Premolars

Tall, laterally compressed premolars first observed in John Day temnocyonines by E.D. Cope are one of the most evident diagnostic traits of the subfamily. Cope's genoholotype mandible of Temnocyon, as well as those of all other plesiomorphic temnocyonines, features tall well-developed p2–4 and P2–3. P1/p1 are small, often low and peglike teeth that lack the height of the more posterior premolars and are not replaced by a second-generation tooth (they belong to the milk dentition as in many other mammals). In Temnocyon and Mammacyon the upper and lower premolars (except of course P4) are typically triangular (in lateral view), narrow, elongate teeth, and are not shortened or widened. This is also true for the only surviving premolar of Rudiocyon. Delotrochanter on the other hand has short, posteriorly wide premolars, distinguishing it from Mammacyon and Temnocyon. Temnocyon and Mammacyon also lack a well-defined posterior accessory cusp on the posterior slope of p3, whereas Delotrochanter has such a cusp. Premolars increase in size as body size increases within all lineages; however in Delotrochanter there is a trend toward short, wide premolars, whereas in Temnocyon and Mammacyon the longer, relatively narrow teeth are maintained (except for a broad p4 in Mammacyon ferocior). Premolars (p1–4) increase in size posteriorly in the toothrow, and p4 is often of striking size relative to m1. These robust premolars accompany crushing carnassials and molars, particularly in the large terminal species of Mammacyon and Delotrochanter, suggesting a durophagous adaptation. The prominent premolars of temnocyonines distinguish them from amphicyonines, which have greatly reduced p2–3/P2–3.

Carnassials

One of the most diagnostic temnocyonine teeth is the upper carnassial. Initially in early John Day temnocyonines this tooth is little different in form from the shearing upper carnassial of a typical plesiomorphic beardog like Daphoenus. However, in Delotrochanter and Mammacyon the upper carnassial evolves into an enlarged, blunt-cusped crushing tooth in which shear is no longer the principal function. This is accomplished by shortening of the metastylar blade (thus metastylar shear is diminished), while the protocone region becomes lingually extended and much enlarged, and the protocone itself transforms into a blunt pestlelike cusp (table 4). The occlusal outline of the upper carnassial thus takes the form of an equilateral triangle, not the more common isosceles triangle found in most arctoid and cynoid carnivorans with a shearing P4. Whether this same trend occurred in Temnocyon remains to be demonstrated because the large end-member species of that genus from the Great Plains (T. macrogenys) is not known from specimens that preserve the upper carnassial. The terminal species of John Day Temnocyon (T. fingeruti, T. ferox) suggest that such modification occurred but was not as pronounced as in the other two genera.

TABLE 4

Temnocyonine P4 Proportions (in mm)

The lower carnassial alters its cusp pattern and proportions in both Delotrochanter and Mammacyon (and in Rudiocyon) but only minor changes occur in Temnocyon. Early John Day temnocyonines (T. altigenis) have lower carnassials that closely approach a plesiomorphic shearing amphicyonid m1 (e.g., Daphoenus vetus). This plesiomorphic pattern in Daphoenus incorporates a prominent tricuspid trigonid with well-developed paraconid-protoconid shear and a distinct metaconid; the talonid lies at a lower level than the trigonid and forms a shallow basin bordered labially by a prominent hypoconid and lingually by a thin enamel rim often with a tiny entoconid cusp. In early John Day temnocyonines the only significant modification of this plesiomorphic pattern is the loss of the entoconid and development of an enlarged hypoconid that occupies nearly the entire talonid. Consequently in the earliest temnocyonines the m1 talonid is already a blunt crushing instrument.

As the three lineages evolve, only two (Delotrochanter, Mammacyon) show significant changes in m1 cusp pattern and form. Temnocyon does not and m1 simply becomes larger: even its terminal latest Arikareean species (T. macrogenys) still retains the plesiomorphic m1 cusp pattern (tall trigonid with metaconid, and low talonid with enlarged blunt hypoconid). The most pronounced modification of m1 occurs in Delotrochanter in which the metaconid is lost even in the oldest species (D. petersoni); the paraconid and hypoconid approach each other in height; and the hypoconid becomes an enlarged bulbous cusp occupying the entire talonid. The talonid becomes enlarged to such a degree that in the terminal species, D. major, the hypoconid exceeds the paraconid in size.

The terminal species of Temnocyon and Delotrochanter (T. macrogenys, D. major), which coexisted in the latest Arikareean (Ar4) of the central Great Plains, demonstrate the divergent extremes attained by the lower carnassial. The former retains an enlarged, massive plesiomorphic m1; the latter a completely transformed crushing tooth. However, the lower carnassials of Mammacyon and Delotrochanter are less readily distinguished since in both genera the metaconid is lost early, and the paraconid and hypoconid tend to approach each other in height. Isolated lower carnassials of the two genera may be identified by the larger, more expanded talonid and hypoconid in Delotrochanter and the form of the labial cingulum, which is straight in Delotrochanter and sinuous in Mammacyon. In addition to loss of the metaconid, the m1 paraconid-protoconid cusps in Mammacyon and Delotrochanter become more rounded and less sectorial. Delotrochanter major represents the extreme expression of this trend: the lower carnassial is little more than three linearly arranged, rounded cusps (with paraconid slightly offset lingually from the protoconid; paraconid-hypoconid nearly equal in height), creating a formidable crushing instrument when coupled with the robust premolars. In Mammacyon ferocior the m1 is modified in similar fashion but fails to attain the size and degree of simplification seen in D. major.

First upper molar

This tooth, together with P4, is particularly diagnostic for Temnocyoninae. Even the most plesiomorphic species from the John Day basin can be identified as members of the subfamily by the form of the M1 protocone isolated on an enamel flat surrounded by a thickened lingual cingulum. In the larger, dentally derived species of Mammacyon and Delotrochanter the protocone evolves to a prominent isolated knoblike cusp; the enamel flat is conspicuously enlarged; and the lingual cingulum is thicker and more prominent. A ratio of the anteroposterior length of the labial margin compared to the length of the lingual half of the tooth measures this change in M1 form in temnocyonines (table 5).

TABLE 5

Measurements (in mm) Demonstrating Expansion of the M1 Protocone Region in Temnocyoninae

The M1 of Temnocyon is unknown in the terminal species, T. macrogenys; it probably did not differ from the plesiomorphic form still retained in T. percussor, one of the youngest species of Temnocyon in which M1 is preserved and the probable predecessor to T. macrogenys. The plesiomorphic M1 of T. altigenis (Logan Butte) is only slightly modified in T. percussor and T. ferox, less so in T. fingeruti. These species lack the blunt knoblike cusp seen in Mammacyon and Delotrochanter, maintaining a preprotocrista diverging from the more normally configured protocone (a postprotocrista may also occur).

Second lower molar

Species of Temnocyon preserve the plesiomorphic cusp pattern of this tooth in which the protoconid and hypoconid are situated at the labial margin. In contrast, the m2 of Mammacyon and Delotrochanter share a derived condition in which a rather blunt protoconid and hypoconid, placed one behind the other, are situated in the center of the tooth so that the former cusp fills the trigonid and the latter occupies the talonid.

In other respects this tooth is one of the more variable elements of the temnocyonine dentition. It is unwise to attribute too much significance to slight differences in m2 form. However, Mammacyon includes large carnivorans with a definitely elongate, rectangular m2 (table 6, ratio E/F), and this trait, coupled with the m1 form and the elongate, narrow premolars, identifies the lineage dentally. Furthermore, a small notch or indentation in the posterolingual cingulum of the Mammacyon m1 is also present at the same location on m2.

TABLE 6

Temnocyonine Dental Ratios^a

The m2 of Delotrochanter is known only in D. oryktes (only the anterior half of m2 survives in D. major) where it is elongated relative to m1 but not to the extent seen in Mammacyon. The m2 in species of Temnocyon is not elongate (table 6, ratio E/F), with values >1.67 for all but T. macrogenys; however, the range of values for T. altigenis itself suggests this is a variable trait. In Rudiocyon, m2 length is quite short relative to m1 length, demonstrating a marked difference from Mammacyon.

In summary, the temnocyonine genera are identified as follows: (1) Temnocyon (a paraphyletic taxon) retains the plesiomorphic dentition in which premolars are robust, tall, narrow, not reduced; m1 retains a tall trigonid with metaconid, a low talonid with enlarged hypoconid; m2 is short, not elongate relative to m1; P4–M1 are similar to the shearing carnassial pair in the dentally unspecialized amphicyonid Daphoenus (except that all teeth are proportionately larger in Temnocyon); (2) Delotrochanter has a derived durophagous dentition in which premolars (p2–4, P2–3) are short, robust, and posteriorly widened; the posterior accessory cusp (PAC) of p4 is centrally placed; m1 has lost the metaconid; the labial cingulum of m1 is straight, not sinuous; there is less height differential between m1 paraconid and hypoconid than in Temnocyon; m1 evolves a large centrally placed hypoconid in the terminal species, D. oryktes and D. major; m2 is slightly elongate in D. oryktes, the only species that preserves m2; in both Delotrochanter and Mammacyon, P4 has lost the typical form of a shearing carnassial and developed a short, rather blunt metastylar blade and an enlarged protocone for crushing occlusion, accompanying the lingual expansion of M1; P4 in Delotrochanter never attains the large size seen in Mammacyon despite the much younger geologic age of D. oryktes and D. major; (3) Mammacyon has a derived durophagous dentition in which premolars (p2–4) are elongate, narrow, with only p4 broadened posteriorly in M. ferocior; the p4 PAC is labially placed; m1 has lost the metaconid and the labial cingulum is sinuous; m1 shows less height differential between paraconid and hypoconid than in Temnocyon, and is similar in this respect to Delotrochanter. Mammacyon develops the longest m2 (relative to m1 length) of any temnocyonine. P4–M1 and m1–2 are massive crushing teeth with an enlarged, bulbous P4 protocone more developed than in Delotrochanter; (4) Rudiocyon amplidens is a large carnivore, possibly a sister taxon to Mammacyon; its m1 is the size of that in D. oryktes or M. ferocior; it has lost the m1 metaconid but differs from these species in retaining a narrow, laterally compressed p4 and short m2 lacking elongation.

Type Species

Temnocyon altigenis Cope, 1878.

Included Species

Temnocyon altigenis Cope, 1878; T. subferox, new species; T. ferox Eyerman, 1896; T. percussor Cook, 1909; T. fingeruti, new species; T. macrogenys, new species.

Distribution

Early Arikareean, Oregon and California; late Arikareean, Oregon and western Nebraska; latest Arikareean, southeastern Wyoming.

Diagnosis

A paraphyletic genus distinguished from other temnocyonines by presence of a metaconid on m1; by plesiomorphic form and proportions of M1–2, m2 (with labially situated proto-and hypoconids), upper and lower carnassials, and premolars (P4–M1 ratios A/B, C/D, table 6). See tables 1–TABLE 2TABLE 3TABLE 45.

Discussion

The genus includes stem temnocyonines that preserve the plesiomorphic form of the cheek teeth, only slightly modified in the younger and larger derivative species of the genus. T. subferox apparently evolved from T. altigenis, the earliest and smallest temnocyonine species. Larger and more dentally derived T. ferox, T. percussor, and T. fingeruti retain a number of primitive dental features and represent larger species evolved from within the T. altigenis–T. subferox group. The enormous T. macrogenys, the last representative of the genus, is most likely derived from T. percussor. The genus ranges in time from the earliest to latest Arikareean.

Temnocyon altigenis Cope, 1878

Figures 4–Fig. 5.Fig. 6.Fig. 7.8, 65

Fig. 4.

Temnocyon altigenis (UCMP 9999) from Logan Butte, Crook Co., Oregon: A, palate with right C, P1–M2; B, C, right mandible with c, p1–m2 in labial and lingual views. Its skull, dentition, and small size establish UCMP 9999 as the most plesiomorphic temnocyonine.

Fig. 5.

Temnocyon altigenis (UCMP 9999), John Day Formation, Logan Butte, Crook Co., Oregon.

Fig. 6.

Holotype of Temnocyon altigenis Cope (AMNH 6855), John Day Formation, Oregon. Right mandible with p2–4, m1–2 in (A) labial and (B) lingual views.

Fig. 7.

Temnocyon altigenis (UCMP 1549), John Day Formation, Morgan's Place, UCMP loc. 874, near the town of Monument, Grant Co., Oregon. (A) lateral and (B) dorsal views of uncrushed cranium.

Fig. 8.

Palate with I1–3, C, P1–4, M1–2 of Temnocyon altigenis (UCMP 1549), John Day Formation, Morgan's Place, UCMP loc. 874, near the town of Monument, Grant Co., Oregon.

Temnocyon altigenis Cope, 1878 (December 3): 6–8; 1879a: 68–70; 1879b: 55–69; 1881: 179; 1883: 238, figs. 2–3; 1884: 903–905, pl. 68, fig. 9, pl. 70, fig. 11.

Temnocyon altigenis: Merriam, 1906: 21–28, text figs. 7–11, pl. 3 (fig. 2).

non Temnocyon altigenis: Thorpe, 1922: 167–168; Hough, 1948: 100–101 (YPM 10065 is removed from T. altigenis).

Type

AMNH 6855, right mandible with p2–m2 and alveoli of the canine and p1, from the John Day beds, Oregon.

Distribution

Early Arikareean, John Day Formation, Oregon.

Diagnosis

Smallest and most plesiomorphic North American temnocyonine species distinguished from all other species of Temnocyon by small size and range of m1 lengths, 17.4–19.5 mm (table 2), and from T. subferox, T. fingeruti, and T. percussor by either M1 or P4 proportions (dental ratios A/B, C/D, table 6); from T. ferox and the huge T. macrogenys by much smaller size. Ratio of m1/m2 lengths, 1.67–1.83, indicating a short m2 relative to m1, thus distinct from species of Mammacyon (1.57, 1.61) but not from other Temnocyon species (ratios of 1.62–1.79) or Delotrochanter (1.64–1.69). Basilar length of skull, ∼16–18 cm relative to 21–30 cm for other Temnocyon species (table 7); frontal sinuses moderately inflated.

Referred Specimens

(1) AMNH 6856, left lower jaw with c, p2, p4–m2, alveoli of p1, p3, and m3, from the John Day beds, Turtle Cove, Grant County, Oregon; (2) AMNH 6857, right lower jaw with p2–m1, roots or alveoli of c, p1, m2–3, partial rostrum anterior to the orbits, with attached left lower jaw fragment, right C, P2–M2, alveolus of P1, left I3–C, P1–3, left c, p1–4, the entire left dentition highly fragmented, and a partial postcranial skeleton, from the John Day beds, Turtle Cove, Grant County, Oregon; (3) USNM 7940, left lower jaw with m1, m2 erupting, and damaged deciduous teeth; right maxilla with C, P1, P2–3 erupting, M1 fully erupted, P4 and M2 erupting, DP2–3, DC, from the John Day beds, Turtle Cove, Grant County, Oregon; (4) UCMP 1549, nearly complete uncrushed skull without basicranium, with right I1–2, alveolus for I3, C, P1–M2 (P4–M2 with parts missing); left I1–P1, P2 roots, P3–M2 (P4–M1 damaged), partial endocast of temporal lobes exposed and undistorted, from John Day beds, Morgan's Place, UCMP loc. 874, Grant County, Oregon, collected by UCMP party, 1899; (5) UCMP 9999, skull with well-preserved basicranium, associated lower jaws, and much of the postcranial skeleton; teeth present include right C–M2, left C–P4, M2 (damaged M1); right C–M2 (broken M3); left C–M3; from the John Day beds, Logan Butte, Crook County, Oregon, collected by Davis and Osmont, 1900.

Description

The lower jaws of the hypodigm are very similar in dental morphology and size and can be described as a group (AMNH 6855, 6856, 6857, USNM 7940, UCMP 9999). In 1884 Cope described and figured AMNH 6855 and the rostrum of AMNH 6857; he briefly mentioned AMNH 6856 and the jaws of AMNH 6857. Although Merriam (1906) mentioned UCMP 1549, it has never been described or figured, but in this same paper considerable attention was given to the associated skull, jaws, and skeleton from Logan Butte (UCMP 9999), which is arguably the most plesiomorphic North American temnocyonine, based upon its dental traits.

The premolars of these lower jaws from the John Day beds are exceedingly similar in form and size: there is a progressive increase in size from p1–4, with p1 a rather small single-rooted tooth relative to p2–4. There are no accessory cusps with the exception of a prominent laterally placed posterior accessory cusp on p4, the plesiomorphic state for the subfamily. The premolars are not crowded except in young animals with newly erupting premolars. In adults the premolars are ranked in a linear series and are narrow with anterior and posterior slopes each traversed by a fine enamel ridge, running down each slope from the tip of the principal cusp to the base of the tooth. Only p4 shows any posterior expansion in width at the base of the tooth, and this is very slight (6.8 mm vs. 5.4 mm anterior width, AMNH 6855).

The molars are also similar in form and size within the hypodigm, with the exception of AMNH 6856, which has a short m2 relative to m1 length (table 6, ratio E/F, 1.83). All lower carnassials conform to a common pattern: the trigonid is high, the three principal cusps present but unequally developed—the tall protoconid is the main cusp, the paraconid and metaconid lower and at about the same height. The metaconid is somewhat reduced and is located both internal to and posterior to the protoconid. The paraconid forms a robust bladelike cusp anterointernally directed. The m1 talonid is short relative to the trigonid: it is much lower relative to the trigonid than in younger, larger species of temnocyonines in which the height inequity is resolved. The m1 talonid is dominated by a large, prominent, centrally placed hypoconid; there is no entoconid. The m2 is characterized by a trigonid somewhat longer than the talonid; there are two main cusps, the protoconid and hypoconid, situated one behind the other toward the labial edge of the tooth. The paraconid is small, occupying the anterointernal corner of m2. The m2 metaconid is absent or in USNM 7940, a juvenile with unworn teeth, a small vestigial metaconid is discernible; a low ridge extends from protoconid to the lingual edge of m2, and in all other specimens except USNM 7940 the ridge is uninterrupted by a metaconid cusp. The anterolabial corner of m2 is slightly protruded in some individuals (AMNH 6856).

The m3 is preserved only in UCMP 9999, where it is a very reduced version of m2. It is rectangular in occlusal outline, not circular. The protoconid and hypoconid are labially placed; there is a small paraconid.

Depth of the lower jaw below the m1 hypoconid (in mm) is: AMNH 6855, 30.7; AMNH 6856, 26.8; AMNH 6857, ∼25.4; USNM 7940, 20.6 (juvenile); UCMP 9999, 23.3.

The rostrum described by Cope (1884: 904–905, pl. 70, fig. 11) is much less well preserved than his description implies. The left upper and lower jaws are still in matrix and the teeth badly shattered so that preparation is unwarranted. The right upper dentition, which Cope figured, is so badly damaged that it would be useless for diagnosis were it not for the associated lower jaws. The only undamaged tooth is M2.

However, the intact upper teeth of UCMP 9999, described by Merriam (1906), can be used to supplement the description of Cope's rostrum. UCMP 9999 from Logan Butte has a gracile dentition relative to the Turtle Cove material, which forms the greater part of the T. altigenis hypodigm. UCMP 9999 may prove eventually to represent a somewhat older and more primitive species but is here considered a gracile female morph of T. altigenis.

The upper premolars of UCMP 9999 are without accessory cusps and were not as crowded nor as tall as the lower premolars. They are triangular, narrow teeth with a slight widening at the posterior base of P3. P4 is a shearing carnassial, although the protocone is enlarged, anticipating the marked change in P4 form that occurs in the younger, more derived temnocyonines. There is little to distinguish the T. altigenis P4 from that of Daphoenus, other than the more prominent protocone in the former. P4 is surrounded by a distinct basal cingulum, more defined lingually.

M1 is intact only in UCMP 9999 and USNM 7940 where it demonstrates the plesiomorphic temnocyonine pattern. Portions of M1 occur in UCMP 1549 and AMNH 6857 where they indicate a tooth of the same type. The protocone is low and anteriorly placed as in Daphoenus, not yet completely isolated on a broad enamel flat as in younger, more derived temnocyonines. The cingulum anterior to the protocone was not swollen as it is in more derived species, an indication that this carnivore retained a highly sectorial carnassial. However, temnocyonine hallmarks have already appeared: the lingual cingulum is slightly swollen, and the paracone and metacone are prominent with a developed parastylar region. Consequently, both the labial and lingual parts of M1 are enlarged, creating the impression that the tooth is somewhat constricted in its middle part in occlusal view. The protocone of M1 is connected by a preprotocrista to the anterior cingulum, a feature retained in most larger, derived species of the genus. A small paraconule is present on the preprotocrista. The postprotocrista is absent and there is no metaconule. The configuration of M1 in UCMP 9999 is the most plesiomorphic of any North American temnocyonine. From this pattern evolved the enormous crushing M1s of the more dentally derived Arikareean species of the subfamily.

M2 is much smaller than M1. Its metacone is more reduced, and its protocone is isolated on an enamel shelf surrounded by a thickened lingual cingulum. Length of the upper toothrow of AMNH 6857 from the anterior alveolar border of P1 to the posterior edge of M2 is 67.2 mm: crushing has to some extent altered the actual distance, but in UCMP 9999 where distortion is less, the same measurement is 67.3 mm. The only significant difference warranting mention between the rostrum of AMNH 6857 and UCMP 9999 is that the upper canine is larger and more robust in AMNH 6857, suggesting it is a male, and more slender and gracile in UCMP 9999, presumably a female.

The Logan Butte skull (UCMP 9999) preserves the basicranium in which the auditory region shows a deeply embayed basioccipital bone containing an enlarged inferior petrosal venous sinus that conforms to the amphicyonid pattern (see Basicranial Anatomy).

I refer to T. altigenis a well-preserved, undistorted skull (figs. 7, 8, UCMP 1549), lacking the basicranium, mentioned by Merriam (1906: 22–24, 29) but not figured. The skull was collected near the town of Monument during the 1899 University of California–Berkeley expedition to the John Day beds. Merriam considered this skull merely an uncrushed version of UCMP 9999 from Logan Butte; its large canines, broad rostrum, and inflated frontal region suggest that it represents an intact male cranium of the species.

In dorsal view UCMP 1549 is a robust skull, similar in overall form to the much larger skull of Mammacyon ferocior (F:AM 54134). The inflated frontals and expanded rostrum swollen around the large canines are believed to be male traits. The braincase of UCMP 1549, measuring 55 mm in greatest width, is surmounted by a low sagittal crest whose height is undetermined (the occiput is missing). Skull width decreases to 35 mm at the postorbital constriction, its narrowest point, and the width of the expanded frontal sinuses at the level of the postorbital processes is 46 mm. Anterior to the frontal region and the orbits, the maxillae are shallowly depressed, but in front of these depressions the snout expands due to the swelling of the maxillary region around the large canine roots: width of rostrum at the base of the canine roots is 40.7 mm. Both zygomatic arches are lost as is the occipital region and basicranium posterior to the basisphenoid. The distance between the infraorbital foramen and the posterior opening of the alisphenoid canal measures 83.4 mm.

I1–3 are preserved on the left side. I2 is somewhat larger than I1, but I3 is considerably larger than I2. I1 and I2 are worn flat at their tips but I3 is grooved obliquely on the posteroexternal face by the lower canine. There is a diastema of 4–5 mm between I3 and the upper canine. Width at the base of I3 (6.1 mm) relative to the more gracile I3 (4.6 mm) of UCMP 9999 reflects the male-female dichotomy.

The canine is robust, its root causing inflation of the maxilla and consequently a widening of the rostrum. Only the canine root and a small segment of the base of each tooth is preserved, so wear grooves cannot be identified. The left canine measures 14.6 mm in anteroposterior length by 9.7 mm in width at the alveolar margin.

About 2 mm posterior to the canine is a small peglike unicuspid P1, measuring 5.8 in length, 4.0 mm in width at the base of the crown.

P2 measures 10.9 mm in length, 4.6 mm in greatest width. It is heavily worn on its posterior surface, is wider behind the main cusp than in front, and has no accessory cusps. The posterior slope is slightly more inclined than the anterior, and a fine enamel ridge runs down each slope.

P3 measures 12.8 mm in length, 5.9 mm in greatest width. It is expanded at its posterior base, has a posterior slope somewhat longer than the anterior, but because of wear it is not possible to decide if accessory cusps were present.

P4 measures 17.8 mm in length of the labial margin, 13.2 mm in greatest width. Although both upper carnassials are damaged, the form of the tooth is discernible, much like P4 of UCMP 9999, although the protocone is somewhat more enlarged than in UCMP 9999. Also, the P4 protocone of UCMP 1549 was somewhat more posteriorly situated relative to the protocone of UCMP 9999, although the difference is quite small. P4 remains a shearing tooth but its height and enlarged protocone foreshadow the change in P4 form that occurs in the more derived temnocyonines from younger rocks.

M1 is 13.5 mm in length, 18.7 mm in width, and is best preserved on the left side. Although the left M1 has a broken metacone, the tooth is nearly intact, whereas most of the right M1 is missing. M1 is worn, the tips of paracone and protocone beveled to flat surfaces. The paracone was larger than the metacone, and the labial portion of M1 bearing the paracone-metacone was elevated, with a developed parastylar region. As in UCMP 9999, the lingual faces of paracone and metacone form a shear surface that descends to an enamel platform forming the lingual half of the tooth. On this enamel flat is situated an apically worn protocone. A distinct ridge (preprotocrista) leads from the protocone to the anterior cingulum; there is no postprotocrista. The protocone region is widened but no more than in UCMP 9999; this expansion is the result of the swollen or thickened posterior and medial parts of the lingual cingulum surrounding the protocone region: there is no expansion of the tooth anterior to the protocone and an anterior cingulum of normal width is developed. Thus a tricuspid m1 trigonid with developed metaconid would be expected in the lower jaw and such a tooth occurs in UCMP 9999. Cope's holotype lower jaw (AMNH 6855) and AMNH 6856, both with an m1 metaconid, occlude well with this skull. A small cuspule on the posterior border of M1 may represent a vestigial metaconule.

M2 measures 8.4 mm in length, 12.0 mm in greatest width. M2 is a small replica of M1 except that the protocone is more centrally located within the enamel flat forming the lingual half of the tooth. The paracone is slightly larger than the metacone. No conules seem to be present but the tooth is well worn. The protocone was apparently a prominent knoblike cusp, which has been worn to a flat surface; it was surrounded by a slightly swollen lingual cingulum. As in UCMP 9999, the M2 is situated on a plane slightly above M1 in order to occlude with the m2, which was elevated and tilted forward on the ascending ramus of the mandible.

No M3 was present. M2 is the most posterior tooth in the maxilla and there is no space available for an M3.

The orbital region of the skull is intact and is characterized by a deeply excavated trough, which in its posterior part contains the sphenorbital fissure and the anterior foramen for the alisphenoid canal (the foramen rotundum in the beardog opens internally into the alisphenoid canal as in Ursus and Canis, hence cannot be seen). They are placed almost side-by-side where they perforate the sidewall of the skull, similar to their location in Ursus. Dorsal and anterior to this is a smaller optic foramen that also lies within the elliptical depression that includes the sphenorbital fissure and foramen for the alisphenoid canal. Leading directly forward beneath the optic foramen is a low horizontal ridge which extends craniad ∼35 mm to end directly above the sphenopalatine foramina. This ridge probably marks the dorsal extent of the pterygoid musculature and corresponds closely to the palatine-orbitosphenoid suture. A second ridge runs upward at a 45° angle to the first, beginning above the sphenorbital fissure and coursing anterodorsad above the optic foramen. It forms the dorsal border of the elliptical depression housing these orbital foramina and terminates ∼18 mm below the postorbital process. This same arrangement of ridges and foramina occurs throughout the Temnocyoninae and is evident in F:AM 54134 (Mammacyon ferocior) and YPM 10065 (Temnocyon subferox), and is suggested in YPM-PU 10787 (T. ferox) in which the orbital region is badly crushed.

Structure in the vicinity of the foramen ovale is similar in other members of the subfamily. The foramen ovale and posterior opening of the alisphenoid canal are contained in a deep common fossa. Anterior to this fossa is a shallow depression for pterygoid musculature, which is even more pronounced in T. subferox (YPM 10065) and Mammacyon (ACM 34–41, F:AM 54134). Medial to the common fossa for the foramen ovale and alisphenoid canal is an oval percussion fracture in the sphenoid bone, evidence that the missing basicranium may have been lost to a scavenger. Lateral to the foramen ovale is a remnant of the glenoid fossa. Little can be inferred about the middle ear cavity and surrounding basicranial structures other than that the anterointernal corner of the auditory region was deep, suggesting the presence of an auditory bulla similar to that known in Temnocyon (YPM 10065) and Mammacyon (ACM 34–41).

The skull of UCMP 1549 measures 146.5 mm in length from a point between the first incisors to the posterior border of the foramen ovale. The approximate distance between the same points on UCMP 9999 is ∼130 mm. Basilar skull length of UCMP 9999 is ∼167 mm. Assuming similar proportions for UCMP 1549, basilar length would be ∼180 mm.

The rostrum of a juvenile animal (USNM 7940) from the John Day beds collected by William Day in 1883 includes part of the deciduous (DP2–3, DC) and permanent (P1, P4–M2 erupted) dentition. This animal agrees most closely with UCMP 1549 in the form of its P4–M1. P4–M1 are somewhat less sectorial than these teeth in UCMP 9999. This is evident in the more expanded protocone region of M1 and in the larger P4 protocone of USNM 7940. The unworn condition of these newly erupted teeth shows that initially the upper carnassial of T. altigenis had a modest cingulum entirely surrounding the tooth that was only weakly developed on the lingual face of the protocone. M1 is encircled by a cingulum as well, which is most pronounced posterolingual to the protocone and on the posterior face of the tooth but remains thin on the anterior face (as in UCMP 9999 and 1549). This condition of the cingulum clearly represents the plesiomorphic state prior to massive enlargement of the lingual cingulum in younger, larger temnocyonines.

Discussion

Temnocyon altigenis Cope is the most dentally plesiomorphic species of North American temnocyonines, and also the smallest in body size. The hypodigm was collected from the John Day beds of Oregon; none are known from the Great Plains or elsewhere in North America. Skulls attributable to the hypodigm indicate carnivorans with basilar lengths of ∼16 to 18 cm (table 7).

The dentition of T. altigenis shares similarities with the teeth of the late Eocene-Oligocene amphicyonid Daphoenus. The principal differences are: (a) T. altigenis has all teeth larger than those of Daphoenus vetus of Orellan age despite similar skull size; (b) T. altigenis has taller premolars than Daphoenus (excluding P1/p1); (c) the T. altigenis P4 is larger and has a more robust lingually extended protocone relative to Orellan Daphoenus; (d) M1–2 of T. altigenis have expanded or swollen lingual cingula not found in Daphoenus; (e) T. altigenis has lost M3 which Daphoenus retains; (f) lower molars of T. altigenis lack entoconids, and have an enlarged m1 hypoconid and only an m2 protoconid and hypoconid relative to Daphoenus, which has a plesiomorphic amphicyonid m1–2; (g) T. altigenis has lost the m2 metaconid that Daphoenus retains.

Merriam's (1906) associated skull, mandibles, and partial postcranial skeleton from Logan Butte (UCMP 9999), is here interpreted as a small female of the T. altigenis hypodigm. The gracile canines, less swollen snout, less expanded frontal region, and smaller overall skull size are believed to indicate the female morph. Logan Butte is an isolated John Day outlier far distant southwest of Turtle Cove where most other T. altigenis fossils were found. Eventually UCMP 9999 may prove a distinct species—the most plesiomorphic North American temnocyonine—but for the present it is placed within T. altigenis as its teeth and dental measurements do not merit separation.

Temnocyon altigenis from the John Day beds shares some dental characteristics with European haplocyonine amphicyonids from the Aquitanian Allier Basin localities of St-Gérand, France. A holotype mandible of Haplocyon elegans (MNHN-SG 395: Bonis, 1966: pl. 3, fig. 4; Viret, 1929: pl. 8, fig. 1) displays teeth and dental measurements much like those of the Logan Butte jaw of T. altigenis (UCMP 9999). The holotype of Haplocyon crucians (MNHN-SG 404: Viret, 1929: pl. 8, fig. 2), a partial ramus with p2–4, and an m1 (MNHN-SG 403) are comparable to Cope's holotype of T. altigenis (AMNH 6855). This similarity between haplocyonine and temnocyonine dentitions had been previously noted by Viret (1929) and Bonis (1973).

In 1992 I was able to study Aquitanian haplocyonines in Paris through the courtesy of L. de Bonis and L. Ginsburg. Of all haplocyonines, Haplocyon elegans and H. crucians from St-Gérand are dentally closest to John Day T. altigenis, which is the most plesiomorphic North American species. However, subtle yet evident dental differences distinguish the St-Gérand forms from John Day T. altigenis—although similar in mandibular length, depth and tooth size, H. elegans (MNHN-SG 395) is dentally more plesiomorphic than UCMP 9999: (1) its m1 metaconid is not as reduced as in UCMP 9999; (2) the more plesiomorphic m2 trigonid is taller and wider, with fully developed metaconid, a low talonid with labially placed hypoconid, and an internal talonid shelf. In UCMP 9999 there is no m2 metaconid, the talonid is nearly equal in height to the trigonid, and the hypoconid nearly fills the talonid, which is not the case in MNHN-SG 395. In all the dental features where UCMP 9999 is more derived than MNHN-SG 395, the small Logan Butte carnivore closely approaches Cope's larger holotype of T. altigenis (AMNH 6855), suggesting a plausible male-female relationship between the two John Day fossils.

Finally, the age of the derived Logan Butte Temnocyon altigenis, found in proximity to tuffs dating from ∼28.8–29.3 Ma, seems in conflict with the probable younger age (within MP29-MN2: ∼26–20.5 Ma) of the more plesiomorphic Haplocyon elegans jaw from the St-Gérand basin.

Temnocyon subferox, new species

Figures 9, 10, 66

Fig. 9.

Temnocyon subferox (YPM 10065), John Day Formation, Oregon. Holotype cranium in (A) lateral and (B) high lateral oblique views. Rostrum restored to the left of dashed line.

Fig. 10.

Holotype of Temnocyon subferox (YPM 10065), John Day Formation, Oregon. A, maxilla with P4–M2; B, basicranium with remnant of ectotympanic bulla (stereopairs with negatives reversed). See figs. 66, 69. For abbreviations, see p. 5.

Temnocyon altigenis: Thorpe, 1922: 167–168.

Temnocyon altigenis (in part): Hough, 1948: 100–101.

Type

YPM 10065, nearly complete skull with basicranium but lacking the rostrum anterior to P3; from the John Day Formation, John Day River, Oregon (a catalog entry at the Peabody Museum, Yale University, reads “Middle John Day”); teeth present include right P3–M2, left P4–M2. Collector or date of collection unknown, but cataloged in 1914.

Distribution

Earlier Arikareean, John Day Formation, Oregon.

Etymology

From the Latin, sub, for “less than, under” and ferox, “fierce,” in the belief that the species precedes Temnocyon ferox.

Diagnosis

Cranium and teeth intermediate in size (basilar length, ∼21 cm) between T. altigenis (UCMP 9999, 1549, basilar lengths ∼16–18 cm) and T. ferox (YPM-PU 10787, basilar length, ∼26 cm) and without strong inflation of the frontal region (table 7). Differs from other species of Temnocyon (except T. altigenis) by (a) smaller size and more plesiomorphic P4–M2 in which a marked expansion of M1 protocone region has not occurred (ratio A/B, 1.44, table 6); (b) P4 sectorial with protocone placed in advance of paracone (more dentally derived species of Temnocyon have protocone directly lingual to paracone). No preprotocrista on M1 (present in T. altigenis). Retains most plesiomorphic auditory bulla of any known temnocyonine (see discussion of Basicranial Anatomy).

Referred Specimens

None.

Description

The skull was first mentioned and briefly described by Thorpe (1922) in a discussion of doglike canids and amphicyonids in the Marsh collection at Yale. Later, Hough (1948: 101—“a specimen in the Peabody Museum, Yale University”) mentioned the auditory region and recognized the presence of a remnant of the auditory bulla (she does not cite a catalog number but her description can apply only to YPM 10065). Neither Hough nor Thorpe illustrated the skull. The basicranium shows the deeply embayed basioccipital bone typical of amphicyonids, thought to contain an enlarged inferior petrosal venous sinus. The auditory region is reviewed and illustrated in the section on Basicranial Anatomy.

Although the skull is uncrushed and its form intact, the bone itself is fractured and altered by diagenesis after burial so that anatomical detail is lacking. The dolichocephalic skull has a gracile appearance with slender zygomatic arches and a long postorbital distance (orbital margin to occpital condyle is 14.3 cm). The frontal region is not inflated, thus similar to T. ferox, contrasting with the expanded frontal region of T. fingeruti. As in T. altigenis, the maxilla is slightly depressed above the prominent infraorbital foramen. Foramina of the orbital region are as described for T. altigenis: an optic foramen, sphenorbital fissure, and anterior foramen for the alisphenoid canal open into the elongate depression that begins 18 mm anterior to the foramen ovale. The posterior opening of the canal and the foramen ovale share a common fossa. As in UCMP 1549 the foramen rotundum must open internally into the enclosed alisphenoid canal. Despite the loss of the rostrum the palate is preserved from P3 to M2; the cheek teeth seem quite small given the palatal area and skull size (greatest palatal width at P4–M1 is 6.5 cm; P4–M2 occupy 58% of this width). The mandibles must have been long and slender because of the modest glenoid fossa for a rather small narrow condyle, the small cheek teeth, and the considerable distance (∼9 cm) from the glenoid fossa to the carnassial.

The left P4–M2 and right P3 survive whereas P1–2, incisors, and canines did not.

P3 is laterally compressed, slightly expanded posteriorly, and is taller than in T. altigenis and T. ferox. A fine enamel ridge runs from the principal cusp down the posterior slope to the basal cusp, the latter more prominent than a tiny posterior accessory cusp 2 mm anterior to the basal cusp. There are weak labial and lingual cingula that meet posteriorly to form the prominent basal cusp. P3 length, 12.9 mm; greatest posterior width, 6.6 mm.

P4 has a well-developed protocone, a sectorial paracone/metastylar blade, and is clearly a shearing carnassial. The metastylar blade is placed at a 45° angle to the anteroposteriorly aligned paracone. Length of the blade, 8 mm; paracone length, 10.3 mm. The protocone, lingual and slightly anterior to the paracone, is enlarged as is the protocone of T. altigenis. In T. subferox there is a broad embrasure between P4 and M1 into which the shearing m1 trigonid fits during carnassial occlusion. Dimensions of this embrasure in T. subferox are similar to those in T. altigenis (UCMP 9999) and represent the plesiomorphic state. Closure of this shearing embrasure between P4 and M1 occurs in the large species of Mammacyon and Delotrochanter where crushing takes precedence over shear.

P4 is encircled by a cingulum that is pronounced at the labial base of the metastylar blade. The cingulum is swollen at the anterolabial corner to form a parastyle, situated 2 mm anterolabial of the base of the enamel ridge descending the anterior face of the paracone. The protocone is a blunt knoblike cusp separated by a valley from the paracone and would have occluded with the heel of p4.

M1 is similar to the T. altigenis M1 except that the protocone region is more derived. A well-developed parastyle contacts the P4 metastylar blade. The labial cingulum is thickest at the parastylar region and thins posteriorly yet is still well defined along the entire labial margin of M1. The paracone is somewhat larger than the metacone. The internal faces of paracone-metacone form a vertical shearing surface that descends steeply to an enamel flat forming the lingual half of the tooth. A knoblike protocone is isolated on this enamel flat: no preprotocrista runs from the protocone across the enamel flat to the anterior cingulum as in the more plesiomorphic T. altigenis. This differs from T. ferox (YPM-PU 10787) that retains not only a pre- but also a weak postprotocrista.

The knoblike M1 protocone and its enamel flat are surrounded by a prominent cingulum, strongly thickened on the lingual margin, slightly less developed on the posterior margin, and thin but distinct anteriorly. Development of a swollen anterolingual M1 cingulum in temnocyonines accompanies loss of the m1 metaconid during closure of the embrasure between P4 and M1. There is no metaconule; a paraconule may be represented by a small cingular cusp on the anterior edge of M1 midway between protocone and paracone.

M2 is much smaller than M1 but remains a fully functional quadrate tooth with a weak parastyle. The paracone and metacone are elevated as in M1 but here the metacone is much smaller than the paracone. The lingual half of the tooth is a low enamel platform with a low blunt protocone. The tooth is surrounded by a prominent cingulum, lingually thickened as in M1.

M3 was not present in life: the maxillary border is smooth and unbroken posterior to M2, and there is no evidence of alveoli or roots for M3.

Discussion

The skull, which is of considerable importance because of the basicranium, has been long ignored except for Hough's (1948) brief mention of its auditory region. It was first described by Thorpe (1922), who placed it in T. altigenis. However, the skull is much larger than the known crania of T. altigenis (UCMP 9999, 1549) and exceeds any reasonable upper limit of skull size predicted for the T. altigenis sample from the John Day region. Unfortunately, Yale University collection records for the skull do not establish an exact location, a collector, or a date of collection, only that it was cataloged in 1914 as YPM 10065, and that it may have come from the “Middle John Day.” Whether the skull was found by collectors employed by O.C. Marsh in the late 19th century is unknown but is a possibility. It is surprising that such a well-preserved and unusual carnivore did not merit a more detailed description of its site of collection, which suggests it was found during one of the early Yale expeditions when exact locality data were rarely obtained. The stage of evolution of the skull relative to other North American temnocyonines indicates that a late Arikareean assignment is unlikely. An earlier Arikareean age is most probable, postdating the T. altigenis hypodigm in the John Day Formation.

The plesiomorphic dentition and weakly inflated frontal region of the skull make it a possible predecessor to T. ferox (YPM-PU 10787) from the “upper John Day beds.” The lack of pre- and postprotocristae on M1 in YPM 10065, and their presence in T. ferox, seem to conflict with this view. However, with acquisition of larger samples, protocristae may prove to be variable features on the molars of these carnivorans. The morphological “distance” between T. subferox and T. ferox is greater than between skulls of T. subferox and T. altigenis, suggesting that substantial time intervened between YPM 10065 and T. ferox.

The skull of YPM 10065 is important to the determination of the broader relationships of temnocyonines since the basicranium and auditory bulla are preserved and largely undistorted. The bulla of T. subferox represents the most plesiomorphic condition found among temnocyonines. The auditory bulla in North American temnocyonines is known only in T. subferox (YPM 10065), in T. fingeruti (NM 280/61), and in Mammacyon obtusidens (ACM 34–41). These bullae are more fully discussed in the section on Basicranial Anatomy.

No mandible was associated with the skull of T. subferox (YPM 10065), and no isolated jaws from the John Day beds occlude satisfactorily, but there are two mandibles of North American temnocyonines that correspond in size: (a) LACM 15908, a partial mandible from the Sharps Formation of South Dakota; (b) LACM 470, a crushed mandible from Kew Quarry, Sespe Formation, Las Posas Hills, California.

The Sharps mandible (LACM 15908) is unlikely to belong to the same species as the holotype skull despite the size correspondence: the m2 protoconid and hypoconid are placed in the center of the tooth, not on the labial margin as in Temnocyon. The m1 metaconid of the Sharps jaw is reduced, more than expected for an m1 that would occlude with the shearing P4 of the holotype skull. This jaw likely represents an early member of the Mammacyon lineage. On the other hand, the Kew Quarry jaw includes a tall p3–p4 that corresponds to the tall P3 in YPM 10065. Here the Kew Quarry mandible is referred to T. cf. T. subferox pending discovery of associated upper and lower dentitions of the John Day species.

Temnocyon cf. T. subferox

Figure 11

Fig. 11.

Temnocyon cf. T. subferox (LACM 470), right mandible with p3–m2, damaged c, p1–2, from Kew Quarry, Sespe Formation, Ventura Co., California. This is the only record of the subfamily in western North America other than fossils from the John Day region.

Temnocyon cf. T. altigenis: Stock, 1933b: 35–37, pl. 1, fig. 6.

Referred Specimen

LACM 470, right mandible with c, single-rooted p1 alveolus, alveoli of p2, damaged p3–m2, including anterior part of ascending ramus, from Kew Quarry, Las Posas Hills, CIT loc. 126, Sespe Formation, Ventura County, California, collected by Thurston, 1930, early Arikareean.

Description

The crushed and fragmented lower jaw was accurately described and figured by Stock (1933b: 35–37). Premolars are closely spaced but not crowded, and p3–4 are tall, more so than in T. altigenis. The m1 is larger than any T. altigenis carnassial (table 2): both trigonid and talonid are broadened relative to T. altigenis. The m1/m2 length ratio is estimated at ∼1.8, similar to T. ferox, taking into account damage to both molars—there is a relatively short plesiomorphic m2. The preserved lower teeth (p3–m2) are all slightly wider and more robust than these teeth in T. altigenis (table 2). The mandible is only 25 mm in depth below the hypoconid of m1, its width much compressed by crushing.

Discussion

The Kew Quarry mandible from the Sespe Formation displays differences in size and proportion of the posterior premolars and molars that argue against its placement in T. altigenis. The upper teeth of Temnocyon subferox (YPM 10065) satisfactorily occlude with the teeth of the Kew Quarry jaw, and indicate a carnivore of approximately the same size. However, there is no associated mandible with the holotype skull of T. subferox (YPM 10065) nor is there any other mandible of appropriate size from the John Day beds that can be referred to the skull. The dentary itself is not diagnostic but is quite shallow as would be predicted for T. subferox.

An important contrast exists between the Kew Quarry mandible and the holotype mandible of Delotrochanter petersoni n. sp. (CM 1603). CM 1603 belonged to a carnivore only slightly larger than the Kew Quarry animal, yet its p2 is short, with a large posterior and a small anterior root, placed very close together. The Kew Quarry p2 is elongate, and its two roots are of about equal size and spaced well apart. Thus the Kew Quarry carnivore does not approach Delotrochanter in premolar form, since the p2 is a key derived trait of the latter genus, nor does its m1 show the straight (as opposed to sinuous) labial cingulum found in Delotrochanter. The Kew Quarry animal cannot be referred to Mammacyon because it lacks shared derived dental traits of that genus: the notched lingual margin of the m1 talonid and the elongate m2. It is more likely that the Kew Quarry mandible belongs to a carnivoran similar to the John Day Temnocyon subferox that retained the shearing dentition typical of the Temnocyon lineage. Fossil mammals found with LACM 470 in Kew Quarry favor an early Arikareean age for the quarry assemblage (see discussion in Age and Correlation).

Temnocyon ferox Eyerman, 1896

Figures 12–Fig. 13.14, 53

Fig. 12.

Holotype of Temnocyon ferox Eyerman (YPM-PU 10787), from “near Maginnis' ranch, Turtle Cove, John Day river, Upper John Day beds, Oregon.” A, Cranium in lateral view; B, maxilla with P3–M2 (white plaster fills voids in teeth); C, Eyerman's (1896) restoration of the holotype cranium and mandible. Note crushing and fragmentation of the cranium in (A).

Fig. 13.

Holotype mandibles of Temnocyon ferox Eyerman (YPM-PU 10787), with both canines and left and right p2–m2. Mandibular bone is heavily fractured. Associated forelimb, figure 53.

Fig. 14.

Occlusal view of the holotype mandibles of Temnocyon ferox Eyerman (YPM-PU 10787), with left p2–4, m1–2, right p3–4, m1–2. An m1 metaconid is evident on both carnassials (stereopair).

Temnocyon ferox Eyerman, 1896: 268–284, pl. 11, figs. 1–9.

Temnocyon ferox: Wortman and Matthew, 1899: 115–118, fig. 3.

Temnocyon ferox: Merriam, 1906: 22–23, 29.

Type

PU 10787, as given in Eyerman (1896), a nearly complete articulated skeleton with associated skull and lower jaws, the basicranium badly damaged. Found in the same block with a partial oreodont skeleton, “near Maginnis' ranch, Turtle Cove, John Day river, Upper John Day beds, Oregon,” collected by L.S. Davis, 1889. The type is now conserved in the Yale Peabody Museum collection and is numbered YPM-PU 10787.

Distribution

Mid- or late Arikareean, John Day Formation, John Day valley, Oregon.

Diagnosis

Wolf-sized species of Temnocyon (basilar skull length, ∼26 cm), distinguished from T. altigenis and T. subferox by much larger size, from T. macrogenys by much smaller size, and from the Mammacyon obtusidens–M. ferocior group by retention of the m1 metaconid and possession of a short m2 relative to m1 (ratio E/F, ∼1.6 in Mammacyon; ∼1.8 in T. ferox). T. percussor is distinguished from T. ferox by its taller, broader p3–4 and by its larger more robust m1–2 and M1 (tables 2, 3). T. ferox and T. percussor share a similar m1–2 form but T. percussor represents a larger species. T. fingeruti differs from T. ferox in its more elongate, narrower m1, longer m2, disproportionate size of M1–2 (small M2 relative to M1 in T. ferox; large M2 in T. fingeruti); dental ratios (table 6); and more inflated frontal region.

Referred Specimens

None.

Description

Eyerman described and figured the type specimen of T. ferox (1896: 268–279, pl. 11), reporting that it came from the upper John Day beds, Turtle Cove, Oregon, and so I include here only relevant observations to facilitate comparison with other temnocyonine species of similar size.

The skull and associated mandibles are crushed and fragmented, held together by an extremely hard, indurated gray tuff that preserved the articulated skeleton. Sutures and other fine anatomical details are obscured and distorted. The skull has a basilar length of ∼26 cm. The snout is of moderate length: preorbital distance, ∼10 cm; postorbital ∼16 cm. Relative to Canis lupus the preorbital distance is proportionately shorter and the postorbital longer. The frontals are little inflated, much less so than in T. fingeruti, a carnivore of about the same size. The basicranium unfortunately is so damaged that no details can be identified.

The dentition reflects the plesiomorphic pattern seen in T. altigenis and T. subferox, is simply scaled to larger size, differing most notably in the form of M1–2. The p1 is represented by a single alveolus; p2 and p3 are not robust as in Mammacyon and Delotrochanter but rather low as seen in T. altigenis from Logan Butte—they carry no posterior accessory cusps and have thin enamel ridges traversing their anterior and posterior faces; p4 is distinctly larger than p3 and has a labially placed posterior accessory cusp—p4 has a squared heel with a centrally placed cingulum cusp and is elongate relative to m1 length. The m1 is a short, rather broad carnassial retaining a slightly reduced metaconid; the m1 talonid is nearly entirely occupied by the massive hypoconid (there is no entoconid); the hypoconid is much lower in height than the m1 paraconid. The m2 is short, a striking feature of the species; m1/m2 length ratio is ∼1.8. The m2 trigonid is dominated by a large protoconid and the talonid by a hypoconid, both labially placed—there is no metaconid and only an extremely reduced paraconid; the trigonid width is 8 mm, and the talonid narrows to a width of 6.6 mm. The m1 hypoconid and m2 protoconid-hypoconid form a linear series of three blunt, crushing cusps.

The upper premolars are also low-crowned, gradually increasing in height from P1–3. Only P3 has a posterior accessory cusp, very small, somewhat labial in position above a slightly expanded heel with a squared posterior border. Although the upper carnassials are damaged, the metastylar blade was apparently short and the protocone modestly enlarged. M1 still retained a parastyle, and a prominent cingulum surrounded the tooth; the lingual cingulum was thickened, enlarging the protocone region, more so than in T. subferox. M1 retains both pre- and postprotocristae and the protocone is situated almost exactly in the center of the expanded lingual half of the tooth. M2 is much smaller than M1 and has markedly reduced the metacone; M3 is absent.

Relative to the wolf the upper and lower canines are larger whereas the incisors are the same size. Each lower incisor shows a single accessory cusp (Nebenzacke) on its labial margin.

A partially articulated postcranial skeleton, including the limbs, fore- and hind feet, was found in association with the skull and mandibles and was described by Eyerman (1896). Although difficult to free from the indurated matrix, the skeleton was prepared to demonstrate its limb proportions and digitigrade paraxonic fore- and hind feet. The limbs are discussed in the section on Postcranial Osteology.

Discussion

Temnocyon ferox is distinguished from T. altigenis and T. subferox by larger size, similar to a living wolf, but shares with those species a plesiomorphic temnocyonine dentition. These three species probably form a morphocline in which body size increases through time. T. ferox, however, is the first species in this lineage to show specialization of its carnassials and molars, while still retaining simple low premolars that distinguish this species from T. percussor, T. macrogenys, and the species of Mammacyon and Delotrochanter that have larger more developed premolars. The presence of the m1 metaconid and short m2 are diagnostic features of the molars distinguishing T. ferox from temnocyonines of similar size such as Mammacyon obtusidens (which has no m1 metaconid and a larger, elongate m2) and Temnocyon fingeruti. T. ferox can be distinguished from T. fingeruti using M1–2: the T. ferox M1 shows a more developed lingual cingulum and M2 is much reduced whereas in T. fingeruti the M1 shows only modest expansion of the lingual cingulum and M2 is large. However, the M1 lingual cingula of both T. ferox and T. fingeruti are much less developed relative to the hypertrophied lingual cingula of Mammacyon and Delotrochanter.

Upper carnassials of T. ferox are damaged but appear to be sectorial teeth like those of T. fingeruti; however, P4 has a prominent protocone more lingually situated than in T. subferox.

The teeth of Temnocyon ferox are similar to the available teeth of T. percussor (p3–m2, M1). The size increase from T. ferox to T. percussor is characterized by development of a taller p3–4 and more robust m1–2 in T. percussor.

Temnocyon ferox is one of the few temnocyonines for which there is a confirmed association of the cranium and an articulated postcranial skeleton. Eyerman (1896) was aware of the evident disproportion between skull and limbs: the skull was larger than that of the average wolf yet the limbs were noticeably shorter. Limb elongation is nonetheless present, although not to the degree seen in the wolf. When compared with the limb and foot skeleton of the cursorial wolf, T. ferox shows a number of parallel adaptations (see Postcranial Osteology).

Forelimb elongation in Temnocyon ferox is evident relative to the plesiomorphic forelimb of T. altigenis and Daphoenus. As in the wolf, the forelimb of T. ferox exhibits a narrow distal humerus with reduced medial epicondyle, a closely apposed distal ulna and radius (i.e., articular process of the ulna reduced to a smooth planar facet pressed against the distal radius), and a digitigrade paraxonic forefoot with narrow carpus (metacarpals 2 and 5 short; metacarpal 1 reduced). In the hind foot there is a narrow elongated tarsus; long paraxonic metatarsals 3–4; shorter metatarsals 2 and 5; and metatarsal 1 a thin rodlike bone 45 mm in length (not as reduced as the minute triangular nubbin, 1 cm in length, that is the vestige of the first metatarsal of the wolf). Paraxonic metapodials of the fore- and hind foot of the wolf are ∼2 to 2.5 cm longer than those of T. ferox. The lengthened radius and ulna and the long paraxonic metacarpals are responsible for the more elongate forelimb of the wolf.

Temnocyon percussor Cook, 1909

Figure 15

Fig. 15.

Temnocyon percussor Cook, Niobrara River valley, Sioux Co., Nebraska. Holotype (AMNH 81005) left partial mandible with p3–m2 (A, B) and right mandibular fragment with p1, p4, and m2 (C, D) from 0.5 mi west of Agate Ranch headquarters, Harrison Formation, Sioux Co., Nebraska. A, D, labial and B, C, lingual views. Referred right m1 (E, AMNH 81054) and right M1 (F, AMNH 81047) from AMNH–Cook Quarry, basal Anderson Ranch Formation. Cook's 1909 drawing of the holotype is a composite of the two mandibular fragments (A to D) illustrated here, both from the same individual.

Temnocyon percussor Cook, 1909: 266, 271, fig. 3.

Type

AMNH 81005, associated right and left partial mandibles, left p3–m2, right p1, p4, m2, roots or alveoli of p2–3, m1, and a partial m3 alveolus, a doubtfully associated right upper canine, comprising Cook's (1909) type of Temnocyon percussor, found “about one-half mile west of Agate, Nebraska, in the Lower Harrison beds ….”, Sioux County, Nebraska (see Discussion for additional information on site of collection). Cook's figure of the type is a composite based on both mandibles, although this is not plainly stated in his publication.

Distribution

Late Arikareean, upper Arikaree Group, Sioux County, Nebraska.

Diagnosis

Differs from T. altigenis and T. subferox in larger size and from T. macrogenys in much smaller size. Distinguished from T. ferox by taller differently proportioned p3–4 and by dimensions of m1–2 and M1 that together exceed the size attributable to a T. ferox population (tables 2, 3). T. percussor differs from T. fingeruti in having a taller more robust, somewhat longer p3–4, shorter m1–2, and no m2 metaconid. M1 lingual cingulum better developed in T. percussor relative to T. fingeruti.

Referred Specimens

(1) AMNH 81054, right m1, American Museum–Cook Quarry (Hunt, 1972), 2 mi north of the Agate Spring Quarries, basal Anderson Ranch Fm., Sioux County, Nebraska, H. Cook, August 1909; (2) AMNH 81047, right M1, American Museum–Cook Quarry, basal Anderson Ranch Fm., Sioux County, Nebraska, H. Cook, July 1908.

Description

AMNH 81005—Cook (1909: 266) only briefly described this individual but presented an accurate illustration of the dentition. This illustration posterior to p3 is based on the left mandibular fragment, and anterior to p3 on the anterior part of the right mandible. It is not possible to estimate jaw depth since the lower border of the mandible has been broken away along the length of the horizontal ramus. However, an estimate of mandibular length, based on length of toothrow (c–m2, 110.9 mm), compares with the mandibular length of Eyerman's type of T. ferox (c–m2, ∼102 mm), an observation complemented by the similar premolar spacing in both animals.

The p1 measures 8.4 mm in length, 5.0 mm in width. It is a small, single-rooted peglike tooth situated on the jaw margin anterointernal to the anterior root of p2. The main cusp is anteriorly inclined, and there are no accessory cusps.

The p2 is not preserved in either mandible; the alveolar measurements of length and width are 14.8 and 7.2 mm. The anterior root is more labially placed than the posterior root, the tooth being directed outward, a trait also seen in several other temnocyonines.

The p3 measures 15.9 mm in length, 7.8 mm in width. There are no posterior accessory or basal cingulum cusps. The tooth in lateral view is nearly an equilateral triangle, the posterior slope being only slightly longer than the anterior slope. Weak enamel ridges run down the anterior and posterior slopes; there is slight development of a lingual and posterior cingulum that is even less well defined labially.

The p4 measures 19.7 mm in length, 9.0 mm in width. It has the same form as p3, only larger and taller, with the addition of a prominent posterior accessory cusp and a shelflike heel at the rear of the tooth. The posterior accessory cusp is situated about half the distance down the posterior slope and is labial in position. A weak cingulum is present on the lingual and posterolabial faces of p4. A small cingular cusp also occurs on the shelflike heel of p4 where the posterior enamel ridge descends to meet it.

The m1 measures 23.5 mm in length, 10.5 mm in width. Trigonid length, 16.2 mm; talonid length, 7.3 mm. The trigonid appears somewhat inflated: paraconid and metaconid are of about equal height, the taller protoconid elevated only ∼1.5 mm above the main p4 cusp. The metaconid is present, reduced, and situated on the posterolingual corner of the massive protoconid about half the distance to its base. The hypoconid is the only talonid cusp: it is centrally positioned, and occupies nearly all the talonid. A cingulum is developed on labial and lingual margins of the talonid. On the posterior slope of the hypoconid there is a tiny cusp apparently derived from the cingulum.

The m2 measures 13.9 mm in length, 9.0 mm in width. The form of the tooth is most like m2 in YPM-PU 10787 (T. ferox) and differs from the elongate m2 of Mammacyon. The prominent m2 protoconid is the same height as the m1 hypoconid. These two cusps, aligned in series with the much smaller hypoconid of m2, are low, blunt, crushing instruments. A low paraconid occupies the anterolingual corner of m2. A low ridge descends the protoconid to the lingual margin of m2. The m2 talonid is occupied solely by the low hypoconid placed directly behind the much larger protoconid. There is no entoconid.

The isolated upper canine measures 12.5 mm in labiolingual width at the enamel base; anteroposterior length, 17.2 mm; height, 61.2 mm. The tooth is recurved and exhibits a flat wear surface, ∼25 mm in length, 6 mm in width, on the anterointernal face near the tip. The tooth is badly cracked, probably from postmortem drying prior to burial.

AMNH 81054—This isolated right m1 compares well with Cook's type m1 (AMNH 81005) but exhibits more wear. Although a vertical shear facet is evident on the labial face of the protoconid-paraconid, the tips of the protoconid, paraconid, and hypoconid are blunted by apical wear, suggesting mastication of hard material. This tooth measures 24.2 mm in length, 10.8 mm in width. The m1 talonid of AMNH 81054 is nearly identical to that of the holotype m1, including the cingular swelling on the posterior slope of the hypoconid.

Because they retain the m1 metaconid, T. percussor carnassials are easily confused with those of Daphoenodon superbus, the amphicyonid that occurs with T. percussor in the American Museum–Cook Quarry. In T. percussor, however, the hypoconid fills almost the entire talonid, but in Daphoenodon the hypoconid occupies only the labial talonid surface, and does not encroach on a narrow lingual talonid shelf. Isolated teeth in the American Museum–Cook Quarry demonstrate the presence of three beardogs (Temnocyon percussor, Daphoenodon superbus, Ysengrinia americana) at the site.

AMNH 81047—This isolated M1 was also found in the American Museum–Cook Quarry. Although not associated with other material of T. percussor, it has the form and predicted cusp morphology that correspond to mandibular teeth of the holotype. The tooth is an enlarged version of the M1 of T. altigenis and T. ferox, measuring 20.9 mm in length, 27.9 mm in width; hence it lacks the extremely swollen lingual cingulum found in Mammacyon and Delotrochanter. The paracone and metacone are conspicuously elevated above the lingual half of the tooth, forming a vertical shear surface. The paracone is somewhat larger than the metacone; both cusps are bordered by a sharply defined labial cingulum continuous with a prominent parastyle. The protocone is centrally situated on a flat enamel platform, and is less knoblike, more crescentic in form than in Mammacyon. A long preprotocrista runs from the protocone to the anterior cingulum, and a short low postprotocrista extends from protocone to posterior cingulum, similar in this respect to M1 of T. ferox.

The plesiomorphic form of the T. percussor M1 (AMNH 81047) is unique in the late Arikareean; the older Mammacyon lineage and contemporary Delotrochanter have more derived upper molars in which the protocone region forming the lingual half of M1 is more expanded, primarily due to the more swollen lingual cingulum. In AMNH 81047 the anteroposterior length of the protocone region is quite narrow relative to the length of the labial margin of the tooth (table 5), and its anterior cingulum is plesiomorphic (as in T. altigenis) in contrast to the enlarged cingulum of Mammacyon. The narrow M1 protocone region and the lack of a swollen anterior cingulum indicate the presence of a metaconid on the mandibular carnassial, and so AMNH 81047 cannot be referred to Mammacyon or Delotrochanter in which the m1 lacks a metaconid.

Discussion

Temnocyon percussor is restricted to upper Arikaree Group sediments in western Nebraska, and has never been found outside of the Niobrara River valley near Agate in Sioux County. The species is more advanced in dental traits than the John Day holotype of T. ferox: T. percussor has broader, more robust teeth, taller premolars (p3–4), and a much larger yet still plesiomorphic M1. Although the two species are similar in dental features, the Great Plains T. percussor has a slightly longer m2 than found in the John Day T. ferox (ratio E/F is ∼1.7 in AMNH 81005, ∼1.8 in YPM-PU 10787, table 6). Eyerman's type (YPM-PU 10787) of T. ferox represents a smaller individual than T. percussor and a less evolved dental stage, suggesting an older age for the John Day carnivore.

Cook (1909) was aware that the holotype, found by him in 1906, was dentally similar to the John Day T. ferox yet represented a larger carnivore with different p3–4 proportions. The M1 later found by Cook in 1908 is reasonably referred to T. percussor, its large size and form in agreement with the m1–2 of the holotype mandible.

If Cook (1909) correctly attributed the holotype of Temnocyon percussor (AMNH 81005) to the Syndyoceras level or “quarry” one-half mile west of Agate, then the holotype material comes from the Harrison Formation about 40–50 ft (12–15 m) below the upper contact of the formation with the Anderson Ranch beds. The isolated molars (AMNH 81054, 81047) come from the American Museum–Cook Quarry in the basal Anderson Ranch Fm., about 50 ft stratigraphically higher than Cook's AMNH 81005. The Syndyoceras level is late Arikareean in age and occurs in proximity to the Agate Ash, initially 21.3 Ma (KA 481, Evernden et al., 1964), now more recently dated by the ⁴⁰Ar/³⁹Ar method at ∼22.9 Ma (Izett and Obradovich, 2001). The similarity of the teeth of T. percussor and T. ferox suggests a relationship: although the age of T. ferox is uncertain (see Age and Correlation), a date older than 22.9 Ma is supported by its less advanced dentition relative to the more robust teeth of T. percussor.

The presence of T. percussor (with m1 metaconid) in western Nebraska in both the Harrison Formation and basal Anderson Ranch beds in the Agate area, and the occurrence of large Delotrochanter oryktes (without m1 metaconid) in these same two formations, also in the Agate area, demonstrates that the temnocyonine dentition was evolving at different rates in these two contemporary lineages. This had been difficult to prove given the fact that two or more temnocyonine species rarely occur in proximity in the same rock unit. However, based on fossils in this study, two temnocyonine genera had already lost the m1 metaconid prior to the late Arikareean (Mammacyon, LACM 9194; Delotrochanter, CM 1603); hence these represent more derived dentitions relative to a geologically younger Temnocyon percussor that retains its m1 metaconid and plesiomorphic M1 into the late Arikareean interval. If T. percussor directly precedes T. macrogenys, which is probable, then Temnocyon maintains this plesiomorphic dental pattern in the Great Plains until its extinction at the end of the Arikareean.

Temnocyon fingeruti, new species

Figures 16–Fig. 17.Fig. 18.19, 68

Fig. 16.

Holotype cranium and mandibles of Temnocyon fingeruti (NM 280/61), from Balm Creek, Haystack Valley Member (revised), John Day Formation, Wheeler Co., Oregon. This skull was found 62 ft (∼19 m) above a tuffaceous horizon dated to ∼23.8 Ma (Hunt and Stepleton, 2004).

Fig. 17.

Holotype palate and upper dentition (A, left and right I1–3, C, P1–M2; B, stereopair with right P3–M2) of Temnocyon fingeruti (NM 280/61), Balm Creek, Haystack Valley Member (revised), John Day Formation, Wheeler Co., Oregon. Note occlusal wear on P3–M2.

Fig. 18.

Holotype left (A) and right (B) mandibles of Temnocyon fingeruti (NM 280/61), from Balm Creek, Haystack Valley Member (revised), John Day Formation, Wheeler Co., Oregon, both in labial view (left c, p2–m3; right c, p1–m3).

Fig. 19.

Occlusal views of the mandibular dentition of the holotype of Temnocyon fingeruti (NM 280/61). Balm Creek, Haystack Valley Member (revised), John Day Formation, Wheeler Co., Oregon. A, left and right p2–m3; B, stereopair of p3–m3. Note apical wear on premolars and carnassials.

Type

NM 280/61, a complete skull and lower jaws in articulation, the rostrum offset ∼3 cm from the posterior cranium by sediment deformation after burial; both left and right I1–3, C, P1–4, M1–2 and left i3, c, p2–4, m1–3, right c, p1–4, m1–3, from the Haystack Valley Member (revised), John Day Formation, Oregon, Lower Beardog Section, E1/2, NE4, SW4, SW4, sec. 27, T8S, R25E, Kimberly 7.5-minute quadrangle, from 96 ft stratigraphic level in appendix 1–6, Hunt and Stepleton (2004: 81); 62 ft above a radioisotopically dated horizon at ∼23.8 Ma.

Distribution

Latest Oligocene, late mid-Arikareean, John Day Formation, Wheeler County, Oregon.

Etymology

The species name recognizes Michael Fingerut, who discovered and collected the holotype in the Balm Creek drainage, north-central Oregon.

Diagnosis

Wolf-sized species of Temnocyon (basilar skull length ∼25 cm), distinguished from T. altigenis and T. subferox by much larger size, from T. macrogenys by smaller size; differs from similarly sized T. ferox by greater inflation of the frontal paranasal sinuses (relative to braincase width, table 7) and by different form and proportions of p4–m2 and M1–2 (tables 2, 3), including a more lingually extended M1–2 with less expansion of M1 protocone region (ratio A/B, table 6), a much larger M2 and m2, a longer m2 (ratio E/F), and by a more robust, wider p4. Temnocyon fingeruti differs from T. percussor in its much smaller M1 (in the upper dentition of T. percussor only M1 survives), more elongate m1–2, stronger m1 metaconid, and larger, rectangular m2 with vestigial metaconid and more developed paraconid, and by the p4/m1 ratio (0.78 in T. fingeruti vs. 0.84 in both T. percussor and T. ferox). Temnocyon fingeruti differs from T. altigenis, T. subferox, and probably T. ferox in P4 proportions (ratio C/D, table 6); P4 is not known in T. percussor and T. macrogenys.

Referred Specimens

None.

Description

The skull of NM 280/61 is elongate (basilar length, 25 cm, table 7) as are the known skulls of T. ferox and T. subferox but differs from the short, broad skull of Delotrochanter oryktes. Cranial size and proportions are similar to those of T. ferox except for the much greater frontal sinus inflation and entirely different form of P4–M2 and p4–m2 in T. fingeruti. Despite smaller size, a narrower snout and clearly dissimilar dentition, the profile of NM 280/61 compares with the skull form of Mammacyon ferocior. NM 280/61 approximates a large male wolf (Canis lupus) in rostral proportions and palatal dimensions but shows a longer postorbital cranium with broader basicranium, greater frontal sinus inflation, and smaller braincase volume. Although the teeth are similar in occlusal dimensions to those of a wolf, NM 280/61 has larger canines, more robust premolars, with a proportionately smaller shearing m1 and a longer m2–3 crushing platform. As with other John Day crania, sutures and anatomical detail are obscured by diagenetic alteration of bone. The orbital region preserves the orbital foramina nested in the elongate depression previously descibed for T. altigenis (UCMP 1549). An orbital foramen, sphenorbital fissure, and anterior opening for the alisphenoid canal (the foramen rotundum opening internally into the canal) are situated as in that plesiomorphic species, and it is apparent that this anatomical pattern is common to all temnocyonines where the orbital region has been preserved. The basicranium and auditory region are exceptional and are discussed in the section on Basicranial Anatomy.

The mandible is gracile and slender (depth beneath m1, 32 mm; below p2, 27 mm) with toothrow length (p1–m3) of 110 mm. These teeth are worn, indicating a mature but not aged individual; the canines, particularly the lower pair, reflect some damage, with the right lower canine beveled to a rounded stub suggesting it had been broken and smoothed by abrasion during life.

Lower incisors of temnocyonines rarely survive but are preserved in T. fingeruti. The i1–3 were quite small; there is little room (∼1 cm) for them between the canines. During feeding these incisors would have been of little use. A somewhat larger i3 (length, 6.4 mm; width, 3.7 mm) contrasts with i2 (length, 5.9 mm; width, 2.9 mm) represented only by a broken root. There is no evidence of i1 that if present was placed in front of and below i2.

The canine measures ∼19 mm in length, ∼10 mm in width, measured at the enamel base; the symphyseal width is narrow, only 30 mm measured across the base of the canines.

The p1 is a small, peglike tooth (length, 6.9 mm; width, 4.0 mm) and is much worn.

The p2 (length, 12.1 mm; width, 5.6 mm) is much larger than p1 with its posterior slope somewhat longer than the anterior slope. A thin enamel ridge travels down both the anterior face to the anterolingual corner and the posterior face to the posterolabial cingulum. There are no accessory cusps.

The p3 (length, 15.3 mm; width, 6.6 mm) is a larger replica of p2 with a small posterior cingular shelf and a tiny basal cusp; the shelf is indented by wear on the left p3, less so on the right. There are no evident accessory cusps.

The p4 (length, 19.3 mm; width, 8.7 mm) is quite long relative to p3. There is a prominent labially situated posterior accessory cusp above a cingular shelf.

The principal cusps of p2–4 are worn flat by apical abrasion. The p2–3 are not tall relative to those in the large durophagous species of Mammacyon and Delotrochanter. They compare with p2–4 of T. altigenis but are much larger teeth.

The m1 (length, 24.7 mm; width, 10.6 mm) is a narrow, shearing carnassial that in this individual shows apical wear on trigonid and talonid cusps. Although the mandible of T. fingeruti and the wolf are of similar size, m1 is much smaller and the premolars larger in the beardog. The paraconid is mesially advanced to form a shear plane with the tall protoconid. A prominent metaconid is situated posterolingual to the protoconid. The talonid is filled by a large hypoconid; this cusp does not attain the height of the paraconid as in Mammacyon and Delotrochanter. A sinuous labial cingulum is present on the talonid but subdued on the trigonid. There is no entoconid yet a very narrow lingual talonid shelf occurs.

The elongate m2 (length, 14.8 mm; width, 8.9 mm) displays a low paraconid anterolingual to the taller, rather blunt, labial protoconid; a small metaconid is present. The m2 talonid has a low, more centrally placed hypoconid; there is no entoconid.

The rectangular m3 (length, 9.8 mm; width, 7.1 mm) has a distinct trigonid somewhat elevated above the talonid. The protoconid is connected by a curvilinear ridge to a very small metaconid. A low hypoconid is the only talonid cusp. A cingulum surrounds both the m2 and m3.

The upper dentition is one of the most complete among temnocyonines. Both toothrows are intact: width of the transverse incisor row, 24.7 mm; width between canines, 24.9 mm; a narrow palate between P1–2 (25.9 mm) widens between P3–4 (36.8 mm), and reaches its maximum breadth (59 mm) between the embrasures of the P4–M1 pair. Palatal length measured from I1 along the midline to the posterior border of the palate is 13 cm. Toothrow length from P1–M2 is 92.1 mm (P2–4, 58.7 mm; M1–2, 27.3 mm).

I1 (length, 5.5 mm; width, 2.4 mm) and I2 (length, 6.5 mm; width, 2.7 mm) are quite small, their apices worn flat. I3 is larger (length, 10.0 mm; width, 5.4 mm), heavily worn to a stub. There are short diastemata between I3 and the canine (4.8 mm), between C–P1 (3.6 mm), P1–2 (3.9 mm), P2–3 (7.0 mm), and P3–4 (2.0 mm). The small I3 of T. fingeruti contrasts in size with the huge I3s of D. oryktes and M. ferocior. Wear on the upper incisors of T. fingeruti shows that they occluded with i1–3 in a grasping or nipping role.

The canine (length, 17.3 mm; width, 12.3 mm, measured at enamel-dentine contact) is broken and worn, but it was more robust than a wolf canine from a similarly sized skull.

P1 is a small peglike tooth (length, 7.6 mm; width, 4.7 mm) with a low vertical anterior face and a posterior slope lacking accessory cusps.

P2 is a larger triangular tooth (length, 13.4 mm; width, 5.7 mm) with a posterior slope slightly longer than the anterior face. There are no accessory cusps.

P3 is somewhat larger (length, 16.1 mm; width, 7.0 mm) than P2 with anterior and posterior faces as in P2 but having a broader heel than P2. Heavy wear on the posterior slope prevents identification of an accessory cusp but it was probably not present.

P4 is a short carnassial (length, 22.0 mm; width, 15.4 mm) with a prominent shear plane beveling the lingual faces of paracone and metastylar blade. A blunt protocone is situated directly lingual to the paracone. The posterior half of the protocone occluded against the m1 paraconid, evidenced by an elliptical wear facet, whereas the anterior face of the protocone was abraded by the p4 accessory cusp. Labial and lingual cingula are present but the anterior face of P4 and the adjacent heel of P3 were deeply worn by the principal and accessory cusps of the large p4.

The principal cusps of P1–3 exhibit flat apical wear facets. This blunting of cusps also occurred on the M1 paracone and metacone (with minor apical wear on M2 paracone) and the paracone and metastylar blade of P4. P4 wear included a shear component that extended posteriorly along the lingual faces of the M1 paracone-metacone and M2 paracone, evidenced by shear facets on these molars.

M1 (length, 17.1 mm; width, 22.4 mm) retains a plesiomorphic occlusal form little different from that of T. altigenis. The paracone and slightly smaller metacone are bordered by a labial cingulum with developed parastyle. The cingulum thins on the margins of M1 at the level of the protocone basin but then thickens to form a prominent posterolingual cingulum around the protocone, which does not equal the strongly swollen M1 lingual cingula observed for Mammacyon and Delotrochanter. The protocone is situated near the anterior margin of the tooth and is more sharp-crested and less bunodont. A preprotocrista runs anterolabially from the protocone to the anterior cingulum, and a short weak postprotocrista intersects the posterior cingulum.

M2 (length, 9.6 mm; width, 17.4 mm) is smaller and of a different form than M1; the anterior margin is convex, the posterior concave. The paracone is much larger than the low reduced metacone yet the lingual face of the two cusps forms a vertical shear plane. The labial and anterior sides of M2 are bordered by a cingulum that thickens to form a wide lingual swelling. A minute protocone situated in the center of an enamel flat retains an anterolabially directed preprotocrista and a tiny postprotocrista. The posterior margin of M2 is worn by the anterior face of m3 since there is no M3.

In T. fingeruti, crushing/shearing occlusion was accomplished by p4–m2 and P3–M2. More anterior teeth served to grasp the food and, via the action of the tongue, passed it back to the carnassials and molars.

Discussion

The skull and mandibles of NM 280/61 represent one of the best-preserved North American temnocyonines. The species preserves the most wolflike dentition of any temnocyonine and exhibits an attritional dental wear pattern much like that known for Canis lupus. Despite some crushing of the skull that collapsed the zygomata and shifted the rostrum laterally relative to the frontal region, the cranium, teeth, and basicranial anatomy are exceptional. NM 280/61 demonstrates that a plesiomorphic temnocyonine dental pattern continued in the Pacific Northwest to very near the Oligocene-Miocene boundary, currently placed at 23.03 Ma (Gradstein et al., 2004). Because NM 280/61 occurs only ∼20 m above a radiometric date of 23.8 Ma, its stratigraphic level places it in the latest Oligocene.

NM 280/61 is the stratigraphically youngest John Day temnocyonine yet discovered; whether Great Plains early Miocene temnocyonines such as T. macrogenys and Delotrochanter existed in the Pacific Northwest is not known. Absence of early Miocene temnocyonines in the John Day beds could simply reflect the more limited representation of sediments of that age, which are often unfossilferous. However, large daphoenine and amphicyonine beardogs have been found in early Miocene John Day strata of early Hemingfordian age (Borocyon, Amphicyon: Hunt and Stepleton, 2004), suggesting that they replaced the temnocyonines.

Temnocyon macrogenys, new species

Figures 20, 21

Fig. 20.

Holotype mandible of Temnocyon macrogenys (F:AM 54139), with left p1–2, p4–m3 (pathologic p3), Anderson Ranch Formation, 2-Mile District, near Guernsey, Platte Co., Wyoming. This is the largest and youngest species of Temnocyon. A, labial and B, lingual views.

Fig. 21.

Occlusal steropair of the holotype dentition (p4–m3) of Temnocyon macrogenys (F:AM 54139). Note presence of m1 and m2 metaconids.

Type

F:AM 54139, left mandible with canine alveolus, p1–2, broken roots of pathologic p3, complete p4–m3, with most of the ascending ramus, articular condyle, and angular process preserved; also two partial metapodials and a fragmentary astragalus, from the Anderson Ranch Fm., “High Daemonelix zone,” 2-Mile District of Charles Falkenbach, near Guernsey, Platte County, Wyoming, collected by C. Falkenbach, 1939.

Distribution

Latest Arikareean, Anderson Ranch Fm., Platte County, Wyoming.

Etymology

From the Greek, macro, “long,” and genys, “jaw,” in reference to the large, elongate mandible of the terminal species of the Temnocyon lineage.

Diagnosis

Largest species of Temnocyon (basilar length, ∼30 cm, table 7), and the only large temnocyonine in which the plesiomorphic cusp pattern of the lower teeth is retained: m1 shear maintained; metaconids present on m1–2; m1/m2 ratio, 1.62, the lowest among Temnocyon species (table 6). Depth of mandible below m1, ∼50 mm; below p2, 46 mm. Mandibular dental measurements highest values among species of Temnocyon: length of toothrow, p1–m3, 135 mm; lengths in mm of p2, 16.6; p4, 22.1; m1, 28.2; m2, 17.4 (table 2).

Referred Specimens

None.

Description

The nearly complete mandible lacks only part of the ventral border of the horizontal ramus and small sections from the labial and lingual sides. From the anterior limit of the jaw to the posterior face of the articular condyle is ∼26 cm; this is a large carnivore with estimated basilar skull length of 30 cm, thus the largest known temnocyonine. The articular condyle and angular process are in close proximity: from the center of the condyle to the process is 24 mm. The distance from the tip of the coronoid process to the articular condyle is 77 mm; from tip of coronoid to base of angular process is 106 mm. The coronoid process is gently recurved posteriorly, with a shallow depression on its lingual face, and a moderately deep masseteric fossa on the buccal face. The articular condyle, mandibular foramen, and coronoid and angular processes are configured as in Canis lupus. Except for the much greater size and mandibular depth of the beardog, the form of the jaw is similar to that of the wolf. As in other temnocyonines, the anterior border of the ascending ramus is gently inclined so that m2–3 are slightly elevated and tilted forward on the edge of the ramus. Depth of the mandible below p2 is 46 mm, beneath m1 ∼50 mm, thus differing from Delotrochanter oryktes (ACM 4804) for which the same measurements, respectively, are 46 and 42 mm, and for Mammacyon obtusidens (LACM 9194), 38 and 35 mm. The mandibular depth beneath m1 for T. macrogenys is nearly twice that of an average wolf.

The mandibular symphysis was united by strong binding ligaments; there is no evidence of symphyseal fusion. The symphysis is widest near its posterior edge. The greatest anteroposterior length is ∼56 mm; greatest dorsoventral height ∼37 mm. The upper part of the symphysis (a rectangular area 22 mm in length, 11 mm in width) is smoother and less rugose than the ventral portion and represents the attachment of a compressible fibrocartilage pad like that described by Scapino (1981; Hunt, 2009). This indicates a flexible mandibular symphysis in T. macrogenys that facilitates close registration of the carnassials during the bite, as in the wolf (fig. 50F).

None of the lower incisors are preserved, however the extremely narrow space between the symphysis and canine alveolus shows that they would have been very small teeth.

The large canine alveolus measures 26.5 mm in length, 20.7 mm in width, with the long axis of the ellipse directed anteriorly and somewhat laterally. There is no intervening space between the canine and p1, their alveolar borders nearly in contact.

Premolars (p2–4) are not crowded and are separated by 2–3 mm diastemata; they are not shortened or posteriorly widened as in Delotrochanter but are relatively narrow, tall yet robust, thus representing a scaled-up version of the premolars of T. percussor.

The single-rooted p1 measures 10.6 mm in length, 7.3 mm in width. It has the form of an oblate cone, anteriorly inclined, whose posterior face has been slightly depressed. There are no accessory or basal cusps. A fine enamel ridge runs from the principal cusp down the posterior slope to the cingulum.

The p2 measures 16.6 mm in length, 8.5 mm in width. In lateral view it approximates an equilateral triangle; there are thin anterior and posterior enamel ridges but no posterior accessory cusp. There is a moderately developed heel with a very small, low basal cusp.

The p3 is represented only by the posterior basal part of the tooth. The tooth was either abnormally triple-rooted, or there was a supernumerary single-rooted premolar inserted in front of a normal p3, this latter alternative more probable. The three alveoli occur in anteroposterior sequence, with a total alveolar length of 24.9 mm. This is longer than p4. If a normal p3 was present, its length based on alveoli would be ∼18 mm.

The p4 measures 22.1 mm in length, 10.4 mm in posterior width, 8.5 mm in anterior width. There is a prominent, labially placed posterior accessory cusp. The posterior border of the tooth is squared and produced as a shelf with a small basal cusp situated below the posterior accessory cusp. Height of p4 is much greater than the m1 paraconid, and I suspect that were p3 present, it also would exceed the paraconid in height.

Length of the premolar series (p1–4) is ∼82 mm. However, because this measurement includes an abnormality (p3), it probably represents a maximum value for the species.

The m1 measures 28.2 mm in length and a talonid width of 13 mm. Length of the trigonid is ∼20 mm, the talonid ∼9.5 mm, emphasizing that the trigonid dominates the tooth and the talonid is not enlarged. The metaconid is well developed; this is the only large temnocyonine in which the metaconid survives. The short talonid is dominated by a large, labially situated hypoconid, directly behind the protoconid. The labial face of the asymmetric hypoconid descends abruptly to the cingulum but the lingual face is more gradually inclined, sloping to the medial edge of the talonid (there is a small lingual talonid shelf). The apex of the hypoconid is an anteroposteriorly aligned ridge that curves posterolingually to become the posterior edge of the talonid. There is weak swelling on the lingual talonid shelf that appears to be a vestigial entoconid.

The m2 measures 17.4 mm in length, with a talonid width of 10.0 mm, and a trigonid width of 10.4 mm. The trigonid is low but higher than the talonid. There is an anterolabial cingulum; the large protoconid is the most prominent cusp and is connected by a smoothly curving ridge to the low, broad paraconid. The small knoblike metaconid lies lingual to the protoconid and is separated from it by a slight constriction. The hypoconid is the only talonid cusp and is labial in position. The talonid surface descends gradually from the hypoconid to the lingual margin.

The m3 measures 10.7 mm in length, ∼8.5 mm in width. The only prominent cusp is a conical protoconid, somewhat anterior to the center of the tooth, flanked lingually by weak enamel ridges descending to barely discernible vestigial paraconid and metaconid swellings. The talonid is much reduced and shows only a vestige of the hypoconid as a swelling on the posterior face of the protoconid.

The toothrow measures as follows: anterior border of canine alveolus to m3, 160.9 mm; p1–m3 length, 135.0 mm; p1–m2 length, 125.4 mm; m1–3 length, 55.5 mm.

Discussion

Temnocyon macrogenys is the largest North American species of the genus and it retains the metaconid on m1–2. The m1 remains a shearing carnassial (i.e., a tall, shearing trigonid and low talonid), and the premolars, although robust, are tall, relatively narrow teeth. These dental features indicate a large predatory carnivore with a more sectorial dentition than found in Mammacyon and Delotrochanter. When compared to the mandibular dentition of the wolf, the canine, premolars, and m2–3 of T. macrogenys are much larger whereas the carnassial is the same length, although a broader tooth. The beardog has incorporated the carnassial as an integral part of a crushing dental battery made up of premolars and posterior molars, while still retaining some shearing ability.

The dentition of T. macrogenys can be derived from that of T. percussor, and there is a particularly strong similarity in cusp pattern to AMNH 81005, Cook's type of the species. This opinion is supported by m1–2 structure and similar m1/m2 length ratios of 1.62 and 1.69 in AMNH 81005 and F:AM 54139 (table 6). Temnocyon macrogenys seems the lineal descendant of T. percussor from the Niobrara River valley at Agate National Monument.

Temnocyon macrogenys (F:AM 54139) comes from the Anderson Ranch beds about 2 mi south of Guernsey, Wyoming; in this district and throughout northwest Nebraska and southeast Wyoming the formation includes widespread flat geomorphic surfaces representing level, early Miocene grasslands (Hunt, 1990). These surfaces are densely packed with fine root casts (rhizoliths) and are cemented by silica derived from diagenesis of volcanic glass shards and unstable minerals in these fine-grained tuffaceous sandstones. Charles Falkenbach, who collected the mandible of T. macrogenys, used the term “High Daemonelix zone” for these silica-cemented geomorphic surfaces, although they have no relationship to rodent burrows first termed “Daimonelix” many years previously by E.H. Barbour (1892). Evidence that Falkenbach in fact collected the mandible from one of the geomorphic surfaces of the Anderson Ranch Formation can be seen in the sediment adhering to the mandible that includes the small siliceous rootlets and pale reddish-brown sand typical of these grassland surfaces. Temnocyon macrogenys was found within one of these siliceous paleosols. Fossil mammals in the Childs Frick collection from the “High Daemonelix” level of the Guernsey collecting district are similar to mammals of the Niobrara Canyon local fauna from the stratotype area of the Anderson Ranch Formation (Niobrara Canyon, Sioux County, Nebraska) and establish the age of T. macrogenys as latest Arikareean.

Rudiocyon, new genus

Type Species

Rudiocyon amplidens, new species.

Included Species

Only the type species.

Distribution

Early or mid-Arikareean of Oregon.

Etymology

From the latinized Greek, kyon, for “dog,” and for Rudio Creek, the locality in north-central Oregon where the holotype was discovered.

Diagnosis

Differs from Temnocyon by absence of the m1 metaconid; from Mammacyon and Delotrochanter by a small, short m2 relative to m1 (table 6, ratio E/F ∼1.9); and from Delotrochanter by a narrow, compressed p4 (not posteriorly broad). See tables 1–2.

Discussion

The genus includes a single species represented by a partial mandible with massive teeth (p4–m2) that define the taxon. A small m2 in Rudiocyon (table 6, ratio E/F, ∼1.9) precludes assignment to Mammacyon (table 6, ratio E/F, 1.57–1.6), a genus defined by a large, elongate m2. Nonetheless, the genus may have evolved from an earlier less derived species that gave rise to both Mammacyon and Rudiocyon. Similarity to cheek teeth of T. ferox suggests a possible alternative derivation for R. amplidens. Rudiocyon amplidens is considered somewhat younger than the age of the Deep Creek tuff (27.9 ± 0.57 Ma); the probable site of collection occurs 10–15 m above a local ash on Rudio Creek identified as a correlative of the Deep Creek tuff.

Rudiocyon amplidens, new species

Figures 22, 23

Fig. 22.

Holotype mandible of Rudiocyon amplidens (LACM 480), with left p4–m2, probably from 1.5 mi south of the Johnson Ranch, from gray beds exposed on the east side of Rudio Creek, Kimberly Member, John Day Formation, Grant Co., Oregon. A, labial and B, lingual views. This is the largest John Day temnocyonine.

Fig. 23.

Occlusal stereopair of the holotype dentition (p4–m2) of Rudiocyon amplidens (LACM 480). Note absence of m1 and m2 metaconids and narrow compressed p4.

Temnocyon large species: Stock, 1933b: 36.

Type

LACM 480, partial left mandible with p4–m2, right m2, right p4 protocone, and a partial braincase, from the John Day Formation, CIT loc. 31, 1.5 to 2 mi south of the Johnson Ranch [now Tom Campbell Ranch] at the mouth of Rudio Creek, from gray beds exposed on the east side of Rudio Creek, south of the North Fork of the John Day River, Grant County, Oregon, collected by E.L. Furlong (∼1930).

Distribution

Early or mid-Arikareean, John Day Formation, Grant County, Oregon.

Etymology

From the Latin, amplus, “large,” and dens, “tooth,” in reference to the robust teeth in this species, the largest of the John Day temnocyonine beardogs.

Diagnosis

Largest John Day temnocyonine species, m1 length, 28.7 mm, m1 without metaconid thereby distinguished from Temnocyon; m1 hypoconid centrally placed on talonid, no entoconid; m1 labial cingulum sinuous, not straight as in Delotrochanter; p4 narrower than in Delotrochanter and Mammacyon ferocior but similar in this respect to M. obtusidens, T. ferox, and T. macrogenys. Relatively short m2 (E/F ratio, ∼1.9, the highest value for any temnocyonine), not elongate as in Mammacyon, with blunt protoconid and hypoconid the principal cusps, aligned one behind the other along the midline of m2; small m2 paraconid occupies anterointernal corner; no metaconid or entoconid. Mandible deep below m1–2 (43 mm), more so than in any other John Day temnocyonine.

Referred Specimens

None.

Description

The partial mandible (length, ∼114 mm) retains p4–m2. The dorsal border of the jaw is damaged anterior to p4 but has alveoli for p3. Depth of mandible: ∼36 mm below p3; 37 mm below the posterior root of p4; 40 mm below the m1 protoconid; and 43 mm below the anterior root of m2; the anterior masseteric fossa is present below the m2 talonid.

The p4 is narrow and similar in form to the p4 of Mammacyon obtusidens and Eyerman's type of T. ferox. The tooth is tall, the principal cusp 5 mm higher than the m1 paraconid. A fine enamel ridge on the anterior slope intersects the cingulum. A large labially situated posterior accessory cusp occurs above a small basal cusp. The anterior (8.1 mm) and posterior (10.0 mm) widths of p4 demonstrate a narrow p4 in Rudiocyon, which compares with these p4 widths of M. obtusidens (7.4, 9.7 mm).

The m1 measures 28.7 mm in length, 13.2 mm in width at the base of the protoconid, 11.8 mm in talonid width, and lacks both a metaconid and entoconid. This is a massive carnassial, the protoconid much taller than the flanking paraconid and hypoconid. The hypoconid is as large as the paraconid and nearly equal in height, indicating the important role played by this pestlelike cusp in crushing. It occupies the center of the talonid, rather than the more labial placement seen in T. macrogenys. Anterior to the blunt hypoconid is a small vertical swelling of enamel on the posterior face of the protoconid. It measures 2.4 mm in width, 3.7 mm in height. Above and lingual to this swelling is a thin enamel ridge also situated on the posterior face of the protoconid. This is not a vestige of the metaconid but rather a remnant of a thin protoconid crest that in a plesiomophic m1 forms an incisure with the metaconid. It remains as a low narrow ridge in temnocyonines that have recently lost the metaconid, and is absent in species that have long been without this cusp. A sinuous (not straight as in Delotrochanter) cingulum surrounds the base of the tooth on the labial side and on the posterior half of the lingual side, but is indistinct on the lingual side of the trigonid. The lingual cingulum medial to the hypoconid contains a small notch.

The m2 is not elongate and narrows posteriorly, differing from Mammacyon: greatest trigonid width is 9.1 mm, talonid width 8.4 mm. The protoconid and smaller hypoconid, both blunt crushing cusps, are anteroposteriorly aligned with the m1 hypoconid. There is variation in the cusp placement of right and left m2: the left m2 shows these two cusps centrally placed, whereas the right m2 shows them closer to the labial margin—it is possible that the latter tooth represents another individual. There is no m2 metaconid or entoconid. However, as in other large temnocyonine species, the anterolabial cingulum of the m2 trigonid is protuberant.

The right P4 protocone, the only remnant of the upper dentition, is similar in size and blunt conical form to that cusp in large species of Mammacyon. It measures 10.9 mm in anteroposterior width and compares closely with the protocone of ACM 34–41, the holotype of M. obtusidens. The protocone was encircled by a thickened cingulum.

Discussion

Rudiocyon amplidens is a carnivore about the same size as Mammacyon ferocior or Delotrochanter oryktes, and is somewhat smaller than T. macrogenys. Thus, the species was one of the larger North American temnocyonines. The m1/m2 length ratio of R. amplidens (∼1.9) distinguishes it from M. ferocior (1.6), T. macrogenys (∼1.6), and the species of Delotrochanter (∼1.6–1.7). In addition, its p4 is not as broad as the M. ferocior p4, nor does it have an m1 metaconid as in T. macrogenys. Delotrochanter oryktes and D. major also differ from R. amplidens in having a short, posteriorly wide p4 in contrast to the more narrow p4 of Rudiocyon.

In its p4–m1, in the form of the mandible, but particularly the P4 protocone, R. amplidens shows an evident similarity to Mammacyon. However, M. obtusidens has a longer m2 and a much smaller m1 than R. amplidens, and M. ferocior has a boader p4 and a much more elongate m2.

Mammacyon mandibular dentitions are distinguished by the elongate m2 and are known only from the Arikareean of the Great Plains. Because the only evidence of Mammacyon in the John Day basin is a maxilla (LACM 5386) from Haystack Valley, the nature of m2 in that species cannot be determined. T. altigenis is the only temnocyonine where m2 variation can be estimated and the fossils are not from a single stratigraphic horizon. The variation in m2 length in the T. altigenis sample indicates that the hypodigm has a range of 1.67–1.83 for the m1/m2 length ratio (table 6). Yet the difference between the m1/m2 length ratio of R. amplidens (1.89) and those of M. obtusidens–M. ferocior (1.57–1.6) exceeds that range. LACM 480 cannot be included in Mammacyon unless the genus is more broadly defined to include more pronounced variation in dental proportions, particularly with regard to m2.

A partial braincase accompanies the lower jaw and presumably was collected with it. It is remarkable how similar most temnocyonine skulls are in braincase width, a proxy for volume of the cranial cavity (table 7): five of the larger species all measure from 62 to 67 mm, R. amplidens among them. The braincase of LACM 480 cannot be compared with temnocyonines with inflated frontal paranasal sinuses because the frontal region anterior to the braincase was not preserved.

The p4–m2 of Rudiocyon shows similarities to these same teeth in T. ferox. These include the squared posterior heel and cusp pattern of p4; the form of m1, which suggests that the reduced metaconid in T. ferox could have finally been lost in LACM 480; and the occlusal form of the relatively short m2 in both species. LACM 480 came from gray tuffaceous sandstone of the Kimberly Member along Rudio Creek a short distance above a supposed correlative ash of the Deep Creek tuff, dated elsewhere at ∼27.9 Ma. If this age for the Rudio Creek ash is correct, it would appear to be a much older carnivore than T. ferox, which is a smaller and possibly much younger species, and in this case an ancestor-descendant relationship would be unlikely. However, if the correlative ash along Rudio Creek, identified by Fisher (Fisher and Rensberger, 1972: 14) as the Deep Creek tuff in fact represents a much younger undated ash-fall event, then the relative ages of R. amplidens and T. ferox are unspecified.

Mammacyon Loomis, 1936

Type Species

Mammacyon obtusidens Loomis, 1936.

Included Species

Mammacyon obtusidens Loomis, 1936; M. ferocior, new species.

Distribution

Early Arikareean of Oregon; early to mid-Arikareean of South Dakota; mid- or early late Arikareean of southeastern Wyoming.

Diagnosis

Distinguished from Temnocyon by absence of the m1 metaconid and different M1 and P4 proportions (table 6, ratios A/B, C/D); from Temnocyon, Rudiocyon, and Delotrochanter by a more elongate m2 (ratio E/F); and from Delotrochanter by M1 proportions (ratio A/B), by P3/p3 without posterior accessory cusp, by p4 with labially (not centrally) placed posterior accessory cusp, by a sinuous (not straight) labial cingulum on m1, and by proportions of p2–4 (short broad p2–3 in Delotrochanter, longer narrow p2–3 in Mammacyon). Mammacyon evolves the most extreme form of crushing P4–M1 of the subfamily. See tables 1–TABLE 2TABLE 3TABLE 45.

Discussion

The genus includes temnocyonines that abandon the plesiomorphic form of the cheek teeth and adopt a durophagous crushing dentition involving not only premolars but also the carnassial-molar battery. The large Mammacyon ferocior apparently evolved from the somewhat smaller M. obtusidens, both species represented by skulls and some postcrania. The genus seems restricted to the Oligocene and ranges in time from the early to mid-Arikareean, possibly continuing into the early late Arikareean but is not known in early Miocene faunas.

Mammacyon obtusidens Loomis, 1936

Figures 24–Fig. 25.Fig. 26.27, 67

Fig. 24.

Holotype rostrum of Mammacyon obtusidens (ACM 34-41) with right I3, P2–M2, left C, P2–M2, from the lower Arikaree Group, Porcupine Creek, Shannon Co., South Dakota. This is one of the earliest durophagous temnocyonines. Focal point of the crushing dentition is centered on the P4–M1 pair. See figs. 67, 69.

Fig. 25.

Referred mandible of Mammacyon obtusidens (LACM 9194) with left canine, p2–m2, LACM loc. 1964, lower Arikaree Group, Wounded Knee area, Shannon Co., South Dakota. A, labial and B, lingual views.

Fig. 26.

Occlusal stereopair of the mandibular dentition (p4–m2) of Mammacyon obtusidens (LACM 9194). Note absence of m1 and m2 metaconids.

Fig. 27.

Referred maxilla of Mammacyon obtusidens (LACM 5386) with right P4–M1, Turtle Cove Member, John Day Formation, Haystack Valley, Wheeler Co., Oregon. Note apical wear on all cusps, indicating crushing action of P4–M1.

Mammacyon obtusidens Loomis, 1936: 44–47, fig. 1.

Mammacyon obtusidens: Macdonald, 1963: 219.

Mammacyon obtusidens: Macdonald, 1970: 63.

Temnocyon percussor: Macdonald, 1970: 61–62.

Type

As given in Loomis (1936: 44), but modified as follows: skull with well-preserved basicranium and dentition, axis vertebra, humerus, radius, both ulnae, four metapodials, tibia, partial scapula, two isolated canines, and an astragalus (ACM 34–41), found “in the summer of 1934 in the lower Miocene Rosebud formation on Porcupine Creek, South Dakota, just above the concretionary layer, or about 100 feet above the base of the beds.” Additional information is found in Macdonald (1963: 157, 219) as to the type locality, “Amherst College Museum Rosebud 2, Monroe Creek Formation or Harrison Formation.” The locality producing the holotype is in the lower Arikaree Group and is of Oligocene age.

Distribution

Early to mid-Arikareean: lower Arikaree Group, Wounded Knee area, ACM Rosebud 2, Shannon County, South Dakota; lower Arikaree Group, LACM loc. 1964, Wounded Knee area, Shannon County, South Dakota; John Day Formation, CIT loc. 29, Wheeler County, Oregon.

Diagnosis

Distinguished from M. ferocior by smaller size of cranium (basilar length: M. obtusidens, 25.5 cm; M. ferocior, 28 cm, table 7), postcrania, and mandibular dentition (toothrow length, ∼98 vs. 120 mm, table 2), including m1–m2 lengths (M. obtusidens, 24.4, 15.5 mm; M. ferocior, 27.7, 17.3 mm) and by a much narrower p3–4 width in M. obtusidens. The two species differ in P4 and M1 dental ratios A/B and C/D (table 6).

Referred Specimens

(1) LACM 9194, left mandible with canine, p1 alveolus, p2–m2, m3 alveolus, isolated right lower canine, right mandibular fragment with m1 talonid and m2 roots, right mandibular fragment with anterior root of p4 and its posterior alveolus, from LACM loc. 1964, Wounded Knee area, “Monroe Creek Formation,” NW4, NW4, sec. 3, T39N, R42W, Shannon County, South Dakota, collected by H. Garbani, June 30, 1964; (2) LACM 5386, partial maxilla with P4–M1, from the John Day Formation, CIT loc. 29, “from a large exposure of green beds on north side of Haystack Valley, Wheeler Co., Oregon.”

Description

ACM 34–41—Loomis (1936) only briefly described the upper dentition of this fine skull, and did not mention the well-preserved basicranium. It is one of only three temnocyonine skulls in which some part of the auditory bulla is preserved, in this case a crescentic anterior portion formed largely or entirely by the ectotympanic (see Basicranial Anatomy).

The left maxilla retains the canine, P2–M2; on the right, I3, P2–M2. An alveolus for a single-rooted P1 is present on the right, but other alveoli are distorted.

The upper canine, blunted at the tip from wear, measures ∼45 mm in length from the tip to the enamel base on the labial face. There is no evident wear groove produced by the lower canine. At the enamel base the canine measures 22.1 mm in anteroposterior length and 14.1 mm in width.

The alveolus of P1 measures 8.7 mm in length, 5.4 in width, and has been slightly distorted by crushing.

P2 measures 18.5 mm in length, the width anterior to the main cusp is 6.8 mm, and the width posterior to this cusp is 7.7 mm. A long posterior slope descends from the main cusp in contrast to the shorter anterior face, both devoid of accessory or cingular cusps. On all upper teeth the cingula are slightly developed basal swellings without sharp or distinct edges. On both P2 and P3 a weak, thin, enamel ridge runs from the main cusp anterolinguad to the cingulum on the anterior slope, and a similar ridge on the posterior slope runs to the posterolabial cingulum.

P3 is an enlarged version of P2, measuring 20.3 mm in length; width anterior to main cusp, 8.2 mm; width posterior to this cusp, 10.6 mm. P3 is somewhat more robust than P2 and much more expanded at its base posterior to the main cusp. There are no accessory or cingular cusps. The anterior and posterior faces of P3 are more nearly equal in length, and the tip of the main cusp shows flat apical wear.

P4 is a robust, massive tooth; M. obtusidens is the first of the larger temnocyonines to develop this crushing carnassial. P4 has a short metastylar blade, an anteroposteriorly elongate stout paracone, and a greatly enlarged protocone that crushes against the p4 heel. Shear is still accomplished to some degree by the lingual surfaces of the paracone and metastylar blade. The metastylar blade is directed outward at a 50° angle from the anteroposterior axis of the paracone. The blade measures 10.5 mm in length; the paracone measures 18.4 mm from the carnassial notch to its anterior base. A weak enamel ridge runs down the anterointernal face of the paracone to end in a tiny cingular cusp. Directly lingual to the paracone is the blunt, knoblike protocone: the protocone is separated from the paracone by a broad valley. Anteroposterior length of the protocone is 11 mm; greatest width of P4 at the level of the protocone is 21.5 mm. Only a narrow embrasure (∼7 mm in length) remains between P4 and M1 to receive the trigonid of m1, and a similarly reduced space also occurs in M. ferocior.

M1 measures 20.4 mm in length, 28.2 mm in width. In M. obtusidens the tooth comprises a labial part dominated by the para- and metacone, and an enlarged lingual half created by swelling of the lingual cingulum around the broad protocone region. Between the lingual and labial parts of the tooth is a prominent constriction, giving the tooth a “waist” at the level of the protocone basin. The paracone is somewhat larger than the metacone; the parastyle, swollen labially and ventrally downturned, is more pronounced than the metastyle. Here, as in other large temnocyonines, M1 has a “folded” appearance as if the entire labial half were inwardly rotated toward the lingual half about an anteroposterior axis through the protocone basin. The protocone is centrally situated in a flat expanse of enamel. A preprotocrista runs from the protocone to the anterior cingulum but there is no postprotocrista nor para- and metaconules. A heavy swollen lingual cingulum, about equally developed throughout its extent, surrounds the protocone region.

M2 measures 10.5 mm in length, 17.9 mm in width. It is much smaller than M1 and fits into the concavity on the posterior margin of the larger molar. The paracone is much larger than the abbreviated metacone, but only slightly taller. The stylar shelf and labial cingulum external to the paracone are better developed than these features labial to the metacone as a result of the reduction of the metacone. The protocone basin occurs at the constricted “waist” of the tooth, and here also, as for M1, the labial half of M2 is rotated or “folded” inward. The protocone is a small, low blunt cusp protruding from the enamel flat forming the lingual half of the tooth. It is surrounded by a swollen cingulum, which is considerably worn. M3 was absent; there is no alveolus and there is no place on the maxilla posterior to M2 for an M3.

The associated limb and vertebral elements are similar in form to skeletal elements of Mammacyon ferocior (F:AM 27562) from north of Keeline, Wyoming. Comparisons of the ulna, humerus, astragalus, and scapula show no differences other than size. The limb elements of M. ferocior, in association with its mandible, indicate a species larger than M. obtusidens (ACM 34–41).

LACM 9194—Only the complete left mandible is described here; the two isolated jaw fragments and canine only duplicate structure observable in the more complete lower jaw. LACM 9194 was initially described as Temnocyon percussor by Macdonald (1970: 61–62).

The jaw is short and massive. Depth under the main cusp of p2 is 38.4 mm, and 35.9 mm below the m1 protoconid. Premolars and molars are large, robust teeth; the molars, especially m2–3, are conspicuously tilted forward on the curvilinear margin of the ascending ramus. The inclined m2–3 are characteristic of temnocyonines, consequently M2 is often dorsally elevated relative to M1 to accommodate this configuration of the lower molars. Such an M2 is present in the holotype (ACM 34-41). The unworn teeth indicate a mature young adult.

The canine is a large, recurved tooth, 38.9 mm in height from tip to enamel base, measured on the labial face. Its anteroposterior width is 19.1 mm measured at the base of the enamel; its labiolingual width is 12.6 mm. A wear groove produced by the upper canine begins near the tip on the posterior surface, continues down the posterior slope, trending slightly labiad, eventually terminating about 9.5 mm above the enamel base on the posterolabial face. Two fine enamel ridges traverse the tooth from tip to enamel base: the first courses down the center of the lingual face, the second traverses the posterolingual face.

The single alveolus for p1, although crushed, measures ∼8 mm in length, 6.2 mm in width. Elongation of the large alveolus suggests that the tooth was directed anterolabially.

The p2 measures 14.4 mm in length, 5.8 mm in anterior width, and 6.4 mm in posterior width. There are no accessory or cingular cusps. A fine enamel ridge descends the somewhat steeper anterior face to the anterolingual corner of the tooth, and a similar ridge occupies the posterior face, ending at the posterolabial corner.

The p3 measures 17.7 mm in length, 6.4 mm in anterior width, and 7.8 mm in posterior width. There are no accessory cusps, but there is a small basal cusp situated at the center of the posterior cingulum. A thin enamel ridge descends the anterior face to the anterolingual corner; a second ridge traverses the posterior face to the basal cingulum cusp.

The p4 measures 20.4 mm in length, 7.5 mm in anterior width, and 9.7 mm in posterior width. A large posterior accessory cusp is labially situated half the distance down the posterior slope. A small basal cusp on the posterior cingulum lies below the posterior accessory cusp. A fine enamel ridge follows the anterior slope and ends in the anterolingual corner as in p2–3, but a posterior ridge is largely obscured by development of the large posterior accessory cusp.

The bases of P3 and p3–4 are posteriorly widened but this is less pronounced relative to M. ferocior or D. oryktes teeth where such widening of the posterior base is strongly developed.

The m1 measures 24.4 mm in length, 11.4 mm in posterior trigonid width, and 10.9 mm in greatest talonid width. Trigonid length is 17.4 mm. This is a robust, short, blunt-cusped carnassial that has lost the metaconid, although a small wear facet on the labial face of the protoconid shows that paraconid-protoconid shear is retained. The short, swollen paraconid blade is angled inward and is much lower in height than the principal cusp of p4. The protoconid is compressed to form a blade running forward and downward to form the incisure with the paraconid. A short enamel ridge 4 mm in length descends the posterior surface of the protoconid to form the incisure with the hypoconid, a massive blunt cusp placed directly behind the protoconid and centrally situated on the talonid. There is also a low enamel swelling on the posterolingual wall of the protoconid that represents the vestige of the incisure once shared with the metaconid. Its presence indicates that the metaconid has probably been lost relatively recently in this lineage. There is no entoconid. The m1 has a swollen, weakly differentiated cingulum, most evident on the labial side where it is sinuous: it rises below the protoconid, descends at the protoconid-hypoconid incisure, and rises again beneath the hypoconid (in contrast, the labial m1 cingulum is straight in Delotrochanter). At the posterolingual corner of the m1 talonid, the cingulum is slightly enlarged, creating a marked indentation of the cingulum anterior to this swelling (also found in M. ferocior).

The elongate m2 is diagnostic of the Mammacyon lineage in the Great Plains. Its length is 15.5 mm; greatest trigonid width, 9.6 mm; greatest talonid width, 8.7 mm: m2 has a rectangular occlusal outline. The labially placed protoconid is the largest cusp. There is no metaconid but a small, distinct paraconid occupies the anterolingual corner of m2. The only talonid cusp is a low, blunt, somewhat labially situated hypoconid. There is no entoconid. The pronounced inclination of m2 (and m3) on the edge of the ascending ramus of the mandible, tilted forward nearly 30° from the horizontal, further emphasizes the short toothrow and massive quality of the teeth.

The m3 is represented by a single alveolus, length 4.7 mm, width 4.1 mm, placed close against the posterior edge of m2.

Length of the toothrow from p2–m2 is 86.2 mm; from the base of the enamel on the posterior face of the canine to the anterior rim of m3 alveolus is 98.3 mm; from anterior rim of p1 alveolus to posterior rim of m3 alveolus is ∼99 mm.

LACM 5386—I refer to M. obtusidens a partial maxilla with right P4–M1 and the alveolus for the posterior root of P3, from CIT Locality 29, Haystack Valley, Wheeler County, Oregon. It was found among unidentified material in the collection of the Los Angeles County Museum and has not been previously described. The teeth are most similar in form to ACM 34-41 (M. obtusidens) and F:AM 54134 (M. ferocior). This is the only specimen from the John Day beds assigned to Mammacyon; it represents a smaller individual than the holotype of M. obtusidens and is closer in size to LACM 9194.

The alveolus (greatest width, 8.2 mm) for the posterior root of P3 is very large, indicating a posteriorly broad premolar at least 9 mm in width. This alveolus indicates that P3 was closely set against the base of P4 as in the holotype.

P4 measures 21.8 mm in length, 18.2 mm in greatest width, and is characterized by the short yet robust metastylar blade relative to the large paracone and protocone. Metastylar blade length is 8.1 mm; width of the blade is 7.4 mm. The upper carnassial is nearly as wide as long, and shows blunted cusps with pronounced apical wear (there is only a weak vertical shear facet on the lingual face of the paracone-metastylar blade). The enormous protocone is responsible for the width of P4: it extends as far toward the midline of the palate as does the M1. An enlarged protocone is characteristic of Mammacyon; in Delotrochanter the P4 protocone does not extend as far toward the midline. A slightly swollen cingulum surrounds P4, and is somewhat more defined around the base of the metastylar blade than elsewhere on the tooth. A conspicuous valley separates the large blunted paracone from the much lower knoblike protocone.

M1 measures 17.8 mm in length, 24.8 mm estimated transverse width (the anterolingual cingulum is broken), and 13.1 mm in anteroposterior width of the protocone region. The form of the tooth is comparable to M1 of ACM 34-41. The embrasure for m1 between P4 and M1 is much reduced as in the holotype, thus there was probably no m1 metaconid. In addition, the M1 of LACM 5386 shows the “folded” appearance in which the labial half of the tooth is rotated inward toward the lingual half as in ACM 34-41. The knoblike protocone of M1 was isolated within the broadened protocone region and surrounded by a swollen lingual cingulum; a weak preprotocrista runs from protocone toward the anterior cingulum but no postprotocrista is apparent. Paracone, metacone, and protocone of M1 are blunt cusps with flat apical wear surfaces. The paracone and metacone were once tall cusps but wear has reduced their height so that evidence of vertical shear on their lingual faces is equivocal. In mature individuals mastication emphasized the processing of hard food items, also suggested by the blunt, knoblike protocones of both M1 and P4.

A vestige of the labial M2 alveolus indicates that the tooth was small and fitted tightly against the posterior margin of M1 in the same location as the holotype.

Discussion

Mammacyon obtusidens is the oldest species that can be confidently attributed to the genus. The maxilla (LACM 5386) from the John Day Formation of Haystack Valley, Oregon, can be conservatively dated at younger than 28.7 Ma based on its probable site of collection close above the Picture Gorge ignimbrite; an age estimate for the locality falls in the interval from ∼28 to 28.7 Ma. The first appearance of M. obtusidens in the Great Plains is based on the referred mandible (LACM 9194) from the “Monroe Creek Formation” of the Wounded Knee area in southwestern South Dakota. The holotype cranium and associated postcranials (ACM 34-41) were attributed to the “Monroe Creek or Harrison Formation” and are also part of the Wounded Knee fauna of Macdonald (1963, 1970). However, a reliable correlation of these rock units with the type Monroe Creek and Harrison formations of Hatcher (1902a) north of the town of Harrison in northwest Nebraska has not been established, and the mammalian fauna from the “Monroe Creek” and “Harrison” of the Wounded Knee area lacks age-diagnostic species found in the stratotype Harrison Formation in Sioux County. The Wounded Knee fauna from the “Monroe Creek-Harrison” interval in South Dakota is an early to mid-Arikareean fauna (see Age and Correlation), hence the holotype of M. obtusidens is assigned to the late Oligocene —the species had attained the specialized durophagous dental traits typical of the genus at least by the earlier late Oligocene, and this dental trend continued in the Great Plains with M. ferocior.

The M. obtusidens mandible (LACM 9194) is from the same Wounded Knee locality as the canid Enhydrocyon pahinsintewakpa, the most plesiomorphic species of its genus (Wang, 1994: 90); this locality (LACM 1964) occurs near the base of the “Monroe Creek” unit and in close proximity to the Sharps Formation, which includes leptauchenine oreodonts in the same area (Macdonald, 1963: 162). These mammals document an early Arikareean age for the stratigraphic level that yielded the mandible of M. obtusidens.

This conclusion is supported by the stratigraphic position of the descendant of M. obtusidens, the larger terminal species of the lineage, M. ferocior. Mammacyon ferocior from the Pine Ridge of Niobrara County, north of Keeline, Wyoming, is a species of larger size in which derived dental features of the M. obtusidens holotype become further accentuated. Thus, in M. ferocior the teeth are similar in form but larger, more massive, with even greater emphasis on the crushing function of the cheek teeth. The fauna from north of Keeline is considered here to be intermediate in age between the Wounded Knee “Monroe Creek” fauna and the fauna of the stratotype Harrison Formation in northwest Nebraska.

The mandibular fragment (LACM 15908) attributed in this study to cf. Mammacyon from the Sharps Formation of South Dakota was collected stratigraphically below and ∼4 mi (6.4 km ) north of the locality that produced the referred M. obtusidens mandible (LACM 9194). It possibly represents the earliest record of Mammacyon in the Great Plains. It has reduced the m1 metaconid to a vestigial remnant and shows the robust teeth and thick mandible that would be expected in the ancestor of M. obtusidens. However, LACM 15908 lacks the elongate m2 relative to m1 (m1/m2 length ratio, ∼1.7) common to M. obtusidens and M. ferocior and is too incomplete to permit confident referral to the genus. The m2 protoconid and hypoconid are centrally situated, a character of Mammacyon and Delotrochanter; these cusps are labially placed in Temnocyon.

Mammacyon ferocior, new species

Figures 28–Fig. 29.Fig. 30.Fig. 30.31

Fig. 28.

Holotype mandible of Mammacyon ferocior (F:AM 27562) with right canine, p1–m3, from the Arikaree Group, north of Keeline, Niobrara Co., Wyoming. The large crushing p4 nearly equals the carnassial in size. A, labial and B, lingual views.

Fig. 29.

Occlusal stereopair of the mandibular dentition (p4–m3) of Mammacyon ferocior (F:AM 27562). Note absence of m1 and m2 metaconids, apical wear on p4–m2 cusps, a labial shear facet on m1 protoconid-paraconid, and the broad p4 (compare fig. 23).

Fig. 30.

Cranium of Mammacyon ferocior (F:AM 54134) in (A) right lateral; (B) dorsal; and (C; opposite page) left lateral (restored) views, from the Arikaree Group, north of Keeline, Niobrara Co., Wyoming, showing (B) width of the frontal paranasal sinuses relative to braincase volume.

Fig. 30.

Continued.

Fig. 31.

Mammacyon ferocior (F:AM 54134) rostrum (A) in palatal view with right and left P4–M2, right P1, and alveoli for I1–3, C, P2–3 , from the Arikaree Group, north of Keeline, Niobrara Co., Wyoming. Stereophotos (B) of crushing P4–M2. This is the largest species of Mammacyon, and represents the climax of durophagy in the genus.

Type

Right partial mandible with i3 root, c–m3, and associated limb elements and vertebrae: atlas, caudal vertebra, sacrum, glenoid of scapula, left humerus, distal right humerus, both ulnae, rib fragments, left metacarpal 2, proximal left and right metacarpal 3, innominate fragments, partial left femur, left tibia, proximal right tibia, both calcanea, left astragalus, proximal right astragalus, right navicular, both cuboids, left ectocuneiform, right metatarsal 2, left distal metatarsal 3, proximal right metatarsal 5 (reduced), from ?Harrison Formation, north of Keeline, Niobrara County, Wyoming, collected by Charles Falkenbach, 1931 (F:AM 27562).

Distribution

Mid- or early late Arikareean, Arikaree Group, Niobrara County, Wyoming.

Etymology

From the Latin, ferocior, “particularly fierce,” in allusion to the large size and presumed disposition of this predator.

Diagnosis

Largest species of the Mammacyon lineage, thus sharing the same dental characters of form and proportion as M. obtusidens but distinguished by larger size. Dental ratio A/B, 1.23; ratio C/D, 1.13 (table 6), lower than all Temnocyon species. Ratio C/D (1.13), lowest of all temnocyonines. Distinguished from other large temnocyonines such as R. amplidens by an elongate m2 (ratio E/F: 1.6 in M. ferocior, ∼1.9 in R. amplidens); from T. ferox, T. percussor, T. fingeruti, and T. macrogenys by absence of the m1 metaconid; from T. macrogenys by much smaller size; and from Delotrochanter oryktes and D. major by a p4 with labially placed posterior accessory cusp (centrally placed in Delotrochanter). M. ferocior p4 posteriorly broader than in M. obtusidens. Cranium with greatly inflated frontal sinuses relative to small braincase volume (table 7).

Referred Specimen

F:AM 54134, a skull with poorly preserved basicranium yet in other features complete and uncrushed. This is the largest known North American temnocyonine skull (basilar length, 28 cm). Upper dentition with alveoli for six incisors, two canines, left P1–3, right P2–3; complete right P1, P4–M2 (P4 damaged), and left P4–M2; M3 not present in the species. From the ?Harrison Formation, north of Keeline, Niobrara County, collected by C. Falkenbach in 1944, from a stratigraphic level “high in the formation” in this area.

Description

FAM 27562—The depth of the mandible (estimated at ∼44 mm) is uncertain because the ventral border of the horizontal ramus is damaged. Length of toothrow from the posterior border of the canine alveolus to the posterior limit of m3 is 126.2 mm. The mandibular symphysis extends from the anterior canine border almost to the posterior border of p2, having a greatest length of 48.8 mm and depth of 36.3 mm.

Lower canine height is 47.5 mm above the labial alveolar border. The tooth is grooved on its posterolabial face by the upper canine, identical to the similarly positioned groove in LACM 9194 (M. obtusidens). At the enamel base the canine measures 14 mm in width, 20.3 mm in anteroposterior length, and is somewhat posteriorly recurved. In addition to the groove worn by the upper canine, there is a deep elliptical wear facet on the anterolingual face produced by the I3 which is much larger than I1–2 based on alveolar dimensions.

The p1 is a small, peglike tooth that measures 9 mm in length, 5.9 mm in width. The main cusp is anteriorly placed as in all temnocyonines, with a long posterior slope and no accessory or cingular cusps.

The p2 is much larger than p1, measuring 17.9 mm in length, 8.1 mm in greatest width posterior to the main cusp. There are no accessory or cingular cusps. The posterior face is longer and more gently inclined relative to the more vertical anterior face. Fine enamel ridges are present on both anterior and posterior faces and, as in M. obtusidens, the anterior ridge descends to the anterolingual cingulum whereas the posterior ridge runs to the posterolabial cingulum. The tooth itself is angled outward; that is, the anterior p2 root is more labial in position than the posterior root. A somewhat swollen cingulum is more defined along the lingual bases of p2, p3, and p4 relative to its less pronounced labial expression.

The p3 is simply a taller, larger example of p2; it measures 19.3 mm in length, 9 mm in width. There are no accessory or cingular cusps despite the broadened heel of p3 relative to p2. Thin enamel ridges run down the anterior and posterior faces to the barely discernible cingulum, the anterior ridge contacting the anterolingual corner, the posterior touching the posterolabial corner where the cingulum is better developed along the posterior margin.

The p4 is more massive and robust than p3. It measures 21.9 mm in length, 10.7 mm in width. A large posterior accessory cusp occurs about halfway between the tip of the main cusp and the crown base. It lies labial not only to the main cusp but also to the basal cusp occupying the center of the broad posterior shelf of p4. The main cusp of p4 is nearly as tall and robust as the protoconid of m1. Both p3 and p4 are particularly tall premolars; their principal cusps are each ∼4 mm taller than the m1 paraconid.

Flat apical wear facets are present on the principal cusps of p1–4 and on the m1 protoconid.

The m1 measures 27.7 mm in length, ∼13 mm in width (the labial talonid is lost). The carnassial is large and robust yet appears rather low and almost dwarfed relative to the tall, massive premolars. The metaconid is absent. Both protoconid and paraconid are massive, broad cusps, the protoconid showing strong apical wear. However, carnassial shear is evidenced by a near-vertical facet on the labial face of the paraconid-protoconid. The paraconid, placed anterolingual to the protoconid, has not rotated into a position directly anterior to the protoconid as in carnivorans with highly developed carnassial shear. The hypoconid is large, blunt, centrally placed on the talonid, and the entoconid is absent. The cingulum is not well defined and appears only as a vague swelling around the base of the tooth. However, a prominent swelling of the enamel occurs at the posterolingual corner of m1 and represents a small cingular shelf; anterior to this swelling on the lingual margin of the talonid, the base of m1 is indented or “notched.” Both the cingular swelling and indentation also occur in M. obtusidens.

The m2 measures 17.3 mm in length, 10.3 mm in width. This is the longest North American temnocyonine m2, yet the m1 of this same individual is exceeded in length by the m1 of three other species (T. macrogenys, D. oryktes, R. amplidens). A low, blunt protoconid occupies the center of the trigonid; a hypoconid of nearly equal height lies directly behind the protoconid in the center of the talonid. There is neither metaconid nor entoconid. A vestigial paraconid appears as a low platform at the anterolingual corner of m2.

The m3 is a small, oval tooth with minimal surface relief. It measures 8.6 mm in length, 7.2 mm in width. Because the enamel is damaged, cusp pattern is uncertain; however, a low protoconid was the principal cusp and a small, reduced talonid is present.

FAM 54134—The skull is largely uncrushed and the largest known for the subfamily. It displays “bearlike” proportions reflected in the swollen muzzle, inflated frontal region, heavy zygomatic arches, and pronounced sagittal and lambdoid crests. The upper teeth represent the most highly specialized durophagous dentition developed by a temnocyonine, a crushing dentition unlike that of bears (Ursinae). The locus of crushing in the upper teeth of F:AM 54134 involves P3–M1/p4–m2 whereas in living ursine bears the M1–M2/m1–m3 serve this function.

Basilar length of skull is 280 mm comparable in size to adult male Ursus americanus. The braincase is proportionately small relative to overall skull size; in fact the expanded frontal sinuses appear to have had a volume greater than the cranial cavity (table 7). The secondary palate is long (14 cm) and narrow for a skull of this length: palatal width measured across the M1s is 8.7 cm whereas only 3 cm of this is occupied by the palatal bone between the teeth, with the remainder taken up by the enlarged molars. The tall, narrow infraorbital foramen (height, 18 mm; width, 9 mm), opens on the maxilla above the upper carnassial. The orbitotemporal region is long: postorbital length is ∼19 cm, preorbital ∼12.5 cm. The left orbital region preserves the elongate depression in alisphenoid and orbitosphenoid bones for the optic foramen, sphenorbital fissure, and anterior aperture of the alisphenoid canal. The canal is 20 mm in length; its posterior opening shares a common fossa with the foramen ovale. The foramen rotundum opens internally into the canal as in other temnocyonine crania.

The basicranium is foreshortened: the length from the common fossa for the posterior opening of the alisphenoid canal and the foramen ovale to the ventral notch of the foramen magnum (an estimate of basicranial length) is only 5.5 cm, approximately one-fifth the basilar skull length. Although the basicranium is damaged there is an evident similarity to the M. obtusidens basicranium (ACM 34-41). Enough of the basicranial axis is preserved to show that, on the right side, the margin of the basioccipital was deeply excavated for an enlarged inferior petrosal venous sinus. This is the most pronounced development of the sinus in any temnocyonine skull and is related to the large size of M. ferocior. The sinus includes a deep central pocket also seen in Temnocyon altigenis (UCMP 9999), penetrating 13 mm into the basioccipital.

The morphology of the upper dentition is comparable to ACM 34-41, however the Keeline skull has larger, more robust teeth, a massive P3–M2 crushing dental battery, close-spaced P2–4 (no diastemata), huge canines, and a full complement of incisors with large I3.

Incisor alveoli show that the upper incisors became larger from I1 to I3: I1 alveolus measures 9.1 mm in length, 4 mm in width; I2 alveolus, 11.2 mm × 6.1 mm; I3 alveolus, 11.5 mm × 10 mm. There is a diastema of ∼9 mm between I3 and the canine alveolus. The canine alveolus measures 24 mm × 15 mm; the maxilla surrounding the canine roots is swollen to accommodate the large canines, indicating a male individual.

The upper premolars were somewhat crowded judging from the placement of P1–3 alveoli. This is not the crowding seen in many young amphicyonids since the wear on cheek teeth shows F:AM 54134 to be a mature adult. As is the case in the mandible, P2 is angled outward and shows the effect of crowding more than any other tooth, its posterior alveolus more lingual than the anterior. This degree of crowding also occurs in the upper teeth of M. obtusidens (ACM 34-41) and must be characteristic of large species of Mammacyon.

P1 is a small rounded peg (7.6 mm in length, 6.4 mm in width) preserved only on the right side close behind the large canine. The alveoli for P2–3 measure 18.6 mm by 9.4 mm, and 21.1 mm by 13.4 mm, respectively. The larger posterior alveolus common to each of these premolars shows that both P2 and P3 were posteriorly broad, especially P3.

The enormous P4 with its short metastylar blade and massive paracone and protocone represents the culmination of the trend toward a crushing dentition within the genus. Its 26 mm length is almost equalled by its 23.1 mm width. Length of the metastylar blade is 10 mm; length of paracone including the parastylar region is 17.4 mm. The embrasure between the protocones of P4 and M1 for the m1 trigonid is reduced to 8.5 mm in anteroposterior length, notably less than the anteroposterior lengths of the protocones themselves (P4 protocone, 12 mm; M1 protocone, 17.4 mm) so that very little of the lower carnassial could have been inserted into this space. Occlusion of the M. ferocior mandible (F:AM 27562) with the upper teeth of the skull demonstrates that the m1 protoconid is arrested at the level of the lingual cingulum of M1 by the narrow embrasure; the m1 paraconid and hypoconid also act as enamel stops against P4 and the M1 protocone basin. Thus, although some shear occurred between upper and lower carnassials, particularly in unworn teeth, a specialized crushing action was the dominant occlusal mode. Following an initial shearing stroke as the mandible brought the lower carnassial into contact with the upper teeth, the final phase of occlusion between p4–m2 and P3–M2 must have been mortar and pestle crushing employing the lingually expanded P4–M1 protocones. Ratios A/B and C/D (table 6) are the lowest in the genus, and are related to the development of a crushing occlusion in M. ferocior.

M1 measures 21.4 mm in length, 29.8 mm in width. Despite its enormous size, it retains the characteristic temnocyonine configuration in which an expanded protocone region is separated from the enlarged labial half of the tooth by a prominent constriction at the level of the protocone basin. The paracone is slightly larger than the metacone. These two cusps are labially bordered by a cingulum, which is more pronounced labial to the paracone and is continuous with a small parastyle at the anterolabial corner of M1. Although the cusps are worn, in all respects they appear as in M. obtusidens. The large protocone is situated in the center of a broad enamel platform more developed than in any other temnocyonine species. The enamel platform is surrounded by an expanded lingual cingulum. A preprotocrista extends from the protocone toward the anterior cingulum, and a weak vestige of a postprotocrista also appears on the surface of the enamel platform trending toward the posterior cingulum. There are no para- or metaconules. Lateral to the protocone itself is a deep protocone basin that receives the m1 hypoconid much as pestle fits against mortar.

M2 is rectangular in occlusal outline with its long axis oriented transversely; it measures 10.8 mm in length, 17.4 mm in width, and is much smaller than M1. The paracone is much larger than the metacone. A strong labial cingulum borders only the paracone; the metacone cingulum is weak. The paracone-metacone region forms an elevated labial rim that overlooks the flat protocone region. The centrally situated protocone on this enamel flat is but a reduced version of the same cusp on M1. Wear has reduced the protocone to a flat surface nearly coplanar with the enamel flat on which it rests.

M3 was no longer present in this species. The maxilla terminates abruptly behind M2 in smooth bone indicating that in life there was no tooth posterior to M2.

Length of the upper toothrow from the posterior border of the canine alveolus to the posterior border of M2 is 104.2 mm. Length of the left P1–3 based on an alveolar measurement is ∼51 mm. Greatest length of P4–M2 is 55.7 mm.

Discussion

Although the mandible (F:AM 27562) and skull (F:AM 54134) here assigned to Mammacyon ferocior were not associated, the teeth in the mandible not only correspond in size and form to the upper teeth but also occlude perfectly. The collector, Charles Falkenbach, did not provide exact locality data for these fossils, but we know that they were derived from the same general locality north of Keeline, Wyoming, from gray volcaniclastic fine-grained sandstone of the Arikaree Group. Recent geologic study of the area north of Keeline and examination of the fossils previously collected from the area by the Frick Laboratory suggest that the fauna including M. ferocior can be considered as mid- to early late Arikareean in age. The fauna lacks a number of mammalian species that typify the late Arikareean (Ar3) Harrison Formation fauna of Sioux County to the east (see Age and Correlation). Although a radioisotopic age is not available for the M. ferocior hypodigm, faunal relationships suggest that the species is older than ∼23 Ma. Mammacyon has not been found in any known latest Arikareean (Ar4) fauna.

Associated limb elements and vertebrae were collected with the M. ferocior mandible (F:AM 27562) in 1931; they are alike in form but larger in size than the postcranials associated with the genoholotype of Mammacyon (ACM 34-41). Additional limb and foot bones (F:AM 107758) collected by Falkenbach in 1950 “from brownish sandstone 15 feet below highest exposure, west end, north of Keeline, Wyo.” belong to a large temnocyonine and also possibly represent limb elements of M. ferocior. Included are a humerus (length, 234 mm; width of distal end, 50.3 mm), partial scapula (only the glenoid and adjacent blade), distal ?tibia, distal ?radius, and two elongate metapodials. M. ferocior retained a postcranial skeleton much like (if not identical to) M. obtusidens, indicative of a digitigrade cursorial gait in which the forelimb was characterized by a narrow distal humerus, a lengthening of radius and ulna with limited ability for pronation/supination, and an elongate tarsus and carpus (see Postcranial Osteology). These are the first large Cenozoic carnivorans to achieve a limb skeleton modified for a digitigrade stance, restricted fore-aft limb excursion, and a striding cursorial gait. This recommends a behavioral mode in which a cursorial habit was coupled with durophagous feeding, an ecology to some extent paralleling spotted hyaenids (Crocuta) in the Old World.

None of the other very large North American temnocyonine species (Temnocyon macrogenys, Delotrochanter major, Rudiocyon amplidens) are known from skulls, although Delotrochanter oryktes is represented by a substantial partial cranium (UNSM 47800) from the carnivore dens at Beardog Hill, Agate National Monument. The skull of M. ferocior (F:AM 54134) is the most complete cranium of a large temnocyonine and demonstrates that the terminal species of the Mammacyon lineage possessed a rather ursidlike profile, strong inflation of the frontal sinuses relative to volume of the cranial cavity, a broad rostrum swollen around the large canine alveoli, and a robust dentition with cheek teeth adapted for crushing bone, fibrous sinew, and muscle.

cf. Mammacyon

Figure 32

Fig. 32.

Tentatively referred mandibular fragment of Mammacyon (LACM 15908) with left m1–2 and partial p4, LACM loc. 1872, Sharps Formation, Shannon Co., South Dakota. A, labial view; B, occlusal stereopair. Note apical wear on m1 cusps.

Sunkahetanka pahinsintewakpa (in part): Macdonald, 1970: 60 (LACM 15908 was initially placed in the canid genus Sunkahetanka).

Referred Specimen

Fragment of mandible with left m1–2, posterior part of p4, and alveolus for m3 (LACM 15908), Sharps Formation, LACM loc. 1872, south of Wolf Ranch, NE1/4, sec. 15, T40N, R42W, Shannon County, South Dakota, collected by J. Harksen, July 1963.

Distribution

Early Arikareean, Sharps Formation, Wounded Knee area, South Dakota.

Comments

A small temnocyonine slightly larger than T. altigenis and with more derived m1–2; LACM 15908 has a more robust ml (length, 20.5 mm; width, 9.9 mm) relative to the T. altigenis hypodigm (m1 length, 17.4–19.5 mm; m1 width, 7.7–9.3 mm); m1 metaconid more reduced relative to T. altigenis (metaconid is a small vestigial cusp on posterolingual slope of protoconid); m2 not elongate (m1/m2 length ratio, ∼1.7) thus differs from M. obtusidens–M. ferocior (m1/m2 length ratios, 1.57, 1.6); m2 protoconid and hypoconid in anteroposterior alignment and more centrally situated than these cusps in T. altigenis; posterior base of p4 only slightly broadened, and posterior accessory cusp labially (not centrally) placed; small m3 present; thick, robust dentary below molars. Lacks an enamel swelling at base of m1 posterolingual to protoconid (present in Delotrochanter petersoni, a carnivore of about the same size).

Description

The fragmentary mandible of LACM 15908 is about the same depth (28 mm) below the m1 as Cope's type of Temnocyon altigenis (29 mm), however, the jaw is thicker and the teeth are larger and more robust (mandibular depth below m2 is similar, ∼31 mm in both species). Delotrochanter petersoni, smallest species of its genus, has a depth below m1 of ∼33 mm; although of about the same size as LACM 15908, its m1 lacks the metaconid and differs in the form of the principal cusps and cingulum. Only the posterior part of p4 is preserved in LACM 15908, anchored in the jaw by the posterior root. This tooth has both a basal cusp and a labially placed posterior accessory cusp. The posterior border of p4 fits tightly against the base of the m1 paraconid. The posterior part of p4 is only somewhat widened, and could be transitional between the plesiomorphic laterally compressed p4 of T. altigenis and the posteriorly broad p4 of M. obtusidens.

The m1 is a small, well-worn carnassial. The placement of the three trigonid cusps is similar to their position in AMNH 6855, Cope's holotype of T. altigenis, however the metaconid is more reduced in LACM 15908. The cusps are more worn than in AMNH 6855, and apical wear is pronounced on all m1 cusps but not present on m2. The broadened talonid of m1 bears a large, rounded hypoconid, nearly centrally situated; lingual to the hypoconid the enamel surface slopes downward to a narrow shelf lacking an entoconid. The labial m1 cingulum is sinuous, not straight as in D. petersoni.

The m2 is not elongated as in M. obtusidens–M. ferocior. The centrally placed, low, blunt protoconid is the largest cusp; a vestigial paraconid is indicated only by the slightly elevated anterolingual corner of the tooth; there is no metaconid. The hypoconid is lower than the protoconid, is placed directly behind it, and is the sole talonid cusp. The m2 paraconid is not as pronounced as in M. obtusidens–M. ferocior, hence LACM 15908 lacks the anteriorly extended m2 trigonid of the latter two species, formed by the paraconid and a protrusion of the anterolabial corner of the tooth. However, the m2 of LACM 15908 could structurally precede the more elongate m2 of M. obtusidens. A single alveolus for m3 is present.

The masseteric fossa terminates anteriorly under the posterior part of m2, and the forward inclination of m2–3 on the edge of the ascending ramus of the mandible is evident in LACM 15908 exactly as in Mammacyon obtusidens.

Discussion

This mandibular fragment, originally identified as a canid, is the earliest evidence of the Temnocyoninae described from the Cenozoic of the North American midcontinent. LACM Locality 1872 occurs in the Sharps Formation with an estimated age of between 28.8 and 29.4 Ma (Tedford et al., 1996, 2004; Macdonald, 1970). The mandible represents a temnocyonine more specialized in its dentition than fossils of Temnocyon altigenis from the John Day beds of Oregon. T. altigenis (UCMP 9999) from Logan Butte is certainly older than ∼28.9 Ma, the date obtained on an ash bed near the top of the John Day Formation at the butte, and more likely falls near 29.2–29.3 Ma if collected near dated tuffs at the base of Logan Butte. UCMP 9999 and LACM 15908 could be approximately contemporaneous species, although a somewhat younger age for LACM 15908 seems more probable.

The teeth of LACM 15908, despite small size, have a robust quality much like the larger durophagous temnocyonines of the later Arikareean. The absence of its anterior premolars makes referral to a genus difficult because the short, wide premolars of Delotrochanter are diagnostic relative to more plesiomorphic premolars of Mammacyon and Temnocyon. Lack of the elongate m2 that distinguishes the Mammacyon lineage prevents certain referral to that genus but the shorter m2 of LACM 15908 could represent the ancestral state. The labial m1 cingulum appears to be slightly sinuous, a trait present in Mammacyon, differing from the straight labial cingulum of Delotrochanter. Referral of LACM 15908 to Temnocyon altigenis is not appropriate because of the strong reduction of the m1 metaconid and the deep mandible and broader teeth, indicating a more derived carnivore. More complete material of the Sharps temnocyonine will be necessary to establish its identity.

Delotrochanter, new genus

Type Species

Delotrochanter oryktes, new species.

Included Species

Delotrochanter petersoni, new species; D. oryktes, new species; D. major, new species.

Distribution

Mid- and late Arikareean of northwest Nebraska; latest Arikareean of Nebraska-Wyoming boundary in vicinity of the Niobrara River.

Etymology

From the Greek, delos, for “evident,” and trochanter, “runner,” to emphasize the cursorial nature of these carnivores.

Diagnosis

Distinguished from Temnocyon by absence of the m1 metaconid; by a centrally placed m2 protoconid and hypoconid; and by proportions of P4 and M1 (table 6, ratios A/B, C/D); from Mammacyon by a short, less elongate skull, by shorter broad p2–3 and less elongate m2 (ratio E/F); and from Rudiocyon by a more elongate m2 (ratio E/F). See tables 1–TABLE 2TABLE 3TABLE 45.

Discussion

The genus includes temnocyonines that abandon the plesiomorphic form of the cheek teeth and adopt a crushing durophagous dentition different from that of Mammacyon in proportions of the carnassial-molar battery. Delotrochanter petersoni appears to be a mid-sized ancestral species evolving to the large D. oryktes and the even larger D. major. The genus ranges in time from the mid- to latest Arikareean but is not certainly known in the early Arikareean interval.

Delotrochanter petersoni, new species

Figure 33

Fig. 33.

Holotype mandibles of Delotrochanter petersoni (CM 1603), with left canine, p1 and m1, and alveoli for p2–p4, from the Monroe Creek beds (fide Peterson, 1907), head of Warbonnet Creek, Sioux Co., Nebraska. A, labial view and B, high oblique lingual view of left mandible and the alveoli of right mandible. The m1 lacks a metaconid.

Family Canidae, gen. et sp. indet: Peterson, 1907: 33–34, fig. 6. (The year of publication usually cited for this paper is 1906; however, a printed erratum in a copy in the American Museum's Osborn Library establishes March 21, 1907, as the publication date.)

Type

CM 1603, associated lower jaws, with only the left canine, p1, and m1 preserved. Also included under the same catalog number are the M1 protocone, the anterolabial corner of left P4, a metastylar fragment of right P4, a premolar, and four canine fragments. The Carnegie Museum field label records the holotype from the “middle Monroe Creek beds,” head of Warbonnet Creek, Sioux County, Nebraska, collected by O.A. Peterson and party, May 1904, but was reported in Peterson's publication (1907: 24) as “from the upper Monroe Creek horizon.”

Distribution

?Mid-Arikareean, Arikaree Group, near head of Warbonnet Creek, Sioux County, Nebraska.

Etymology

The species name recognizes the paleontologist Olaf Peterson who collected the holotype in northwest Nebraska in 1904.

Diagnosis

Smallest recognized species of Delotrochanter with m1 length of 22.3 mm; differs in size from the much larger D. oryktes and D. major. Distinguished from Temnocyon and Mammacyon by nearly straight (not sinuous) labial cingulum on m1 and by short, posteriorly broad p2–3. Differs from Temnocyon by absence of m1 metaconid; by its m1 paraconid-protoconid-hypoconid triad low, blunt, not specialized for shear and in direct anteroposterior alignment; and from Rudiocyon by much smaller size.

Referred Specimens

None.

Description

The canine measures 33.3 mm in height from tip to enamel base on the labial face, 12.0 mm in width, and 16.0 mm in length at the enamel base. A wear groove is developed on the posterolabial surface by the action of the upper canine.

The p1 measures 7.7 mm in length, 5.6 mm in width. It is a simple, conical tooth with an anteriorly placed main cusp and no accessory cusps.

Only roots of p2–4 are present, however they demonstrate that the premolars were closely spaced but not crowded. Labial placement of the anterior root of p2 suggests that the front of the tooth was rotated slightly outward. An important dental trait of Delotrochanter petersoni is that p2–3 are short with small-diameter anterior roots, a specialization of the anterior premolars also found in Delotrochanter oryktes and D. major, thus a defining characteristic of the genus. In Mammacyon and Temnocyon, p2–3 are elongate teeth with anterior roots of larger diameter.

The m1 measures 22.2 mm in length, 11.3 mm in greatest width (at protoconid), and 10 mm in transverse width across the hypoconid. There is no metaconid; the paraconid is short and robust, much lower than the protoconid, with a paraconid blade 3.7 mm in length. The wide, robust protoconid is 10.1 mm in length measured between the incisures it forms with the paraconid and hypoconid. Its talonid is dominated by a massive, blunt, centrally placed hypoconid that functions as a crushing cusp when applied to the protocone basin of M1. From the hypoconid-protoconid incisure to the rear of the talonid is 7.6 mm. An indistinct cingulum surrounds the base of m1, but this cingulum on the labial side is nearly straight, not sinuous, and is a key feature linking D. petersoni to D. oryktes.

The m2 was not preserved and its size cannot be accurately determined since both mandibles have been damaged in this area. Peterson (1907: 34, fig. 6) in his description of the specimen stated that “judging from the specimen, m2 was of considerable size.” Peterson figured two alveoli for the roots of m2: the anterior alveolus is represented by a root, and the posterior alveolus by a circular depression where a root tip once existed. Estimating the length of m2 from these two alveolar remnants gives an m2 length of ∼13 mm, and an m1/m2 ratio of ∼1.7, similar to that of D. oryktes (ratio E/F, 1.69, table 6).

Little can be said concerning the four isolated canine fragments that accompany CM 1603. However, these fragments and the intact canine in the mandible show that the canines are large teeth relative to the dimensions of the mandible.

The fragment of the anterolabial corner of the left P4 indicates that the upper carnassial had a slightly swollen basal cingulum, and that a fine enamel ridge ran from the tip of the paracone to a point just medial to the anterolabial corner, a typical feature of temnocyonines. In addition, the paracone was a low, blunt cusp as are the m1 cusps. This tooth fragment is configured as in Delotrochanter where the anterolabial corner is not anteriorly extended as it is in large Mammacyon. A fragment of the metastylar blade of the right P4 shows that the blade was short and broad, measuring 9 mm in length, about 8.2 mm in width. It compares well with the metastylar blade of D. oryktes (ACM 4804). An indistinct cingulum surrounds the blade. The fragment is similar in size to the fragmentary P4 metastylar blade of T. ferox (YPM-PU 10787) but is more robust. In fact, the narrow blade of T. ferox represents a more sectorial carnassial.

A fragment of the left M1 protocone shows that the protocone region was already anteroposteriorly widened. The protocone fragment measures 12.5 mm in anteroposterior width but in life the tooth was wider since the posterior part of the thickened lingual cingulum is missing. The M1 protocone of CM 1603 was knoblike and isolated in an enamel flat surrounded by an expanded lingual cingulum.

The depth of the mandible below m1 is 32.8 mm and below p2 is estimated at 30 mm. This is a rather shallow jaw relative to its thickness (12.9 mm below m1).

Note that Peterson's figure (1907: fig. 6) gives CM 1506 for this specimen, whereas his text reads CM 1603: this last is the correct catalog number and is the only number on the specimens. The specimen labels show that the mandibles initially were given CM 1506 and the isolated teeth fragments assigned CM 1505; all this material later was united under CM 1603.

Discussion

This mandible, collected in 1904, was the first record of Temnocyoninae in the North American midcontinent, and was believed by Peterson (1907) to be a canid (within the broader meaning of that term as used at that time, i.e., canids and amphicyonids). The few intact teeth present in the left mandible (canine, p1, m1) made identification difficult. However, the following traits suggest that this is the earliest North American representative of Delotrochanter:

(1) absence of m1 metaconid, straight labial cingulum on m1, and large centrally placed m1 hypoconid filling the talonid; (2) short p2–3, with small-diameter anterior roots and larger-diameter posterior roots; (3) unexpanded anterolabial corner of P4 like that of D. oryktes; low, short P4 metastylar blade; (4) slender, shallow horizontal ramus of the mandible.

Fine-grained gray volcaniclastic sandstone adheres to the mandibles, leaving no doubt as to the sediment in which the fossil was found. Along the Pine Ridge escarpment in Sioux County near the head of Warbonnet Creek, this matrix is typical of the Arikaree Group, and indicates gray volcaniclastic sandstone exposures of the middle and upper part of the escarpment. Peterson's confusing attribution to “middle” Monroe Creek (on field labels) and later on, “upper” (in his published article), at least indicates that CM 1603 came from beds stratigraphically between the lower Arikaree fluvial sandstones at the base of the Pine Ridge and the eolian Harrison Formation sandstones forming the upper part of the escarpment, but prevents an exact stratigraphic designation.

The Pine Ridge escarpment immediately east of Warbonnet Creek includes the stratotypes for the Monroe Creek Formation and Harrison Formation of Hatcher (1902a) at Monroe Creek Canyon. Because Peterson did not attribute the fossil to the Harrison Formation, one can be reasonably confident that CM 1603 did not come from that rock unit as understood by Peterson and Hatcher (i.e., the upper ∼200 ft of the Pine Ridge Arikaree escarpment). Hatcher in 1902 described the Monroe Creek Formation as “300 feet of very light-colored, fine-grained, not very hard, but firm and massive sandstones.” Based on Peterson's published attribution to “upper Monroe Creek,” one would predict that CM 1603 came from fine-grained gray tuffaceous sandstones of the Pine Ridge escarpment ∼200–400 ft below the terminal paleosol of the Harrison Formation (Hunt, 1985). These sandstones exhibit large-scale eolian cross-strata, in which CM 1603 was possibly found, that constitute much of the middle part of the Pine Ridge stratigraphic section in Sioux County from Monroe Creek west to Warbonnet Creek. Fossil mammals are rarely encountered in these beds.

Previously I attributed the temnocyonine here named Delotrochanter petersoni to an early Arikareean chronofauna (Hunt, 1985: 192) but Peterson's (1907: 24) “upper Monroe Creek” allocation, which places CM 1603 in association with the oreodonts Promerycochoerus and Phenacocoelus, cannot be ruled out and suggests a younger age. It is at least certain D. petersoni occurs in a fauna older than that typical of the Harrison Formation of Peterson. This is supported by the size relationship of D. petersoni to its presumed descendant, D. oryktes. The latter species occurs in the Harrison Formation in the Niobrara River valley at Agate National Monument, and its predecessor, D. petersoni, is of the smaller size and dental features one would predict in “middle to upper Monroe Creek” sediments of the Pine Ridge.

Delotrochanter oryktes, new species

Figures 34–Fig. 35.Fig. 36.Fig. 37.Fig. 38.39

Fig. 34.

Holotype mandible of Delotrochanter oryktes (ACM 4804) with left canine, p4–m2, alveoli of p1–3, from Stenomylus Quarry, Harrison Formation, Agate Fossil Beds National Monument, Sioux Co., Nebraska. A, labial and B, lingual views. Male individual found with baculum.

Fig. 35.

Holotype mandible of Delotrochanter oryktes (ACM 4804) with left canine, p2–m2 (p2–3 inserted in mandibular alveoli to demonstrate small diameter of anterior roots).

Fig. 36.

Occusal stereopairs of the holotype mandibles of (A, C) Delotrochanter oryktes (ACM 4804) and (B, D) Delotrochanter major (F:AM 27561). D. oryktes, right p4–m2, alveoli of p1–3; D. major, left p2–m2. Note absence of m1 metaconids and relative anterior and posterior alveolar diameters of p2 in D. oryktes. The p4–m2 are crushing teeth in both species.

Fig. 37.

A, Delotrochanter oryktes (UNSM 47800), cranium with P1–M2 of presumed female from carnivore den, Beardog Hill, Agate Fossil Beds National Monument, Sioux Co., Nebraska. The short, deep hyenalike skull is unique among temnocyonines. B, Plan map of dens showing location of burrow (den 2) that contained the skull and foot bones of D. oryktes (UNSM 47800). The beardog Daphoenodon superbus was the principal occupant of dens 1, 3, and 4; mustelids were found in dens 5 and 6.

Fig. 38.

Left maxilla of the Delotrochanter oryktes cranium (UNSM 47800) with P1–M2, presumed female, Beardog Hill, Agate Fossil Beds National Monument, Sioux Co., Nebraska. The crushing P4–M1 is more gracile than these teeth in the male (ACM 4804, fig. 39).

Fig. 39.

Right maxilla with P4–M2 of Delotrochanter oryktes (ACM 4804, male), Stenomylus Quarry, Harrison Formation, Agate Fossil Beds National Monument, Sioux Co., Nebraska.

“The major part of a skeleton of Daphoenodon superbus Peterson ….”: Loomis, 1910: 298.

Type

ACM 4804, partial skull with right I2–3, C, P1–2, P4–M2, and left I2–3, C (partial), P1–2, P4–M1, labial half of M2; right mandible with c, p4–m2; left mandible with c, p2–4, m1–2; and much of the postcranial skeleton, from the Harrison Formation, Stenomylus Quarry, Agate Fossil Beds National Monument, Sioux County, Nebraska, collected by F.B. Loomis, 1908.

Distribution

Late Arikareean, Harrison Formation and basal Anderson Ranch Formation, Agate Fossil Beds National Monument, Sioux County, Nebraska.

Etymology

From the Greek, oryktes, for “digger” or “excavator,” in allusion to the discovery of one of the individuals of the species in a burrow.

Diagnosis

Distinguished from D. petersoni by larger size (D. petersoni m1 length, 22.3 mm; D. oryktes m1 length, 27.1 mm). Dentition clearly antecedent to that of the much larger D. major (tables 2, 3), but not as massive and robust. Differs from Temnocyon by loss of the m1 metaconid and by short, posteriorly wide p2–3 and P2–3; from Mammacyon by a posterior accessory cusp on p3, a large centrally placed posterior accessory cusp on p4, and by a smaller P4 with lingually abbreviated protocone region; and from Rudiocyon by the form of p4.

Referred Specimens

(1) UNSM 47800 (field no. 8-81), partial skull with left P2–M2, lingual half of right M1, right I2–3, and petrosal fragment; associated left calcaneum, left metatarsals 2–4, left metacarpals 3–5 with proximal phalanx and 2 sesamoids, left magnum; found in place within a burrow, Carnegie Quarry 3, Beardog Hill den site, Agate Fossil Beds National Monument, Sioux County, Nebraska, collected September 14, 1981, by J. Kaufman and R.M. Hunt (UNSM field no. 9-81 was given to the fifth metacarpal, found near the skull, presumably from the same individual); (1a) CM 1589b, right metatarsal 5, Carnegie Quarry 3, Beardog Hill den site, collected by O.A. Peterson, 1904–1905, probably the same individual as UNSM 47800; (2) YPM-PU 24872, left calcaneum, removed from Princeton University quarry block (now YPM-PU 12213), Princeton Expedition of 1914, loc. 1002A, Carnegie Hill waterhole bonebed, Agate Fossil Beds National Monument, Sioux County, Nebraska.

Description

Both individuals referred to this species (ACM 4804, UNSM 47800) come from Agate Fossil Beds National Monument, Sioux County, Nebraska, and were collected 73 years apart. ACM 4804, designated here as the holotype, includes a nearly complete skeleton discovered by Loomis at Stenomylus Quarry in 1908. It was originally referred to Daphoenodon superbus: Loomis (1910) did not recognize it as a temnocyonine, a group almost unknown at that time. A second individual (UNSM 47800) is represented by a skull (fig. 37) and bones of the fore- and hind feet, found within a carnivore burrow (front cover) during the reopening of Carnegie Quarry 3 by the University of Nebraska in 1981 (Hunt et al., 1983; Hunt, 1990: 106–107).

The mandible and lower dentition are known in ACM 4804 but were not recovered with UNSM 47800. The following description is based on ACM 4804; it is selected here as the holotype of the species because of the association of maxillary and mandibular dentition with the postcranial skeleton (see Postcranial Osteology). Appropriate comparisons are made with the larger derivative species, D. major, known only from teeth.

The mandible is deep yet thin, with a distinctive recurved coronoid process and a wide (35.8 mm) articular condyle. The anterior margin of the ascending ramus curves gradually downward beneath the rear molars so that m2–3 are forwardly inclined as in several other temnocyonines. The nearly uniform depth of the horizontal ramus is demonstrated by similar measurements beneath m1 (45.7 mm) and p2 (41 mm). Length of the mandible measured from the symphysis (between the first incisors) and the posterior surface of the articular condyle is 205 mm.

Lower incisors did not survive in either individual referred to this species. Alveoli for i1–3 are present in the left mandible of ACM 4804: the single alveolus for i3 (width, 4.5 mm) shows that it was the largest of the lower incisors. A smaller i2 alveolus is situated medial to i3, and a very small i1 alveolus is placed slightly internal and ventral to i2.

The left lower canine of ACM 4804 is not as strongly recurved as in some species of Temnocyon (UCMP 9999, YPM-PU 10787). Canine height from the base of the enamel to the unworn tip on the labial side is 36.8 mm. Length and width measured at the base of the enamel are 18.5 mm and 12.5 mm, respectively. A pronounced enamel ridge runs from the tip to the enamel base on the posterior edge, and a second ridge travels down the anterolingual face. Although the m1 of a wolf is about the same size as the D. oryktes m1, the beardog canine is much larger.

The p1 is not preserved in ACM 4804; the single large alveolus measures 8.2 mm in length, 6 mm in width, and is crowded between the canine and p2 alveoli, and is in contact with both. Closely spaced alveoli are common to the entire lower toothrow. The slight crowding of premolar alveoli is due to the youth of this carnivore (all teeth are only in early wear).

The p2 is a tall, yet anteroposteriorly short, double-rooted tooth, 14.9 mm in length, 8.2 mm in greatest width. There is no posterior accessory cusp. The anterior face is steeply inclined and is traversed by a fine enamel ridge that extends from tip to anterolingual cingulum. The posterior face is more gradually inclined with an enamel ridge running from tip to posterolabial cingulum. The anterior root of p2 is much smaller (length, 4.9 mm) than the robust posterior root (length, 8.5 mm), a distinguishing trait of the species.

The p3 measures 16.6 mm in length, 9.4 mm in posterior width, and 6.9 mm in anterior width. The tooth is short and posteriorly wide, which is evident when compared with the respective dimensions of p3 in Mammacyon ferocior (19.3 mm, 9.0 mm, 6.8 mm). There is a posterior accessory cusp, somewhat labially placed. A slightly swollen basal cingulum is more pronounced on the lingual and posterior sides. Just as for p2, the anterior root is smaller than the posterior root.

The p4 is robust, with a wide heel, measuring 21.0 mm in length, 10.8 mm in posterior width, and is much larger than p3. Its form is diagnostic of the genus, notably the large posterior accessory cusp centrally placed on an expanded heel behind the principal cusp. Delotrochanter is the only temnocyonine that has evolved a centrally placed posterior accessory cusp on p4. In Temnocyon and Mammacyon the accessory cusp is labially placed.

The height of the principal cusps of p2–4 is pronounced, so much so that the linear serial alignment of the premolar cusps is continued along the toothrow by the principal cusps of m1 and m2, and this cusp alignment is further emphasized by the elevation of m2–3 on the rising margin of the ascending ramus.

The m1 is a crushing carnassial, 27.1 mm in length, 12.7 mm in greatest width, formed by three aligned bunodont cusps, the paraconid-protoconid-hypoconid. There is no metaconid or entoconid. The thickened labial cingulum is essentially straight, not sinuous as in Temnocyon and Mammacyon. The hypoconid is centrally placed on the talonid and in labial view is as tall as the paraconid (as in Mammacyon but not Temnocyon). The nearly equal development of hypoconid and paraconid flanking the protoconid creates a profile typical of large species of Mammacyon and Delotrochanter, but a somewhat larger hypoconid in Delotrochanter emphasizes the cusp alignment adapted for crushing.

The m2 is rectangular in occlusal view, measuring 16.0 mm in length, 10.0 mm in greatest width (m1/m2 length ratio, ∼1.7, table 6). The protoconid and hypoconid are placed in direct anteroposterior alignment behind the m1 hypoconid. These three cusps are about the same height and form a crushing battery posterior to the carnassial blade. There is a low vestigial paraconid. The m2 is not as elongate (relative to m1 length) as in Mammacyon, but the cusps of m2 are centrally placed as in that genus. Hence, one of the hallmarks of Delotrochanter is the central placement of all cusps from p4 to m3 (with slight lingual diversion of the m1 paraconid).

The m3 is not preserved but is represented in ACM 4804 by a single large alveolus, length 5.3 mm, width 5 mm.

In ACM 4804, p1–m3 length is 104.5 mm; p2–m2 length, 89.6 mm; p1–4 length, 59.7 mm; m1–3 length, 48.2 mm.

The skull of ACM 4804 was damaged during preparation and mounting of the specimen at Amherst College shortly after it was collected in 1908. With the aid of Drs. Margery and Walter Coombs, we were able to recover the skull and restore part of the rostrum and palate and a few basicranial fragments, including the right petrosal and a partial basioccipital.

The central incisors (I1–2) of ACM 4804 are much smaller than I3. I1 is not preserved in ACM 4804, however the alveolus measures 9.3 mm in length, 4.1 mm in width. I2 is present with a vertical anterior face, a sloping posterior surface, and a prominent auxiliary cusp (Nebenzacke) on its lateral margin. I2 is 9.8 mm in length, 5.7 mm in width at the enamel base. I3 is a large, recurved caniniform tooth measuring 23.2 mm in height from tip to posterolabial enamel border, 13.1 mm in length, 10.1 mm in width at the base of the enamel. I3 has a thin enamel ridge from tip to posterolabial base and another on the lingual face from tip to enamel base. I2–3 in UNSM 47800 show no differences from those incisors in ACM 4804.

The upper canine measures 37.9 mm in height from its posterolabial enamel base to its unworn tip. Length and width at the base of the enamel are 19.3 and 13.4 mm. An enamel ridge runs from the tip to the enamel base on the posterior surface; a second ridge runs from tip to base on the anterolingual surface.

P1 is a small, conical tooth 9.4 mm in length, 6.9 mm in greatest width. There is a single principal cusp and no accessory cusp. The principal cusp is placed forward of center, and from this cusp a fine enamel ridge extends down both the anterior and posterior faces of the tooth.

P2 is much larger than P1, measuring 15.8 mm in length, 8.7 mm in posterior width (in UNSM 47800, P2 measures 15.9 mm and 8.0 mm). P2 is tall, triangular in lateral view, with a vertical, nearly rectilinear anterior face and a more gradually sloping posterior face. Of importance is the small anterior root relative to the larger posterior root, a diagnostic feature of Delotrochanter, indicating that the shortening of p2 in the lower jaw is reciprocated in the upper dentition. Alveolar length of the P2 anterior root is 5.6 mm; posterior root, 10.0 mm; length of the roots themselves are, respectively, 5.2 and 8.8 mm. The posterior heel of P2 is widened and a weak accessory cusp is labially placed low on the broad heel. Thin enamel ridges are present on the anterior and posterior faces of P2, running to the respective anterolingual and posterolabial corners.

P3 is lost from the skull of ACM 4804 but is present in UNSM 47800 (length, 18.2 mm; width, 9.8 mm) where it is about the same height as P2 but posteriorly wider with a small accessory cusp slightly labially situated on the posterior slope.

P4 measures 21.4 mm in labial length, 18.8 mm in greatest transverse width across the protocone. This massive tooth has a short metastylar blade 9 mm in length that forms a moderately developed shear surface with the large paracone. Small vertical wear facets are developed on the lingual faces of metastylar blade and paracone but the predominant crushing mode of occlusion is evidenced in this young individual by blunt-cusp wear on the paracone. As in the larger temnocyonines, the rounded blunt protocone is enlarged for crushing but is not as massive as in species of Mammacyon. The protocone in Delotrochanter does not extend as far toward the midline of the palate as it does in Mammacyon, and does not reach the lingual margin of M1. P4 length is 22.8 mm and width is 17.9 mm in UNSM 47800, too young an individual to develop shear facets on P4.

M1 measures 17.7 mm in length of the labial margin, 25.4 mm in greatest transverse width, and 14.7 mm in anteroposterior width of the protocone region. In UNSM 47800, M1 length is 17.5 mm; greatest width is 23.9 mm; and width across the protocone is 14.1 mm, thus a smaller molar relative to the M1 in ACM 4804. An M1 that is constricted at the level of the protocone basin so that the tooth has a “waist” is typical of Delotrochanter and Mammacyon. The labial half of M1 is formed by a prominent, yet rather blunt, paracone and somewhat smaller metacone bordered by a labial cingulum thickened in the parastylar region. The lingual faces of the paracone and metacone form a vertical shear surface but these cusps are lower and shear is not as developed as in large species of Temnocyon. The low protocone is isolated within an enamel platform, creating a flat crushing surface surrounded by a much-thickened lingual cingulum. A preprotocrista runs from the protocone to the anterior cingulum at the base of the paracone but there is no postprotocrista. On the M1s of both ACM 4804 and UNSM 47800 the lingual cingulum thins abruptly at the level of the protocone basin. The lingual cingulum of UNSM 47800 is not as developed as in ACM 4804.

M2 is a rectangular tooth, much smaller than M1, measuring 9.3 mm in labial length, 17.4 mm in greatest width (M2 length, 8.9 mm; width, 15.3 mm in UNSM 47800). The paracone is much larger than the strongly reduced metacone, both bordered by a labial cingulum thickened in the parastylar region. The lingual faces of para- and metacone form a low shear surface that continues the paracone- metacone shear plane of M1 (this plane less evident in UNSM 47800 in which M2 is more reduced). The M2 protocone is a low, blunt cusp isolated on an enamel platform surrounded by a slightly thickened lingual cingulum. There is no evidence of M3 although that part of the maxilla where M3 would occur is not preserved. However, UNSM 47800 has an intact maxilla in which M3 is certainly absent.

The skull (UNSM 47800) found in 1981 within a burrow at Beardog Hill, Agate Fossil Beds National Monument preserves the left P1–M2 (figs. 37–38). This is the only known skull of Delotrochanter. The left side of the skull and upper dentition are largely intact, whereas the right side was destroyed by plant roots that invaded the burrow. The skull is remarkably similar in form to the skull of the hyena, Crocuta crocuta. Both skulls are broad and rather short with laterally extended zygomatic arches and large temporal fossae indicating well-developed temporal jaw musculature. Both have blunt, short snouts of similar length: the preorbital proportion of skull length in UNSM 47800 is 40% and in the hyena 35%. UNSM 47800 differs from Crocuta in possessing a more inflated frontal region and more massive snout with larger canines. The infraorbital foramen of UNSM 47800 is much larger than this foramen in the hyena, suggesting a more developed neurovascular supply to the nasal region. The braincase of UNSM 47800, despite the rather short skull, is the largest of any known temnocyonine (table 7).

The upper teeth are extremely similar in form to those of ACM 4804 but are somewhat smaller and more gracile. P4 is longer and its protocone not as developed, and the M1–2 are smaller, not as hypertrophied as in ACM 4804. The crushing function of P4–M2 is better developed in ACM 4804. ACM 4804 is a male (a baculum was found with the skeleton) and, based on the gracile teeth, UNSM 47800 could be a female. Sexual dimorphism is common among amphicyonids and it would not be surprising to find it present in temnocyonines. Dimorphism in temnocyonine species is difficult to demonstrate because of the lack of multiple individuals from a single locality.

Discussion

Delotrochanter is defined on the basis of the holotype dentition and skeleton of D. oryktes (ACM 4804) supplemented by information derived from the individual (UNSM 47800) found in the carnivore dens at Agate National Monument. These two individuals are clearly closely related and are assigned here to a single species, diagnosed by short, posteriorly wide premolars (p2–4, P2–3) with centrally (not labially) placed posterior accessory cusp on p4. These premolars differ from the longer premolars and labially placed p4 posterior accessory cusp found in Temnocyon and Mammacyon. Although the p4 of Mammacyon ferocior is broader than that of M. obtusidens, the posterior accessory cusp in both species remains labial in position. Also, Delotrochanter oryktes does not have as prominent a protocone on P4 as in Mammacyon nor is its m2 as elongate; and it lacks the m1 metaconid found in Temnocyon. These differences, however, are most readily observed in relatively complete dentitions: isolated teeth will in many instances be difficult to assign.

The mandible of ACM 4804 is only ∼2 cm longer than that of a mature female Alaskan wolf (ZM 17459: mandibular length, 18.5 cm; basilar length, 22 cm); yet mandibular depth and the height of the coronoid process are much greater and the canine, premolars, and m2 are conspicuously larger and more robust. The m1 is about the same length, but is a wider, blunt-cusped tooth, its protoconid-paraconid lacking the sharp edges seen on the m1 shearing blade of the wolf. The feeding strategy of D. oryktes relied on powerful jaw musculature applied to a deep, rigid mandibular “beam” carrying the heavy, blunt premolar and molar crushing dentition with enormous canines.

The four individuals of D. oryktes from Agate National Monument occur at three superposed stratigraphic levels: (1) ACM 4804, the holotype, was found ∼12–15 m above the two principal camel-producing horizons at Stenomylus Quarry in the Harrison Formation (a second individual, ACM 4804A, a juvenile, was found with ACM 4804 but is known only from postcranials); (2) The calcaneum (YPM-PU 24872) indicates the presence of D. oryktes in the Agate waterhole bonebed (basal Anderson Ranch Fm.), which occurs stratigraphically between the levels of ACM 4804 and UNSM 48700; (3) The adult (UNSM 47800) discovered in the carnivore dens at Beardog Hill (Hunt et al., 1983: fig. 1, den 2, burrow C) also occurs in the basal Anderson Ranch beds (the dens are excavated into the waterhole bonebed). The amount of time intervening between these stratigraphic levels remains uncertain. The age of ACM 4804 is approximated by the Agate Ash, recently redated at ∼22.9 Ma (Izett and Obradovich, 2001).

Despite its younger stratigraphic age, UNSM 47800 is the smaller, more gracile animal relative to ACM 4804. Given the baculum found with ACM 4804, the size difference between UNSM 47800 and ACM 4804 is more likely due to sexual dimorphism and not relative stratigraphic position.

The postcranial skeleton of ACM 4804 is discussed under Postcranial Osetology.

Delotrochanter major, new species

Figures 36, 40–Fig. 41.42

Fig. 40.

Holotype dentition of Delotrochanter major (F:AM 27561), left p2–m2 (p2, p3, and m2 damaged), Anderson Ranch Formation, near Van Tassell, Niobrara Co., Wyoming. A, labial and B, lingual views.

Fig. 41.

Holotype dentition of Delotrochanter major (F:AM 27561), right p3–m1, Anderson Ranch Formation, near Van Tassell, Niobrara Co., Wyoming. A, labial and B, lingual views. Note large p4 relative to the carnassial in this terminal species of Delotrochanter.

Fig. 42.

Holotype maxillary teeth of Delotrochanter major (F:AM 27561). A, lingual view of right P2–4. B, stereopair of right P3–4; enamel was etched by plant roots.

Type

F:AM 27561, right and left lower canines; partial p2–3, complete p4–m1, m2 trigonid, all of the left side; right p3–m1; an upper canine, right P2–4, maxilla fragment with right I1–3, left I1–2, and right P1 in matrix, from the Anderson Ranch Formation, collected by Paul Miller in 1927 from the vicinity of Van Tassell, Niobrara County, Wyoming.

Distribution

Latest Arikareean, Anderson Ranch Formation, near Van Tassell, Niobrara County, Wyoming.

Etymology

From the Latin, major, “greater,” in reference to the large size of the terminal species of this genus.

Diagnosis

Distinguished from D. petersoni by larger size (m1 length: D. petersoni, 22.3 mm; D. major, 28.0 mm). All D. major teeth are larger and more robust than the equivalent teeth of its predecessor D. oryktes (tables 2, 3). Differs from Temnocyon by loss of the m1 metaconid and by short, posteriorly wide p2–3 and P2–3; from Mammacyon by a posterior accessory cusp on p3, a large centrally placed posterior accessory cusp on p4, and by a smaller P4 with lingually abbreviated protocone region; from Rudiocyon by the form of p4.

Referred Specimens

None.

Description

Neither mandible of D. major (F:AM 27561) survived. The dentition is embedded in series in sediment without mandibular bone supporting the teeth. Thus only D. oryktes (ACM 4804) and D. petersoni demonstrate the form of the mandible in Delotrochanter.

In F:AM 27561, the left lower canine measures 36.6 mm in height from the slightly worn apex to the labial enamel-dentine junction (estimated unworn height, ∼40 mm); length and width at the base of the enamel are respectively 18.4 mm and 12.8 mm. The posterolabial surface is grooved by the upper canine; a second wear facet made by I3 occurs at the base of the anterolingual face. In other respects the tooth is similar to ACM 4804.

The p1 is not preserved, and only the main cusp of the left p2 remains; this indicates a tall principal cusp—there are no evident differences from p2 of ACM 4804.

The p3 measures 17.7 mm in length, 10.2 mm in greatest width, and is nearly identical to ACM 4804 except in its larger size and greater posterior width. A centrally placed posterior accessory cusp is present.

The p4 measures 21.6 mm in length, 9.2 mm in anterior width, 12.3 mm in posterior width; its posterior accessory cusp is very large, centrally placed, and surrounded by a well-defined cingulum, which becomes less conspicuous anteriorly on both lingual and labial sides. The heel of p4 is much wider than in ACM 4804 as is true for p3.

The m1 is similar in all respects to that of ACM 4804 but is more massive and robust, measuring 28.0 mm in length, 14.1 mm in greatest width. Although these lower carnassials were undoubtedly capable of limited shear (vertical wear facets occur on the paraconid-protoconid blade), the premolars, m2–3, and m1 functioned primarily as crushing instruments. The m1 hypoconid and the protoconid of m2 are blunt, low cusps, and in this species could have served no role other than crushing.

The m2 is represented only by the trigonid. It is similar to the m2 trigonid of D. oryktes but is more massive, having a very large, blunt protoconid and small vestigial paraconid with a strong labial cingulum; there is no metaconid. Trigonid width is 10.6 mm. Based on similarity of the cheek teeth in F:AM 27561 and ACM 4804, the complete m2 of D. major was a larger version of the m2 in D. oryktes.

The m3 was not preserved but would have been present.

I1–I2 are of equal size and similar cusp form. Both have a vertical anterior face and a curvilinear posterior surface, and both possess prominent auxiliary cusps (Nebenzacken) on their lateral margins as in D. oryktes. I2 measures 9.9 mm in length, 5.6 mm in width; I1 measures 9.1 mm in length, 5 mm in width. The large I3 measures 22.8 mm in height, 13.3 mm in length, and about 10 mm in width at the enamel base. There is a wear groove on the posterolabial face of I3 made by the upper canine.

P1 has the same subconical form as in ACM 4804 and measures 9.4 mm in length by 6.9 mm in width.

P2 is much larger than P1; it is tall, subtriangular in labial view, with near-vertical anterior and more inclined posterior faces, and measures 17.0 by 9.0 mm. The heel of P2 is broad; the thin enamel ridge on the posterior slope ends in a minute cingular cuspule at the posterolingual corner.

P3 is tall, triangular in lateral view, and posteriorly broad, measuring 18.7 mm in length, 11.9 mm in posterior width. The posterior face has a small, labially situated, posterior accessory cusp. P3 has fine enamel ridges on its anterior and posterior faces; the anterior ridge terminates at the anterolingual margin of these teeth; the posterior ridge at the posterolabial corner. The expansion of the heel of both P2 and P3 is more pronounced than in D. oryktes.

P4 measures 23.5 mm in length, 19.8 mm in width across the protocone, and is similar in form to the P4 of ACM 4804 but is larger and slightly more robust. The metastylar blade is blunt as is the protocone; both are essentially unworn, and the slightly worn paracone shows only a flat apical wear facet. Vertical shear, although possible, appears to have been secondary to a crushing action.

The upper molars of D. major were not recovered, but based on the remaining teeth they would be similar in form yet somewhat larger than M1–2 of D. oryktes.

Discussion

F:AM 27561 differs from D. oryktes in larger size and in the greater accentuation of many defining features of the cheek teeth. Not only are its teeth larger but cingula are more pronounced, the cusps more massive, and p3–4 and P2–3 are posteriorly broadened. The defining dental traits of Delotrochanter are evident and indicate that D. major is clearly derived from D. oryktes. Paul Miller, the collector, reported that F:AM 27561 came from the “vicinity of Van Tassell,” Wyoming. Near the town of Van Tassell only the Harrison Formation and the Anderson Ranch beds outcrop in the area where Miller worked. Fortunately, sediment adhering to F:AM 27561 shows the pale reddish-brown patina of the Anderson Ranch sandstones. The advanced features of its dentition indicate that Miller may have obtained F:AM 27561 stratigraphically high in the Anderson Ranch beds in the stratotype area of the formation in the Niobrara Canyon (see Hunt, 2002b: 35–39), which is in proximity to the town of Van Tassell. F:AM 27561 is of latest Arikareean age and represents the youngest occurrence of Delotrochanter. More than 80 years of exploration of Anderson Ranch Formation outcrops in northwest Nebraska and Wyoming have failed to yield additional remains of this rare temnocyonine since Miller's fortunate discovery.

The stratigraphic levels that produced D. oryktes are well documented, and the horizon that yielded D. major can be reliably estimated. The stage of evolution of these two species of Delotrochanter, which occur in late to latest Arikareeean time, can be contrasted with that of Mammacyon. The occurrence of the terminal and largest species, M. ferocior, preceded the earliest occurrence of D. oryktes in the Harrison Formation at Stenomylus Quarry. Thus, the Mammacyon lineage reached its apex at a time before 23 Ma, when the coeval species of Delotrochanter must have been a much smaller carnivore in the same geographic region.

Dentition

The upper cheek teeth (P3–M2) best show this progressive shift in dental pattern (fig. 43). In Temnocyon altigenis (UCMP 9999, fig. 4A) and T. subferox (YPM 10065, figs. 10A, 43A) the shearing carnassial retains a bladelike form with the protocone advanced in the anterior direction, leaving a large embrasure between P4 and M1 for the lower carnassial. This is the primitive amphicyonid condition, also common to primitive arctoid and cynoid carnivorans. M1 is essentially a triangular tooth in occlusal view, its lingual half somewhat broader than in Daphoenus, and flanked by a small elliptical M2. M3 is lost in all temnocyonines, unlike most daphoenine and amphicyonine beardogs. P3 remains a narrow premolariform tooth.

Fig. 43.

Maxillary dentitions showing the development of a durophagous dental battery in temnocyonines: A, Temnocyon subferox; B, Temnocyon ferox; C, Mammacyon ferocior; D, Delotrochanter major.

In younger species of Temnocyon a subtle change in the cheek teeth has taken place. In Temnocyon ferox (fig. 43B), body size has increased and the form of P4 and M1 has been modified. A shearing carnassial is still present, yet the tooth is wider and the protocone retracted posteriorly and enlarged relative to T. altigenis and T. subferox. M1 is less triangular, more quadrate in form, and the protocone region has begun to expand, although the P4–M1 embrasure remains fully functional. A shearing dental battery is still evident. M1 of T. percussor (fig. 15F) could be derived from the smaller T. ferox M1, and is lengthened lingually, whereas the M1 of T. fingeruti is little different from a scaled-up M1 of the Logan Butte T. altigenis. The M1 of T. fingeruti remains a relatively plesiomorphic molar (fig. 17).

The terminal species, T. macrogenys, displays enormous cheek teeth but the carnassial and m2 do not adopt the more modified occlusal form found in Mammacyon and Delotrochanter. The m1 and m2 of T. macrogenys retain metaconids and labially placed hypoconids (figs. 21, 44C), primitive features lost in the other genera (figs. 44A, B).

Fig. 44.

Mandibular dentitions of the three terminal species of the genera Temnocyon, Mammacyon, and Delotrochanter: A, Delotrochanter major, F:AM 27561; B, Mammacyon ferocior, F:AM 27562; C, Temnocyon macrogenys, F:AM 54139.

Species of Temnocyon remain dentally closer to the plesiomorphic dentition than other temnocyonines. On this basis they have been grouped together. The progressive size increase in these species as they undergo a gradual transformation of dental pattern over time lends support to this paraphyletic classification.

In species of Mammacyon (M. obtusidens, M. ferocior, figs. 24, 27, 31, 43C) where the upper cheek teeth are preserved, the P3–M2 transform from the plesiomorphic state to a full-blown crushing dentition. In M. obtusidens some P4 shear remains, which is much less evident in M. ferocior. In these two species the P4 metastylar blade is shortened, the paracone becomes a tall, blunt cone, and the protocone a rounded, broad pestlelike cusp. M1 is now a crushing platform with a broadened lingual half (the protocone region), surmounted by a blunt, knoblike protocone. The P4–M1 embrasure is almost obliterated; hence the lower carnassial in these species chiefly crushes food against the P4 protocone/M1 lingual platform, although early in life some shear is present. This configuration is complemented by the shape of P3, whose heel widens (relative to species of Temnocyon) to form a platform for occlusion with the main cusp of p4.

The occurrence of the maxilla of Mammacyon (LACM 5386, fig. 27) in proximity to the dated Picture Gorge ignimbrite (∼28.7 Ma) in Haystack Valley, Oregon, demonstrates that a crushing dental pattern had already developed in the early Arikareean, predating many of the less dentally derived species of Temnocyon in the John Day region.

Delotrochanter underwent a similar dental transformation resulting in crushing cheek teeth like those seen in Mammacyon (figs. 38, 39, 43D, 44A, 45A). However, upper cheek teeth of Delotrochanter are unknown in the geologic record until D. oryktes occurs in the Harrison Formation at Agate National Monument, dated at ∼22.9 Ma. D. oryktes and D. major share upper carnassials that duplicate the general form of the Mammacyon P4, together with the crushing platform on the adjacent P3 (the P4 protocone in Delotrochanter is never quite as developed as in Mammacyon). The Delotrochanter M1 is known only in D. oryktes (figs. 38, 39) where the lingual half of the tooth is broadened as in M. obtusidens but not to the extent evident in Mammacyon ferocior. Nonetheless, the dental batteries of both Delotrochanter and Mammacyon represent the parallel evolution of a type of crushing cheek teeth unknown in other living and extinct Carnivora. These durophagous dental batteries of Mammacyon and Delotrochanter are approached but not equalled by a few dentally similar European haplocyonine beardogs (Peigné and Heizmann, 2003; Bonis, 1973; Hürzeler, 1940; Helbing, 1928).

Fig. 45.

Comparison of the lower carnassials of Delotrochanter petersoni and Temnocyon ferox, temnocyonines of similar size. The m1 metaconid has already been lost in D. petersoni (A), the earliest species of the genus, but is present in T. ferox (B). Note p2 alveolar diameters in (A).

Measurements of the teeth support recognition of the four temnocyonine genera as distinct dental groups. Dimensions of M1 (fig. 46) reflect the expansion of the protocone region (relative to M1 length) that distinguishes Mammacyon and Delotrochanter from the more plesiomorphic Temnocyon. This same relationship is also evident when graphing M1 length against width (fig. 46). Also, P4 remains a narrower shearing carnassial in Temnocyon species, whereas it becomes a large crushing tooth, broadened by an enlarged protocone, in species of Mammacyon and Delotrochanter (fig. 47).

Fig. 46.

Bivariate graphs of M1 dimensions (in mm) in North American temnocyonines. Lingual enlargement of the protocone region is interpreted as a durophagous adaptation distinguishing Mammacyon and Delotrochanter from Temnocyon.

Fig. 47.

Bivariate graph of P4 dimensions (in mm) of North American temnocyonines. Development of a crushing protocone increases P4 width in durophagous species of Mammacyon and Delotrochanter.

In the mandible, m1 dimensions suggest that this carnassial in species of Delotrochanter is slightly broader than in Mammacyon (fig. 48A); the m1s of the geologically youngest species of Temnocyon (T. fingeruti, T. percussor, T. macrogenys) seem narrow relative to length, reflecting retention of their primary role as shearing carnassials (table 2). Mammacyon obtusidens and M. ferocior mandibles are conspicuous for their elongated second molars. This is demonstrated by m2 length plotted against m1 length (fig. 48B) compared with the known examples of Delotrochanter and Temnocyon but especially with Rudiocyon.

Fig. 48.

Bivariate graphs of (A) m1 dimensions and (B) m1 length relative to m2 length (in mm) for North American temnocyonines, demonstrating a size increase during the Arikareean. Carnassial size (>27 mm) for the largest species of the four genera is similar in (A).

These genera are also distinguished by premolar dimensions, notably the widths of p2, p3, and p4 (fig. 49). Invariably, broader premolars differentiate species of Delotrochanter from those of Mammacyon. The large species of Temnocyon generally fall between species of Mammacyon and Delotrochanter, and the dimensions of p2 and p4 seem to indicate an evident regression of length on width for Temnocyon species, a trend less obvious for p3. Van Valkenburgh (1989) noted the relationship of broad, robust lower premolars (measured as p4 “roundness”) to a diet of meat/bone, a diet less dependent on meat alone.

Fig. 49.

Bivariate graphs of mandibular premolar dimensions (in mm) in North American temnocyonines. A, p2; B, p3; C, p4. Symbols for the genera are as in figure 48.

The teeth of temnocyonines are unlike those of any living carnivore. Whereas the small stem temnocyonine, T. altigenis, differs little in dentition and body size from the plesiomorphic amphicyonid Daphoenus vetus, the larger species evolve a unique combination of dental features. Large temnocyonines possess prominent canines and incisors, deep jaws, and broad snouts, features reported to be typical of large living canids that specialize in hunting prey larger than themselves (Van Valkenburgh et al., 2003). When combined with their tall, developed premolars, these traits argue for an ability to grasp and hold large prey. The deep jaws, strong mandibular symphysis, and canines are evidence of the strength and leverage of the bite. Rather blunt carnassials and broad molars add the capability to crush and grind hard food items during mastication.

Canines

Temnocyonines have prominent canine teeth, capable of stabbing and lacerating. Temnocyon altigenis, the most plesiomorphic species, with a skull much smaller (75% of basilar length) than that of the living wolf, has already evolved proportionately large upper canines. The upper canine of T. altigenis (UCMP 9999) is long and robust (height 83% and basal area 89% of that of the wolf), whereas the lower canine falls at 83% and 62%, respectively.

Van Valkenburgh and Ruff (1987) have reported that the forces involved in the killing bite and handling of large prey are reflected in canine shape. I measured temnocyonine canines and compared this data with values obtained by Gittleman and Van Valkenburgh (1997) for large living carnivores of similar body mass (Canis lupus, Crocuta crocuta, Panthera leo, P. tigris, Ursus arctos). Measurements of height and basal canine diameters were taken on the lower canines of temnocyonines, since these, rather than the upper canines, were more often complete.

Canine bending strength was determined using the formulas of Van Valkenburgh and Ruff (1987) that estimate bending resistance around both the anteroposterior (AP) and mediolateral (ML) axes of the lower canine with canine height incorporated in these calculations: nine temnocyonines, three from each genus, were represented by intact teeth. Bending strength around the mediolateral axis (S_y) is commonly greater relative to the anteroposterior axis (S_x) in large living carnivores and also in temnocyonines (table 9). Species of Temnocyon show the lowest values, less than those of wolves for the small T. altigenis (the large T. macrogenys canine was not preserved). Species of Delotrochanter, regardless of body size, correspond closely to bending strength values of Crocuta, in keeping with the hyenalike skull form of D. oryktes. Species of Mammacyon fall between the spotted hyena and the large felids, indicating considerable strength of the lower canines.

TABLE 9

Dimensions and Bending Resistance of the Upper and Lower Canines of Temnocyonines Relative to Living Carnivorans (in mm) (S_x  =  log₁₀ estimate of bending strength about anteroposterior axis of the canine; S_y  =  log₁₀ estimate of bending strength about mediolateral axis of the canine.

Calculated bending strength of the few intact temnocyonine upper canines yielded similar estimates, suggesting comparable stresses to these teeth during predation and feeding. Four species preserve the upper canine (table 9). The small Temnocyon altigenis is similar to the much larger Canis lupus, particularly in resisting force applied to the sides of the canine; T. ferox closely parallels Crocuta crocuta in resisting both AP and ML stresses, whereas the upper canines of both Delotrochanter oryktes and Mammacyon obtusidens have S_x and S_y values that fall between those of the spotted hyena and the large cats, indicating strong resistance to forces applied at the symphysis.

Furthermore, many of the remaining upper canines, although broken off at the tip, retain the base in the alveolus. A proxy for canine shape was calculated using only the AP and ML diameters measured at the enamel-dentine junction at the base of the upper canine. Van Valkenburgh and Ruff (1987: fig. 3) plotted AP against ML diameters for living canids, felids, and hyaenids. As body size increased, the more mediolaterally compressed canines of large canids diverged from the more robust, less compressed canines of large felids and hyaenids. AP–ML upper canine diameters for temnocyonines plot with large felids and hyaenids, not canids (fig. 50A), suggesting ample resistance of their canines to mediolateral bending stresses. These authors thought that the greater strength of felid canines was related to their killing style, and for hyaenids to their bone-crushing habit. Canines of felids-hyaenids may strike bone during the bite, and then “be subjected to large oblique or mediolateral bending stresses” as the prey is subdued (Van Valkenburgh and Ruff, 1987: 391). Thus, both the upper and lower canines of temnocyonines, especially in the larger species, seem well adapted to stresses occurring during prey capture and feeding.

Fig. 50.

(A) Log-log plot of upper canine anteroposterior diameter (AP) against mediolateral diameter (ML) as calculated by Van Valkenburgh and Ruff (1987: fig. 3) from samples of living felids, hyaenids, and canids. Temnocyonines plot with felids and hyaenids, not canids; (B) Log-log plot of mandibular depth against width in large living carnivores and temnocyonines (table 10): measured below m1–m2 interdental gap in canids and temnocyonines; below m1 in felids and Crocuta; below m2 in ursids.

Fig. 50

(continued). (C) Log-log plot of dentary length relative to working side lever arm length (articular condyle to m1/m2 interdental gap for canids and temnocyonines; to p4/m1 for hyaenids). Canid-hyaenid data and regression from Biknevicius and Ruff (1992a: fig. 9A).

Fig. 50

(continued). (D) Magnitude of dorsoventral force (Zx/L) applied at interdental loci along the mandible. Zx, the section modulus, measures parasagittal bending strength relative to length (L) of the working side lever arm. (E) Magnitude of labiolingual force (Zy/L) applied at interdental loci along the mandible. Zy, the section modulus, measures transverse bending strength relative to length (L) of the working side lever arm. Diagrams of caniform and feliform mandibles show location of linear measurements taken at interdental gaps. Data for canid, felid, and hyaenid from Therrien (2005).

Fig. 50

(continued). (F) Mandibular symphyses of Delotrochanter oryktes (top) and Temnocyon macrogenys (bottom). Interdigitating rugose bone and ligaments strengthen these mandibular symphyses against forces encountered during prey capture and feeding; note area for fibrocartilage pad (fc) in T. macrogenys.

Fig. 50

(continued). (G) Mandibular force profiles (Zx/Zy) of caniform and feliform carnivorans. Zx/Zy values are nearly all >1.0, indicating greater bending resistance to dorsoventral relative to labiolingual stresses applied at mandibular loci during prey capture and food processing. The temnocyonines show exceptional parasagittal bending resistance below the molars that reflects their deep, narrow mandibles at this locus. Zx/Zy canine ratios (1.2 to ∼1.5) of the three amphicyonids, Crocuta crocuta, and Panthera leo are comparable, indicating similar resistance to stresses at the mandibular symphysis.

Mandible

Mandibular dimensions of temnocyonines indicate considerable resistance of the jaws to bending stresses arising during mastication of hard materials and to forces applied at the front of the jaws during the kill. Jaws of some large temnocyonines such as Temnocyon macrogenys had apparently developed a parasagittal bending strength at the molars much greater than that of any living canid and similar to that of Panthera leo and Crocuta crocuta.

The resistance of the mandible to forces applied during feeding and prey capture has been examined in living carnivorans using several methods (Biknevicius and Ruff, 1992a, b; Van Valkenburgh and Koepfli, 1993; Biknevicius and Van Valkenburgh, 1996; Therrien, 2005). Biknevicius and Ruff (1992b) modeled the mandible as a hollow beam hinged at the jaw articulation, in which the distribution of cortical bone was calculated from biplanar radiographs. Mandibular cross-sectional geometry was first determined by examining actual cross-sections cut from the jaw, which were found to best approximate a hollow, asymmetrical cortical bone model. Using the radiographs, Biknevicius and Ruff (1992a) then calculated section moduli for canid, felid, and hyaenid mandibles to be used in determining mandibular bending strength. Alternatively, Therrien (2005) modeled the carnivoran mandible as a solid beam, recognizing that the solid beam method overestimates cortical bone cross-sectional area, which is better approximated using the method of Biknevicius and Ruff (1992b). Therrien (2005) showed that the solid beam approach, which lends itself to fossil material, could be used to represent relative differences in biomechanical properties along the mandible. Hence, differences among fossil and living species in terms of relative bending resistance along the mandible can be estimated and compared. The method assumes that the amount and distribution of cross-sectional cortical bone in such comparisons are similar, and that the individuals are of approximately the same body mass.

These studies employed measurements taken from intact dentaries of living species. Temnocyonine mandibles are commonly incomplete, and almost all lack the ascending ramus with its processes and articular condyle that are essential to measurement of dentary length and the calculation of mandibular lever arms used to determine the mechanical advantage of jaw muscles (Van Valkenburgh and Koepfli, 1993; Van Valkenburgh et al., 2003). However, it was possible to approximate bending strength along the length of partial temnocyonine mandibles that preserved undamaged bone beneath the cheek teeth, using the solid beam model and external cross-sectional diameters measured at various points along the toothrow.

Mandibles providing the essential conditions for calculation of bending resistance survived in six temnocyonine species. Measurements were taken from two complete and four partial dentaries (appendix 2): Relative to dentaries in living carnivorans, estimated temnocyonine dentary lengths indicate that the small T. altigenis is similar to Canis latrans; Temnocyon ferox, T. fingeruti, and Mammacyon obtusidens approximate Canis lupus; and the large T. macrogenys slightly exceeds Panthera leo, whereas D. oryktes is most like Crocuta in form but of greater length. Calculation of bending resistance depends on the length of the working side lever arm (the distance from the point of application of force along the toothrow to the jaw articulation). This length (L) was estimated for the partial dentaries and accurately determined for the two complete jaws. The m1m2 interdental gap was selected as the bite locus (point of applied force). Temnocyonines have a proportionately longer working side lever arm than those of large living canids (fig. 50C). This appears to be the result of lengthened crushing surfaces of molars used to process hard material, a probable adaptation for durophagy.

Many temnocyonines seem to have dorsoventrally deep, labiolingually narrow mandibles. When temnocyonine mandibular width is plotted relative to depth, it is apparent that these beardogs possess deep, narrow mandibles when compared with those of living carnivores of similar body size (fig. 50B). When bite force is applied to the principal cheek teeth during mastication, these temnocyonine mandibles are able to resist substantial dorsoventral bending stress. Jaw depth measured below m1–m2 for large species of Temnocyon (∼35–53 mm), Mammacyon (∼38–44 mm), and Delotrochanter (∼46 mm) corresponds to, and in some cases exceeds, values (29.8–43.3 mm) observed for Canis lupus, Panthera leo, Crocuta crocuta, and Ursus arctos at functionally analogous points on the mandible (table 10).

TABLE 10

Mandibular Measurements of Temnocyonines and Living Carnivorans (in mm)

To further explore the relationship between mandibular structure and jaw function, the two intact temnocyonine mandibles (Temnocyon macrogenys, F:AM 54139; Delotrochanter oryktes, ACM 4804) were compared with a representative canid (Canis lupus), hyaenid (Crocuta crocuta), and felid (Panthera leo), carnivores that approximate the size range of these large temnocyonines (appendix 3). Included here for its heuristic value is the largest known daphoenine amphicyonid, Daphoenodon (Borocyon) robustum, from the early Miocene Runningwater Formation of Nebraska, a carnivore comparable in body size to the large temnocyonines (Hunt, 2009).

The two temnocyonine mandibles were oriented and measured according to Therrien's (2005: 252f.) method of beam analysis. Mandibular depth and width were measured in centimeters with dial calipers at interdental foci between p3p4, p4m1, m1m2, m2m3, and directly behind m3 (post-m3). Depth of the mandibular symphysis was measured as a vertical posterior to the canine; mandibular width was measured from a point on the labial face of the mandible directly behind the canine to the posterior termination of the symphysis (Therrien, 2005: fig. 2, canine). Working side lever arms were calculated as the distance from the articular condyle to the various interdental points along the mandibular corpus. Dentary length was taken as the distance from articular condyle to the mandibular terminus directly anterior to the canine (Biknevicius and Ruff, 1992a: fig. 1). These measurements on temnocyonine jaws were then compared with sample means for the various species of canids, hyaenids, and felids provided by Therrien (2005: figs. 3–7, appendices 1, 2).

Despite its limitations in overestimating the amount of cortical bone in cross-sections, the solid beam model allowed comparison of relative patterns of bending strength along the mandibles of both the fossil and extant species. Second moments of area (Iy, Ix), which estimate the cross-sectional distribution of cortical bone about the dorsoventral and labiolingual axes, were calculated from the formulas of Therrien (2005: 250). From these were derived the section moduli, Zx and Zy, which are respective measures of parasagittal and transverse bending strength. These moduli are a reflection of bending moments applied to the mandible during prey capture and feeding.

The bending resistance of a point along the mandible during the bite is a function of the distance (L) from the articular condyle to a particular dental locus, which is then expressed here as a ratio to derive parasagittal (Zx/L, fig. 50D) or transverse (Zy/L, fig. 50E) bending strength values. These graphs compare patterns of bending strength along the various caniform and feliform mandibular corpora, providing insight into relative bending moments typical of these species.

Living carnivores studied by Therrien (2005: figs. 3–7) typically show parasagittal bending strengths at the various post-canine loci along the mandibular corpus to be greater than transverse values. The temnocyonines and the daphoenine Borocyon robustum show clear parallels here to the large living canids in bending strength values distributed along the mandible, which rise rapidly distad along the mandible from p3p4 to directly behind m3. The daphoenine and temnocyonines share with these canids (Canis lupus, Lycaon pictus, Canis rufus) a crushing molar platform (m1 talonid, m2–3) and similar ZxL and ZyL force profiles in which maximum bending strength occurs in proximity to the molars at the back of the jaw (figs. 50D, E). These amphicyonids, however, show much greater bending strength at the molars and a longer crushing platform relative to canids studied by Therrien (2005: fig. 6).

The mandibular force profile of Borocyon robustum far exceeds the values for any living canid, and also the extinct Canis dirus (Therrien, 2005: fig. 6). Borocyon robustum is a much larger animal with more massive jaw musculature than a typical wolf, thus the greater bending strength values are likely due in part to the size of this daphoenine. However, resistance to bending at the canines is much greater than in the wolf, evident in the beardog's more strongly buttressed mandibular symphysis (Hunt, 2009: fig. 38), resistant at the canines to both parasagittal and transverse forces applied during prey capture and feeding. Known from its skeleton to be a cursorial predator, B. robustum appears to have had an ecologic role as a large wolflike hypercarnivore able to easily process bone and other tough fibrous foods (Hunt, 2009).

Higher Zx/L and Zy/L values for Temnocyon macrogenys and Delotrochanter oryktes reflect their greater body size and more massive jaw musculature relative to the large living canids. The largely parallel patterns of bending resistance along the toothrow, however, indicate a similar distribution of applied force along the mandibular corpus during feeding, except for the canines and mandibular symphysis. Zy/L values show temnocyonine mandibles to be stronger at the canines when subjected to transverse stresses, than at the molars, which Therrien (2005) interpreted as an indication of substantial torsion at the symphysis.

Temnocyonines and Borocyon are exceptional in their Zx/L and Zy/L values for parasagittal and transverse bending strength at the canines/mandibular symphysis. They compare with the spotted hyena (figs. 50D, E), yet cannot match the considerable parasagittal and transverse strength achieved by the lion. The wolf and other large living canids have lower values for canine/symphyseal bending strength, possibly explained by the quick, sharp nipping bites used by social pack-hunters to bring down large prey without the stresses encountered by lion and spotted hyena in capturing and feeding on large ungulates. The reinforced symphysis of these temnocyonines (fig. 50F) and Borocyon, composed of two rugose interlocking mandibular plates joined by strong ligaments, better resists torsional forces than the symphysis of the wolf. The broader, more rectangular form of the T. macrogenys and B. robustum symphyses is evident in their length/width ratios (1.51, 1.48, respectively; Mammacyon ferocior, 1.41; M. obtusidens, 1.45) relative to the narrower, elliptical symphyses of Delotrochanter oryktes (2.15) and Crocuta crocuta (∼2.7–2.9).

Zx/Zy values are a measure of relative bending strength about the sagittal and transverse planes of the mandibular corpus (Biknevicius and Ruff, 1992a: 484), considered by Therrien (2005) as mandibular force profiles. These Zx/Zy ratios (fig. 50G, appendix 3) estimate the relative differences between parasagittal and transverse loadings along a single mandible with the advantage that these patterns can then be compared among carnivoran species. The temnocyonine pattern of Zx/Zy ratios along the mandibular corpus indicates the ability to tolerate heavy dorsoventral loading at the molar platform. Thus, temnocyonines combine great strength at the canines and symphysis to hold and dispatch prey with powerful forces exerted at the molar battery to grind and crush hard food material during feeding. Living large canids have been observed chewing hard food items, including bone, at this molar locus (Mech, 1970; Haynes, 1982; Van Valkenburgh, 1996), and temnocyonines also possessed this capability.

Addition of bone (buttressing) is to be expected in the mandibular corpus reinforcing dorsoventral and labiolingual as well as torsional bending strength when strong forces are applied to the teeth during feeding and prey capture. Particularly strong loadings are anticipated where teeth are used to comminute hard, tough foods. In their analysis of living canids and hyaenids, Biknevicius and Ruff (1992a: 497) observed that the location along the mandible of teeth used to chew hard foods (primarily in bone-processing) seemed to determine durophagous adaptations: increased bending strength materialized directly caudal to bone-processing teeth. In temnocyonines, precarnassial teeth seem less involved in masticating bone or other hard food items, evidenced by less apical wear on anterior premolars (P1–2, p1–3), while cusps of P3–M2 and p4–m3 show blunt wear, a situation similar to that known in wolves. The posterior cheek teeth of temnocyonines, notably in the large dentally derived species, were uniquely specialized for crushing and grinding hard material. Bone processing by temnocyonines, relative to living canids, may have been favored by the longer working side lever arm at the molars (fig. 50C) that apparently accommodated a somewhat wider gape.

The two temnocyonines (T. macrogenys, D. oryktes) in which intact mandibles survived must have differed considerably in external appearance (figs. 20, 34). Although much larger, the elongate mandible of T. macrogenys is shaped like that of Canis lupus and indicates a rather long “doglike” skull. Temnocyon macrogenys must have been an enormous wolflike predator with a long-legged striding gait and other predictable cursorial features. The shape of the D. oryktes mandible is hyenalike; the skull short with broad snout and deep zygomata much as in Crocuta, although lacking the wider palate of the hyena. The form of its blunt-cusped cheek teeth are unlike those of living canids and hyaenids, yet doubtless served a crushing function, in agreement with pronounced mandibular buttressing along the ventral margin of the jaw below the cheek teeth. The large species of Delotrochanter appear to be temnocyonines that, to some extent, craniodentally parallelled the bone-crushing hyaenids.

Scapula

Delotrochanter oryktes (ACM 4804) preserves the only relatively complete scapula known for temnocyonines (fig. 57). The scapula of ACM 4804 is most similar in form to those of other contemporary amphicyonids (e.g., Daphoenodon superbus, Borocyon niobrarensis; Hunt, 2009: fig. 20A). Among the living Carnivora its shape approaches the scapular form of felids. Schlain (1980) established the presence of a postscapular fossa for the subscapularis muscle in this species, a feature typical of amphicyonids. The supraspinous fossa occupies proportionately more surface area than the infraspinous fossa, and acromnion and metacromnion processes are present. Schlain (1980) noted the narrow glenoid cavity implying restriction of the movement of humerus on scapula to a more fore-aft path. The spinodeltoid, which flexes the shoulder, runs from the scapular spine to the deltoid crest of the humerus; this crest in temnocyonines is a low arcuate ridge that traverses the upper two-thirds of the diaphysis.

In Temnocyon altigenis (UCMP 9999) only the glenoid of the scapula survives. Temnocyon ferox, the only other species of this genus that is represented by most of the postcranial skeleton, also retains a glenoid remnant as does Mammacyon ferocior (F:AM 27562). The holotype of Mammacyon (M. obtusidens, ACM 34-41) preserves the glenoid and adjoining portion of the scapular blade including the scapular spine. The spine terminates anteriorly in a very small, blunt acromnion process (there is no metacromnion process), much less developed than even that of the wolf. This suggests a reduced acromiodeltoid muscle, a forelimb abductor less necessary in a cursor. In this respect the scapula of M. obtusidens differs from the more plesiomorphic scapula of D. oryktes with its prominent acromnion and metacromnion processes.

Humerus

The humerus is complete in Temnocyon altigenis (UCMP 9999) and already displays a more specialized distal anatomy than seen in D. vetus (fig. 51). While these two species have humeri of equal length, the distal humerus of UCMP 9999 is transversely narrower, shows reduction of the medial epicondyle, and possesses a deeper olecranon fossa for reception of the ulna. These traits accompany a close congruent registration of the humeroulnar articulation, effective in restricting the ulna to flexion-extension on the humerus when the lower limb is extended, and thus favoring a parasagittal trajectory for the lower forelimb. A rotational torque at the elbow is resisted by contact of a facet on the lateral face of the anconeal process of the ulna with the lateral margin of the olecranon fossa of the humerus, assisted by the tight contact of the ulna's coronoid process with the humeral trochlea. This type of humero-ulnar joint registration, which occurs in temnocyonines, was previously noted by Jenkins (1973) who observed that reduction of the medial epicondyle and its associated flexor musculature decreases torque at the elbow. These features are a prelude to a more erect forelimb stance, evident in both the Logan Butte T. altigenis and T. ferox.

Fig. 51.

Comparison of the distal humerus of a stem daphoenine (A, C: Daphoenus vetus) and a plesiomorphic temnocyonine (B, D: Temnocyon altigenis). The daphoenine amphicyonid humerus is distally broad for developed flexors arising from the medial condyle and has an asymmetric olecranon fossa indicating an everted elbow. Temnocyonines have evolved the narrow distal humerus of cursorial carnivores with reduced medial condyle and symmetric olecranon fossa corresponding to an elbow with a more fore-aft alignment.

Both Mammacyon obtusidens and M. ferocior preserve the humerus, damaged in the former species, nearly complete in the latter. In both species the form of the bone is identical except for greater size in M. ferocior. As in Temnocyon, the distal humerus is specialized for fore-aft excursion of ulna on humerus, as evidenced by a narrow transverse distal width, reduction of the medial epicondyle, and a deep symmetrical olecranon fossa (fig. 52). All temnocyonine humeri exhibit a more rounded capitulum and deeply grooved trochlea than seen in those of Daphoenus vetus and D. superbus, a condition seemingly related to the narrow distal articular width of this bone. This differs from the more cylinderlike, wider articular surface of Daphoenodon superbus and D. vetus, which lacks the rounded capitulum.

Fig. 52.

All temnocyonines have developed a narrow distal humerus (with reduced medial condyle, symmetric olecranon fossa) in which the forelimb shows little or no eversion at the elbow as also seen in the wolf. These forelimbs adopt an erect digitigrade stance. Upper row, anterior view; lower row, posterior view. A, Delotrochanter oryktes; B, Mammacyon ferocior; C, Temnocyon ferox; D, Canis lupus. A, B: left humerus; C, D: right humerus.

The humerus of Delotrochanter oryktes (ACM 4804), ∼22 cm in length, is much the same size as the humerus of M. obtusidens, and is nearly as large as the humerus of M. ferocior (∼24.5 cm). The humerus is straight-shafted, a trait common to all larger temnocyonines (figs. 53–Fig. 54.55); the bone has a slimmer profile in lateral view due to the reduction of the plesiomorphic amphicyonid deltopectoral crest and its restriction to the proximal humerus. The proximal placement and reduction of this crest also occurs in Canis lupus. The distal anatomy of the humerus is identical to that seen in other temnocyonines (fig. 52), indicating a similar upright stance and fore-aft excursion of ulna on humerus.

Fig. 53.

Articulated right forelimb of the holotype of Temnocyon ferox (YPM-PU 10787), John Day Formation, Oregon. The forefoot shows elongation of the paraxonic metacarpals 3, 4, and their digits, with reduction of metacarpals 1, 2, and 5 (metacarpal 5 is hidden from view). A partial femur (fe) and tibia (ti) of an oreodont were entombed with the beardog. Abbreviations: sc, scapula; hu, humerus, ul, ulna; ra, radius; cp, carpus; nos. 1–4, metacarpals; pr, proximal phalanx; ip, intermediate phalanx.

Fig. 54.

Associated humerus and ulna of (A) Mammacyon ferocior and (B) Canis lupus showing elongation of the ulna (and radius) and pronounced reduction of the distal ulna in these cursors.

Fig. 55.

Humerus and radius of (A) Delotrochanter oryktes and (B) Canis lupus. Anatomy of the D. oryktes forelimb and elongation of the radius and ulna are similar to these features in the wolf. Left humeri, anterior view; left radii, posterior view.

Radioulna

The radius is known in species of Temnocyon, Mammacyon, and Delotrochanter. In the Logan Butte Temnocyon altigenis both radii are present but lack their distal ends. In Temnocyon ferox the forelimb preserves a fully articulated radius, ulna, and humerus (fig. 53) but these bones are so fragmented that details of the articular surfaces are obscure. The forelimb does not show significant elongation in T. altigenis, however with the younger T. ferox some elongation of the forelimb has taken place (table 11). The articulated forelimb of T. ferox makes possible a reliable estimate of limb proportions in this carnivore.

Forelimb proportions were also calculated from an associated humerus, radius, and ulna for T. altigenis, T. ferox, M. obtusidens, and D. oryktes and compared with other carnivorans (table 12). For M. ferocior where no radius was found, an associated humerus and ulna determine the length of the radius. The forelimb proportions for Mammacyon and Delotrochanter show that elongation of the radioulna contributes to lengthening of the limb.

Although no radius was collected with M. ferocior, its humerus and ulna (fig. 54) are identical to those of M. obtusidens and so a similar radius can be inferred; these two species differ only in the larger size of M. ferocior. Here the narrow distal humerus, best preserved in M. ferocior, confines the ulna to fore-aft flexion and extension. When the radioulna is extended on the humerus, the movement of the lower limb is plainly fore-aft in direct parasagittal alignment with the humerus. The rounded capitulum and grooved trochlea of the distal humerus are congruent and register exactly with the circular, concave radial head and the semilunar notch and coronoid process of the ulna—these joint surfaces correspond closely in form to those of the wolf (fig. 54). However the radial head is not as limited in its rotation in the radial notch of the ulna as it is in the wolf, and the humeral trochlea not quite as narrow. Similarly, the bicipital tuberosity of the radius in Mammacyon remains better developed than in the wolf, suggesting that some ability to pronate/supinate the lower limb and forepaw was still retained.

Delotrochanter oryktes (ACM 4804) is represented by a humerus, radius, and ulna (fig. 55) that show anatomical modifications like those of Mammacyon. In particular, the proximal ulna is a mediolaterally compressed, thin bone as in the wolf, and the semilunar notch is proximodistally divided by a low ridge, which fits into the groove of the trochlea of the humerus, aiding in humeroulnar registration in wolf and D. oryktes. These features are not found in the less cursorially patterned elbow joint of Daphoenodon superbus.

Moreover, the head of the radius has a precise articulation with the radial notch of the ulna in Delotrochanter and Mammacyon that differs from this articulation in the wolf: in the wolf when the radial head is placed in the radial notch, the radial head is positioned somewhat in front of the ulna. However in Mammacyon and Delotrochanter, when the radial head is placed in the laterally facing notch, the radial head is situated lateral to the ulna, which corresponds to the placement of the capitulum of the humerus directly lateral to the trochlea. This more side-by-side articular relationship of the proximal radius and ulna seems exclusive to temnocyonines. But most important for fore-aft excursion of the limb is that the shaft of the radius in these temnocyonines aligns more in parallel to the shaft of the ulna, as in the wolf. In D. oryktes and M. ferocior, the radius lies in close apposition to the ulna, much more so than in Daphoenodon superbus where the shaft of the radius is aligned at a greater angle to the shaft of the ulna.

Furthermore, in temnocyonines the extent of the articular facet on the radial head suggests that rotation of the radius on the ulna was limited. In the wolf a limit to the amount of rotation of the radius in the radial notch of the ulna is evident from the shape of a well-defined articular facet on the radial head that incorporates a bony stop on its lateral edge. The articular facet ends sharply where it articulates with the lateral extension of the coronoid process of the ulna, which has been reduced to a thin flange in both temnocyonines and the wolf. The even greater reduction of this flange in temnocyonines seems to be due to the more lateral placement of the radial head relative to the ulna in these beardogs.

In temnocyonines where the distal ulna is preserved, the distal articulation for the carpal cuneiform is present, but the articular process contacting the radius is much reduced. This parallels the condition in cursorial wolves and coyotes, in which rotation of radius on ulna is restricted.

Forefoot (carpus, metacarpus, phalanges)

Bones of the forefoot rarely survive among temnocyonines. In T. altigenis from Logan Butte no forefoot elements are known. The forelimb of Temnocyon ferox (YPM-PU 10787), however, includes a partially articulated paraxonic forefoot with carpals, metacarpals, and phalanges (fig. 53) that demonstrates the relative proportions of these elements. Metacarpals 1 to 5 of T. ferox are all slightly elongated relative to those of Daphoenodon superbus, and the paraxonic metacarpals 3–4 are more robust as well as longer. In T. ferox the trapezium, first metacarpal, and its phalanx indicate a much reduced, thin first digit in the forefoot. However, this degree of reduction of the first digit does not equal the more extreme state seen in the wolf. Although the metacarpals of T. ferox are not tightly appressed as in the wolf, they are in closer contact than in Daphoenodon.

In the forefoot of T. ferox the proximal phalanges of digits 3–4 are long (both ∼31 mm), those of digits 2 and 5 shorter (∼27–29 mm), and that of metacarpal 1 reduced (length, 21.8 mm). Intermediate phalanges of digits 3–4 are short (length, ∼19 mm). Only in digit 4 has the ungual been preserved; it is short (length, ∼13 mm) and rather blunt. The phalanges of paraxonic digits 3–4 are more robust than those of D. superbus. If T. ferox is representative of the subfamily, temnocyonines lack the extreme elongation of the proximal and intermediate phalanges seen in the wolf.

The scapholunar, unciform and trapezium of Temnocyon ferox are proportionately smaller than those of D. superbus and are most like those of the wolf (a small carpal cuneiform is obscured by sediment). Thus the carpus is transversely narrow in order to accommodate a distally narrow radius and ulna, as in the wolf.

However, the carpus lacks some anatomical features evident in wolves. The unciform serves as a case in point. The temnocyonine unciform from the Live Oak site in Florida (table 13) confirms that this bone was low and broad, much different in form from the tall, narrow unciforms seen in living ursids and felids. The Live Oak unciform shows a marked similarity to that of Canis lupus (the wolf is 23% smaller) except that a stop facet on the wolf unciform for the scapholunar is absent in the Live Oak beardog (Hunt, 2009: fig. 25, Canis lupus, #3). In both temnocyonine and wolf the unciform articulates with metacarpals 4–5, but in the wolf the carpal cuneiform, in close contact with the unciform, bends down around the unciform and articulates by a defined concave facet with the head of metacarpal 5. There is no evidence of such a specialized carpal cuneiform in temnocyonines, and it is not present in the articulated carpus of Temnocyon ferox or in Delotrochanter oryktes (ACM 4804). The canine carpal cuneiform strengthens the carpal-metacarpal joint where it extends outward in the wrist between the fifth metacarpal and unciform. In the wolf, the fifth metacarpal is a large bone but it is much reduced in Temnocyon ferox and Delotrochanter oryktes.

TABLE 13

Measurements (in mm) of the Unciform in Temnocyon, Delotrochanter, Daphoenodon, and the Live Oak Carnivore

The few metacarpals of Delotrochanter and Mammacyon that are known (table 8) conform anatomically to those of T. ferox and indicate elongation of the paraxonic forefoot (fig. 56). The considerable length of paraxonic metacarpals 3–4, and the shorter metacapal 5 evident in D. oryktes (UNSM 47800), together with metacarpals 2 and 5 in ACM 4804, show that the flanking metacarpals 2 and 5 were of similar length and were more reduced than those of the wolf, in which a robust metacarpal 5 plays a more prominent role. Only ACM 4804 preserves metacarpal 1, a small, thin, elongate bone (length, 37–38 mm), not as reduced as in the wolf.

Fig. 56.

Forefoot of Delotrochanter oryktes (UNSM 48700), from carnivore den, Beardog Hill, Agate Fossil Beds National Monument, Sioux Co., Nebraska. Metacarpals 3, 4, 5 (A, anterior view; B, posterior view); C, proximal phalanx of paraxonic digit of forefoot; D, magnum.

Innominate

The pelvis with sacrum in direct articulation survives in Temnocyon ferox and a damaged innominate is present in Delotrochanter oryktes, however a complete pelvis is not preserved in other temnocyonines. Only a partial innominate of the Logan Butte T. altigenis and fragments of the innominate of M. ferocior are known; these do not differ from the innominates of T. ferox and Daphoenodon superbus except in size. The innominate of Delotrochanter oryktes (ACM 4804) lacks only the terminus of the ischium and shares the same form and proportions as in T. ferox (fig. 57).

Fig. 57.

Scapula (A) and innominate (B) of Delotrochanter oryktes (ACM 4804, ischium restored in part). The temnocyonine scapula did not often survive but in this species it is most similar to the plesiomorphic scapulae of Daphoenus and Daphoenodon superbus. The innominate, also known in Daphoenodon superbus and in Temnocyon ferox, has an extended ischium that presumably provided a mechanical advantage for the action of hamstring muscles.

During excavation of the carnivore dens at Agate National Monument in 1985, we recovered a nearly complete innominate of D. superbus with ischium and pubis intact. The most striking difference in the pelvis of Daphoenodon and the temnocyonines relative to that of a wolf is the much longer ischial region relative to the ilium—the postacetabular distance in the wolf is ∼68% of the preacetabular distance, whereas in T. ferox this is 84%, in D. oryktes 83%–87%, and in D. superbus 98%. In the wolf the posterior ischium is rotated outward, presumably to increase the mechanical advantage of the hamstring muscles. This ischial eversion is absent in temnocyonines, but the simple lengthening of the ischium may have provided a similar mechanical advantage.

The temnocyonine sacrum also differs from that of the wolf in its narrow, elongate shape compared to the short, broad sacrum of Canis lupus. These sacra all include three fused vertebrae but the transverse processes are more elongated and developed in the wolf, giving the sacrum its broader appearance. In the Logan Butte Temnocyon altigenis, T. ferox, and Mammacyon ferocior, and in Daphoenodon superbus, the width of the sacrum (measured at the 2nd sacral vertebra) is 44–46% of the length but in the wolf this value is 88%. Nonetheless, the articular surface of the ilium with the sacrum is of similar proportions in all these carnivores.

Femur-Tibia

An associated femur and tibia occur in Temnocyon ferox, Mammacyon ferocior, and Delotrochanter oryktes (fig. 58). Thus, hind limb proportions can be estimated for representatives of the three principal temnocyonine genera (tables 11–12). In these three species the femorotibial proportion varies from 91% to 96%, approaching the ratio found in wolves (∼96%–101%). Lower indices of 67%–76% in ursids, and 83%–89% in felids, reflect the shorter tibiae in these living carnivores. The indices for the daphoenine amphicyonids (89%–96%), while similar to those of temnocyonines (91%–96%), are lower for the large, terrestrial Daphoenus vetus (∼89%–90%) and higher for the small scansorial D. hartshornianus and the Chadronian forms (∼93%–96%). Despite the proportionately longer hind limbs in these latter two groups, their forelimbs remain quite short (80%–83%, ratio R/H, table 11) and show that D. hartshornianus and the Chadronian daphoenines differ from the temnocyonines who evolved longer forelimbs (∼90%–95%, ratio R/H) relative to the hind limb. The earliest temnocyonines probably retained short, plesiomorphic forelimb proportions, since the forelimb of the Logan Butte Temnocyon altigenis lacks conspicuous elongation.

Fig. 58.

Femur and tibiae of Delotrochanter oryktes, Stenomylus Quarry, Harrison Formation, Agate Fossil Beds National Monument, Nebraska. The long slender femur (A) and tibiae (B, right; C, left) are similar in proportion and form to those of the wolf whose tibia is ∼8% longer. An arrow (1) marks the sulcus muscularis in B. Muscle scars (2, stippled) on the posterior tibia (C) indicate proximally situated posterior tibial and long digital flexor muscles. The femur (D) of Temnocyon altigenis, the smallest and most primitive temnocyonine, retains a lesser trochanter (3), gluteal flange, and third trochanter (4). A, B in anterior view; C, D, posterior view.

The temnocyonine femorotibial ratios suggest that other morphological correlates of an elongate cursorial hind limb might be present, and such is the case. For both Delotrochanter oryktes and Mammacon ferocior a prominent groove, the sulcus muscularis for the tendon of the long digital extensor, indents the lateral margin of the tibial platform immediately anterior to the lateral condyle (fig. 58B). From its origin on the femur, the tendon travels distad to extend the digits and participate in tarsal dorsiflexion. Although common in artiodactyls and perissodactyls, the sulcus is well developed among living Carnivora only in canine canids and the cheetah, and is not found in other arctoid carnivorans (ursids, procyonids, mustelids) or in Daphoenus and Daphoenodon. The sulcus is not quite as deep in Mammacyon ferocior and Delotrochanter oryktes as in the wolf. The tibia of T. ferox does not have a sulcus muscularis; thus it may have been acquired independently in Mammacyon and Delotrochanter (the proximal tibia was not recovered in T. altigenis). Except for size, the tibiae of Temnocyon ferox, Mammacyon ferocior, and Delotrochanter oryktes correspond closely in form and anatomy; the diaphyses are somewhat more robust when compared to the more slender tibia of the wolf.

Mammacyon ferocior and Delotrochanter orkytes also compare closely with the wolf in the pattern of muscle scars on the posterior surface of the tibia (fig 58C). The pattern indicates that the posterior tibial (PT) and long digital flexor (FDL) muscles have been reduced, and are proximally situated, a condition seen in the wolf (Evans, 1993) where the hind foot is specialized for fore-aft movement on level ground. The large temnocyonines, Canis lupus, and the cheetah Acinonyx exhibit this PT-FDL reduction more so than the large, living felids and ursids. For M. ferocior, tibial length (265 mm) relative to the width of the PT-FDL scar (∼5 mm) corresponds to the values found for the wolf and cheetah (Hunt, 2009: fig. 30). However, temnocyonines differ from the wolf in the shallower registration of the astragular trochlea in the distal tibia, which is like that of the cheetah; deeper penetration of the astragalus in the wolf creates a more tightly registered joint.

Delotrochanter oryktes (ACM 4804) preserves both femora, described by Schlain (1980), who noted their resemblance to those of Canis. The femur is long, slender, slightly curved in the sagittal plane as in the wolf, but with the femoral head and neck almost at a right angle to the diaphysis (fig. 58A). This orientation of femoral head, neck, and diaphysis is typical of most amphicyonids and contributes to the adduction of the limb. The distal femur is transversely narrow and the condyles extend posteriorly as in the large living canids. In all these features the femur of D. oryktes is nearly identical to the femur of the cursorial daphoenine amphicyonid Borocyon robustum (Hunt, 2009).

The width of the femur across the distal condyles is narrow relative to femoral length in cursorial carnivores, and is similar among temnocyonines and the wolf, between 16% and 18% (table 14). However, the width of the groove for the patella relative to condylar width differs, for in temnocyonines the groove is from 41% to 46.5% of condylar width and in the wolf only 28%, indicating a narrower patellar groove in Canis lupus.

TABLE 14

Dimensions of Femoral Condyles and Patellar Groove Relative to Femoral Length in Temnocyonines and Canis lupus (in mm)

The femur of Temnocyon altigenis from Logan Butte (fig. 58D) can be articulated with the corresponding innominate to estimate the alignment of the hind limb in that species. Here the femur appears to be abducted ∼20°–25° from the vertical, whereas abduction of 10° has been reported as the characteristic posture of living canids (Jenkins and Camzine, 1977). In T. altigenis the pit (fovea capitis) for the femoral ligament on the femoral head is situated as in the larger canids (Canis lupus, Lycaon pictus) measured by Jenkins and Camazine (1977: fig. 9) and does not adopt the more extreme position found in living ursids and procyonids. In Temnocyon ferox, Mammacyon ferocior, and Delotrochanter oryktes the fovea capitis is similar in its placement to the fovea of T. altigenis.

The extent and orientation of the articular surface of the femoral head as it registers in the acetabulum provides a measure of the range of motion of the femur. Jenkins and Camazine (1977) measured the angle formed by a line drawn along the margin of the anterior articular surface of the femoral head and the femoral diaphysis; this angle in D. oryktes and M. ferocior (∼38°–40°) was similar to that of Canis lupus (∼40°), which suggests that the temnocyonines hold the femur in a more abducted position than in small cats and foxes, but with much less abduction than in procyonids and the large living ursids.

Jenkins and Camazine (1977) observed only modest extension of the articular surface of the femoral head in the proximal and posterior directions in canids. In T. altigenis and T. ferox the articular surface of the femoral head approximates that seen in the wolf, and does not extend downward on the posterior surface as in carnivores such as procyonids and ursids with greater femoral excursion employed in actions such as climbing, where a greater range of abduction-adduction occurs. In D. oryktes and apparently in M. ferocior the extent of the articular surface on the posterior femoral head does somewhat exceed that of the wolf so that the excursion of the femoral head in the acetabulum may have been more similar to the conformation in the large living cats (Schlain, 1980).

The femur of T. altigenis is unusual in having a gluteal ridge distal to the greater trochanter, forming a pronounced flange that continues along the outer diaphysis (Merriam, 1906: 27). The flange thickens to form a third trochanter ∼35 mm distal to the greater trochanter for insertion of the superficial gluteal (fig. 58D, #4). This gluteal flange also occurs in Daphoenus and in archaic carnivores such as the mesonychid Pachyaena ossifraga (O'Leary and Rose, 1995) and the miacid Miacis petilus (Heinrich and Rose, 1995). The presence of the gluteal flange, large third trochanter, and prominent lesser trochanter suggests retention of some mediolateral mobility in the hip joint of T. altigenis indicative of more abduction-adduction than found in small living canids. This is also supported by the prominent knoblike lesser trochanter in T. altigenis (fig. 58D, #3) that projects posteromedially as in terrestrial small carnivorans in which some abduction and lateral femoral rotation occur.

Tarsus

The astragalus and calcaneum in temnocyonines, together with rare distal tarsals in Temnocyon altigenis, T. ferox, Mammacyon ferocior, and Delotrochanter oryktes, demonstrate a modified daphoenine tarsal pattern evolving toward a hind foot similar to the canine type, adapted for fore-aft flexion and extension and an erect digitigrade stance.

The form of astragalus and calcaneum approach that of living large canids, most closely in T. ferox, M. ferocior, and D. oryktes. The temnocyonine calcaneum is both proximally and distally narrow relative to Daphoenus, Daphoenodon superbus, and amphicyonines (fig. 59, table 15). Calcaneal width at the sustentaculum is ∼35%–42% of calcaneal length in temnocyonines, but in short-limbed daphoenines this measures from 46.6% to 52.1% (fig. 60). This reduction in width is already evident in Temnocyon altigenis. Also in temnocyonines, the sustentaculum appears proximally elevated because the distal calcaneum is lengthened below the sustentaculum, and its posterior surface becomes deeply grooved for the deep digital flexor (flexor digitorum profundus), active during the stance phase in dogs (Evans, 1993). A calcaneum of this type occurs in the wolf and is also common to living felids whose proximal tarsals are similarly lengthened distad contributing to hind foot elongation.

Fig. 59.

Calcaneum of Delotrochanter oryktes (UNSM 47800), from temnocyonine carnivore den, Beardog Hill, Agate Fossil Beds National Monument, Sioux Co., Nebraska.

Fig. 60.

Bivariate graph of calcaneal dimensions of temnocyonines relative to calcanea of daphoenine amphicyonids. A narrow, distally extended calcaneum similar to that of living digitigrade felids is characteristic of temnocyonines.

TABLE 15

Calcaneal Dimensions of Temnocyonine and Daphoenine Amphicyonids (in mm)

The temnocyonine astragalus (table 16) shows three derived features that correspond to the form of the calcaneum. The distal process for articulation with the navicular is no longer tranversely aligned as in Daphoenus vetus and Daphoenodon superbus but has shifted to a more angled alignment (a long axis directed anterolaterad to posteromesad) beneath the trochlea. This type of astragalus is known in Temnocyon altigenis, T. ferox, Mammacyon obtusidens, M. ferocior, and Delotrochanter oryktes. These temnocyonines also share a derived placement of the sustentacular facet situated proximally on the neck whereas in D. superbus and amphicyonines this facet extends far distad. Temnocyonines also possess a relatively narrow astragalar trochlea when compared to Daphoenodon superbus (table 16). By articulating the calcaneum and astragalus with the tibia in T. ferox, M. ferocior, and D. oryktes, the alignment of joint surfaces suggests a more erect hind foot stance as in large living cats and Canis lupus.

TABLE 16

Dimensions of the Astragalus in Temnocyonine and Daphoenine Amphicyonids (in mm)

The complete articulated tarsus of Temnocyon ferox was illustrated by Eyerman (1896). The plesiomorphic tarsal pattern of daphoenines such as Daphoenus vetus and Daphoenodon superbus differs from the derived pattern evident in Temnocyon ferox (fig. 61), in which a mediolaterally narrow tarsus had evolved, paralleling the cursorial wolf. This derived state was also present in Mammacyon and Delotrochanter.

Fig. 61.

Hind foot of (A) Temnocyon ferox (YPM-PU 10787) and (B) Daphoenodon superbus (CM 1589). The narrow columnar tarsus and appressed metatarsals in the cursorial temnocyonine differ from the more splayed and mobile foot of the daphoenine. T. ferox after Eyerman (1896); D. superbus after Peterson (1910).

Distal tarsals of temnocyonines have survived in M. ferocior (F:AM 27562: associated navicular, cuboid, and ectocuneiform) and in D. oryktes (ACM 4804: navicular and ectocuneiform). In primitive daphoenines the cuboid shows the expected facets for the ectocuneiform and navicular but also retains a plesiomorphic contact with the astragalus, whereas in the wolf and temnocyonines this astragalar contact is lost (fig. 62). Furthermore, in temnocyonines where the articulations among calcaneum, astragalus, navicular, and cuboid can be determined from facets, the distal calcaneum contacts the navicular, but in Daphoenus and Daphoenodon superbus the astragalus excludes the navicular from contact with the calcaneum. The calcaneal facet on the navicular is well developed in M. ferocior and D. oryktes and appears to be present in Temnocyon altigenis and T. ferox.

Fig. 62.

Comparison of the cuboids of (A) Daphoenodon superbus, (B) Mammacyon ferocior, (C) Canis lupus, and (D) Amphicyon galushai. In temnocyonines and daphoenines, the proximal (upper) ectocuneiform facet (ec) is well separated from the navicular facet (n) but in Amphicyon they are confluent. The wolf cuboid has two navicular facets, one in direct contact with the ectocuneiform facet. In D. superbus the cuboid also contacts the astragalus (a), likely the plesiomorphic state in both daphoenines and temnocyonines. A, left cuboid; B, C, D, right.

In wolf, Mammacyon, and Delotrochanter, the cuboid and navicular are in close contact, and their articular surfaces for the proximal tarsals lie in an essentially horizontal plane; the cuboid surface is only slightly convex and the navicular modestly concave. As a result of this nearly coplanar articulation of the proximal tarsals (calcaneum-astragalus) with the cuboid and navicular, the motion possible at the intratarsal joint limits rotation around the long axis of the hind foot, yet still allows some limited inversion/eversion as the foot accommodates to uneven ground. Other than this adjustment to the substrate, the hind foot of wolf and temnocyonines is committed to fore-aft flexion/extension at the ankle joint (1° of rotational freedom at the tibio-astragalar articulation).

Metatarsals and Phalanges

A remarkable modification of the metatarsals of the hind foot can be documented for Temnocyon and Delotrochanter and is likely in Mammacyon. This involves the near loss of the first metatarsal and marked reduction of metatarsals 2 and 5, resulting in a strongly paraxonic hind foot in which metatarsals 3–4 bear the weight in the erect digitigrade stance (fig. 61A). In Temnocyon ferox, Mammacyon obtusidens, and Delotrochanter oryktes, the metatarsals are proximally appressed as is evident in large living canines (fig. 63).

Fig. 63.

Metatarsals 2–5 of Canis lupus (A) and Delotrochanter oryktes (B). Paraxonic metatarsals 3–4 are flanked by markedly reduced metatarsals 2 and 5 in D. oryktes (UNSM 47800), indicating a narrow hind foot.

Representation of temnocyonine metatarsals is shown in table 8: In Temnocyon altigenis (UCMP 9999) only metatarsal 2 is complete, with proximal parts of metatarsals 3–4 and distal metatarsal 1; in T. ferox (YPM-PU 10787) metatarsals 1–5 survived; in Mammacyon obtusidens (ACM 34-41) metatarsals 3–5; and in M. ferocior (F:AM 27562) only a complete metatarsal 2, a proximal metatarsal 5, and metatarsal 3 lacking the proximal end were recovered. For D. oryktes (UNSM 47800, CM 1589b) there are metatarsals 2–5. Feet of the large species of Temnocyon are unknown but similar cursorial adaptations are likely, given the derived hind foot anatomy of T. ferox.

All of these temnocyonine metatarsals indicate a paraxonic hind foot with elongate metatarsals 3–4 and shorter, more slender metatarsals 2 and 5. The most complete hind foot (fig. 61), that of T. ferox, shows metatarsals 2 and 5 to be ∼1 cm shorter than metatarsals 3–4, with metatarsal 5 even more reduced than metatarsal 2. Metatarsal 1 also shows marked reduction (length, 45 mm; mid-shaft width, 4.8 mm). Delotrochanter oryktes exhibits even greater reduction of metatarsals 2 and 5, more so than in the wolf (fig. 63). Although the first metatarsal was not recovered, the scar for it on metatarsal 2 shows it was quite reduced, nonfunctional, yet not as small as the minute bone known in the canid. Metatarsal 2 of M. ferocior lacks an obvious scar for metatarsal 1 and so reduction of the latter bone in that species likely exceeded the state in D. oryktes and possibly equalled that found in the wolf. Reduction of the metatarsals flanking the paraxonic metatarsals 3–4 in D. oryktes is the most pronounced in the subfamily; consequently this reduction also probably occurred in D. major.

The lengths of paraxonic metatarsals relative to metacarpals are similar to those proportions in the wolf. In T. ferox where this can be measured, metatarsals 3–4 are 16%–17% longer than metacarpals 3–4, and in D. oryktes (UNSM 47800) this is 9%–10%, whereas in the wolf the difference is 10%–13%, a clear indication of the elongation of the forefoot in Delotrochanter.

Metapodials of these temnocyonines have prominent, sharp distal keels that fit into a groove at the base of each proximal phalanx (fig. 56). A pair of sesamoid bones within the tendons of each interosseous muscle would have bordered each keel in the articulated hind foot of Temnocyon ferox and together with the grooved phalanx would help to align and stabilize the digits. The position of the keel permits extension but little flexion at the metapodial-phalangeal joint. The metapodial-phalangeal joint is prevented from hyperextension in the erect digitigrade stance by contraction of the interosseous and digital flexor muscles.

Reduction of the side-toes in these temnocyonines (Temnocyon ferox, Delotrochanter oryktes, and less complete remains of Mammacyon) parallels reduction in the wolf and other large cursorial canines and is the first example of this type of paraxonic hind foot to evolve in the Carnivora.

No phalanges certainly attributable to the hind foot of temnocyonines were recovered with the exception of much reduced, intermediate (length, 15 mm) and ungual (length, ∼13 mm) phalanges of digit 5 in T. ferox. This suggests, however, that the intermediate phalanges in the hind foot were quite short, as in the forefoot. Eleven proximal and six intermediate phalanges and one ungual were recovered with the skeleton of D. oryktes (ACM 4804), but attribution to fore- or hind feet was not recorded (Schlain, 1980). These proximal phalanges are shorter, not as slender, and more robust than those of the wolf. Intermediate phalanges are uniformly short and somewhat asymmetrical; the distal trochlea extends proximad on the dorsal surface, indicating a moderate ungual retractility similar to that of the wolf.

The claws of T. ferox do not appear to have been very long. In the wolf, the keratin sheath covering the ungual phalanx adds ∼50% to its length so that in T. ferox, if similar, the functional claw on digit 5 of the hindfoot would measure ∼21 mm in length. This does not give us the length of the claw on a paraxonic digit of the hind foot; however, digit 4 of the forefoot also retains an ungual of similar size, suggesting that the claw length might have been about the same on the paraxonic digits of both fore- and hind feet. This situation is in conformity with the wolf where claws on fore- and hind feet are of similar lengths.