frontal areas of the frontal sinus, and the horse has a very small olfactory extension into the nasal (not frontal) part. Negus (1958: fig. 147) reported that a rhinoceros (species not indicated) retained very large nasal and frontal parts of an enlarged frontal sinus but that these spaces were nonolfactory. However, the horse, tapir, and rhinoceros all have large sphenoidal sinuses invaded by olfactory tissue. In his survey of paranasal sinuses, Negus (1958) remarked on the presence in many ungulates of ‘‘supracranial cavities’’ in the frontal, parietal, and occipital areas of the skull that did not contain ethmoturbinals but that communicated with the nasal region. Based on the limited anatomical evidence, invasion of the chalicothere dome by olfactory ethmoturbinals seems unlikely, but it is possible that a passage from the dome into the nasal cavity existed. Such a connection between the frontal sinus and nasal cavity occurs in many ungulates ( Sisson and Grossman, 1953; Negus, 1958: fig. 156). The broad enlarged snout of chalicotheres suggests that the ethmoturbinals may have been sufficiently well developed within the nasal fossa such that no need existed for ethmoturbinal invasion of the cranial dome.

Evidence reported above demonstrates that large entelodonts scavenged the remains of these chalicotheres (only entelodont teeth match bite marks in these bones). It is significant that no complete chalicothere limb bones, despite their large size and durability, survived intact in the sands and gravels of this fluvial system. All limb bones have been broken, and several show green­bone fractures that would have required considerable force to produce (fig. 16). The record of entelodont tooth marks on chalicothere limb bones has now been documented at three localities of early Miocene age in the Great Plains: (1) the late Arikareean waterhole bone bed at Carnegie Hill, Agate Fossil Beds National Monument, Sioux County, Nebraska ( Hunt, 1990: fig. 25); (2) the latest Arikareean Lay Ranch paleovalley, Anderson Ranch Formation at Spoon Butte, Goshen County, Wyoming (fig. 18); and (3) early Hemingfordian fluvial sediments of the Carpenter Ranch Formation, Goshen County, Wyoming (fig. 14).

Coombs (1979: 5) visualized Tylocephalonyx as a browsing ungulate of the temperate forests of northwestern North America with ‘‘a significant distribution onto the Great Plains, perhaps associated with riparian communities’’. The presence of a dome­skulled chalicothere apparently referable to Tylocephalonyx in association with abundant silicified wood in fluvial sediments of a major early Miocene paleovalley in the central Great Plains, supports her hypothesis. Chalicotheres in the Miocene of the Great Plains are commonly found in fluvial and waterhole settings, where they have been regarded as water­dependent mammals in recent field studies ( Hunt, 1990).

AGE OF THE DOME­SKULLED CHALICOTHERE

Coombs (1979) and Munthe and Coombs (1979) attributed all known fossils of the dome­skulled Tylocephalonyx to the mid­ Miocene, that is, to the late Hemingfordian and early Barstovian land mammal ‘‘ages’’. They believed that none of these remains was older than;17 Ma. The fauna associated with the dome­skulled chalicothere in the Carpenter Ranch Formation is of early Hemingfordian age, and is representative of the earliest assemblage of Hemingfordian mammals yet identified in the Great Plains. Mammalian faunas of similar age are now known from western Nebraska (Northeast of Agate local fauna, MacFadden and Hunt, 1998; Tedford et al., 1987) and northeastern Colorado (Martin Canyon local fauna, Tedford, 1999), and from less well­known assemblages from near Wheatland in southeastern Wyoming (Hunt, unpublished observations), and from the Troublesome Formation in northcentral Colorado ( Izett, 1968). These strata record a regional tectonic/climatic event identified by the first appearance of Hemingfordian faunas in the Great Plains and adjacent Rocky Mountain basins. At this time an influx of coarse sands and granitic gravels indicates increased competence of streams bringing Rocky Mountain detritus from the uplifts eastward onto the Great Plains.

The Northeast of Agate local fauna has been paleomagnetically calibrated and is considered to fall in Chron C5E of the Magnetic Polarity Time Scale ( Berggren et al., 1995), dated at;18.2 to 18.8 Ma. This local fauna comes from the stratigraphically lowest exposures of the Runningwater Formation in western Nebraska ( MacFadden and Hunt, 1998), best developed northeast of the Agate post office in central Sioux County where these pale orange­brown sandstones yield a fauna of earliest Hemingfordian aspect. A key index taxon in this assemblage is the large oreodont Merycochoerus magnus , which occurs both in the Northeast of Agate local fauna and in the Carpenter Ranch mammal assemblage.

These earliest Hemingfordian faunas in western North America are considered to fall within the 18.2–18.8 Ma interval, hence old­ er than any of the previously reported occurrences of dome­skulled chalicotheres.

Here I summarize mammals contributing to the age assessment of the Carpenter Ranch beds: the oreodont Merycochoerus , a dromomerycid Aletomeryx , the moschid Pseudoblastomeryx , the small canid Phlaocyon , and the amphicyonid carnivore Daphoenodon .

ORDER ARTIODACTYLA

FAMILY OREODONTIDAE