LOCALITIES

The new material studied herein comes from 12 localities in Chadronian and Chadronian/Orellan deposits of southwest Montana and southeast Wyoming (fig. 1). McCarty's Mountain in the Beaverhead Basin is the oldest of all localities in Montana described in this paper, and has the longest history of exploration, being discovered by Earl Douglass in July 1903 (Tabrum et al., 2001). It includes an 850 ft thick sequence, originally divided by Douglass into successive layers designated Q to Z. The sequence crops out over an area of approximately 0.25 mi² in section 28, T4S R8W in Madison County, about 5 mi south-southeast of McCarty's (now McCartney) Mountain and north of the Big Hole River (Tabrum et al., 1996; Tabrum et al., 2001). The locality was included in the Renova Formation and the exact time span represented by the sediment series was eventually established as early Chadronian, based on the faunal assemblage (Tabrum et al., 1996, 2001; Tabrum and Fostowicz-Frelik, 2009).

FIG. 2.

Generalized stratigraphic sections of the White River Formation in SE Wyoming (Douglas and Lusk areas), showing distribution of Litolagus molidens (marked as gray rabbit ) and Limitolagus roosevelti, new species (marked as black rabbit) in the sediment series near the Eocene-Oligocene boundary. P.W.L. used as correlation level 0; all data after Skinner (MS).

FIG. 3.

Skull measurements based on Lepus; ventral (A, C, closeup of palatine region), and dorsal (B) views of cranium, and mandible in lateral view (D). Abbreviations: BL, bulla length; ChW, choanae width; CpH, condylar process height; FrL, frontal length; FrW, minimal frontal width; InW, distal incisive width; JuW, jugal width of the skull; LDL, lower diastema length; LTR, lower toothrow length; MH, mandible height; ML, mandible length; MxW, maxillar width of the skull (at alveolar processes of maxilla); NaL, nasal length; PaL, hard palate length; PalL, palatine length; ParL, parietal length; PaW, palate width (measured between tooth row at the level of P4/M1); SL, skull length; SqW, squamosal width of the skull; UDL, upper diastema length; UTR, upper toothrow length.

FIG. 4.

Tooth measurements (occlusal surface). P2 (A), p3 (B), P3–M2 (C), and p4–m3 (D). Abbreviations: ant, anterior direction; bucc, buccal direction; L, length; ling, lingual direction; TalW, talonid width; TriW, trigonid width; W, width.

The Pipestone Springs area in Jefferson County Montana, includes three small local faunas: the Pipestone Springs Main Pocket, Pipestone Springs Fence Pocket, and Little Pipestone Creek. The first two local faunas are dated to the middle Chadronian, while Little Pipestone Creek is considered to be slightly younger, middle to early late Chadronian (Tabrum et al., 2001; Dawson, 2008; Janis et al., 2008). The Pipestone Springs area lies in the Jefferson Basin and the sediments are part of the Climbing Arrow Member of the Renova Formation. The Main Pocket of Pipestone Springs is one of the classic localities discovered and studied by Douglass (1901) and is the type locality for Palaeolagus temnodon.

Two other localities in southwest Montana, 10N locality of the Dunbar Creek Formation and Little Spring Gulch local fauna, which is regarded as coming from the lowest part of the Cook Ranch Formation, are of late Chadronian origin (Tabrum et al., 2001). The material from the 10N area was collected at several exposures located on the east side of U.S. Route 287 (formerly U.S highway 10N) in the Three Forks Basin, near the connection with Interstate 90 (Tabrum et al., 2001; Asher et al., 2002). The fossiliferous beds of Little Spring Gulch crop out on the north side of the gulch in the Sage Creek Basin, southwestern Montana, and are located a few miles north of the Cook Ranch type area, NE of Lima (Tabrum et al., 1996, 2001; Prothero and Emry, 2004).

The eight localities in Wyoming that yielded the new specimens described here, are located in the vicinity of Douglas, Converse County (Reno Ranch, south of tower, Herman Wulff Ranch, and an unnamed locality 8 mi SE of Douglas) and Lusk, Niobrara County (Bonner brothers' Ranch—Pass area, Seaman Hills, S side of Jim Christian Hill, Lower Spring Draw, and an unnamed locality 2 mi NE of Indian Creek), the classic area of the Frick Laboratory expeditions led by Morris Skinner.

The specimens were recovered during the 1953–1963 field seasons, according to Skinner's field journal (Archives of the Department of Paleontology, AMNH). They all come from the upper Chadronian and the Chadronian-Orellan transition part of the White River Formation sequence in that area (Emry et al., 1987; Evanoff et al., 1992; Prothero and Emry, 2004).

The detailed field journals kept by Skinner give an exact position of each finding with reference to a so-called persistent white layer (see fig. 2), otherwise known as “purple-white layer” of Schultz and Stout (1955, 1961), “tuff 5” of Evanoff (1990), or “Upper Purplish White layer” of Zanazzi et al. (2009), and dated by the ⁴⁰Ar/³⁹Ar method to 33.7 ± 0.7 Ma (Evanoff et al., 1992), which falls within the interval of the Eocene/Oligocene boundary (Prothero and Emry, 2004). Evanoff et al. (1992) give correlations for some localities studied herein, while others were approximately correlated on the basis of Skinner's unpublished material (Skinner, MS).

The stratigraphy and fauna of the Flagstaff Rim and Bates Hole areas in Wyoming, which yielded the type series of Chadrolagus emryi, were described in detail by Emry (1973, 1992), Gawne (1978), and Emry and Gawne (1986), who established the age of these remains as Middle Chadronian (see also Prothero and Emry, 2004). The fauna from the Pipestone Springs area used for comparison is of middle Chadronian age (Tabrum et al., 2001; Prothero and Emry, 2004).

MORPHOLOGY OF THE SKULL

SKULL IN GENERAL: The skull of Litolagus molidens is longer than that of Chadrolagus emryi, Palaeolagus temnodon, P. burkei, and P. haydeni. The measurements for the specimens described here are given in table 9. The skull AMNH FM 143955 (figs. 25–27) is slightly smaller and more delicate and gracile than the skulls of Megalagus, and P. intermedius, being comparable to Archaeolagus ennisianus, but slightly more slender. The skull of Litolagus molidens is relatively higher and more angled than the skulls of Palaeolagus and Megalagus, but the basicranial bending is less profound than in Archaeolagus. Moreover, the braincase structure with elongated, flat parietals, well-pronounced sagittal crest, and prominent dorsally exposed mastoids seems to be more similar to that of Palaeolagus.

The muzzle of Litolagus molidens is long and slender with a relatively elongate diastema (compared to the length of the upper tooth row; table 2), which is similarly long in Litolagus molidens, Palaeolagus temnodon and P. haydeni. Moreover, the palate area is wider in Litolagus than in Palaeolagus temnodon and P. haydeni, which have similar length of the upper tooth row, but is visibly narrower than in Archaeolagus (table 2).

FIG. 22.

The holotype of Litolagus molidens (LACM-CIT 1568) in a slab.

FIG. 23.

Skull of the holotype of Litolagus molidens (LACM-CIT 1568) in ventral (A), and lateral (B) views.

Both skulls of Litolagus described here (AMNH F:AM 105991 and AMNH FM 143955) have strongly marked ornamentation expressed as a series of minute concavities covering nasal and frontal bones, especially the region above the olfactory lobes, running in two plies along the sagittal suture on parietals, and concentrating at the parietal-mastoid suture (figs. 25–26, 29). That kind of ornamentation is weaker in all other palaeolagine lagomorphs, although perforation of the frontals can be observed in Chadrolagus (fig. 10C) and Palaeolagus. It is also common in modern leporids, but it concentrates mostly on the parietals and in most cases is not extensive. The only extant species with such well-developed ornamentation covering most of the skull roof is the volcano rabbit Romerolagus diazi (personal obs.; see also Wible, 2007: fig. 1B). The original description of Litolagus lacks any comment on this character, but the material presented by Dawson (1958) was almost completely devoid of the skull roof and the holotype specimen is in major part still embedded in the matrix (figs. 22–23).

FIG. 24.

Mandibles of the holotype of Litolagus molidens (LACM-CIT 1568). Left mandible in buccal (A), lingual (B), and occlusal (F) views. Right mandible in buccal (C), lingual (D), and occlusal (F) views.

NASALS: The nasals are long and generally equally wide over the entire length, not tapering visibly as in Chadrolagus emryi. They are located relatively high, not gently arched as in Palaeo- lagus, forming a more spacious nasal cavity. The anterior part of the nasals, where the bones detach from the premaxillae and form the roof of the piriform foramen (the external nasal opening), is higher than in Palaeolagus haydeni and P. temnodon. Two-thirds of the posterior part of the nasals have a visible ornamentation in the form of minute irregular concavities, located especially along the external margins of the nasal bones. This kind of ornamentation is displayed more distinctly in AMNH F:AM 105991, where the posterior part of the right nasal bone is heavily perforated. Similar perforation of the nasal bones is also observed in a specimen of P. temnodon (CM 8620), but is much weaker in P. haydeni.

PREMAXILLA: The premaxilla accommodates the upper incisors, which are embedded in the dorsal part of the body of premaxilla, although slightly more ventrally than in Palaeolagus. The length of the incisor alveolus is shorter in Litolagus (as well as in Palaeolagus and Chadrolagus) than in modern leporids, although the distal opening of the root ends similarly at the premaxillarymaxillary suture, but it is located higher laterodorsally (in the upper 40% of the dorsoventral extent of the body of the premaxilla), whereas in modern lagomorphs the end of dI2 is located at the ventral side of premaxilla, in the vicinity of the premaxillary-maxillary suture.

The alveolar part of the body of the premaxilla in Litolagus is not as elongated as in modern leporids (e.g., Lepus or Sylvilagus), and the thin crests at the base of the frontal processes of premaxilla (the posterodorsal process sensu Wible, 2007) that frame the lateral margins of the piriform aperture anteriorly descend to the very anterior margin of the incisor alveoli, as in Palaeolagus and Chadrolagus. Nevertheless, the body of the premaxilla is slightly elongated in Litolagus in comparison to those of Palaeolagus and Chadrolagus. The frontal process of the premaxilla in Litolagus is long and thin, reaching as far as the nasal-frontal contact at the lateral side of the skull, only ≤1.0 mm shorter than the greatest posterior extent of the nasals, as in Chadrolagus, Palaeolagus, and Archaeolagus.

MAXILLA: The anterior part of the maxilla is heavily perforated, forming, as is characteristic of Lagomorpha, lacework bone on both sides of the muzzle (Meng and Wyss, 2001; Meng et al., 2003; Asher et al., 2005; Wible, 2007). The fenestration is similar in Litolagus to that in Chadrolagus and Palaeolagus. The bone ventral to the lacework is solid and forms an acute-angled process that cuts anteriorly into the premaxilla. This prominent process is sharper, narrower, and longer in Litolagus than in Chadrolagus, but shorter than in Palaeolagus haydeni. It is oriented more laterally in Litolagus, higher at the muzzle than in Chadrolagus emryi, as in Palaeolagus haydeni. The remainder of the premaxillary-maxillary suture is located at the ventral side of the muzzle and forms a deeply overlapping serrated bone contact. The infraorbital foramen is slightly smaller than in P. haydeni and P. temnodon and positioned over the anterior part of the alveolus for P2, like in Chadrolagus emryi, and slightly more anteriorly than in Palaeolagus.

The frontal process of the maxilla is straighter and more erect than in Palaeolagus haydeni, in which it reclines distally ca. 45°, and in Chadrolagus emryi, which expresses a slightly less inclination of this process. In Litolagus it forms a gentle arch framing the anterior part of the orbit.

In ventral view, the incisive foramen bordered by the premaxilla and maxilla is long in Litolagus. It extends to the distal part of the P3 alveolus, whereas in Chadrolagus emryi it reaches the anterior margin of P2 alveolus; in Palaeolagus temnodon, the midlength of P2 alveolus; and in P. haydeni, the P2/P3 contact. This shortens the palatal bridge in Litolagus more than in North American stem lagomorphs (fig. 30, table 2). Moreover, the choanae reach the M1/M2 in Lito- lagus as in Chadrolagus emryi, whereas in P. haydeni they open closer to the middle of M2. In Litolagus the maxillary part of the palate formed by the palatine process of the maxilla occupies 57% of the length of the palatal bridge, comparable to Archaeolagus ennisianus (53%), whereas in Chadrolagus emryi, Palaeolagus temnodon, and P. haydeni it constitutes only 36%–49%, 42%, and 38%–46% of the palate, respectively.

FIG. 25.

Skull of Litolagus molidens (AMNH FM 143955) from Herman Wulff Ranch, Converse County, Wyoming in dorsal (A), ventral (B), and lateral (C) views. See figure 26 for explanations.

FIG. 26.

Line drawing of skull of Litolagus molidens (AMNH FM 143955) in dorsal (A), ventral (B), and lateral (C) views. Abbreviations: bo, basioccipital; bs, basisphenoid; eam, external acoustic meatus; ec, ectotympanic; fm, foramen magnum; fpp, frontal process of premaxilla; fr, frontal; inf, incisive foramen; iof, infraorbital foramen; ju, jugal; me, mastoid exposure of petrosal; mx, maxilla; na, nasal; oc, occipital condyle; pa, parietal; pal, palatine; pmx, premaxilla; pt, pterygoids (ento- and ectopterygoid crests); sc, sagittal crest; so, supraoccipital; sq, squamosal. Hatched areas correspond to missing or damaged bone.

FIG. 27.

A, Occipital region of Litolagus molidens skull (AMNH FM 143955) in caudal view, and B, line drawing. See figure 26 for abbreviations; nc, nuchal crest. Hatched areas correspond to missing or damaged bone.

The alveolar process of the maxilla is deep and occupies the lower part of the orbit as in other palaeolagines and modern lagomorphs (Craigie, 1945; Wible, 2007). The lateral border of the alveolar process on the left side of the skull in AMNH FM 143955 is complete and frames a partly destroyed dentition. It allows view of the spatial relation of the zygomatic process and the ventral margin of the alveolar process of the maxilla, which are closer to each other in Litolagus molidens than in Palaeolagus haydeni, creating simultaneously a more spacious and round orbit.

The morphology of the zygomatic process of the maxilla was one of the characters differentiating Litolagus from Palaeolagus emphasized by Dawson (1958). In ventral view the zygomatic process of Litolagus, especially the masseteric spine is less expanded laterally than that of Palaeolagus (P. temnodon, P. haydeni, and P. burkei), Chadrolagus emryi, and Archaeolagus ennisianus (fig. 30). It is also longer and resembles the shape of this structure in the European rabbit (Oryctolagus cuniculus), although it is located closer to the alveolar process of the maxilla. On the other hand, in Palaeolagus the masseteric spine is shorter and more compact in ventral view, but flares more laterally, similar to the condition found in the mountain hare (Lepus timidus). In lateral view, the anterior tip of the masseteric spine in Litolagus forms a sharp apex (as in Chadrolagus emryi) pointing anteroventrally, whereas in Palaeolagus the tip is blunt, directed more ventrally, and the zygomatic process is swept somewhat backward.

FIG. 28.

Mandible of Litolagus molidens (AMNH FM 143955) in lateral (A), and lingual (B) views.

In Litolagus molidens a shallow fossa occurs anterior to the zygomatic process of the maxilla, forming a nascent depression almost completely lacking in Palaeolagus and Chadrolagus emryi, but well marked, although shallow, in Megalagus, deeper in Archaeolagus, and well developed and deep in extant leporids (in some species perforated). The fossa was mistakenly marked as the infraorbital foramen in Romerolagus, by Wible (2007: fig. 2B), but in fact the infraorbital foramen in leporids lies within the ventral part of the fenestrated area of the maxilla (Craigie, 1945: 179, fig. 79) and is obscured by the extensive fenestration of the whole region.

JUGAL: The zygomatic arch in Litolagus molidens is positioned relatively lower on the skull than in Palaeolagus, which creates more space in the orbit. This is a result of simultaneous lowering of the position of the zygomatic process of the maxilla (moving it closer to the occlusal tooth surface), and stronger bending ventrally of the zygomatic process of the squamosal, unlike the condition observed in Palaeolagus, Megalagus, and Archaeolagus ennisianus. In Chadrolagus emryi the zygomatic process of the maxilla lies equally low (fig. 10B) as in Litolagus, but the squamosal zygomatic process is not bent ventrally more than in Palaeolagus.

FIG. 29.

Skull of Litolagus molidens (AMNH F:AM 105991) in dorsal view. Note the strongly perforated frontal.

The exact position of the bone contact between the zygomatic process of the maxilla and the jugal bone is not clear in any of the specimens of Litolagus molidens because this suture is obliterated very early in ontogeny (Craigie, 1945). The lamina of the jugal bone is relatively broad in Litolagus and positioned exactly parallel to the sagittal plane. It has neither the masseter margin gently inclined outward nor both the jugal plates swung aside backward, flaring slightly, increasing the distance from the orbit, as it is in Palaeolagus. The jugal bone of Litolagus is broadest at its anterior end near the zygomatic process of the maxilla; it then narrows slightly toward the posterior terminus. The lateral surface of the jugal is flatter in Litolagus (and Chadrolagus) than in Palaeolagus. Although a shallow zygomatic fossa occurs in both species, it is not emphasized by the additional lateral bending of the orbital and masseter margins observed in Palaeolagus haydeni and Archaeolagus ennisianus (expressed more weakly in P. burkei). In Litolagus molidens the zygomatic fossa is a single, well-defined oval depression about 3.5 mm long and 2.2 mm high, unlike Palaeolagus haydeni, which has a spade-shaped zygomatic fossa shallowing distad.

The posterior part of the jugal arises slightly upward to join the zygomatic process of the squamosal. In lagomorphs the part of the jugal plate along the jugal-squamosal suture is generally narrower than the rest of the bone plate. In Litolagus molidens this part is relatively longer and less narrow than in Palaeolagus haydeni, and the distal end of the jugal seems blunter than in Palaeolagus.

PALATINE AND PTERYGOID: The palatine bone forms the posterior part of the hard palate and surrounds the opening of the internal nares. The palatine portion of the palatal bridge is relatively short in Litolagus molidens, in comparison to Chadrolagus emryi and Palaeolagus. The anteromedial part of the maxillo-palatine suture in Litolagus molidens is profoundly flattened, apart from the minute anterior spinelike apex, and is perpendicular to the sagittal plane differentiating this species from Chadrolagus emryi, Palaeolagus, and Archaeolagus ennisianus, which have the maxillo-palatine contact rounded with the convex margin directed anteriorly (figs. 26, 30; Gawne, 1978: fig. 2a). One pair of major and one pair of minor palatine foramina are located closer to the maxillo-palatine suture, and more laterally in Litolagus than in Chadrolagus and Palaeolagus. The foramina are also relatively smaller in Litolagus than in Palaeolagus or Chadrolagus.

The distal margin of the palatal bridge forming the rim of the choanae has a weakly developed tubercle, which can be homologized with the posterior spine of the palatine, not detected in Chadrolagus and Palaeolagus, but well developed in more advanced lagomorphs, including archaeolagine leporids (see Dawson, 1958: fig. 27a; Fostowicz-Frelik, 2007: fig. 8a) and some leporine genera (Craigie, 1945, for Oryctolagus; personal obs. for Sylvilagus).

The choanae are wider in Litolagus molidens in relation to the width of the palate than in Palaeolagus haydeni and P. burkei or Chadrolagus emryi, but slightly narrower than in Archaeolagus ennisianus, P. intermedius, and Megalagus turgidus (table 2). The medial walls of the basipharyngeal canal (sensu Evans, 1993), formed anteriorly by the palatine and posteriorly by the entopterygoid crests (Wible, 2007), are parallel and show no tendency toward narrowing in the midlength, as can be observed in some Palaeolagus haydeni specimens (e.g., AMNH FM 143956), where the choanae are somewhat heart shaped. The entopterygoid crests are positioned almost vertically in Litolagus molidens, with the ectopterygoid crests (lateral laminae of pterygoid processes sensu Craigie, 1945: fig. 81) spread out laterally, at an oblique angle of about 45°. The entopterygoid crests in Litolagus molidens are much more vertical than those of Chadrolagus emryi and Palaeolagus. The hamuli of the pterygoids are partly preserved in AMNH FM 143955. They lie in the plane of the entopterygoid crests (sagittally), project ventrodistally, and do not deviate laterally as in some species of extant lagomorphs, e.g., Oryctolagus (Craigie, 1945) or Sylvilagus (personal obs.).

BASISPHENOID: The basisphenoid is well preserved in specimen AMNH FM 143955, although its anterior part is broken and contact with the pterygoid is obscured. The basisphenoid in Litolagus molidens appears relatively narrow in comparison with Palaeolagus and extant Lepus and Sylvilagus, and this narrowing seems rather more natural than artificial because the entire distal part of the skull of AMNH FM 143955 is not crushed or distorted significantly (fig. 25–27). The bone is also gently grooved longitudinally. The craniopharyngeal canal (Evans, 1993; Wible, 2007: fig. 4B), or foramen cavernosum (Craigie, 1945), is visible in Litolagus molidens but hard to locate even in the best-preserved specimens of P. haydeni. The intersphenoidal synchondrosis is well marked in Litolagus and flanked by the auditory bullae, unlike in modern leporids, where it occurs anterior to the bullae. The bone, immediately anterior to the contact with the basiooccipital, forms two gently swollen tuberosities, indicating the position of the synchondrosis.

OCCIPITAL: The ventral part of the occipital region, the basioccipital, in Litolagus molidens is relatively long compared with Palaeolagus haydeni. The posterior part bears a thin spine that continues as a conelike structure located at the ventral side of the intercondyloid notch, a structure found in Palaeolagus but very poorly developed in modern leporids. Although the supraoccipital in Litolagus molidens (fig. 27) is partly crushed, it seems higher and more vertically positioned than in Palaeolagus haydeni and P. burkei, and even more than in extant leporids, in which the distal surface of the skull is strongly inclined backward and faces ventrodistally. The exoccipital in Litolagus molidens is similar to that of Palaeolagus and the jugular process or paracondylar process of the exoccipital (sensu Wible, 2007: fig. 5B) has a length similar to that in P. haydeni, but is longer than in P. burkei. However, they all differ markedly from the structure in Archaeolagus ennisianus and extant leporids. The paired exoccipital bones in Litolagus and Palaeolagus are much smaller and narrower than in extant species and Archaeolagus. They constitute relatively narrow lateral margins of the foramen magnum with the L-shaped jugal processes detaching lateroventrad, with their tips pointing slightly mesiad. In leporids the exoccipital bones are much wider, not L-shaped but semicircular, with a flared, round dorsolateral margin, covering a significant part of the mastoid exposure of the petrosal in distal aspect.

FIG. 30.

Structure of hard palate. A, Litolagus molidens (AMNH FM 143955); B, Megalagus turgidus (AMNH FM 141012); C, Archaeolagus ennisianus (AMNH FM 7190); D, Palaeolagus haydeni (AMNH FM 143957); E, Chadrolagus emryi (AMNH F:AM 99108); F, Palaeolagus burkei (AMNH FM 8704).

The articular surface of the occipital condyle is slightly broader in Litolagus molidens than in Palaeolagus haydeni and the intercondyloid notch is deeper and narrower. The whole structure of the condyles seems stronger in Litolagus, which is not surprising, given the larger skull compared with Palaeolagus.

FRONTAL: The skull roof in AMNH FM 143955 is partially crushed, although most of the frontals are intact (fig. 25A). The frontal region is less complete in skull AMNH F:AM 105991, but the parietal region is better preserved (fig. 29).

The anterior part of the frontal bones in Litolagus is heavily ornamented, especially in the region covering the olfactory lobes (figs. 25A, 26A, 29). The ornamentation is stronger in Litolagus than in any species of Palaeolagus or Chadrolagus emryi. The nasal spine of the frontal is blunted and irregular at its anterior margin, wedged between the nasal bones. It does not form such a triangular wedge as in Palaeolagus and Chadrolagus. The supraobital processes are lacking in both specimens, as well as in the holotype (Dawson, 1958), although the presumption is that only the posterior supraorbital processes were present and they were similarly developed as in Palaeolagus or Chadrolagus (Wood, 1940; Dawson, 1958; Gawne, 1978). The orbital part of the frontal is significantly higher in Litolagus than in Palaeolagus or Chadrolagus resulting in elevating the skull roof.

Two distinct gently ornamented frontal ridges follow along the orbital margins (fig. 25A), cross the frontoparietal suture, and meet medially midway along of the parietal (figs. 25A, 29). The ridges are much better developed in Litolagus than in any species of Palaeolagus, where only delicate lines can be discerned, or in Chadrolagus emryi, where they are lacking entirely. The frontals join with the parietals by means of an irregular suture generally perpendicular to the sagittal plane. This suture in both specimens of Litolagus molidens described here is similar in its appearance to that of Chadrolagus emryi, more irregular and flatter than the suture of Palaeolagus haydeni, which shows two symmetrically located processes wedged into the parietals.

PARIETAL: The parietal region of Litolagus molidens is similarly structured as in Palaeolaginae (Dawson, 1958: pls. 1, 2). The anterior part is slightly convex dorsally, whereas the distal part flattens and extends caudally The morphology markedly distinguishes Litolagus from modern leporids and Archaeolagus ennisianus, whose braincases have a shorter and rounded parietal portion. The skull of Litolagus has visible narrow V-shaped ornamented crests terminating distally at the sagittal suture on the parietals (figs. 25A, 29). From the place of their fusion, the sagittal crest arises and continues distally to the suture with the supraoccipital. The sagittal crest is more pronounced in both specimens of Litolagus than in any Palaeolagus haydeni and P. burkei. The condition of this character is unknown in Chadrolagus, whereas in Archaeolagus ennisianus, although the crest is not present, the parietals narrow slightly and form a longitudinal ridgelike eminence, not as sharp as a crest, but visible.

SQUAMOSAL: The parietal joins the squamosal via a strongly serrated suture. The ventralmost parts of both squamosals are missing in both skulls of Litolagus, but the zygomatic processes of the squamosal are well preserved. They join the jugal dorsally forming a relatively long, straight suture in the horizontal plane. The zygomatic process of the squamosal is formed as a rightangled triangle with its right angle directed distally and the base joined with the jugal. In Litolagus molidens the structure is more slender and elongate (with its sides either gently convex or concave) resembling the processes or Lepus or Sylvilagus more than those of Palaeolagus haydeni, P. temnodon, P. burkei, Chadrolagus emryi, and Archaeolagus ennisianus, in which it is sturdier and fully triangular. Moreover, the base of the zygomatic process, which connects the zygomatic arch with the skull, is more strongly bent ventrally in Litolagus molidens (as in modern leporids) than in Chadrolagus emryi and Palaeolagus, resulting in a lower position of the jugal arch in relation to the skull roof, and creating more space in the orbit. Also, the bending of the zygomatic processes in Litolagus brings the plate of the jugal closer to the skull.

ORBIT REGION: The orbit in Litolagus is markedly larger and more round than in all species of Palaeolagus analyzed here, Megalagus, and Chadrolagus emryi. This is caused by the lowering of the zygomatic arch, the more vertical position of the frontal process of the maxilla, the higher orbital portion of the frontal, and the raised orbital margin of the frontal at the supraorbital process. Size of the orbit in Litolagus molidens is comparable to that of Archaeolagus ennisianus or slightly smaller, but AMNH FM 7190, a specimen of Archaeolagus, has one of the supraorbital processes of its frontal preserved, and it defines the orbit more accurately than can be determined for any of the Litolagus specimens (all of which lack the supraorbital processes).

The orbitosphenoid is mostly preserved, and sutures with the frontal, squamosal, and parietal are clearly visible. The optical foramen is large, and the position of the alisphenoid canal is well marked, especially on the right side of the skull in AMNH FM 143955. However, the alisphenoid is mostly damaged on both sides of the skull (figs. 25–26).

The lacrimal bone, which is usually reduced in Lagomorpha and forms a small tubercle connected to the anterolateral extremity of the frontal (Craigie, 1945), is mostly lacking in the fossil material. The skull of Litolagus molidens (AMNH FM 143955) has a small protuberance at the orbital margin of the frontal process of the maxilla, possibly the remains of the lacrimal bone, not different in size from that in modern taxa.

PETROMASTOID AND ECTOTYMPANIC: The ear region in lagomorphs consists of two distinct elements: the ectotympanic, which forms the auditory bullae, and the petromastoid, which encapsulates the inner ear structures (Craigie, 1945). The auditory bullae are crushed or embedded in the matrix in all specimens of Litolagus, although the partially preserved auditory bullae of AMNH FM 143955 give some insight into the external morphology of the ear region (fig. 25). The auditory bullae are relatively large in Litolagus compared to Palaeolagus haydeni, Archaeolagus ennisianus (table 2), and extant leporids. They are similarly enlarged in P. burkei in relation to the skull size, although in Litolagus molidens the bullae were probably more dorsoventrally extended than in P. burkei, which has a generally flattened skull and bullae. However, they are relatively smaller than in P. hypsodus (see Dawson, 1958: p1. 2:1a–d). The auditory bullae have, as far as can be ascertained after taking into account alteration caused by the sedimentation processes, their long axes directed more parallel to the sagittal plane than in Palaeolagus, in which the long axis of the auditory bulla forms an angle of about 45° with the sagittal plane of the skull. The external meatus of Litolagus molidens is equally long in Palaeolagus haydeni, and the position of the stylomastoid foramen is clear in Litolagus molidens AMNH FM 143955, on the left side of the skull.

The dorsal exposure of petromastoid (Craigie, 1945), or the mastoid exposure of the petrosal (Meng et al., 2003), is massive in Litolagus molidens, somewhat swollen, and covered with delicate ornamentation that is finer than that of the frontal (fig. 25); this ornamentation, while typical for this element in lagomorphs (Craigie, 1945), is better expressed in Litolagus than in Palaeolagus. It protrudes over the surface of the parietal as a noticeable, right-angled triangle with ventrally directed mastoid process (Craigie, 1945), otherwise called the paroccipital process of petrosal (sensu Wible, 2007). The general outline of the mastoid exposure of petrosal is similar in all lagomorphs, but the exact position of this element in the skull differs between modern leporids and Paleogene genera discussed herein. In modern leporids it is bent ventrally together with the entire occipital region and part of the parietals, whereas in palaeolagines and Litolagus it is oriented much more horizontally. Thus, the right angle of the triangular form of the mastoid exposure is directed distad, with the mastoid process pointing ventrally with slight anterior inclination (figs. 25–26).

MANDIBLE: The mandibles of Litolagus molidens (AMNH FM 143955) are almost complete (fig. 28), the left one having a partly damaged condylar process. In AMNH FAM 105991 only mandible bodies with dentition are preserved, and the left mandible produced its impression in the underlying matrix. The whole mandible is larger than that of Chadrolagus emryi and Palaeolagus haydeni and more slender. Moreover, the mandible of Litolagus molidens has a higher and more delicate condylar process and slightly longer diastema (almost equal in length to the LTR) than in Palaeolagus and Megalagus (fig. 21, table 2). The ratio of the LTR to the LDL in Litolagus is closest to that of Archaeolagus ennisianus (table 2).

The articular head in Litolagus molidens is wider and proportionally shorter than that of Palaeolagus. The coronoid process is of similar size in both species, although in Litolagus it seems to be less projecting than in Palaeolagus. The coronoid crest is strongly developed and similar in both species, although the mandibular notch is slightly deeper and more flared out in Litolagus than in Palaeolagus.

The angular process in Litolagus is visibly longer than in Palaeolagus, protruding more distally The distal margin of the mandibular ramus is incised more deeply, producing an angular process that is more pointed. The masseteric line has a similar course and the masseteric fossa in Litolagus is similarly formed as in Palaeolagus, occupying almost the whole lateral area of the mandible ramus. The anterior margin of the masseteric fossa has a distinct triangular tubercle also seen in Megalagus and Mytonolagus. The tubercle is located more posteriorly in Litolagus (under m3) than in Palaeolagus (under the m2/m3 contact).

The ventral margin of the mandible body below m3 is not as concave as in Palaeolagus; thus, it seems to gently increase in depth distally, whereas in Palaeolagus haydeni the greatest depth is at the level of p4. The lower incisor in Litolagus molidens reaches only the anterior part of p4, similar to that in Chadrolagus emryi and Limitolagus roosevelti, and unlike in Palaeolagus haydeni, P. temnodon, and P. burkei, where the di2 reaches to the level of m1. Additionally, the lower incisor is less procumbent in Palaeolagus, erupting from the alveolus at a greater angle than in Litolagus, which has relatively less recurved incisor, although its distal end is slightly bent upward, so it is embedded farther from the ventral margin of the mandible than in Limitolagus roosevelti. The p3 in Litolagus is erect in its alveolus and its distal end produces a gentle impression at the lateral side of the mandibular body. In Palaeolagus the p3 is gently bent anteriorly and no such impression can be recognized.

TABLE 8.

Lower tooth measurements (mm) of Limitolagus roosevelti.

The arrangement and number of the mental foramina is variable in lagomorphs, especially in palaeolagines. Whereas the first (anterior) mental foramen is located at the diastemal part of the mandible body, the second can be either clearly developed in the area near the p3–p4 alveolus or it can be indistinguishable from other perforations on the lateral surface of the mandibular body. In Litolagus molidens (AMNH FM 143955), the first mental foramen is located at twothirds of diastema length, more anteriorly and medially than in some specimens of Palaeolagus haydeni (e.g., AMNH FM 143956), whereas in AMNH FAM 105991 it is oriented more laterally on the mandibular body. The second mental foramen appears below the talonid of p3 in AMNH FM 143955, accompanied by a few smaller fenestrae and is completely obscured by a series of crude ornamentations spreading between the distal part of the diastema and the alveolus of p4 at the ventrolateral part of the mandible body.

DENTITION: The upper incisors (dI2) are relatively wider and flatter than in Chadrolagus emryi, Palaeolagus haydeni, and P. temnodon, but relatively less flattened than in Archaeolagus ennisianus. Also, the mesial lobe in Litolagus is wider (almost equal in width to the lateral lobe) than in the other species. The lower incisor (di2) resembles that of Palaeolagus haydeni, in that it is relatively narrow and high, but with a markedly wider anterior margin.

The upper and lower teeth of Litolagus molidens (figs. 31–32; tables 9–10) show fully hypsodont morphology and relatively early in ontogeny form the lingual enamel bridges, thus expressing some ontogenetic acceleration, in which respect they resemble Chadrolagus (figs. 6, 8). Nevertheless, the development of the occlusal enamel pattern, especially of P2, P3, and p3 differentiates this species from Chadrolagus and the fully hypsodont Palaeolagus species.

The early stages of wear of P2 are not known, but at a moderate stage of wear it is bilobate, showing no trace of the buccal lobe. The P2 of Litolagus is more oval than those of Chadrolagus or Palaeolagus, with the longer axis positioned transversely, as in Limitolagus (fig. 20) and P. burkei. The lingual lobe is full and round, larger than the buccal one. Heavily worn specimens lose the anterior reentrant and their occlusal surface becomes completely smooth and gently concave (fig. 31).

The P3 is relatively slender and compressed anterodistally in comparison with Chadrolagus or Palaeolagus haydeni. Its buccal side is oblique and the distalmost part forms a long projection. In Litolagus, as in Chadrolagus, there is an asymmetrical structure of the P3 shaft with pronounced distal ridge, homologous to the distal root in the semihypsodont species. The P3 closes and loses a crescent very quickly but preserves open and deep hypostriae for a long time in ontogeny. The crescent is already lost in the holotype specimen (LACM-CIT 1568) described by Dawson (1958), which is a subadult specimen with long bones showing incomplete fusion of the epiphyses. All other specimens of Litolagus have open and deep hypostriae, which are similarly deep on P3–M2, resembling in this respect Palaeolagus burkei and P. hypsodus rather than Chadrolagus, which maintains the deepest hypostria on P4 at comparable stage of wear (fig. 6). The M3 is slightly smaller than in Chadrolagus and Palaeolagus and its occlusal surface is rounder than in these two genera. The enamel pattern of M3 is completely obliterated in all known specimens of Litolagus (fig. 31; see also Dawson, 1958: fig. 15).

The lower dentition in two newly described skulls (AMNH FM 143955 and FAM 105991) does not depart in morphology from the holotype described by Dawson (1958: fig. 16). The p3 shows simple morphology, even in the holotype, which is a young individual (Dawson, 1958: 33). The tooth is relatively short in comparison to Palaeolagus or even Chadrolagus (figs. 15, 32, table 10). It has a single buccal reentrant filled with cement and no traces of the lingual reentrant or a lake. Dawson (1958) stated that the lingual reentrant is lost early in ontogeny, but in fact it is not certain that it was ever present, because in all known specimens no trace of it can be found. The trigonid of p3 in Litolagus molidens is round with a smooth outline and is markedly narrower than the talonid, which is distinctively short in comparison with other contemporary lagomorphs. The lingual margin of the tooth in the holotype bears a visible concavity flanked by an eminent metaconid anteriorly and well-developed entoconid distally (fig. 24E; see also Dawson, 1958: fig. 16). The new specimens described herein display slightly more advanced wear and this concavity was either completely or greatly eradicated, but the proportions of the trigonid and talonid remain the same (fig. 32).

FIG. 31.

Incisors and upper cheek teeth of Litolagus molidens, the Chadronian-Orellan transition, Wyoming. A, right tooth row (AMNH F:AM 105991) in occlusal view, Reno Ranch; B, right and partial left tooth rows (AMNH FM 143955) in occlusal view, Herman Wulff Ranch; right P3 (AMNH F:AM 105994) in buccal (C), occlusal (D), and ventral (E) views, Seaman Hills; F, left upper and G, left lower incisors of AMNH FM 143955.

The lower molariform teeth, p4–m2, are very similar and form thin enamel bridges very early in ontogeny, similarly to Chadrolagus and Limitolagus. The cement is well developed on both upper and lower teeth and fills the hypostriae and space between the trigonids and talonids.

The m3 of Litolagus molidens does not differ markedly from that of Chadrolagus, although it is more buccolingually extended and has a visibly wider trigonid (fig. 32). The m3 forms a wide enamel bridge and persists in this form for a long time, whereas its wear is marked only by the shallowing of the buccal reentrant, which does not form a lake as seen in Palaeolagus temnodon.

COMMENTS: Litolagus molidens is an extremely rare species, known only from a few specimens found in a limited area a few miles SE of Douglas, Wyoming (Dawson, 1958). The original description of the species included only two specimens, the holotype (LACM CIT 1568) and an adult skull FH 10283 (Dawson, 1958). Prothero and Whittlesey (1998) refer to five known specimens from the Lusk and Douglas area, but did not give a precise description of the material or collection numbers, although it is possible that they refer to the specimens collected by the Frick Laboratory of AMNH, which are revisited in this paper.

Both skulls assigned to Litolagus molidens, described herein (AMNH FM 143955 and F:AM 105991) come from sediments above the P.W.L., roughly marking the Eocene-Oligocene transition (Evanoff, 1990; Evanoff et al., 1992), and the same applies to both specimens mentioned by Dawson (1958). The only specimen known from a different horizon is an isolated P3 (AMNH F:AM 105994) found in Channel Quarry (SE Seaman Hills), which according to Skinner (MS) lies below the Chadronian-Orellan boundary. Prothero and Whittlesey (1998) stated that one of the specimens from the Douglas area attributed to Litolagus was found ca. 10 m below “the 5 tuff” correlated to P.W.L., but revision of the material excavated by the Frick Laboratory revealed that all specimens below P.W.L. from the Douglas vicinity were either Limitolagus, which otherwise is similar in size to Litolagus (tables 6–10), or Palaeolagus temnodon.

DENTITION

TABLE 9.

Skull and upper tooth measurements (in mm) of Litolagus molidens. For abbreviations, see figure 3.

TABLE 10.

Mandible and lower tooth measurements (in mm) of Litolagus molidens from Wyoming. For abbreviations, see figures 3 and 4.

CRANIUM

MANDIBLE

PHYLOGENETIC CONSIDERATIONS

The present work focuses on the North American lagomorphs near the Eocene-Oligocene transition and their interrelationships at the species level. A cladistic analysis (fig. 33) rejects the immediate relationship between Chadrolagus emryi and Litolagus molidens, doubted by Gawne (1978) but recently implied by Dawson (2008). Chadrolagus, Limitolagus, and Litolagus do not form a monophyletic group, despite similar dental adaptations that are due to the accelerated development of the occlusal pattern. These similarities include early closure of crescents, early obliteration of all occlusal structures on the upper teeth, except persistently open hypostria, and early formation of the lingual dental bridges on lower cheek teeth. Chadrolagus emryi and Limitolagus roosevelti seem to be closely related considering the dental characters (fig. 33), which might be misleading because of generally high homoplasy of these characters in lagomorphs and the lack of skull material for Limitolagus, which prevents incorporation of cranial characters into the data matrix. On the other hand, the detailed analysis of ontogeny of the dental pattern has proved a reliable tool for deciphering phylogenetic relationships in the case of some middle Eocene (Fostowicz-Frelik and Tabrum, 2009) and Miocene stem lagomorphs (López Martínez, 1986, 1989; Fostowicz-Frelik et al., 2012a), and the development of dental pattern in Chadrolagus and Limitolagus is strikingly similar (figs. 6, 8, 12, 19, 20). Furthermore, the present cladistic analysis places Chadrolagus emryi closest to the Palaeolagus temnodon-P. haydeni cluster recognized by Dawson (1958, 2008). This proximity of Chadrolagus to Palaeolagus was suggested by Gawne (1978), who stated that Chadrolagus seems “to be an offshoot of the evolutionary line leading to Palaeolagus….” Later, Emry and Gawne (1986) did not rule out that P. primus (recovered from the same sediments as Chadrolagus but preceding this species stratigraphically) could be tentatively considered ancestral to Chadrolagus, although they preferred a common ancestor near Mytonolagus wyomingensis. At present, the origins and possible descendant lineage of Chadrolagus are unknown, especially because this genus is rare and appears in the early Chadronian already showing a combination of primitive and derived cranial and dental characters, including fully developed hypsodonty. Interestingly, scarce dental remains of Duchesnean age from the Black Rabbit Locality at Diamond O Ranch in Montana show development of the occlusal surface of upper teeth similar to Chadrolagus (Fostowicz-Frelik and Tabrum, 2009: fig. 12). Moreover, the buccal roots are slightly weaker in these scarce specimens (attributed to ?Palaeolagus sp.) than in P. temnodon (Fostowicz-Frelik and Tabrum, 2009). Emry and Gawne (1986) mentioned that P. primus does not have a hypostrial lake on P3, the condition found also in Chadrolagus emryi. Thus, noting the similarities and earlier occurrence of Palaeolagus primus in the White River Formation at Flagstaff Rim, it seems plausible that Chadrolagus emryi and Palaeolagus primus share a common ancestor dating back at least to the Duchesnean.

The origins and relationships of Litolagus molidens are even more enigmatic than those of Chadrolagus. Litolagus is known only from a few specimens; however, thanks to the exceptionally well-preserved skull described herein, which yielded abundant data, its phylogenetic position can be established with certainty as an evolutionarily advanced species, more closely related to Miocene Archaeolagus than to the remaining Eocene and Oligocene stem taxa. Such a placement within the phylogenetic tree removes Litolagus from the Palaeolagidae, a group that at present contains most of the Eocene to Miocene stem lagomorphs (see Gureev, 1964; Erbajeva, 1988; Fostowicz-Frelik et al., 2012a, 2012b), and suggests its closest affinity to a more advanced family, Leporidae. Dawson (1958) noted that the skull structure of Litolagus molidens shows a mixture of primitive and advanced characters. The material described here allows for a more detailed analysis. The most evident feature is a highly progressive structure of the hard palate, which shows a relatively reduced palatine portion, mentioned already by Dawson (1958). Further, the skull has visibly stronger basicranial angulation than the other coeval species, and in general is higher, due to the enlargement of the orbital portion of frontal and uplifting of the skull roof. Dawson (1958) argued that the structure of the zygomatic arch, namely the masseteric spine at the maxillary zygomatic process, is primitive in Litolagus, in which it is not markedly extended laterally. However, the structure and extent of the masseteric spine is a highly variable character in lagomorphs and can differ within a genus (e.g., in Lepus, personal obs.). A relatively long and narrow (in ventral view) masseteric spine can be observed also in archaeolagines (e.g., Hypolagus beremendensis, see Fostowicz-Frelik, 2007) and some extant rabbits, e.g., the European rabbit (Oryctolagus cuniculus). It is difficult to establish whether this character in the extant species is plesiomorphic or not, as most Eocene taxa are known mainly from dental remains. Nevertheless, according to the present author's observations, the structure of the anterior part of the jugal arch in Litolagus molidens does not show a plesiomorphic character, especially when compared with Dawsonolagus (Li et al., 2007) or Desmatolagus (Lopatin, 1998; Meng et al., 2005). On the other hand, the parietal and occipital regions (especially the exoccipital structure) in Litolagus show plesiomorphic conditions. The parietals are positioned relatively horizontally and not bent ventrally as in extant leporids, and the occipital face is almost vertical, not inclined ventrally. This particular structure of the braincase is probably one of the most conservative morphological regions of the skull and is shared not only by the Paleogene stem lagomorphs and Litolagus, but also mimotonids, including their basalmost genus, Gomphos (Li and Ting, 1993; Asher et al., 2005: fig. 2A–B).

However, looking at derived cranial characters of Litolagus, such as the larger basicranial angle, shorter palate, wider choanae, and enlarged auditory bullae, it becomes obvious that they are adaptations to more open and arid habitats. The increased angulation, shorter palatal bridge, and relatively wider choanae have functional meaning, as they facilitate balancing of the head and ventilation (respectively) during locomotion, thus increasing the capability of an animal to withstand greater mechanical effort, resulting in quicker and/or prolonged running. Adaptations of that kind can be observed in most extant nonsylvan species of Leporidae (Fostowicz-Frelik, 2007). Additionally, the enlargement of the auditory bullae is a common adaptation among relatively small mammals that entered open environments (e.g., Cainotheriids; Theodor, 2010), and in the light of other Litolagus molidens adaptations it is the most plausible explanation.

The preliminary cladistic analysis of North American Paleogene lagomorphs shows extended paraphyly of some genera and a split within Palaeolagus, which disintegrates into at least two groups: the “old core” P. temnodon-P. haydeni (incidentally, the latter is the type species; Leidy, 1856) clade and P. burkei, which is separated from the former clade by Chadrolagus and Limitolagus (fig. 33). The detailed phylogeny of Palaeolagus is beyond the scope of this paper and will be investigated further. Nevertheless, distinctive structure of the skull and dentition of Palaeolagus burkei needs a brief comment. This species appears as the last of the analyzed taxa after the EoceneOligocene transition (Prothero and Whittlesey, 1998; Dawson, 2008). Although its dentition shows a simplified occlusal pattern similar to Litolagus molidens, the skull structure expresses quite different specialization displayed, again, in the inflation of the auditory bullae, but also in flattening of the whole skull. In general form and adaptations, the skull of Palaeolagus burkei resembles that of extant Ochotona, inhabitant of steppes and mountains and a frequent burrower (Nowak, 1999). This analogy is even more surprising when the dental characters are taken into account, as some structural and developmental features are also strikingly similar between these two very distant relatives. The dental features of Palaeolagus burkei can be characterized as pedomorphic, the trend opposite to that observed in Chadrolagus, Limitolagus, and Litolagus. The ontogenetic delay in molarization of P3 and no formation of the lingual enamel bridges on p4-m2 even in an advanced age parallel dental structures (nonmolarized P3 and lack of lingual bridges) in the crown ochotonids (Erbajeva, 1988). In any case, even the preliminary phylogenetic analysis presented herein shows the paraphyletic arrangement of stem taxa. P. temnodon-P. haydeni form the core of “typical” Palaeolagus, while Litolagus molidens, the most derived of the Paleogene lagomorph species, links to Archeaolagus ennisianus as a sister taxon. Whether this position reflects real relationships or choice of taxa, or is yet another result of the convergent evolution (and thus cranial and dental homoplasies) frequently observed in lagomorphs, remains to be seen.

FIG. 33.

Phylogenetic position of Limitolagus roosevelti, new genus and species within Paleogene North American lagomorph taxa.

DIVERSIFICATION OF NORTH AMERICAN LAGOMORPH FAUNA AT THE EOCENE-OLIGOCENE BOUNDARY

The Eocene-Oligocene transition about 34.0 Ma was the time of one of the most profound global climate changes (Zanazzi et al., 2007; Lear et al., 2008; Liu et al., 2009) during the entire Cenozoic, and major faunal turnover coincided with the Eocene-Oligocene boundary worldwide (e.g., Evanoff et al., 1992; Prothero and Emry, 2004; Kraatz and Geisler, 2010). In North America, the drying and cooling of the climate gradually turned moist forest habitats of subtropical character into broad-leaved deciduous woodlands, dry and open woodlands, bushlike environments of warm temperate zone, and finally the first Paleogene grasslands (Evanoff et al., 1992; Retallack, 1992; Prothero and Emry, 1996; Sheldon and Retallack, 2004; Woodburne, 2004; Meyer, 2005; Hembree and Hasiotis, 2007; Boyle et al., 2008; Leopold et al., 2008; Mathis and McFadden, 2010).

FIG. 34.

Lagomorph biostratigraphy in the Chadronian-Orellan sediment series of the studied areas, Montana (Renova Formation) and Wyoming (White River Formation). Data after Dawson, 2008 and Janis et al., 2008; correlation after Prothero and Emry (2004: fig. 5.2). Solid lines: actual presence in studied localities; dashed lines: ranges inferred from data from other parts of the White River Group. LADs of Megalagus turgidus (1), Palaeolagus intermedius (3), and P. burkei (4) established on late Whitneyan-earliest Arikareean (Wh2-Ar1) in Harris Ranch LF (= Harris Ranch Badlands Unit E); LAD of Palaeolagus haydeni (2) noted in Lower and Upper Fauna III (= lower Sharps Formation), early Arikareean (Ar1). Presence of P. burkei in E Wyoming (Whitneyan deposits of Sherrill Hills) after Dawson (2008), FAD of this species in the White River Group falls to the mid-Chron C12r (after Prothero and Whittlesey, 1998). Abbreviations: EOB, Eocene-Oligocene boundary; Whit., Whitneyan.

The early Chadronian to early Orellan interval, as documented by one of the most complete late Eocene to early Oligocene vertebrate records yielded by the sediment series of the White River Group, is the time of noticeable high biodiversity among Lagomorpha in North America (fig. 34; Evanoff et al., 1992; Prothero and Whittlesey, 1998). The extinction followed by the radiation events can be noted for lagomorphs near the Eocene-Oligocene boundary (Korth, 1989; Prothero and Whittlesey, 1998), similar to those observed for other herbivorous mammals (Janis et al., 2008; Mathis and McFadden, 2010).

The first species disappearing from the fossil record is Chadrolagus emryi, one of the smallest and first fully hypsodont, of North American Paleogene lagomorphs (Gawne, 1978). Taking into account the material described herein, the stratigraphic range of this species covers the early to late Chadronian with the latest occurrences in 10N highway (Dunbar Creek Formation) and Little Spring Gulch (Cook Ranch Formation) localities in Montana. It appears that despite fully developed hypsodonty, a specialization typical for grazers, the dentition of Chadrolagus was probably not well adapted to feeding on harder, more xerophilous vegetation, as its occlusal enamel pattern was obliterated early. Future microwear study can shed light on this hypothesis. Thus, it can only be said that apparently Chadrolagus was not well suited to increasingly drying habitats, either because of dental adaptations or other possibly paleoecological factors such as predation or competition with other species, such as Palaeolagus temnodon, as suggested by Gawne (1978). Chadrolagus emryi disappears from the fossil record in Flagstaff Rim a little above Ash G bed and well before the change of the sedimentation type from fluvial to eolian, noticeable near Ash I bed (Evanoff et al., 1992), which implies greater susceptibility of this species to subsequent change. Although Evanoff et al. (1992) mentioned that dry land deposition had a west to east progression, the more localized nature of sedimentation in the Montana basins and the higher elevation could be mitigating factors, producing more favorable conditions for Chadrolagus, which found a temporary refuge there. Similar buffering processes are known to have operated at the Eocene-Oligocene transition in the northeastern Colorado White River Formation, as detected in paleosols (Hembree and Hasiotis, 2007).

Two other Chadronian lagomorphs, Palaeolagus temnodon and Megalagus brachyodon, outlasted Chadrolagus, reaching or even crossing (P. temnodon) the Eocene-Oligocene boundary. However, precise LAD of P. temnodon is obscured by the presence of Palaeolagus hemirhizis. Prothero and Whittlesey (1998) presumed that the latter species is a nomen dubium and results from mixing of P. temnodon and P. haydeni, which are closely related and morphologically similar species (Dawson, 1958: fig. 26) in early Orellan localities. They argued that, notwith standing similar size of P. temnodon and P. haydeni, they can be easily recognized by the presence or absence of roots in their upper premolars and molars (Prothero and Whittlesey, 1998). In any case, despite the putative validity of Palaeolagus hemirhizis, the “hemirhizis” zone in Orella A demonstrates that there is a species close to P. temnodon, with not fully hypsodont teeth, in the early Orellan. Probably, those remains belong to P. temnodon, as no actual difference between P. temnodon and the rooted specimens of P. hemirhizis was provided (Korth and Hageman, 1988). I presume that P. temnodon survived major environmental change at the Eocene-Oligocene boundary. The same cannot be said of Megalagus brachyodon, which apparently became extinct prior to the Eocene-Oligocene transition (Prothero and Whittlesey, 1998). However, it is sometimes problematic to distinguish this species from Megalagus turgidus or Palaeolagus intermedius (especially if the roots are hidden in alveoli), both thought to be of Orellan origin (Dawson, 1958; Prothero and Whittlesey, 1998). The stratigraphic range of Megalagus brachyodon is well recognized in the Flagstaff Rim area, where it mostly overlaps with Chadrolagus emryi, but disappears ca. 25 feet lower in the section than Chadrolagus (Gawne, 1978: fig. 7; Emry, 1992). Dawson (1958, 2008) suggested that although Megalagus and Palaeolagus are not immediately related, the former mimics dental adaptations of the latter. The development of the occlusal surface shows a similar sequence of structures, including early formation of the hypostrial lakes and long persistence of the crescents (Wood, 1940; Dawson, 1958). Such dental adaptations suggest a browser type of foraging, as the occlusal surface, similar to the condition in Palaeolagus temnodon, gains a “bunodont” character due to the presence of persistent lakes (at last two of them: hypostrial and crescential). This pattern stands in opposition to the “lophodont-like” type of typical grazers, imitated by Chadrolagus, Limitolagus, and Litolagus, which all have their upper tooth enamel organized in loops; in this case a single loop of relatively deep hypostria. Nevertheless, where Megalagus brachyodon failed, Palaeolagus temnodon succeeded, although it did not last beyond the “hemirhizis zone,” roughly correlated with the Orella A Member (Korth, 1989), and was replaced by Palaeolagus haydeni, which shows a similar occlusal pattern but is fully hypsodont (Wood, 1940; Dawson, 1958, 2008).

Following the evolutionary patterns in North American lagomorphs at the Eocene-Oligocene boundary two general observations can be made: first, disappearance of the Chadronian semihypsodont species and second, advent of many fully hypsodont species. In the case of the semihypsodont species, basically they were replaced by closely related ecological counterparts: Palaeolagus temnodon by P. haydeni, and Megalagus brachyodon by M. turgidus (see Dawson, 2008). A different situation is presented in the fully hypsodont genera, especially those showing acceleration in the development of the occlusal dental pattern in comparison with both species of Megalagus and all coeval Palaeolagus. Three species have a very similar pattern of dental wear: Chadrolagus emryi, Limitolagus roosevelti, and Litolagus molidens. From the standpoint of functional morphology, their dental characters imply grazer-type dietary adaptations, as they parallel similar lophodont dental structures of some grazers and most extant leporids, which feed mainly on grasslike vegetation and herbs (Nowak, 1999). However, taking a closer look at the extant hares and rabbits, it is highly probable that the relatively simple enamel pattern in Chadrolagus, Limitolagus, and Litolagus was insufficiently abrasive for new post-Eocene vegetation. Today, the most extreme dental adaptations to a hard-leaved diet are seen in the Amami rabbit (Pentalagus furnessi), which feeds on bamboo shoots (Nowak, 1999). This species has a very complicated enamel pattern with all folds strongly crenulated; the crenulation is also quite considerable in some other extant genera, e.g., Lepus or Pronolagus. However, in the three Paleogene taxa discussed above it is comparatively very simple. Additional argument confirming this hypothesis is the strong wear of the occlusal surfaces, which can be observed in relatively young specimens, even subadults in the case of Litolagus molidens, which owes its specific name to that particular character (Dawson, 1958).

The lack of evolutionary success in Chadrolagus, Limitolagus, and Litolagus is further evidenced by their restricted stratigraphic and geographic ranges, especially in the case of Limitolagus and Litolagus (known only from the Douglas and Lusk areas, Wyoming). Chadrolagus, the most widespread of them, went extinct in the late Chadronian. Limitolagus is not known later than the Eocene-Oligocene boundary and Litolagus was reported from the sediments above P.W.L., with the exception of an isolated P3 found 10 feet below P.W.L. in Channel Quarry, Seaman Hills, Niobrara Co. Furthermore, none of them left identifiable descendants.

The picture of lagomorph faunal turnover at the Eocene-Oligocene boundary in North America is completed by the appearance of Palaeolagus burkei in the Orella C horizon of the Orella Member, Brule Formation (Prothero and Whittlesey, 1998). This relatively small species mimics some ochotonid adaptations in skull structure, which suggest similar lifestyle and possible burrowing adaptations not surprising in increasingly open environments. Interestingly, there is no decrease in lagomorph species richness across the Eocene-Oligocene boundary in North America. Five species are known from the latest Chadronian, and the same number from the earliest Orellan.