Pseudaelurus marshi Thorpe, 1922

ROTHWELL, TOM, 2003, Phylogenetic Systematics of North American Pseudaelurus (Carnivora: Felidae), American Museum Novitates 3403, pp. 1-64 : 38-47

Pseudaelurus marshi Thorpe, 1922

HOLOTYPE: YPM 12865 ( AMNH 13506 cast), articulated left and right rami with left i2, i3, left and right c, left alveolus of p2, right p2, left and right p3–m1, from Valentine Formation, Niobrara River , near mouth of Minnechaduza Creek (late Barstovian) ( Webb, 1969), Cherry County, Nebraska.

PARATYPE: YPM 12815 ( AMNH 133505 cast), left ramus with broken c, alveolus of p2, p3–p4, broken m1, from Gerry’s Ranch, Ogallala Group (early Barstovian), Weld County, Colorado .

REFERRED SPECIMENS: Lower Snake Creek Fauna, Olcott Formation (early Barstovian), Sioux County, Nebraska: F:AM 61808, left ramus with c, alveolus of p2, p3–m1, Echo Quarry; AMNH 22401, left partial ramus with alveolus of c, p3–broken m1, Ashbrook Pasture; AMNH 22404, right partial ramus with alveolus c and alveolus p2, broken p3– m1, Ashbrook Pasture; F:AM 61807, left partial ramus with alveolus of c, p2, p3–m1, Quarry 2; F:AM 61809, left partial ramus with alveoli of c and p2, p3–m1, Quarry 2; AMNH 22398, partial right ramus with alveoli of i1–p4, m1, Ashbrook Pasture.

Sand Canyon Formation (Early Barstovi­ an), Dawes County, Nebraska: F:AM 61846, partial left ramus with alveolus of i3, c, alveolus of p2, p3–broken m1, Observation Quarry.

Green Hills Fauna, Barstow Formation (early Barstovian), San Bernardino County, California: F:AM 61926, partial left ramus with p3–m1, Upper Steepside Quarry; F:AM 61921, right ramus with c, alveolus of p2, p3–m1, Upper Steepside Quarry; F:AM 61924, partial left ramus with broken c, p2 alveolus, p3–m1, Upper Steepside Quarry; F: AM 61918, partial left ramus with broken c, p3–m1, Upper Steepside Quarry; F:AM 61938, partial left ramus with i3–c, p3–m1, Turbin Quarry; F:AM 27337, left and right rami with c, p2–m1, Rak Division.

Second Division Fauna, Barstow Formation (early Barstovian), San Bernardino County, California: F:AM 18008, crushed partial skull with upper dentition, occipital fragment, left partial ramus with c, alveolus of p2, p3–m1 and partial right ramus with c, p2–m1,Valley View Quarry.

Trinity River Local Fauna, Fleming Formation (early Barstovian), San Jacinto County, Texas: F:AM 69331, partial left ramus with broken alveolus of c, alveolus p2, p3 and p4, m1, pit no. 1.

First Division Fauna, Barstow Formation (late Barstovian), San Bernardino County, California: F:AM 61934, articulated lower jaw fragment with i3, alveolus of p2, p3–m1, Leader Quarry; F:AM 61916, partial right ramus with broken c, p3–m1, Leader Quarry; F:AM 61936, partial right ramus with c, alveolus of p2, p3–m1, Black Hill Quarry; F: AM 27317, left and right partial rami with left and right c, left and right alveolus of p2, broken p3–m1, White Layer, 0.5 mile below cabin; F:AM 61940, crushed skull with left and right I1–M1 and right ramus with i1–c, alveoli p2–p4, broken m1, Hidden Hollow Quarry.

Pawnee Creek Formation, Big Springs Pit, Weld County, Colorado: F:AM 61841, partial right ramus with c, alveolus of p2, p3– m1.

Cerro Conejo Member, Zia Formation (late Barstovian; Tedford and Barghoorn, 1997), Sandoval County, northern Albuquerque Basin, New Mexico: F:AM 62144, partial skeleton with crushed skull with I1–C, alveolus of P2, P3–M1, partial left ramus with c, alveolus of p2, p3–m1, left and right humerus, radius, ulna, femur and tibia, partial scapulae, vertebrae, articulated pelvis, left and right calcaneus, assorted left and right metatarsal bones, and assorted phalanges, Rincon Quarry.

Pojoaque Member, Tesuque Formation (late Barstovian), Santa Fé and Rio Arriba counties, NM: F:AM 27457, partial articulated lower jaw with left and right lower dentition, left and right maxillary fragments with I3, C, P1, P2–M1, right calcaneus, partial humerus, Santa Cruz; F:AM 62135, skull fragment, left ramus with i2–c, p2 alveolus–m1, right partial ramus with separated c and p3– m1, vertebrae in matrix, partial femur, and assorted metapodials, south side of the Lobato branch of the South Fork of 3 Sand Hills Wash; F:AM 27453, partial skeleton, skull with upper dentition and intact basicranium, left dentary with i3, c, alveolus of p2, p3– m1, vertebrae, one rib, partial left articulated tarsus and metatarsus, Santa Cruz, field no. 13.

DISTRIBUTION: Early Barstovian of Texas, early and late Barstovian of Nebraska, California , and Colorado, late Barstovian of New Mexico .

DIAGNOSIS: Differs from all other species by combination of large size, short c–p3 length and slender and sloping coronoid process. P. marshi has m1 length that ranges from 14–19 mm, overlapping only with P. validus and P. intrepidus in North America and P. quadridentatus in Europe. P. marshi can be differentiated from P. validus by its much shorter c–p3 length and a slender and sloping coronoid process. P. marshi can be differentiated from P. intrepidus by shorter c–p3 length and dentary that is narrower and shallower below the tooth row. P. marshi can be differentiated from the European P. quadridentatus by its shorter c–p3 length and m1 with larger talonid and greater incidence of metaconid.

DESCRIPTION AND COMPARISONS: Thorpe’s (1922) description of the type lower jaw (YPM 12865) listed most characters that differentiate P. marshi from other large species in the genus: smaller size and a more slender and shallower dentary below the tooth row. However, he did not mention the shorter c– p3 length that differentiates P. marshi from other large species of Pseudaelurus (fig. 50). There is no evidence of p1 on either ramus. A single­rooted p2 can be seen on the right ramus, an alveolus on the left ramus. The p3 is smaller and shorter than p4, and both p3 and p4 have prominent posterior accessory cusps. The m1 has a reduced, yet distinct metaconid and talonid. The paratype partial left ramus (YPM 12815) was originally described as having an m2 alveolus. I examined this specimen and a cast (AMNH 133505) (fig. 51). The posterior root of the broken m1 is exposed, forming a triangular area posterior to the broken m1 crown. I suggest that the supposed m2 alveolus is this cavity formed behind the exposed portion of the m1 root.

The upper dentition of P. marshi does not differ appreciably from that of P. intrepidus . The best preserved specimen is F:AM 27453, from the late Barstovian Cerro Conejo Member of New Mexico (figs. 52–54). This skull has only three pairs of upper premolars. The Frick­AMNH collection has five specimens of P. marshi with upper dentitions. F:AM 27453 is the only specimen not having evidence of four upper premolars. I think that the most anterior premolar present in F:AM 27453 is P2. The crushed skull was reconstructed by Frick Laboratory preparators who located this P2 anterior to a point midway between C and P3. The suspected P2 is long, peglike and single­rooted in F:AM 27453. P3 is large, with a primary cusp that is similar in size to the paracone of P4. The width of P3 is greater in the posterior half of the tooth. P4 is similar to that of P. intrepidus . There is a prominent protocone that projects anterolingually. Anteriorly, at the base of the para­ cone of P4, a distinct parastyle is present, forming a deep secondary carnassial notch. On the anterolateral surface of the parastyle, a small accessory cusp is present. All of this P4 morphology resembles other species of North American Pseudaelurus . M1 is com­ pressed anteroposteriorly. M1 of P. marshi is nonvestigial. All North American specimens of Pseudaelurus studied have a functional, occluding, nonvestigial M1. In modern felids, however, M1 is vestigial. The upper component of the carnassial apparatus in­ cludes P3 and P 4 in extant felids, but does not involve M1.

F:AM 27453 provides well­preserved information about the skull of P. marshi (figs. 52–55). However, the dorsum of the cranium has been crushed, thus losing information on the sagittal and nuchal crests. The basicranium is better preserved. The left auditory bulla was prepared (fig. 54). Anteriorly, the septum bullae can be seen at the anterior margin of the petrosal, forming the lateral wall of the most anterior portion of the caudal entotympanic chamber. Posterolaterally, a sliver of the septum bullae can be seen arching toward the petrosal promontorium. This morphology of the septum bullae and its relationship to the petrosal promontorium agrees with that of P. validus ( Hunt, 1998; Rothwell, 2001). However, the caudal entotympanic of the P. marshi skull has expand­ ed more anteriorly when compared to the P. validus skull. The anterior expansion of the caudal entotympanic of P. marshi has reached the level of the most anterior margin of the ectotympanic chamber. The caudal entotympanic chamber terminates only slightly posterior to the anteromedial process of the auditory bulla. The caudal entotympanic has invaded, emarginated, and thinned out the ectotympanic, where it articulates with the basioccipital and basisphenoid. This differs from the much thicker, less invaded ectotympanic attachment seen in P. validus . The most anterior point of the P. validus caudal entotympanic is slightly posterior to the anterior apex of the ectotympanic chamber. This degree of anterior expansion in the P. marshi auditory region approaches the level seen in modern felids. On the medial surface of the petrosal, a medial extension or process of the promontorium can be seen. It rests upon the edge of the basioccipital. This process is smaller than the one described in a skull assigned to P. validus ( Hunt, 1998) . Possible correlation of depth of an expanding caudal entotympanic with disappearance of the medial process of the petrosal promontorium in modern felids has been suggested ( Hunt, 1998).

The close relationships of the bullae of F: AM 27453 with the mastoid and paroccipital processes are similar to those described in the P. validus skeleton from the Nambé Member of the Tesuque Formation ( Rothwell, 2001). Medial to each bulla, the ridge formed by the expanded caudal entotympanic indents the basioccipital. Posteromedial to each bulla, the hypoglossal foramen is visible within the same depression as the posterior lacerate foramen (figs. 54, 55). In F:AM 27453, the hypoglossal foramen is at the posteromedial border of the depression that it shares with the posterior lacerate foramen. Because of its position on the rim of a depression, the hypoglossal foramen does not open in a ventral direction as in Proailurus

TABLE 2 Greatest Length Measurements (mm) of Components of F:AM 62144 Skeleton, Referred to Pseudaelurus marshi (this specimen has no metacarpal bones)

lemanensis (fig. 29). The hypoglossal canal in P. marshi has shifted toward the horizontal, opening in an anteroventral direction, similar to modern felids. The alisphenoid canal is present on both sides of the P. marshi skull. However, crushing of this specimen prevents documentation of the presence of a foramen rotundum.

Postcranial information on P. marshi is provided by F:AM 62144, a partial skeleton from the late Barstovian Cerro Coñejo Formation (table 2). The limb elements do not differ from those previously referred to P. validus and P. intrepidus . Unfortunately, F: AM 62144 does not include any metacarpal bones. This hinders comparison of the front limb elements of P. marshi with Proailurus lemanensis , Pseudaelurus validus , Pseudaelurus intrepidus , and modern felids ( Rothwell, 2001).

DISCUSSION: P. marshi is the most abundant Miocene felid in the Frick­AMNH collection. I have studied 80 Pseudaelurus low­ er jaws that I have been able to assign to one of six North American species. Twenty­eight specimens, or 35%, are referred to P. marshi . Twenty­two are assigned to the similar­sized and temporally equivalent species P. intrepidus . Specimens of P. marshi are found in virtually every felid­producing locality in the early and late Barstovian of North America. P. marshi and P. intrepidus account for 75% of the felid lower jaws from North American Barstovian localities. Pseudaelurus marshi , similar in size to P. intrepidus , differs primarily in the c–p3 length and the height of the dentary.

The possibility that these two fossil taxa are sexually dimorphic members of the same species must be considered. One definition of sexual dimorphism is ‘‘when the means for fully developed males and females differ in a statistically significant way. The dimorphism is partial when the standard ranges of males and females overlap, and complete when they do not’’ ( Kurtén, 1955: 7). One hypothesis is that lower jaw specimens assigned to P. marshi , based primarily on the character of short c–p3 length and dentary height, are female members of P. intrepidus . To test this hypothesis, I studied modern felids. I examined modern skulls of four felid species in the Department of Mammology at the American Museum. Skeletons of Lynx canadensis , Lynx rufus , Felis concolor , and Panthera leo , killed in the wild and labeled as either male or female, were examined, measured, and scored for 40 dental and cranial characters (table 6). The limited number of skeletons labeled as to sex did not make it possible to select specimens in a random manner. To facilitate comparison of c–p3 length in different­sized felids, I calculated the mean of a c–p3 index for extinct and modern felids (table 4). To compute the c– p3 index, I divided the c–p3 length by the m1 length for each specimen studied (table 5). The length of the lower carnassial tooth has been demonstrated to be a good proxy for body size in felids ( Legendre and Roth, 1988).

Panthera leo and Lynx canadensis displayed sexual dimorphism with respect to the c–p3 length; Felis concolor and Lynx rufus did not (table 3). I then grouped P. intrepidus and P. marshi into one hypothetical extinct species. I treated the data from these two taxa as partitions of the same species. The P. marshi and P. intrepidus data present as either two separate species or as sexually dimorphic members of the same species (table 3). In other words, the data of P. intrepidus and P. marshi present as either male and female felids whose c–p3 length resembles somewhat the lion or the lynx, or as two similarsized but separate species which have the c– p3 length form of the mountain lion or bobcat.

To further compare P. intrepidus and P. marshi with modern felids, I plotted the dentary height versus m1 length (fig. 56). The height of the dentary below the carnassial

TABLE 3 Results of Analysis of Variance Performed on c­p3 Length Data of Extinct and Living Felids

tooth differs significantly (p = 0.0004) in P. intrepidus and P. marshi . Again, the length of m1 was used as an analog for body size. These data can be compared to the same information from modern specimens of F. concolor killed in the wild (fig. 57). The male and female F. concolor data resemble the P. intrepidus and P. marshi chart. Data from figures 56 and 57 are combined in figure 58. Felis concolor does display sexual dimorphism with respect to height of the dentary (p = 0.0475). This felid does not display sexual dimorphism in the c–p3 length (p = 0.28).

TABLE 4 Data of c­p3 Length in Felid Lower Jaws TABLE 5 Data of m1 Length in Felid Lower Jaws

In summary, there are arguments for and against sexual dimorphism being the explanation for character differences between P. intrepidus and P. marshi . Supporting arguments include: (1) Sexual dimorphism is seen in modern felid species, at the very least in body size and skull length in small felids ( Dayan and Simberloff, 1996). (2) P. skinneri and P. stouti display differences in c–p3 length that could also be interpreted as sexual dimorphism. (3) P. intrepidus and P. marshi are size and temporally equivalent. Specimens of the two species are sometimes found in the same localities. Counter arguments include: (1) A similar range in c–p3 length and dentary height and width is not seen in P. validus . P. validus and some modern species do not exhibit sexual dimorphism in the low­

TABLE 6 Character/Taxon Matrix Used to Resolve Felid Phylogeny (question marks indicate missing data)

TABLE 7 Metacarpal 3 Contribution to Total length of Front Limb in Six Felids Total front limb length was computed from combined lengths of humerus, ulna, and metacarpal 3.

er jaw. (2) Modern assemblages of felids sometimes contain at least two species that are indistinguishable in jaw length ( Kiltie, 1988). (3) The p error for the hypothetical P. intrepidus – P. marshi species (0.0001) is far outside the range of p error seen in the four modern species studied (0.28–0.008).

The results appear inconclusive, primarily because of the variability of the sexual dimorphism displayed by different species of living felids. However, if P. intrepidus and P. marshi were indeed sexually dimorphic members of the same species, the males and females differed far more than any of the four living felids. For this reason, I choose to consider these two samples as separate species.