Introduction

In addition to their historical importance as domestic animals (horses), or more recently as icons of endangered wildlife (rhinos), perissodactyls, the order of mammals containing modern tapirs, rhinos, and horses, are noteworthy for their high levels of past diversity, which contrasts with their diminished modern diversity. With the possible exception of horses (Simpson, 1951), interest in perissodactyl evolution waned in the mid–20th century and despite renewed interest in perissodactyl systematics (MacFadden, 1992; Hooker, 1994; Cerdeño, 1995; Holbrook, 1999; Froehlich, 1999, 2002; Antoine, 2002; Lucas and Holbrook, 2004; Prothero, 2005), various groups remain in a state of neglect. Among the most neglected perissodactyls are the Brontotheriidae, an extinct family that is confined to the Eocene of North America, Asia, and Europe. Brontotheres are notable for having evolved bony frontonasal protuberances (or “horns”) and body sizes approximating those of extant rhinos and elephants. Brontotheres are one of the more diverse groups of ungulates and during the Eocene they exceeded other perissodactyl families in terms of overall diversity. Brontotheres are among the most abundantly represented ungulates in Eocene fossil-bearing terrestrial deposits. The bulk of the known brontothere fossil record is from central Asia and western North America (Osborn, 1929a, 1929b; Granger and Gregory, 1943; Yanovskaya, 1980; Wang, 1982). Brontothere records are fewer and more fragmentary in other regions, but it is evident that this family had an essentially Holarctic distribution with the exception of Western Europe. Brontothere fossils have also been recovered from eastern Europe (Nikolov and Heissig, 1985; Lucas and Schoch, 1989a), eastern Russia (Yanovskaya, 1957), Kazakstan (Yanovskaya, 1953; Emry et al., 1998; Mihlbachler et al., 2004a), Pakistan (Dehm and Oettingen-Spielberg, 1958; West, 1980; Kumar and Sahni, 1985), southeast Asia (Pilgrim, 1925; Colbert, 1938; Holroyd and Ciochon, 2000; Tsubamoto et al., 2000; Qi and Beard, 1996), Korea (Takai, 1939), Japan (Miyata and Tomida, 2003), the southeastern United States (Gazin and Sullivan, 1942), and in northern regions of Canada (Eberle and Storer, 1999; Eberle, 2006).

Brontotheres have humble beginnings as relatively rare, small, unspecialized animals in the early Eocene (e.g., Eotitanops), but quickly radiated into an assortment of species that developed unique and bizarre skulls. Some achieved enormous body sizes in comparison to other contemporary perissodactyls such as horses, tapirs, and rhinos, which, by comparison, were relatively small throughout the Eocene (fig. 1). With massive bodies and robust graviportal limbs, the largest brontotheres superficially resemble rhinos. However, brontothere horns were made of bone, not keratin; they retained four front digits, whereas modern rhinos have only three. Additionally, brontotheres show subtle differences in their overall limb proportions in comparison to modern rhinos (Mihlbachler et al., 2004a).

Figure 1

Reconstructions of assorted North American brontotheriids and the brontotheriioid Lambdotherium popoagicum from Osborn (1929a). Synonyms used in this paper are as follows: Brontotherium platyceras and Brontotherium leidyi ( =  Megacerops coloradensis sensu Mihlbachler et al., 2004b), Manteoceras manteoceras ( =  Telmatherium validus), Dolichorhinus hyognathus ( =  Dolichorhinus hyognathus), Mesatirhinus petersoni ( =  Mesatirhinus junius), Palaeosyops leidyi ( =  Palaeosyops robustus sensu Gunnell and Yarbrough, 2000), Eotitanops princeps ( =  Eotitanops borealis sensu Gunnell and Yarbrough, 2000), and Eotitanops gregoryi ( =  Eotitanops minimus sensu Gunnell and Yarbrough, 2000).

The exact phylogenetic position of brontotheres within the Perissodactyla is unresolved. Brontotheres are traditionally grouped within the suborder Hippomorpha along with horses, palaeotheres, and sometimes chalicotheres (Wood, 1937; Simpson, 1945; Scott, 1941; Radinsky, 1969). Froehlich (1999) suggested that brontotheres are actually a subclade of the Palaeotheriidae. Others have included them with tapirs and rhinos in the Ceratomorpha (McKenna and Bell, 1997), while others still have excluded them from other major perissodactyl clades (Prothero and Schoch, 1989) and considered them the sister taxon of all other perissodactyls (Hooker, 1984), or placed them at an unresolved polytomy with other major radiations of perissodactyls (Janis et al., 1998).

Characters that typify brontotheres include an elongated postorbital cranium and an abbreviated face (Osborn, 1929a; Mader, 1989, 1998). In some species (e.g., Duchesneodus uintensis) this condition is extreme, with the orbits positioned near the very front of the elongate skull. Some brontotheres evolved extremely dolichocephalic (elongated) skulls (e.g., Dolichorhinus), while others evolved massive laterally paired sexually dimorphic bony horns composed of the frontal and nasal bone, and deeply concave saddle-shaped skulls with grossly expanded zygomatic arches (e.g., Megacerops). (While recognizing that the various cranial appendages seen in ungulates, such as giraffe ossicones, cervid antlers, bovid horns, rhino horns, and brontothere frontonasal protuberances are non-homologous, as a matter of convenience I adopt a generalized usage of the term “horn” for cranial appendages of brontotheres and other ungulates.) Brontothere horns resemble the secondary sex characters (e.g., horns, tusks) commonly seen in many sexually dimorphic extant ungulates and seem to suggest a polygynous and possibly gregarious form of sociality (Jarman, 1983; Loison et al., 1999). Other brontotheres, such as Embolotherium, had acquired a single, massive battering ram–like process extending upward from the face, suggesting expanded and highly modified narial morphologies (Mihlbachler and Solounias, 2004) (fig. 2). Though superficially similar to rhinos, brontothere horns are composed of bone, not keratin, and therefore preserve as fossils, unlike rhino horns. Brontothere horns are most similar to the ossicones of modern giraffes; they lack grooves for nutrient blood vessels, tend to show evidence of secondary bone grown (probably due to head clashes), and were likely to have been covered by skin (Mihlbachler et al., 2004b).

Figure 2

Reconstructions of the heads of central Asian brontothere species. (A) Metatitan primus, (B) Embolotherium andrewsi, (C) and Embolotherium grangeri.

Despite the variety of bizarre head shapes seen among brontotheres, this group is most easily characterized by its distinctive upper molar pattern, which consists of a W-shaped ectoloph that is at least half the labiolingual width of the entire upper molar crown, with a strong parastyle and mesostyle, and isolated lingual cusps (protocone and hypocone). Transversely (labiolingually) oriented upper molar lophs (protoloph, metaloph) and other cusps (paraconules and metaconules) are either vestigial or absent (Osborn, 1929a; Radinsky, 1969; Mader, 1989, 1998). The distinctive brontothere molar represents a crown-type that is completely nonexistent today, although a similar kind of bunolophodont dental morphology evolved in another extinct perissodactyl family, the Chalicotheriidae (Radinsky, 1969; Coombs, 1998). Brontothere teeth have been previously characterized as weak or mechanically inferior (Osborn, 1929a; Mader 1998) largely because they did not evolve extreme hypsodonty unlike many other lineages of ungulates (e.g., horses). Functionally, brontothere dentitions have been described as having a dual function of shearing (ectoloph) and crushing (lingual cusps) (Osborn, 1929a; Radinsky, 1969; Mader, 1998). This bunolophodont dental morphology has been interpreted to indicate a diet of selective folivory/frugivory (Janis, 2000). However, the immense size of many species suggests that at least some species must have been nonselective feeders, despite their dental morphology. Indeed, microwear patterns of brontothere molars suggest a strictly folivorous diet that is similar to that of the black rhinoceros (Diceros bicornis) or moose (Alces alces), both very large herbivores with diets consisting mostly of leaves and other fibrous woody material (Mihlbachler and Solounias, 2002).

The brontothere dental battery is not evolutionarily static despite notions to the contrary (Radinsky, 1969); early species had short crowns and thick enamel, however a transformation from this thick-enameled brachydont type of tooth to a more hypsodont and thin enameled type of molar did occur among brontotheres (fig. 3). Functionally, this transition seems related to more emphasis being placed on the vertical shearing action of the labial wall of enamel of the ectoloph. Surprisingly, microwear patterns do not suggest a significant change in diet associated with this morphological transformation (Mihlbachler and Solounias, 2002). Other conspicuous dental transformations in brontotheres include the reduction of the anterior dentition to a vestigial state and the semimolarization of the premolars (Osborn, 1929a; Radinsky, 1969).

Figure 3

Anterior views from Osborn (1929a) of (A) the relatively plesiomorphic upper molar of Palaeosyops, and (B) the relatively apomorphic upper molar of the Brontotheriine, Megacerops coloradensis (sensu Mihlbachler et al., 2004b).

The majority of brontotheres (all except Eotitanops and Palaeosyops) have the more advanced molar condition. One species, Bunobrontops savagei, is clearly an intermediate form; it retains the very brachydont crown, but shows some derived molar characters. In an earlier analysis of brontothere phylogeny, those brontotheres, including B. savagei, showing derived molar characters related to this functional transformation form a robust and well-supported clade (Mihlbachler 2003a). In this paper, I apply the subfamily name, Brontotheriinae, for this clade of brontotheres showing advanced dental characters. More formal definitions of Brontotheriidae, Brontotheriinae, and other higher brontothere taxa, are described near the end of this paper, following the phylogenetic analysis.

For the most part, brontotheres have been neglected by systematists and other paleontologists, despite their apparent taxic diversity, their large size, wide variety of appealing cranial morphologies, their conspicuous presence among the exhibits of most major natural history museums, and the overwhelming number of storage units in many museum collections that are filled with brontothere fossils. In contrast to the intense interest in brontotheres showed by early paleontologists such as J. Leidy, E.D. Cope, O.C. Marsh, and particularly, H.F. Osborn, (see review in Osborn, 1929a and in Prothero and Schoch, 2002) there is a dearth of recent literature on the systematics, phylogeny, or overall paleobiology of the Brontotheriidae. Nonetheless, Osborn's (1929a) exhaustive monograph on brontotheres, which remains the primary reference for this family, is almost universally recognized as deeply flawed, with severely oversplit alpha-level taxonomy, paraphyletic higher taxa, and discredited theories regarding evolution (Scott, 1941; Radinsky, 1969; Prothero, 1994; Prothero and Schoch, 2002; Alroy, 2003). Nevertheless, despite the prevalence of brontotheres in Holarctic Eocene mammal faunas, little effort has been focused on rectifying the seriously flawed state of brontothere systematics. Unfortunately, subsequent brontothere taxonomists such as Granger and Gregory (1943) and Yanovskaya (1980) used methods similar to Osborn's and only compounded the problem by erecting more problematic taxa and generating questionable phylogenetic hypotheses. In the recent literature, only a few short papers on the genus-level taxonomy of North American brontotheres (Mader, 1989, 1998) and on the species level phylogeny of Asian horned brontotheres (Mihlbachler et al., 2004a) have made serious attempts to rectify the chaotic state of brontothere phylogenetic systematics. However, the main task of overhauling the species taxonomy of brontotheres and developing a cladistically derived hypothesis that includes both Asian and North American species has not been accomplished until now.

Brontothere Research in the Osbornian and Post-Osbornian Eras

The early history of brontothere research in North America is recounted by Osborn (1929a) with additional information in Prothero and Schoch (2002), Mihlbachler et al. (2004b), and Lucas (2004), and is only briefly recounted here. Brontothere fossils were among the first fossil vertebrates from the western United States to be described. Joseph Leidy was the first to provide scientific descriptions of brontothere fossils, although the material he examined was mostly fragmentary and did not allow him to develop of good understanding of what a brontothere was or looked like. However, Cope and Marsh were the first to examine entire skulls, and, in the case of Marsh, entire skeletons (Marsh, 1889). Marsh (1873) first recognized brontotheres as perissodactyls, coined the term Brontotheridae (later amended to Brontotheriidae) and wrote papers outlining the principle characteristics of this group (Marsh, 1874, 1876). Marsh's concept of the Brontotheriidae was originally confined to horned varieties of the White River Formation of North America; the earlier hornless varieties known to him were referred to as the “Limnohyidae” Marsh (1875), although he did observe a phylogenetic connection between the earlier hornless species and the later horned species via the discovery of Diplacodon elatus, an essentially intermediate form between the horned and hornless varieties that were known at the time (Marsh, 1875). In addition to Brontotheriidae Marsh (1873) and Limnohyidae Marsh (1875), several alternative familial names were coined, including Titanotheriidae Flower (1876), Menodontidae Cope (1881), and with Osborn unknowingly coining a term twice, Titanotheriidae Osborn (1889). Cope (1887) and Earle (1892) presented early ideas about brontothere phylogeny, although the diversity of brontotheres was not well understood at the time.

As a New York social elite, founder of the Department of Mammal Paleontology (later to be expanded to include all vertebrates), and, thereafter, president of the American Museum of Natural History, Henry Fairfield Osborn devoted the extensive resources that he had at his disposal to amassing one of the world's most scientifically valuable collections of fossil vertebrates, including the world's largest and most comprehensive collection of brontotheres from North America and Asia. Osborn spent a large portion of his scientific career studying brontotheres (Osborn et al., 1878; Scott and Osborn, 1887; Osborn and Wortman, 1895; Osborn, 1895, 1896, 1897, 1902a, 1908a, 1913, 1914, 1916, 1919, 1923, 1925, 1929b) and formulating theories on evolution and extinction derived from these studies (Osborn, 1902b, 1902c, 1908b, 1911, 1931, 1934). Many early paleontologists, particularly those who worked on large vertebrates, are regarded today as poor taxonomists with the tendency to oversplit taxa (Alroy, 2003). Osborn is no exception, but he stands out because he wrote extensively on his own evolutionary theories. It is therefore possible to understand how his evolutionary views influenced his taxonomic decisions, which reveals how fundamentally incompatible his work, and that of those who followed him, is with current phylogenetic theory and practice.

Osborn's ultimate work on brontotheres (or “titanotheres” as Osborn and many others have called them) culminated in his massive two-volume monograph, “Titanotheres of Ancient Wyoming and Nebraska” (Osborn, 1929a). The monograph largely represents collaboration between William King Gregory and Osborn himself. The titanothere monograph is exhaustive in scope, and includes a meticulous recounting of the history of early research, the original descriptions and figures of all previously named species (as of about 1920, when the monograph was mostly completed), complete descriptions of skulls, dentition, and postcrania of every brontothere known at the time, biostratigraphic summaries of the major fossil bearing strata of the North American Tertiary, and a preliminary summary of Asian brontotheres that were being recovered by the then ongoing central Asiatic expeditions of the American Museum of Natural History. Moreover, hundreds of pages are devoted to general methodological principles, biomechanics, soft tissue reconstruction, functional morphology, adaptation, paleoecology, evolutionary theory, and extinction theory.

Initially, Osborn (1896) had adopted a straightforward “monophyletic” interpretation of brontothere phylogeny that is roughly compatible with more recent evolutionary theory in the sense that organisms with similar characters are initially hypothesized as closely related. However, Osborn's evolutionary theories, which, he claimed, were largely developed through pure induction (Osborn, 1930), evolved though the years (Osborn, 1902c, 1421908, 1911). In 1914, Osborn proposed his “polyphyletic” hypothesis of brontothere phylogeny, by which he recognized eleven subfamilies. In Osborn's polyphyletic interpretation, which was retained in the 1929 monograph, evolution was thought to be driven by an internal driving force (a hereditary germ) that caused lineages to evolve in parallel, thus resulting in groups with phylogenetically disparate origins giving rise to nearly identical, but distantly related taxa. Osborn drew analogies between the evolution of lineages (which he called “aristogenesis”) with ontogenesis and related extinction to ideas of overadaptation or racial senescence. Osborn thought that a variety of processes caused evolution, including saltation, Darwinian natural selection, and germ-motivated orthogenesis. However, he dismissed most of these factors, including natural selection, as insignificant evolutionary driving forces and instead considered the germ the dominant evolutionary force, driving change in a predetermined direction.

To Osborn, evolution consisted of two types of “biocharacters”, rectigradations (new characters) and allometrons (changes in proportions). In practice, Osborn (1929a) was preoccupied with locating “rudimentary” characters, particularly horns and tooth cusps and documenting changes within perceived lineages in the proportions of dentition, skulls, and limb segments via a series of indices, such as the cranial index (skull length/skull width). By his admission, the rudimentary characters that were deemed important rectigradations were often imperceptible on actual specimens. Osborn (1929a) described and labeled rudimentary horns and cusps on many of the specimens figured in his monograph, diagnosed taxa, and identified perceived lineages with these structures, although in many cases direct examination of the fossils reveals no such clear structures. Perceived lineages were identified not only by their characters, but also by rates of “phylogenetic progress” as judged by changes in proportions. Sister lineages would continue to evolve in parallel because of their shared evolutionary destiny, but at different rates. Perceived lineages could be identified by their rates of evolutionary progress. In Osborn's species concept, species were arbitrary divisions along a path of germinally motivated evolutionary progress, so that nearly every minor variant would be labeled a new species. Osborn (1929a) paid lip service to recognizing sexual dimorphism, ontogenetic change, and the effects of taphonomic distortion, and he attempted to determine the sex of nearly all the specimens that he described. However, minor variations in characters that were to be deemed by him to be phylogenetically important, particularly horn, dental, and cranial index characters, were considered to indicate separate species or even separate lineages. For instance, based on Osborn's methods, minor variations in characters that were perceived as phylogenetically significant among specimens from the same stratigraphic level would indicate multiple lineages evolving in parallel at separate rates. The possibility that such differences were due to intraspecific variation was not generally considered.

Like most other paleontologists whose work preceded the “Modern Synthesis” of evolutionary theory that took place in the 1930s and 1940s, Osborn's taxonomy and phylogenetic hypotheses, which are wedded to discredited evolutionary theories, can be rejected outright. However, the problem generated by Osborn's work is compounded by the fact that the methods and results presented in his 1929 monograph have been used, often uncritically, by post–“Modern Synthesis” paleontologists as the primary reference work for nearly all subsequent research on brontotheres, leading to equally dubious results by others who have dabbled in brontothere taxonomy and other aspects of paleobiology in the post-Osbornian era. Unfortunately, Osborn's (1929a) practice of figuring, describing, and placing emphasis on “rudimentary” structures that are not always clearly present on the specimens has misled several researchers (Lane, 1932; Wilson, 1977; McCarroll et al., 1996a). For example, Wilson (1977) named a new species of horned brontothere, Sthenodectes australis. The referral of this new species to the genus Sthenodectes is clearly erroneous and was based on Osborn's (1929a) inaccurate description of the holotype skull of S. incisivum (CMNH 2398), which he claimed had a horn (and even labeled a horn on his figure of that specimen), even when there is plainly no such structure on the actual specimen. No specimen of Sthenodectes incisivum has a horn, and Wilson's (1977) horned species is actually more comparable to Protitanotherium, an early horned brontothere (see remarks under Protitanotherium emarginatum).

William Berryman Scott (1941), a colleague and lifelong friend of Osborn's, rejected Osborn's hypersplit taxonomy outright, but did not provide a definitive solution to the brontothere problem. However, in 1945 Scott described a large sample of associated Duchesneodus uintensis material from a single fossil quarry. Peterson (1931) preliminarily reported this quarry. Scott readily accepted the differences in the specimens as intraspecific variation, ontogeny, and taphonomic distortion. The discovery of the presumably monospecific quarry sample of Duchesneodus uintensis and Scott's description of it is significant because it was among the first of several apparently monospecific brontothere death assemblages now known that appear to document intraspecific variability, including variability in the very characters that Osborn and others had used to split nearly every specimen into separate species (also see Lucas and Schoch, 1989b).

Few followed Scott's (1941, 1945) lead in recognizing the hypersplit nature of brontothere taxonomy. New species of North American brontotheres and minor taxonomic revisions have trickled into the literature over the years following Osborn's (1929a) monograph, which was essentially completed in 1920 (Cook, 1926; Peterson, 1931, 1934; Stock, 1935, 2011936; Russell, 1940; Stovall, 1948; West and Dawson, 1975; Wilson, 1977; Lucas and Schoch, 1982, 1989b; Mader and Alexander, 1995; Gunnell and Yarborough, 2000; Mader, 2000). Nonetheless, little attempt has been made to extensively revise Osborn's deeply flawed taxonomy, except for Mader's (1989, 1998) work on the higher-level taxonomy of North American brontotheres. However, Mader's revision of higher taxa (genera) was not substantiated by a published reworking of brontothere species taxonomy.

Although North American brontothere diversity had been well documented by the beginning of the 20th century, it was not until 1912 that the first brontotheres were reported from Asia, from the Pondaung Formation of Myanmar (Burma) (Pilgrim and Cotter, 1916; Pilgrim, 1925). In the 1920s and early 1930s, field crews from the American Museum of Natural History revealed a diversity of brontotheres in the Eocene deposits of central and southeast Asia (Osborn, 1923, 1925, 1929a, 1929b; Colbert, 1938; Granger and Gregory, 1938, 1943) that rivaled the known diversity of brontotheres in North America. Osborn (1923, 1925, 1929a, 1929b) published a series of papers with short and hurried descriptions of some of this material, in some cases actually describing and figuring the specimens while still in their plaster encasings. Despite his tendency to oversplit North American brontotheres, Osborn adopted the peculiar practice of lumping most of the newly discovered Asian brontotheres into various North American genera (Telmatherium, Manteoceras, Dolichorhinus, Protitanotherium), despite that fact that many of the Asian specimens obviously represent very different taxa. Following Osborn's death, Granger and Gregory (1943) extensively revised the taxonomy of central Asian brontotheres. Recognizing problems with Osborn's genus-level assignments of Asian brontotheres, Granger and Gregory (1943) erected many new genus names and assigned Asian taxa to new subfamilies. However, they adopted the general methodology of Osborn (1929a) and oversplit species based on barely perceptible differences in size, proportions, often when the differences are clearly attributable to taphonomic distortion or ontogeny. They perceived such differences as evolutionary stages, and, like Osborn, they did not consider the possibility that such minor differences simply represented random intraspecific variation, as Scott (1945) demonstrated shortly thereafter.

Although it has remained the primary reference for Asian brontotheres, the revision of central Asian brontotheres (Granger and Gregory, 1943) was clearly done in a hurried fashion and it has numerous problems. Some specimens were mistakenly referred to multiple species. Specimens that are obviously juveniles were apparently confused as adults, thus resulting in extremely erroneous taxon diagnoses. Species diagnoses tended to describe single specimens, while the lists of referred specimens provided by Granger and Gregory (1943) were carelessly assembled without consideration of whether they possessed the diagnostic features of the particular species to which they were assigned. In some instances, specimens that Osborn had correctly identified were actually misidentified by Granger and Gregory (1943). Granger and Gregory (1943) admitted that many of the species included in their revision were not even valid.

Subsequently, many paleontologists, primarily Chinese and Russian workers, have continued to name Asian brontothere taxa and a few have reevaluated Asian brontothere taxonomy on a limited scale (Dehm and Oettingen-Spielberg, 1958; Chow and Hu, 1959; Xu and Chiu, 1962; Chow et al., 1974; Yanovskaya, 1953, 1957, 1976; Dashzeveg, 1975; Wang, 1978, 1982; Miao, 1982; Ye, 1983; Qi, 1987; Qi and Beard, 1996; Wang, 1997; Holroyd and Ciochon, 2000; Mihlbachler et al., 2004a). However, the AMNH collection from the Central Asiatic Expeditions of the 1920s and 1930s is still the largest collection of Asian brontotheres in the world and contains the majority of holotypes, but for the most part this large but poorly documented collection has remained unutilized until recently (Mihlbachler et al., 2004a). The problematic work of Granger and Gregory (1943), which represents all that subsequent Chinese, Mongolian, and Russian workers had to go on, has only compounded the problem of Asian brontothere systematics. For instance, Yanovskaya (1980) contributed a major monograph on Mongolian brontotheres and described several new species. However, much of Yanovskaya's work is unreliable because she did not have direct access to the AMNH collection of Asian brontotheres. As a result the majority of the material reported in her monograph was misidentified (Mihlbachler et al., 2004a). Recently, Wang et al. (1999) provided a review of Chinese brontotheres, although no attempt was made to revise the taxonomy or evaluate phylogeny.

Granger and Gregory (1943) were the first to propose an explicit phylogenetic hypothesis of Asian brontotheres. They largely viewed the radiation of Asian brontotheres as separate from those of North America, but noted some similarities between North American and Asian brontotheres during what is now considered the middle Eocene. Yanovskaya (1980) and Wang (1982) proposed similar phylogenies for Asian brontotheres but used outdated methods similar to those of the 1940's. Mihlbachler et al. (2004a) recently developed a phylogenetic hypothesis of Asian horned brontotheres, but noted that Asian brontotheres, although commonly viewed as being a separate radiation, probably do not constitute a truly monophyletic assemblage. However, the interrelationships of North American and Asian brontotheres have not been rigorously examined, due to the lack of published phylogenetic analyses that incorporate taxa from both continents.

Methods of Taxonomic Revision

The aim of this work is to provide a cladistically generated phylogenetic hypothesis of the Brontotheriidae that includes all the known valid species. Because of the presently dubious nature of many brontothere species and lack of reliable published descriptions and figures, it was necessary to first evaluate the taxonomic validity of most species and provide adequate descriptions that (1) allow for the unambiguous diagnosis of terminal taxa (species), (2) examine patterns of intraspecific variability, and (3) facilitate the development of phylogenetic character data. Teeth, skulls, and mandibles are incorporated into this study. Postcrania were not used, largely due to a lack of associated cranial and postcranial material for most species. For the most part actual specimens or casts were examined with some exceptions. In several instances references were made to published figures and archived photographs of specimens that were lost or were otherwise inaccessible.

I was able to examine and describe the craniodental morphology of nearly every taxon that had ever been named. Two recently named species, Aktautitan hippopotamopus Mihlbachler et al. (2004a) and Eubrontotherium clarnoensis (Mihlbachler, 2007), were named and described using the methods consistent with those adopted here and are a part of this study, but they were published separately. As far as I am aware, all known Asian taxa are addressed here with the exception of a yet undescribed brontothere from Japan (Miyata and Tomida, 2003). All North American brontotheres were included in the phylogenic analysis, but some of them were not considered for taxonomic revision; these include: Eotitanops Osborn (1907), an unnamed species referred to as cf. Eotitanops by Eberle (2006), Palaeosyops Leidy (1870a), and Megacerops (sensu Mihlbachler et al. 2004b). Gunnell and Yarborough (2000) recently examined Eotitanops and Palaeosyops. Eotitanops and Palaeosyops, which are presumably the most basal brontotheres, have already been incorporated into general analyses of perissodactyl phylogeny (Froehlich, 1999; Lucas and Holbrook, 2004) and are not as problematic as most other brontothere taxa.

Late Eocene brontotheres (Megacerops sensu Mihlbachler et al., 2004b) primarily from the Chadron Formation have previously been assigned to a number of genera, including Menodus Pomel (1849), Titanotherium Leidy (1852), Megacerops Leidy (1870b), Brontotherium Marsh (1873), Symborodon Cope (1873b), Brontops Marsh (1887), Menops Marsh (1887), Titanops Marsh (1887), Allops Marsh (1887), Diploclonus Marsh (1890), Ateleodon Schlaikjer (1935), and other genera were not revised here. These brontotheres consist of hundreds of complete skulls that, according to Osborn (1929a), represent at least 37 species. Others consider them to represent far fewer than 37 species (Scott, 1941; Clark et al., 1967; Mader, 1989, 1998; Mihlbachler et al., 2004b). A revision of these particular brontotheres is beyond the scope of this paper but is preliminarily addressed in Mihlbachler et al. (2004b) and will be addressed more extensively in a future publication. Many of the invalid taxa, including nomina dubia are discussed with other taxa when appropriate, although there is a separate section on miscellaneous dubious species and other problematic taxa. Dubious taxa not considered in this paper are those that were already established as dubious by Osborn (1929a).

In the following revision, I provide only diagnoses and descriptions of species taxa. Higher taxa are discussed following the results of the cladistic analysis. Within the descriptions, intraspecific variation is meticulously documented using all of the specimens assigned to a particular species. Terminology for the upper and lower premolars and molars used in the descriptions is shown in figure 4.

Figure 4

Dental terminology for upper and lower cheek teeth. (A) Generalized brontothere upper molar, (B) Lambdotherium upper molar, showing characters that are either vestigial or absent in some or all brontotheriids, (C) generalized brontothere upper premolars (P1–P4), (D) brontothere lower molar. Figures of teeth are taken from Osborn (1929a).

It is traditional to provide a diagnosis for each species. While it is agreed that a species diagnosis should contain information that validates a taxon, there is not complete agreement on how this should be done (e.g., differential diagnosis versus apomorphy-based or autapomorphy-based diagnoses). It is obvious that the main function of the diagnosis, taxon validation, is also served by the coded character data (table 13). If the character data demonstrate a unique combination of states, then the species is validated. The traditional diagnoses given in this paper consist of two parts: (1) an abbreviated description pointing out some of the more obvious characters used in identification and (2) one or a few sentences that explicitly validates the species by using the minimal amount of information necessary to differentiate it from the most similar taxa.

Table 13

Phylogenetic character matrix of Brontotheriidae and outgroup taxa

Species Concept and Species Delimitation

Despite the large body of literature on theoretical species concepts, paleontologists do not usually explicitly state or explain their species concepts when diagnosing new taxa or revising previously named taxa or performing phylogenetic analyses with species. Because of the many deterministic and stochastic processes associated with speciation, species boundaries are often fuzzy and different methods are widely known to yield discordant results (Sites et al., 2004). Species-level taxonomy in paleontology is often viewed as a subjective process and species are often based on minor size differences between specimens or other minor differences found between specimens, despite the fact that size and shape should be expected to vary to some degree in all species. For instance, body-size fluctuations among modern species are clearly documented by the Pleistocene and Holocene record (e.g., Mihlbachler et al., 2002). Species boundaries are also sometimes placed between specimens or groups of specimens that are from different localities or different stratigraphic levels, even when there may be substantial size overlap between these samples and/or when other clear diagnostic characters are not present. The taxonomic relevance of such distinctions is ambiguous and taxa delineated in such terms are not necessarily suited to cladistic analysis if differing combinations of character states and/or other autapomorphies cannot characterize them.

Given the amount of taxonomic revision presented in this work, it seems appropriate to devote some discussion to species delimitation, how it was accomplished, and suggestions for future approaches. Because of the sometimes fuzzy nature of species boundaries and/or limitations in the available data, complete objectivity is not always possible when delimitating species; those represented by fossils tend to be more problematic due to the inevitably highly fragmented nature of at least some of the material. However, I was able to apply a reasonably objective set of operational criteria for delimiting species that is theoretically consistent with the phylogenetic species concept, which defines species operationally as the smallest diagnosable clusters of specimens (Cracraft, 1989; Nixon and Wheeler, 1990; Wheeler and Platnick, 2000). No two fossils are exactly alike, and those practicing taxonomy and/or working with species as terminals in phylogenetic analyses must ultimately draw a distinction between tokogenic character variation (sexual dimorphism, ontogeny, or other population variation) and phylogenetic character variation (species differences). Making this distinction is particularly problematic for fossil species because information on populations is severely limited. Nevertheless, considerable specimen-to-specimen variability was encountered among brontothere fossils. Treating all of these differences as taxonomically significant would have resulted in hundreds of obviously invalid species, with most represented by only a single specimen. This is essentially the same pitfall that misled earlier taxonomists into erecting unrealistic numbers of species.

To develop a set of taxon delimitation criteria, characters with tokogenic tendencies were identified and differentiated from species-level characters using a method comparable to population aggregation analysis (Davis and Nixon, 1992; Sites et al., 2004), but modified in ways to deal with the more restricted information about fossil populations. In population-aggregation analysis, (1) all individuals sampled from a population are assumed to be conspecific and (2) two or more populations are considered conspecific when the individuals drawn from these populations show identical character attributes, while populations that show one or more different character attributes are delimited. In true population-aggregation analysis, population boundaries are drawn around clusters of specimens (e.g., specimens collected from one locality), which are then assumed to be conspecific. In comparison to extant species whose individuals are sampled over a single geological interval, it is more difficult to draw population boundaries around clusters of fossils, even if they occur in a single geographic locality because the specimens are often separated by geological time. Moreover, many of the important fossil specimens are in older museum collections and lack adequate locality and stratigraphic data. In these instances, the degree to which such specimens are separated by time is very poorly constrained. Nonetheless, it is important to consider all specimens that show different combinations of character attributes. However, the addition of the time dimension increases the difficulty of determining whether differences between individual fossil specimens represent intraspecific polymorphism (taxonomically insignificant tokogenic variation), whereby specimens would be grouped into one species, or whether the differences are a consequence of evolution (taxonomically significant phylogenetic differences), in which case the specimens would be subdivided among multiple species.

One way of testing for the taxonomic significance of character differences between two or more temporally asynchronous fossil specimens would be by examining the populations of closely related species for which population information is available. For instance, suppose that two fossil specimens that are temporally asynchronous and/or from different localities show two different states for a single character, but that character is found to have polymorphic tendencies (i.e., exhibits both states) within one or more of the populations of the species comprising the extant phylogenetic bracket of the two fossil specimens. In this case, the conspecificity of these specimens cannot be falsified. One the other hand, if the same character shows monomorphic tendencies (exhibits one state) within all of the populations of the species comprising the extant phylogenetic bracket, the two specimens in question can be considered separate species. An extant phylogenetic bracket approach is perhaps an ideal way to approach the problem of species delimitation and would work best when dealing with taxa (actually, specimens) with a more refined phylogenetic background. However, this approach was not completely feasible with brontotheres; they are all, of course, extinct, with no close living relatives. Moreover, prior to this study, the interrelationships of brontotheres were poorly understood, and their position in relation to other perissodactyls remains ambiguous.

Fortunately, however, population information is available for a few brontothere species that are known from quarry samples containing assemblages of associated individuals who died instantaneously or during a very brief interval of time. Such mass death assemblages offer rare glimpses into the populations of extinct species (Voorhies, 1969; Turnbull and Martill, 1988; Lucas and Schoch, 1989b; Berger, 1983; Berger et al. 2001; Mihlbachler, 2003b, 2005; Prothero, 2005). The existing brontothere mass death assemblages (all assumed to be monospecific) were used as guidelines to preliminarily screen for characters with intraspecific polymorphic (tokogenic) tendencies that were potentially uninformative of species boundaries. Mass death assemblages of Metarhinus sp., Duchesneodus uintensis, and Megacerops coloradensis (Turnbull and Martill, 1988; Scott, 1945; Gregory and Cook, 1928) were examined. A few other species, such as Eubrontotherium clarnoensis, are known from more limited numbers of associated individuals but were also valuable in identifying characters with polymorphic tendencies. It was found that all of these samples contained common patterns of polymorphism. All samples revealed conspicuous intraspecific variability in premolar morphology, canine size, the size and presence of the M3 hypocone, horn size and shape, and overall skull robusticity. (These characters and their polymorphic tendencies are discussed in more detail in the individual species descriptions.) The consistency of this pattern from assemblage to assemblage suggests a widespread pattern of intraspecific polymorphism throughout the Brontotheriidae. The apparent tokogenic variation in horns, canine size, and overall skull robusticity most likely relates to sexual dimorphism (Mihlbachler et al., 2004b). Notably, other tokogenic variation, particularly the rampant variability of premolar morphology is also apparent in other perissodactyls, most notably ceratomorphs including both rhinocerotoids and tapiroids (Matthew, 1931; Radinsky, 1963, 1967a, 1967b; Prothero, 2005). Before paleontologists like Matthew and Simpson began thinking in terms of population variability, rampant premolar variation led to similar taxonomic oversplitting in these groups. Apparently, this pattern of tokogenic variability is widespread throughout the Perissodactyla.

In the next level of analysis, individual brontothere specimens were examined considering only those characters that did not show polymorphic tendencies in the mass death assemblages. Groups of specimens showing identical attributes were considered conspecific, while specimens showing one or more different character attributes among characters showing monomorphic tendencies were considered different species. Finally, once the specimens had been aggregated into species-specific groups, the rules of zoological nomenclature were used to determine the proper species name.

Utilization of this modified population-aggregation analysis method led to the synonomization of many species and the discovery a few new ones, while many other species were rendered dubious or otherwise problematic. The dubious status of many species relates largely to that fact that many specimens were not directly comparable because they were not known from anatomically overlapping parts (e.g., a jaw and a skull). In instances where two or more specimens representing non-overlapping parts were of similar size and from similar time periods and geographically similar regions, conspecificity could not be falsified. In such instances, if two or more noncomparable specimens were a type, I considered no more than one of these to be valid, while the remaining were considered nomina dubia.

Measurements

Morphometric data were compiled during this study. Unfortunately, quantitative methods (e.g., principal-components analysis) did not prove to be very useful in delimiting species; the distorted condition of most brontothere specimens would have severely complicated any interpretation made from such analyses. Likewise, variable dental wear limited morphometric analysis of dentitions. Nor was overall size an important factor in delimiting species. Many brontothere species overlap in body size. There were a few exceptions where two morphologically similar species were found to be vastly different in body size; in these cases, the difference is obvious without examining quantitative data. More kinds of measurements were taken than are actually reported, but many of the cranial and mandibular measurements are easily influenced by distortion. Therefore, I have limited the included measurements to basic dental measurements and a few basic cranial measurements. All dental measurements and cranial measurements of small skulls were made with digital calipers; cranial measurements of larger specimens were taken with a combination of large wooden calipers or a metric tape measure. Measuring methods differ due to the unmanageable size of many specimens, some of which could not easily be removed from their storage units. For those species known from adequate samples, basic summary statistics (mean, standard deviation, coefficient of variation) of a few variables have been provided in tables to document general size variability within species. The remaining morphometric data on specimens used in this study is provided in appendix 1 in the hope that it will be useful in the identification and/or comparison of future specimens. Although most of these data were not directly utilized in the study, it is nevertheless important to document size. Moreover, the morphometric data were mostly gathered by a single observer (myself) using consistent landmarks and are more reliable than the vast pool of morphometric data on brontotheres from other published sources, which were collected by many different observers, often on heavily distorted specimens, and without well-defined landmarks. The measurements are described below. For the dental abbreviations, upper case letters refer to upper teeth and lower case letters refer to lower teeth.

Institutional Abbreviations

Systematic Paleontology

class mammalia linneaus, 1785

order perissodactyla owen, 1848

family brontotheriidae marsh, 1873

subfamily brontotheriinae marsh, 1873

Bunobrontops savagei Holroyd and Ciochon, 2000

Holotype

UCMP 128416, a right M2.

Type Locality

Pondaung Formation, approximately 3 miles (4.8 km) north of Bahin village, Pale Township, Myanmar.

Age

Middle Eocene (Sharamurunian land mammal “age”).

Referred Specimens

(all from the Pondaung Formation of Myanmar) UCMP 128391, a left M2; UCMP 128414, a right m3; UCMP 147045, a right M3; UCMP 147046, a right molar protocone; UCMP 147047, a left m1 or m2 talonid; UCMP 147048, a left m1 or m2 trigonid; NMMP-KU 0312, a left M3; NMMP-KU 0313, a partial right upper molar (M3?); NMMP-KU 0319, a right molar fragment with a protocone and paracone; NMMP-KU 0333, a left m1 or m2 talonid.

Diagnosis

Bunobrontops savagei exhibits a unique and highly distinctive combination of molar characteristics that differentiate it from primitive brontotheres such as Palaeosyops as well as all more derived species of Brontotheriinae. Molar plesiomorphies retained by B. savagei include a brachydont ectoloph, strong upper and lower molar ribs (most clearly visible in specimens with minimal dental wear), rounded lingual sides of the paracone and metacone, retention of small paraconules and a metalophlike ridge of enamel on the (M2) hypocone, and a rather short and broad m3. Bunobrontops savagei lacks both a central molar fossa and an anterolingual cingular cusp on the upper molars. Brontotheriine molar apomorphies exhibited by B. savagei include the lack of a cingular shelf on the parastyle, and a lingually directed ectoloph. The lingual enamel of the upper molar ectoloph is similar in thickness to the labial ectoloph enamel, a condition intermediate between the thickened enamel of Palaeosyops and the thinned enamel of most other brontotheriines. Finally, the upper and lower molars of Bunobrontops savagei tend to have deep, conspicuous crenulations, an autapomorphic condition.

Description

Upper dentition

Bunobrontops savagei is known only from isolated upper and lower molars, but because of the distinctive molar morphology of this species, even partial molars retain diagnostic features (fig. 5). The upper molars of Bunobrontops savagei exhibit the strongly W-shaped molar wear facet that is characteristic of brontotheriines, but the molar morphology of this species is intermediate between more primitive brontotheres such as Palaeosyops and Eotitanops and more highly derived brontotheriines. The upper molar parastyle lacks a strong anterior cingulum that in more primitive brontotheres such as Palaeosyops forms a small anterolabial occlusal shelf. However, like Palaeosyops, the upper molar ectoloph is very brachydont and not much taller than the lingual cusps. The ectoloph is shorter than any other brontotheriine. However, the labial wall of the ectoloph is nearly flat and is angled in a lingual direction. This resembles other brontotheriines and contrasts with Palaeosyops in which the labial walls of the paracone and metacone are strongly concave, leading to more erect cusps. Holroyd and Ciochon (2000) described the labial ribs of the upper molars as slightly developed; however, this seems to be an artifact of wear in the holotype (UCMP 128416, fig. 5a), because a nearly unworn molar, NMMP-KU 0313 (fig. 5c), has wide and prominent labial ribs that are similar to those of Palaeosyops but are broader and more distinct than those seen in more advanced brontotheriines.

Figure 5

Selected molars of Bunobrontops savagei. (A) A right M2 (UCMP 128416, holotype), (B) a right M3 (UCMP 147045), (C) occlusal and (D) anterior views of a right upper molar (M3?) (cast of NMMP-KU 0313), (E) a right m3 (UCMP 128414), and (F) a left partial molar (cast of NMMP-KU 0333).

The lingual sides of the paracone and metacone of Bunobrontops savagei are rounded as seen in Palaeosyops. This contrasts with more derived brontotheriines in which the inner sides of the paracone and paracone are acutely wedged, particularly toward the distal apex of the cusps. The enamel on the lingual margins of the paracone and metacone is about as thick as that of the labial enamel; however, the enamel between these cusps is slightly thinner on the lingual side of the ectoloph. Overall, the lingual ectoloph enamel in Bunobrontops savagei seems intermediate in thickness. The lingual ectoloph enamel is thinner in comparison to Palaeosyops. However, the lingual ectoloph enamel of other brontotheriines is much thinner than that of Bunobrontops savagei.

The upper molars of Bunobrontops savagei lack some apomorphies that are distinctive of more advanced brontotheriines, such as a central molar fossa and an anterolingual cingular cusp. Several plesiomorphic traits are retained. For instance, there is a small anteriorly positioned paraconule. The upper molars also retain a prominent metalophlike ridge that runs down the labial slope of the hypocone in an anterolabial direction. One M3, UCMP 147045 (fig. 5b) lacks a hypocone, although the posterolingual corner of the tooth bears a thick, crenulated, raised cingulum. Another M3, NMMP-KU 0312 (not shown), has a similar raised posterolingual cingulum and a metalophlike ridge of enamel situated just lingual to the metacone. The labial cingulum of the molars is weak. There is no lingual cingulum, although the posterior cingulum wraps around the lingual side of the hypocone in M2.

The dentition of Bunobrontops savagei is peculiar in that the enamel is crenulated, with narrow grooves running along the lingual bases of the ectolophs in the valleys between the protocones and hypocones, and on the anterior and distal sides of the molar crowns. The enamel of heavily worn specimens tends to be much smoother than the roughened and heavily crenulated enamel of unworn or lightly worn specimens.

Lower dentition

The single complete lower molar, UCMP 128414 (fig. 5e), a right m3, is identifiable as B. savagei largely because of the peculiar enamel crenulations in the trigonid and talonid occlusal valleys; these crenulations characteristically resemble those of the upper molars. The basins of the trigonid and talonid are very shallow. The m3 is only slightly longer than those of Palaeosyops with a length/width ratio of 2.0, but it is clearly shorter and broader than most brontotheriines whose length/width ratios are always above 2.0. The deep crenulations in the occlusal basins tend to obscure the lingual lower molar ribs of the protoconid and hypoconid. However, a lingual lower molar rib is prominent in NMMP-KU 0333 (fig. 5f), an unworn lower molar fragment. In UCMP 128414, the cristid obliqua intersects the protolophid in the middle of the tooth, rather than meeting the metaconid. This character is unusual, and seems to be limited to this specimen. Another specimen, NMMP-KU 0333, an m1 or m2 fragment, has a cristid obliqua that joins the metaconid on the lingual side of the tooth. A slight cingulid surrounds the base of the hypoconulid of the m3 (Holroyd and Ciochon, 2000), but cingulids are generally very weak or absent in the lower molars of Bunobrontops savagei.

Remarks

Holroyd and Ciochon (2000) based Bunobrontops savagei on isolated teeth from the Pondaung Formation, Myanmar. In the same year, Tsubamoto et al. (2000) reported several additional specimens, all isolated teeth, from the Pondaung Formation that clearly represent the same species. Despite the extremely fragmentary nature of brontothere material from the Pondaung Formation, Bunobrontops savagei can clearly be distinguished from other Pondaung brontotheres, and, indeed, all other brontotheres due to a unique combination of molar characters. These include plesiomorphic characters shared with basal brontotheres like Palaeosyops, such as a brachydont ectoloph, prominent labial ribs, and rounded lingual sides of the paracone and metacone. Derived characters shared with other brontotheriines such as Mesatirhinus junius include a lingually angled ectoloph, unthickened enamel on the lingual sides of the paracone and metacone, and the lack of a cingular parastyle shelf. Holroyd and Ciochon (2000) recognized that Bunobrontops savagei represents a clear morphological link for what is otherwise an abrupt morphological transition from the more bunodont molar condition seen in Palaeosyops and Eotitanops, to the more derived molars of advanced brontotheres.

Mesatirhinus junius (Leidy, 1872)

Holotype

ANSP 10349, a right p4; ANSP 10348, a partial left m3, apparently from the same individual.

Type Locality

Bridger Basin, Wyoming.

Synonyms

Mesatirhinus megarhinus (Earle, 1891); Mesatirhinus petersoni Osborn, 1908a.

Age

Middle Eocene (Bridgerian and possibly early Uintan land mammal “ages”)

Referred Specimens

(From the Bridger Basin, Wyoming) AMNH 1509, a skull with right and left P1–M3; AMNH 1523, a palate with right P1–M3, left P1, and P3–M3; AMNH 1520, a left mandibular ramus with c–m3; AMNH 1551, a partial left mandibular ramus with c and p2–m2; AMNH 1567, a mandible with right c, p2–m3 and left c–m3; AMNH 1627, a mandible fragment with right m3; AMNH 12184 (holotype of Mesatirhinus petersoni), a skull missing the occiput and right zygomatic arch with right P2, P3, P4–M3 (partial), left P2–P4, M1–M2 (partial), and M3; AMNH 12191, a right maxilla fragment with M2–M3 and a left mandibular fragment with p2–m3; AMNH 12199, a mandible with right and left p3–m3; AMNH 12202, a skull missing the premaxillae with right C–M3 and left P2–M3; AMNH 12206, a skull fragment with right C–P4, M1–M2 (partial), left P1–P4, and M1–M3 (partial); AMNH 12211, a mandible with right i3(?), p2, p3, p4 (partial) and left p2–m1; AMNH 12686, maxillary fragments with right M2 and M3; BMNH.M10471 (formerly AMNH 1556), a skull with right P1–M3 and left C–M3; UCM 72430, a crushed skull fragment; USNM 26111, a mandible with right p2–p4, m3, left p3–p4, m1 (partial), and m2–m3; USNM 26116, a skull with right P2–M3 and left P4–M3; USNM 26123, a fragmented skull; USNM 26136, a skull with right C, P3–M3 and left C–M3; USNM 26148, a left mandibular ramus with p2–m3; USNM 26151, a mandible with right, left p3–m3, and fragmentary canines and incisors; USNM 363884, a left mandibular ramus with p2–m2, isolated incisors and p1; YPM PU 10242, a left maxilla fragment and associated upper teeth, including right P2–M1, left M2, and M3; YPM 11070, a partial skull with right P2–M3; YPM 11148, a skull with right P1–M3, left P3, M1–M3; YPM 11149, a skull with right and left M3; YPM 16420, a partial skull with left P2–P3, M1–M3; YPM 16423, a right maxilla with P2–M3; YPM 16421, a partial mandible with p4–m3, YPM 16722, a right maxilla fragment with P4–M2, M3 (partial), and left maxilla fragment with M1–M3; YPM 16725, a left mandibular ramus with p3–m3; YPM 16732, a left mandibular ramus with c, p2–m3; YPM 16743, a mandible fragment with left m2–m3; YPM PU10118, a mandible with right and left p3–m3; YPM PU10184, a partial mandible with right p3–m3 and left m1–m3; (from the Washakie Basin of Wyoming) AMNH 1512, a mandible with right and left p2–m3; AMNH 1513, an anterior half of skull with right P2, P3, P4, M3 (partial), and left C–M2, M3 (partial); AMNH 1514, a premaxilla-maxilla with I1–M3; AMNH 1571, a palate with right I1–M3, left I2, C, P1–P3, and P4 (partial); AMNH 1575, a partial mandible with right i2–c, left i2–c, and p2–m3; AMNH 1577, a mandible with right and left p2–m3; AMNH 1651a, a left maxilla fragment with M1–M3; AMNH 13178, a partial mandible with right c, p2–p3 (partial), left i2 (?), c, p2, p4–m3; FMNH PM1676, a crushed skull with right P2, P4–M1 and left P4–M1; FMNH PM27939, a skull with C–M3; FMNH PM36045, a skull with right P2–M3, left C–P4, and M1–M3 (all partial); FMNH PM39945, a partially prepared palate with crushed cheek teeth; FMNH PM54864, a crushed skull fragment with left P2–M3; YPM PU10008 (holotype of Mesatirhinus megarhinus), a partial skull with very poorly preserved molars; YPM PU10041, a posterior part of a cranium; (from the Sand Wash Basin of Moffat County, Colorado) DMNH 8103, a right maxilla with C–M3; DMNH 9687, a left mandibular ramus with p3–m3; DMNH 29950, a partial skull (in two pieces) missing nasals and frontals with right P4–M1 (partial), and M2–M3; (no locality data) YPM PU25021, a crushed skull with right P2–M3, left C, P2–P4, M1–M3 (all partial), and a mandible with right and left p3–m3.

Diagnosis

Mesatirhinus junius is a small hornless brontothere in which the frontal bone does not overlap or intrude into the nasal bone. The nasal incision extends posteriorly to between the anterior margin of P4 and the posterolateral root of M1. The nasal process is horizontal, unelevated, of relatively constant transverse width, narrow, with thin and deep lateral walls, and without a well-defined or strongly rounded distal margin. The orbits do not protrude laterally and are positioned over the M2 with the anterior lateral root of M2 and posterior lateral root of M1 below the anterior orbital rim. There is a prominent infraorbital process on the jugal. The premaxillomaxillary rostrum deepens posteriorly and it is not enclosed dorsally by bone. Other cranial characteristics include a flat or convex dorsal cranial surface, a sagittal crest, thin and weakly curved zygomatic arches, a ventrally open external auditory pseudomeatus, and wide occipital condyles.

Dentally, Mesatirhinus junius is characterized by large subcaniniform upper incisors, a postcanine diastema, a simple P1, a distinct P2 metacone, weak premolar preprotocristae and/or paraconules, and short crests extending posteriorly from the protocones of the premolars. Premolar hypocones are absent. The molars of M. junius have tall, lingually angled ectolophs with weak labial ribs, and somewhat thinned lingual ectoloph enamel with wedge-shaped paracones and metacones. A cingular parastyle shelf is absent. Mesatirhinus junius molars retain vestigial paraconules and metalophs, and they lack central fossae and anterolingual cingular cusps. The lower dentition of M. junius is characterized by large semispatulate incisors that are all of a similar size, a short p1–p2 diastema, an elongate p2 trigonid, a metaconid on p4 but not on p2 and p3, shallow molar basins, and a slender m3.

Mesatirhinus junius is very similar to Sphenocoelus uintensis but is clearly distinct from that species due to its smaller size, more brachycephalic proportions, and lack of paired ventral sphenoidal fossae. M. junius is similar in size to Metarhinus, but can be differentiated from it by its unspecialized rostrum.

Description

Skull

Many skulls of Mesatirhinus junius are known, but because each one is damaged or distorted in some way it is necessary to refer to several specimens for a full description. The following description is primarily based on AMNH 1509 (fig. 6), AMNH 12202 (fig. 7a, fig. 8a), USNM 26116 (fig. 7b), and AMNH 12184 (fig. 7c). M. junius is a small (table 1) hornless brontothere similar in size to Metarhinus, but it is smaller than other late Bridgerian taxa, e.g., Telmatherium and Palaeosyops, that are contemporaneous with it. The general morphology of the skull most closely resembles Sphenocoelus uintensis, though it is smaller and more brachycephalic.

Figure 6

A skull referred to Mesatirhinus junius (AMNH 1509). (A) Right view, (B) dorsal view.

Figure 7

Selected skulls referred to Mesatirhinus junius. (A) Dorsal view of AMNH 12202 showing thinner sagittal crest than AMNH 1509, (B) posterior view of USNM 26116 showing undistorted occiput, (C) dorsal view of anterior part of AMNH 12184 showing complete nasal process.

Figure 8

Ventral surface of the skull and upper teeth of Mesatirhinus junius. (A) Ventral view of AMNH 12202, (B) right premolars of AMNH 12202, (C) right molars of AMNH 12202, (D) lingual view of left incisors of AMNH 1514, (E) labial incisor view of left incisors of AMNH 1514.

Table 1

Summary statistics for selected morphometric variables of Mesatirhinus junius See Methods for measurement definitions

The frontonasal suture is not clear in most of the specimens. Osborn's (1929a: fig. 328) illustration of AMNH 1556 (now in the British Museum with number BMNH.M10471) seems to portray a suture that looks more like a crack on the actual specimen. The frontonasal suture on the left side of the skull of AMNH 12184, also figured by Osborn (1929a: fig. 327), appears as a deeply convoluted line. This represents a true frontonasal suture and indicates the absence of a frontal process overlapping the nasal bone like that seen in Telmatherium validus.

The nasal bones tend to be poorly fused together. The nasal process is shorter than the premaxillomaxillary rostrum and is transversely narrower. The nasal process is slightly angled downward as it projects from the skull. The sides of the nasal process form relatively thin and deep vertical walls. From a dorsal view, the nasal process appears to be slightly constricted proximally in AMNH 1509, but in other specimens, such as AMNH 12184, the nasal process is more nearly constant in width. The anterior margin of the nasal process is flat from a dorsal view, except for a median notch. The anterior edge of the nasal process is thin, roughened, and angled downward only slightly.

The nasal incision of Mesatirhinus junius is long and shallow. The position of the posterior margin of the nasal incision fluctuates slightly from the anterior margin of M1 (e.g., AMNH 1509, AMNH 12184) to the posterolateral root of M1 (YPM 11070). The orbits tend to be positioned more or less over the M2 although its position with respect to the molars varies slightly. For instance, in YPM 10008, the alveolus for the anterior lateral root of M1 is directly below the anterior edge of the orbital rim; this is somewhat more anteriorly positioned than the majority of M. junius specimens where the anterior lateral root of M2 and posterior lateral root of M1 are situated below the anterior orbital rim.

A shallow facial concavity exists between the nasal incision and orbit. The facial concavity is bordered superiorly by a prominent rim of bone that abruptly meets the nearly flat dorsal surface of the skull. The rim gives this part of the skull a somewhat thickened appearance. Osborn (1929a) described this species as having incipient horns, although Mader (1989) could not corroborate the presence of incipient horns. Later, McCarroll et al. (1996a) reported an incipient horn on FMNH PM27939, although it seemed to occur only on one side (right) of the specimen. Although, the prominent rim of bone above the facial concavity tends to gives this area of the face a thickened appearance, I agree with Mader's (1989) judgment that there is no distinct hornlike structure on any specimen of M. junius.

From a lateral view, the dorsal margin of the premaxillomaxillary rostrum is steeply sloped, so that the posterior notch of the nasal incision is level with the top of the orbit. The premaxillae most often are not fused at the symphysis; however, they are occasionally fully co-ossified (e.g., AMNH 1571). The premaxillomaxillary suture is not always clear, but it is clearly visible on USNM 26136 (not figured). In this specimen, the ascending nasal process of the premaxilla does not reach the posterior notch of the nasal incision, and, therefore, does not contact the nasal bone. The premaxillomaxillary rostral cavity and the nasal cavity form a single continuous osteological cavity; this condition is normal, but differs substantially from Dolichorhinus and Metarhinus, in which the premaxillomaxillary rostral cavity is completely enclosed in bone.

Among the least distorted skulls of Mesatirhinus junius, the dorsal surface of the skull above the orbits is flat or slightly convex from a lateral view. The lateral profile of the dorsal surface of the posterior half of the skull varies from flat to slightly convex. The parasagittal ridges tend to be thin but distinct, and join medially to form a sagittal crest.

The zygomatic arches of Mesatirhinus junius are thin and weakly curved, like those of Sphenocoelus and Dolichorhinus. From a lateral view, the jugal zygomatic process is horizontal while the zygomatic process of the squamosal is slightly angled posterodorsally, giving the zygomatics a weak curvature. From a dorsal view, the zygomatic arches are slightly bowed laterally. M. junius possesses a large laterally projecting infraorbital process on the jugal. The size of this process varies (compare AMNH 1509 to AMNH 12184), although they tend to be large and conspicuous like those of Sphenocoelus, and they are more distinct than those of Metarhinus.

From dorsal views of the skulls, the nuchal crest is anteromedially angled. From lateral views, the occiput is slightly tilted backward. The occiput of Mesatirhinus junius is best preserved in USNM 26116. From the posterior view, the nuchal crest is arched dorsally. The dorsal portion of the occiput is nearly as wide as the posterior portion and the occiput is constricted in the middle. Occipital pillars are evident, but they are weak, and the central depression in the occiput is shallow. The occipital condyles are very wide, a condition shared with Sphenocoelus uintensis and Dolichorhinus hyognathus.

Unlike many other brontotheres, there is no horseshoe-shaped emargination around the opening of the posterior nares of Mesatirhinus junius. A small medial process typically projects posteriorly from the anterior rim of the posterior nares (although it is not preserved on AMNH 12202). The anterior margin of the posterior nares varies in position from between the M2 protocones (AMNH 12202) to between the M3 hypocones (AMNH 36045). Most frequently, the position of the posterior nares is anterior to the posterior margin of the M2. The elongate posterior narial canal formed by the posterior nares does not extend into the body of the sphenoid as in Sphenocoelus uintensis. The mastoid process does not contact the postglenoid process, thus the opening of the external auditory pseudomeatus is unconstricted ventrally. Other aspects of the basicranium of M. junius are typical of other brontotheres, such as a widely separated foramen ovale and foramen lacerum.

Upper Dentition

Few specimens of Mesatirhinus junius have preserved upper incisors; a partial skull, AMNH 1514, includes a complete row of three left incisors (fig. 8d, e). The incisors are large and arch anterior to the canines. The tips of I1 and I2 are worn off, but these incisors appear to have had conular crowns. The I3 crown is somewhat larger than those of I1 and I2 and is more nearly caniniform in shape. Each of the canines has a distinct lingual cingulum. The canines of Mesatirhinus junius are generally of moderate size, like those of Dolichorhinus hyognathus. The incisors of AMNH 1514 are separated by very short gaps, although this may have not been the case in other specimens. Finally, there are both short precanine and postcanine diastemata. The postcanine diastema is shorter than P2.

Complete cheektooth rows are preserved in many specimens of Mesatirhinus junius. The following description of the premolars and molars of M. junius is primarily of AMNH 12202 (fig. 8a–c) but information from other specimens is provided. P1 is slightly shorter than P2, although it is much narrower and with a simpler crown. The P1 crown has a single cusp and an elongate posterior heel. There is no P1–P2 diastema.

The anterior side of P2 is more steeply angled posterolingually than P3 and P4, thus giving the P2 crown a more oblique shape and a relatively narrower lingual side. P3 and P4 are progressively less oblique in outline; their anterior and posterior sides are nearly parallel. The parastyle and metastyle of P2 are angled slightly lingually. The P3 parastyle and metastyle are nearly straight, while those of P4 are slightly angled labially. The labial paracone rib of P2 is broad while those of P3 and P4 are progressively narrower and less distinct. The metacone of P2 is positioned somewhat more lingually in comparison to P3 and P4. Because of these differences the labial side of P2 is rounder than the labial sides of P3 and P4.

In AMNH 12202 a short lingual crest extends posteriorly from the protocone on P2–P4, but in some specimens this lingual crest is absent (e.g., AMNH 1514). Each premolar exhibits a small but distinct preprotocrista that extends from the anterior slope of the protocone and meets the ectoloph at a point anterior to the lingual base of the paracone. In AMNH 12202, the preprotocrista of P2 tends to be wider, while those of more posterior premolars (P3 and P4) are thinner and sometimes have a small but distinct paraconule. The preprotocrista of P2 is discontinuous with the small anterior cingulum. However, this morphology is highly variable. For instance, in AMNH 1523, the lingual side of the P2 crown is severely shifted posteriorly; the protocone is posterior to the metacone, and the preprotocrista attaches to the lingual base of the metacone. AMNH 1514 shows an entirely different morphology whereby the preprotocrista contacts the lingual base of the ectoloph at a point anterior to the protocone.

No hypocone is found on any of the premolars of Mesatirhinus junius. The anterior and posterior cingula of P2–P4 stretch around the anterolingual and posterolingual corners of the crown but they do not join on the lingual side. The labial premolar cingula of the P2 and P3 connect to the posterior slope of the paracone ribs, thus forming a ridge that runs from the posterolabial base of the crown to the occlusal peak of the paracone. The P4 labial cingulum stretches across the proximal base of the crown and does not join the paracone rib. However, in some specimens (e.g., AMNH 1513) the labial cingulum of P4 is incomplete and does not stretch across the base of the paracone.

The molars of Mesatirhinus junius exhibit numerous apomorphies shared with other brontotheriines. For instance, the anterior cingulum is very thin labially and does not form a thick shelf at the peak of the parastyle. The ectoloph is taller than the lingual cusps. The labial paracone and metacone ribs are thin and indistinct. Overall, the labial side of the ectoloph is strongly angled lingually. The lingual wall of the ectoloph enamel thickens slightly around the lingual sides of the paracone and metacone, but not to the degree seen in Palaeosyops or Eotitanops. The lingual ectoloph enamel that stretches between the cusps is about as thick as the labial sheet of enamel. In unworn teeth, such as the M3 of AMNH 12202, the lingual sides of the paracone and metacone are wedge-shaped, but in more worn molars such as the M2 and M1 of the same specimen the lingual sides of these cusps become progressively rounded near the proximal base of the cusp.

The molars of Mesatirhinus junius retain small and essentially vestigial paraconules. The presence of paraconules appears to be a fixed condition. There is a very small but distinct crest of enamel that runs along the anterolabial slope of the hypocone of M1 and M2 that connects to the inner base of the metacone. Sometimes this structure is long and crestlike as on the M2 of AMNH 12202, and other times it is shorter and cusp-like as on the M1 of the same specimen. This crest seems to be a vestigial remnant of a metaloph. It should be noted that evidence of the paraconules and the metaloph fades with wear, and in heavily worn molars the evidence for these structures is erased.

The molars of Mesatirhinus junius lack central molar fossae and anterolingual cingular cusps. A hypocone is not present on the M3, although a thick cingulum wraps around the posterolingual corner of the crown. Rarely (e.g., USNM 26136), a very small cusp is seen near the posterolingual corner of the M3 that could be interpreted as a rudimentary hypocone. Lingual cingula are absent on the molars, while the labial molar cingula are faint and discontinuous around the mesostyles.

Mandible and Lower Dentition

Four skulls of Mesatirhinus junius are associated with mandibles (AMNH 12191, FMNH PM36045, USNM 26116, and YPM PU25021). Unfortunately, these mandibles lack incisors and have incomplete, heavily worn sets of cheek teeth. However, several mandibles recovered from the Upper Bridger (Bridger C–D) and the lower parts of the Washakie Basin (Washakie A of Granger 1909 and TWKK–TWKA1 of McCarroll et al., 1996b) are consistent in size and morphology with the few mandibles that are directly associated with M. junius skulls. Among these jaws, the cheek teeth, particularly M3, are far too slender and thin-enameled for any species of Palaeosyops. Telmatherium validus co-occurs with M. junius in the upper Bridger and lower Washakie formations and has similar lower dentition. Other than their smaller size, M. junius mandibles and lower dentition lack any obvious characters that clearly distinguish them from Telmatherium validus. However, those referred to M. junius are well below the lower size range of T. validus. There are no other small brontotheres from upper Bridger and/or lower Washakie sediments to whom these mandibles could belong.

Pictured is a complete mandible, AMNH 1567 (fig. 9a, b). The mandible is similar in proportion to most other Bridgerian and Uintan hornless brontotheres except Dolichorhinus, whose mandible is more slender. In Mesatirhinus junius the position of the posterior margin of the symphysis fluctuates slightly but it is generally positioned between the talonid of p2 and the trigonid of p3. The inferior angle of the mandible tends to be quite steep (≥ 45°).

Figure 9

Selected mandibles and lower teeth of Mesatirhinus junius. (A) Left view of AMNH 1567, (B) dorsal view of AMNH 1567, (C) left premolars of AMNH 1520, (D–E) left lower incisors of USNM 363884, (F) anterior dentition of AMNH 1575.

Complete sets of relatively unworn lower incisors of Mesatirhinus junius are rare, although the alveolar surface suggests that the incisor row was arched and extended anterior to the canines. One partial jaw, USNM 363884, includes a complete set of nearly unworn but isolated left incisors (fig. 9d, e). The crowns are large, of similar size, and bilaterally asymmetrical. The i1 and i2 have rounded apices and are semispatulate. The i3 is more mesiodistally elongate than i1 or i2 is more lingually curved. Each incisor has a thin but distinct lingual cingulid. Another specimen, AMNH 1575 (fig. 9f), has a partial incisor row with more damaged incisors, but with essentially the same morphology. The i3 of this specimen is more subcaniniform in shape than the i2, with a short, lingually curved crown. There are no diastemata between the incisors or canines.

The following description of the lower premolars is based primarily on AMNH 1520 (fig. 9c), a specimen with nearly unworn premolars, but other specimens provide additional information on variation. The p1 is isolated by a postcanine diastema and a p1–p2 diastema. The postcanine diastema tends to be shorter than the p2. The p1–p2 diastema is even shorter. The p1 has a small simple crown with a single cusp and a very short talonid heel.

The trigonid of p2 is nearly twice as long as the talonid; the p3 trigonid is also distinctly longer than the talonid. However, the p4 trigonid is slightly shorter than the talonid. The talonid and trigonid are of nearly equal width in p2 and p3, although in most specimens (e.g., YPM 16725 and YPM PU10184) the p3 talonid is slightly wider than the trigonid. The p4 talonid is always wider than the trigonid. The paralophids of the p2 and p3 are angled in a slightly lingual direction, creating a small lingual trigonid notch, while the paralophid of p4 curves fully lingually, creating a much larger lingual trigonid notch. The protolophids of p2 and p3 are straight but are angled slightly lingually; in p4 the paralophid arches 90° lingually and is fully molariform. Metaconids are absent on p2 and p3; however, there is a large lingually positioned molariform metaconid on p4. The talonid of p2 tends to have a short and/or poorly developed cristid obliqua and hypolophid, and the lingual side of the p2 talonid is a slightly concave sloped surface. However, p3 and p4 have more or less basin-shaped talonids with longer cristids obliqua and hypolophids. Lingual premolar cingulids are absent and labial premolar cingulids are generally weak.

The molars of Mesatirhinus junius are typical with relatively thin enamel, shallow trigonid and talonid basins, and with an elongate m3. Labial molar cingulids are generally weak and lingual molar cingulids are absent. A thin beaded cingulid wraps around to the distal end of the hypoconulid of the m3 of some specimens.

Remarks

Leidy (1872) originally assigned Mesatirhinus junius to the genus Palaeosyops Leidy (1870a). This species was named from portions of a lower jaw from an unspecified stratigraphic level from Fort Bridger, Wyoming. The material was later described as “several small fragments of the right side of a lower jaw, together with a sketch of a larger fragment of the left side, containing the last premolar and the succeeding molars” (Leidy, 1873: p. 57). Leidy (1872; 1873) did not figure any of these specimens or provide specimen numbers, but he did provide measurements for the specimen he had described as Palaeosyops junius. Later, Osborn noted, “Of this type material only p4 (right) and the posterior half of m3 (right) were located in the collection of the Academy of Natural Sciences of Philadelphia” (Osborn, 1929a: 159); he considered these specimens (fig. 10) cotypes. However, if they are from a single individual, as Leidy's (1873) description seems to imply, they are not cotypes, but a single holotype. The original measurements provided by Leidy (1873) for the p4 and m3 are consistent with ANSP 10349 (p4) and ANSP 10348 (m3 fragment): breadth (anteroposterior length) of last premolar  =  8 lines (16.9 mm), thickness (width) of last premolar  =  5.5 lines (11.6 mm), thickness of third molar at middle  =  7 lines (14.8 mm). Spamer et al. (1995) list these specimens as the syntypes of Mesatirhinus junius (fig. 10).

Figure 10

Holotype of Mesatirhinus junius. (A) Right p4 (ANSP 10349), (B) partial right m3 (ANSP 10348). Illustrations from Osborn (1929a).

Earle (1891) erected another species, Mesatirhinus megarhinus, which he also assigned to the genus Palaeosyops. Palaeosyops megarhinus was based on a nearly complete skull, YPM PU10008, with poorly preserved teeth from an unspecified stratigraphic level of the Washakie Basin, Wyoming. It is worth noting that Earle's (1891, 1892) figures of the holotype (YPM PU10008) show a complete left zygomatic arch, a complete nasal process, and a complete occiput. Currently, the distal end of the nasal process is not preserved, the zygomatic arches are incomplete, and the occiput is missing in the holotype.

Osborn (1908a) recognized that Earle's Palaeosyops megarhinus differed from other Palaeosyops species and erected the genus name Mesatirhinus for it. Mesatirhinus megarhinus was diagnosed by its infraorbital process, sagittal crest, and upper molars with flattened outer cusps and reduced conules. In that same paper Osborn (1908a) erected a new species, M. petersoni, based on a partial skull (AMNH 12184) from Bridger D. Osborn (1908a) did not state how M. petersoni was distinct from M. megarhinus in clear morphological terms. He noted that the preorbital facial region was more elongate, and that the grinding teeth occupy more space. However, considering the fact that specimens are damaged and distorted to varying degrees, the subtle distinctions between these two species is clearly suspect.

Ultimately, Osborn (1929a) accepted three species of Mesatirhinus: M. junius (Leidy), M. megarhinus (Earle), and M. petersoni Osborn. Osborn (1929a) considered Mesatirhinus junius the earliest form of Mesatirhinus and suggested that the holotype was probably from Bridger B. Osborn states, “The type lower molar of M. junius, according to Leidy's description, was found near Fort Bridger, Wyo., at a geological level that Granger places in Bridger B” (1929a: 388). However, Leidy's description states that the fossils were “received from Dr. J. Van A. Carter, of Fort Bridger, Wyoming” (Leidy, 1892: 277). Leidy (1872, 1873) did not state that the fossil was “found near Fort Bridger” (contra Osborn, 1929a). Therefore, the stratigraphic provenience of the holotype of M. junius is unknown. There is no direct evidence of the occurrence of Mesatirhinus in Bridger B, although it is relatively common in Bridger C–D.

Mader (1998) recognized only a single species, Mesatirhinus megarhinus. He considered M. junius to be a nomen dubium but suggested that M. junius is a potential senior synonym of M. megarhinus. In contrast, Gunnell and Yarborough (2000) considered M. junius to be a synonym of Palaeosyops paludosus. Nonetheless, it can be shown that M. junius is a valid name, that it is distinct from Palaeosyops, and that it is the senior synonym of both M. megarhinus and M. petersoni. Among brontotheres that occur in the Bridgerian, the m3 of the holotype of M. junius (ANSP 10348) is smaller than Palaeosyops from Bridger B, or from Bridger C–D, and, more importantly, it is far too slender and thin enameled for a m3 of Palaeosyops. It is also much too large for Eotitanops. Strictly speaking, the morphology and proportions of the holotype of Mesatirhinus junius are not different from Telmatherium validus; however, it is considerably smaller than all known specimens of T. validus. Moreover, the holotype of M. junius falls within the size range of other materials that have been referred to as Mesatirhinus (table 1). Therefore, M. junius is the earliest name that pertains to Mesatirhinus. M. megarhinus (Earle) and M. petersoni Osborn appear to represent the same species and are considered junior synonyms of M. junius (Leidy).

The material referred to Mesatirhinus junius consists mostly of specimens from late Bridgerian deposits of the Bridger Basin (Bridger C–D) and Washakie Basin (Washakie A of Granger [1909] and TKWW–TWKA1 of McCarroll et al. [1996b]). There is evidence, however, that Mesatirhinus junius ranges into the early Uintan. A crushed anterior cranial fragment (FMNH PM54864) is from early Uintan deposits (Washakie Basin, TWKA2). Another specimen, FMNH PM1676, a crushed skull, is recorded from the “upper Washakie”. Generally, “upper Washakie” refers to the Uintan portion of the Washakie deposits (see Granger, 1909), but the exact stratigraphic position of this specimen is uncertain. There are two probable specimens of M. junius from the Sand Wash Basin: DMNH 8103, a left maxilla; and DMNH 29950, a left partial skull. These specimens possess a prominent infraorbital process, a deep nasal incision, and they lack central fossae on the upper molars. This combination of characters is consistent with M. junius and rules out all other Uintan brontotheres except Sphenocoelus uintensis Osborn (1895). These specimens are considerably smaller than S. uintensis. They are slightly larger than Bridgerian specimens of Mesatirhinus junius, but only by millimeters. It is therefore probable that (1) these specimens are a new species slightly larger than the M. junius from the Bridgerian land mammal “age”, or (2) M. junius survived into the early Uintan but was slightly larger at that particular time and place. Because the material at hand does not clearly indicate a morphologically distinct species, the latter possibility is preferable; therefore, these specimens are referred to M. junius.

Desmatotitan tukhumensis Granger and Gregory, 1943

Holotype

AMNH 21606, a partial mandible with right i1–c, p2–m3, left i1, i3, and c.

Type Locality

Ulan Shireh Formation, four miles north of Tukhum Lamasery, Inner Mongolia, China.

Age

Middle Eocene (Irdinmanhan land mammal “age”).

Diagnosis

Desmatotitan tukhumensis is an intermediate-sized brontothere with three pairs of large incisors that strongly arch anterior to the canines. The i1 and i2 are semispatulate while the i3 is subcaniniform. The posterior margin of the symphysis extends to the p3 metaconid. There is both a postcanine diastema and a short p1–p2 diastema. The p1 is anchored obliquely in the jaw. Metaconids are absent on p2–p3 but present on p4. The molar ribs are weak, the lingual enamel is thinner than the labial enamel, and m3 is more elongate than those of Palaeosyops and Bunobrontops.

Desmatotitan tukhumensis is one of a few brontotheres with a p1–p2 diastema. Among these, D. tukhumensis is most similar in size and morphology to Mesatirhinus junius. It shares with M. junius a semispatulate i1 and i2. However, the premolars of D. tukhumensis are broader than those of M. junius, the crescents of the lower molars are more strongly rounded, and the p1 is anchored obliquely in the jaw.

Description

Mandible and Lower Dentition

The holotype jaw of Desmatotitan tukhumensis (AMNH 21606) includes a symphysis and right ramus with a complete set of teeth that are not heavily worn (fig. 11). The ramus is very thin and the roots of the cheek teeth are exposed, possibly reflecting poor health near the time of death. The left side of the symphysis is not preserved but it has been reconstructed with plaster. Judging by the intact right side, the proportions of the symphysis as it is reconstructed is accurate. The angle of the inferior margin of the symphysis is somewhat less then 45°. The symphysis extended about to the middle of the p3.

Figure 11

The holotype of Desmatotitan tukhumensis (AMNH 21606). (A) Right view, (B) dorsal view, (C) right premolars (p2–p4), (D) right molars, (E) lingual view of anterior dentition, (F) labial view of anterior dentition.

The lower dental formula (3-1-4-3) is unreduced. The incisor row forms a strong arch anterior to the canines. Overall, the incisors are large and are all roughly of similar size. The crowns of the i1 and i2 are semispatulate and i3 is more subcaniniform and mesiodistally elongate but with a slightly shorter crown. The apices of both i1s are worn off. The distal apex of i2 forms a dull rounded tip. A distinct lingual cingulid is present on each incisor. The lingual cingulid of the i3 is less distinct than those of i1 and i2. The canine is both large in diameter and in height.

There are no diastemata between any of the incisors or canines. However, there is a relatively long postcanine diastema. On the alveolar surface, between the canine and p2, an elongate p1 alveolus extends into the mandible in a posteroventral direction, indicating that the root of p1 was anchored obliquely in the mandible. The position of the p1 alveolus indicates a short p1–p2 diastema.

The p2 trigonid is nearly twice as long as the talonid. Moreover, the trigonids of p3 and p4 are somewhat longer than their talonids. The lingual side of the p2 is incomplete although the trigonid and talonid of that tooth appear to have been similar in width. However, the talonids of p3 and p4 are distinctly wider than the trigonids. The paralophid of p2 curves slightly lingually, creating a small lingual trigonid basin. The protolophid is not curved although it is slightly lingually angled. The paralophid and protolophid of p3 have orientations similar to those of p2, but the protolophid is more curved and longer in p3, resulting in a shallow but broad lingual trigonid notch. The paralophid and protolophid of the p4 each arch fully lingually, creating a nearly molariform trigonid notch. The p2 and p3 lack metaconids, while p4 has a large lingually positioned metaconid. The p2 talonid has a short but distinct cristid obliqua and hypolophid; the lingual side of the p2 talonid is a slightly convex sloped surface. The cristids obliqua and hypolophids of p3 and p4 are longer, creating broader and more nearly molariform talonid basins. The labial premolar cingulids are strong, but they are discontinuous around the proximal bases of the protoconids.

The basins of the molars are shallow, the labial ribs are very weak, and the lingual enamel is thinner than the labial enamel. The m3 is proportionately longer than those of Palaeosyops or Bunobrontops, and is similar in proportions to Sthenodectes incisivum, but it is relatively short in comparison to most other brontotheres. The double crescent-shaped lophids of the lower molars, particularly that of m3, are very rounded in shape. The labial molar cingulids are distinct and continuous around the proximal bases of the cusps. There is an additional strong beaded cingulid on the lingual side of the m3 hypoconulid, although lingual cingulids are otherwise absent in the holotype of D. tukhumensis. The m2 shows the tendency for the protoconid and hypoconid to form rounded exposures of dentin in early stages of wear, a plesiomorphic wear pattern seen in basal brontotheres such as Palaeosyops and Eotitanops. In more progressive states of wear, characterized by the m1, this plesiomorphic wear pattern is not notable.

Remarks

Granger and Gregory (1943) recognized AMNH 21606 as a distinct species and concluded that it was a close ally of Metatelmatherium Granger and Gregory (1938). With only a single mandible, it is difficult to confirm Granger and Gregory's conclusion. Characteristics of the molars, such as the weak labial ribs, and the thinner lingual enamel are different from Palaeosyops and Bunobrontops and suggest a close relationship with more advanced brontotheres. Among more advanced brontotheres, a p1–p2 diastema is seen in Dolichorhinus, Mesatirhinus, Telmatherium, and Metatelmatherium. Among these, only Metatelmatherium occurs in Asia. However, the incisors of Desmatotitan tukhumensis are decidedly less subcaniniform than those of Metatelmatherium. In this respect, D. tukhumensis more closely resembles Mesatirhinus. However, the relatively broad premolars and very rounded lower molar crescents are out of character with North American specimens of Mesatirhinus junius. In this respect D. tukhumensis seems to resemble Microtitan mongoliensis, although that species differs from D. tukhumensis in having a much more elongate m3, narrower premolars, much smaller body size, and absence of a p1–p2 diastema. For this reason Desmatotitan tukhumensis is accepted, although more material is clearly needed to generate a more thorough diagnosis of this species.

Acrotitan ulanshirehensis Ye, 1983

Holotype

IVPP V6686, a partial mandible with left p3 and p4.

Type Locality

Ulan Shireh Formation, Inner Mongolia, China.

Age

Middle Eocene (Irdinmanhan land mammal “age”)

Diagnosis

Acrotitan ulanshirehensis is a small brontothere with two pairs of large incisors that arch anterior to the canines. The posterior margin of the symphysis extends to the anterior margin of the p3. There is a short postcanine diastema and a long p1–p2 diastema. The molar ribs are weak and the lingual enamel is thinner than the labial enamel.

Acrotitan ulanshirehensis is one of two brontotheres known to have a reduced number of lower incisors. The other, Megacerops coloradensis, is substantially larger and lacks mandibular diastemata. Additionally, Acrotitan ulanshirehensis has an elongate p1–p2 diastema and an obliquely oriented p1.

Description

Mandible and Lower Dentition

The holotype of Acrotitan ulanshirehensis (IVPP V6686) is a partial mandible of a small brontothere. The mandibular symphysis is elongate and narrow. The inferior margin of the mandibular symphysis is shallow and forms an angle less than 45°. The posterior margin of the symphysis extends to the anterior margin of the p3. The crowns of the anterior dentition are not preserved; however, the alveoli indicate a reduced number of incisors (two pairs). Justification for the dental formula as interpreted here (2-1-4-?) is labeled in fig. 12. From the anterior view of the specimen, two pairs of small alveoli are seen between a pair of very large alveoli. The right pair of small alveoli are empty, although the roots remain in the left pair. The large pair of alveoli also contain roots. The two pairs of small alveoli at the front of the jaw are clearly incisor alveoli. The pair of much larger alveoli that are lateral to the smaller incisor alveoli could either represent a third pair of very enlarged incisors, or, more likely, the canines. Immediately behind the canines is a smaller anteriorly angled dental alveolus that undoubtedly held the p1, which, in brontotheres, is a small, single rooted tooth. Moving posteriorly, the additional alveoli held the anterior and posterior roots of the p2.

Figure 12

The holotype mandible of Acrotitan ulanshirehensis (IVPP V6686) with revised interpretation of dentition. (A) Left view, (B) dorsal view of left p3 and p4, (C) anterior view showing dental fragments and alveoli of anterior dentition, (D) dorsal view.

The orientation and size of the incisor alveoli indicate relatively small and procumbent incisors that are ovoid in cross section and were probably arranged in a small arch. The central incisors are separated by a short median diastema. The canines were relatively large with rounded cross sections. The postcanine diastema was rather short. The p1 alveolus indicates a simple, single-rooted tooth with a somewhat anteriorly directed orientation. There is a diastema between the p1and p2 that is substantially longer than the postcanine diastema.

The trigonids of p3 and p4 are slightly shorter and narrower than their respective talonids. The paralophid of p3 is short and angled anterolingually. The protolophid is straight and angled about equally lingually and posteriorly. The paralophid and protolophid of p4 are each angled more strongly lingually, creating a large molariform trigonid notch. The lingual enamel of the premolars is thinner than the labial enamel. A portion of the lingual enamel of the p3 is broken off and it is difficult to discern whether this tooth had a metaconid. Typically, brontothere premolars that have strongly lingually angled protolophids, such as this p3, also have metaconids, but it is difficult to confirm that in this specimen. Likewise, the lingual enamel of p4 is broken away. However, from the lateral view of the tooth, the protolophid rises lingually as it stretches from the protoconid toward the junction with the cristid obliqua. This strongly suggests that a p4 metaconid was present. The cristids obliqua and hypolophids of p3 and p4 are long and obliquely angled, thus creating broad and nearly molariform talonid basins. The labial premolar cingulids are distinct although lingual cingulids appear to be absent.

Remarks

Acrotitan ulanshirehensis is based on a small and rather peculiar brontothere mandible (IVPP V6686). No other specimens have ever been assigned to this species. The brief description provided by Ye (1983) of A. ulanshirehensis is puzzling. Ye (1983) described the posterior margin of the symphysis as terminating anterior to the p2, although the symphysis clearly extends to the anterior margin of the p3. More puzzlingly, the dental formula was described as 3-1-4-3, although there only appear to be two pairs of lower incisors. Moreover, no molars are preserved with the specimen. Thus, the dental formula is 2-1-4-?. Among those brontotheres for which lower incisors (or mandibles with alveoli) are known, only one other brontothere, the large North American Chadronian (late Eocene) species Megacerops coloradensis, is characterized by a similar mandibular dental formula.

In addition to Acrotitan, the following taxa retain a p1–p2 diastema: Eotitanops, Palaeosyops, Mesatirhinus junius, Dolichorhinus hyognathus, and Desmatotitan tukhumensis. Additionally a p1–p2 diastema is occasionally present in Metatelmatherium ultimum and Telmatherium validus. The postcanine diastema of IVPP V6686 is rather long (16 mm), while the postcanine diastemata of these other taxa are typically shorter (usually less than 10 mm), despite the much larger size of most of these species. The p1–p2 diastema of Dolichorhinus, however, ranges up to 19 mm. However, diastema length is quite variable within brontothere species. For example p1–p2 diastema length ranges from 7 mm to 18 mm within D. hyognathus. Therefore, the unusual length of the postcanine diastema of IVPP V6686 might be a characteristic of this particular individual rather than a character of the species, Acrotitan ulanshirehensis. Brontotheres that are similar in size to Acrotitan ulanshirehensis, such as Microtitan, Metatitan, and Pygmaetitan, all lack p1–p2 diastemata.

In addition to its small size and reduced number of incisors, a potentially diagnostically important characteristic of Acrotitan ulanshirehensis is the oblique orientation of p1. In nearly all brontotheres, p1 is typically rooted vertically in the jaw. However in IVPP V6686, the alveolus of p1 is slanted, indicating a posteroventrally rooted p1. The holotype specimen of Desmatotitan tukhumensis (AMNH 21606), a larger brontothere with three incisors, has a similarly angled p1, suggesting these taxa may have a sister relationship.

Dolichorhinus hyognathus (Osborn, 1889)

Holotype

YPM PU10273, a mandible with right i1–c, left i3, c, p2, m2 (partial), and m3.

Type Locality

Adobe Town Member (Washakie B of Granger, 1909) of the Washakie Formation, Sweetwater County, Wyoming.

Synonyms

Dolichorhinus cornutum (Osborn, 1895); D. intermedius Osborn, 1908a; D. heterodon Douglass, 1909; D. longiceps Douglass, 1909; D. superior (Riggs, 1912); D. fluminalis Riggs, 1912.

Age

Middle Eocene (early Uintan land mammal “age”).

Referred Specimens

(From the Wagonhound Member of the Uinta Basin, Utah) AMNH 1832, skull fragments; AMNH 1836, a right mandibular ramus with p2–m3; AMNH 1837 (holotype of Dolichorhinus intermedius), a skull with left C–M3; AMNH 1840, a partial left mandibular ramus with p2–m3; AMNH 1843, an anterior portion of skull with partial cheek teeth; AMNH 1845, a partial skull missing the nasal and premaxillae with right P4–M3 and left M2–M3; AMNH 1847, a dorsal surface of a skull; AMNH 1849, a skull fragment; AMNH 1850, a skull with right and left C–M3; AMNH 1851 (holotype of D. cornutum), a skull with complete right and left dentition; AMNH 1852, a skull with right P1–M3 and left P2–M3; AMNH 1854, a partial mandible with right i2–c, p2–m1, left c, and p2–m1; AMNH 1856, a complete mandible with complete dentition; AMNH 1857, a mandible with heavily worn teeth including right i1–i2, p2–m3, left i1–i2, c, and p1–p4; AMNH 1858, a juvenile mandible with right c (erupting), p1, p4 (?), left p1, p4 (erupting), m1–m2, and m3 (erupting); CMNH 2340 (holotype of D. heterodon), a skull with right P3–M3 and left partial cheek teeth; CMNH 2347, a skull with left and right P2–M3; CMNH 2865, a skull with left P4–M3, partial right ramus with m2–m3; CMNH 2964, a skull fragment; CMNH 3095, a partial skull with no dentition; CMNH 3096, a skull with right M1–M3 (heavily worn); CMNH 3117, a posterior half of a skull with right and left M2–M3; CMNH 3119, a skull fragment; CMNH 11071, a complete skeleton; CMNH 11080, partial skull with right C–M3; CMNH 11081 (holotype of D. longiceps), a skull with right C, P2–M3 and left P2–M3; CMNH 11083, a skull (partially prepared); CMNH 11091, a skull, broken into two large fragments with right M1–M3, and a partial right mandibular ramus with m2–m3; CMNH 11092, a mandible with left p1, p2 (partial), and p3–m3; FMNH P12167, a skull with right C–M3, left I1–I2, C, P2–M3, and a mandible with right and left p2–m3; FMNH P12175, a skull with right and left C–M3; FMNH P12176, a skull with a left M3; FMNH P12182, a skull with right P2–P3 (partial), P4–M3, and left P3–M2; FMNH P12168 (holotype of D. superior), a skull with right C–P4, M3, and left C, P2–M3; FMNH P12193, a skull with right and left P1–M3; FMNH P12200, a skull with right and left P3–M3; FMNH P12205 (holotype of D. fluminalis), a skull with right I1–M3 and left I1–P2, M1 (partial), M2–M3; FMNH P12215, a skull with complete dentition; LACM 128402, a skull with right P3–M3 (partial) and left P1–M3; UCMP 31845, a partial skull with left P1–M3; UCMP 31846, a partial skull with partial dentition; USNM 6702, a partial mandible with right p4–m2 and left c–m2; USNM 6703, a skull with right P2–P4 and left P1–2, P4–M3; USNM (uncatalogued), premaxillomaxillary rostrum; YPM PU11241, a skull with right P1–M3 and left C (partial), P1–M3; (from the Adobe Town Member [Washakie B of Granger, 1909] of the Washakie Basin, Wyoming) AMNH 13164, a complete skull with complete right dentition, left C–M3, and a mandible with right i1–p4, m2–m3 (partial), and left i1–m2, m3 (partial); CMNH 9413, a skull with right P1–M3 and left I2–M3; FMNH PM3870, a skull with right and left I3–M3; FMNH PM3873, a skull with right and left P2–M3; FMNH PM26100, a mandible with damaged teeth.

Diagnosis

Dolichorhinus hyognathus is an intermediate-sized brontothere. The frontal bone does not overlap or protrude into the nasal bone. A small superorbital protuberance, primarily on the nasal bone, is seen in some specimens. The cranium of Dolichorhinus hyognathus is highly dolichocephalic. The nasal incision extends to the posterior margin of the P4. The nasal process is horizontal, unelevated, of relatively constant transverse width, narrow, with thin and relatively deep lateral walls, and without a strongly rounded distal margin. The orbits do not project laterally and are positioned directly over the posterior half of M2 and the anterior half of M3, with the anterolateral root of M2 and the posterolateral root of M1 directly below the anterior orbital rim. The prominent infraorbital process of the jugal also extends onto the maxillary, often to form a double flange. The premaxilla is robust and does not contact the nasal bone. The premaxillomaxillary rostrum does not deepen proximally and it is enclosed dorsally by bone. Dolichorhinus hyognathus lacks a sagittal crest, but the parasagittal ridges strongly constrict the dorsal surface of the skull posteriorly. Other cranial characteristics include a strongly dorsally arched cranium, thin and weakly curved zygomatic arches, a ventrally open and posteromedially angled external auditory pseudomeatus, and disproportionately wide occipital condyles. The functional posterior nares are shifted posteriorly by a bony palatal extension and posteriorly extended maxilloturbinates.

Dentally, Dolichorhinus hyognathus has large subcaniniform upper incisors, a postcanine diastema, a simple P1, a distinct P2 metacone, weak premolar preprotocristae, and short crests extending posteriorly from the premolar protocones. Premolar hypocones are exceedingly rare and inconspicuous when present. The molars have tall, lingually angled ectolophs with weak labial ribs, and thinned lingual ectoloph enamel with wedge-shaped paracones and metacones. A cingular parastyle shelf and an anterolingual cingular cusp are absent. Central molar fossae are present. D. hyognathus molars occasionally retain vestigial paraconules. All traces of a metaloph are lost. The lower dentition of D. hyognathus includes large subcaniniform incisors all of similar size, a postcanine diastema, a p1–p2 diastema of variable length, an elongate p2 trigonid, a metaconid on p4 but not on p2 and p3, shallow molar basins, and a slender m3.

Dolichorhinus hyognathus is most similar to Sphenocoelus uintensis in its hyperdolichocephalic skull, but it can be differentiated by the absence of both large fossae in the sphenoid and a sagittal crest, and by the presence of a highly specialized premaxillomaxillary rostrum and a more dorsally arched cranium.

Description

Skull

The following description of the skull of Dolichorhinus hyognathus is primarily based on AMNH 1845, AMNH 1851, AMNH 13164, and FMNH P12175, although other specimens provide additional information on variation. FMNH P12175 (fig. 13a, c,) and AMNH 1851 (fig. 13b, d) are complete and undistorted skulls. AMNH 1845 is another skull that has an intact occiput (fig. 13e). Finally, the skull of AMNH 13164 (figs. 13f, 14a) is slightly crushed dorsoventrally, obscuring some aspects of the shape of the cranium, but otherwise the preservation of this specimen is exquisite and reveals details of the rostrum and ventral surface that are not well-preserved in other skulls.

Figure 13

Selected skulls referred to Dolichorhinus hyognathus. (A) Right lateral view of FMNH P12175, (B) right lateral view of orbit and nasal horn of AMNH 1851, (C) dorsal view of FMNH P12175, (D) dorsal view of superorbital region of AMNH 1851, (E) posterior view of AMNH 1845, (F) oblique view of AMNH 13164 showing details of premaxillomaxillary rostrum.

Figure 14

The ventral surface and upper dentition of Dolichorhinus hyognathus, AMNH 13164. (A) Ventral view of skull, (B) right molars, (C) left premolars, (D) lingual view of right incisors, (E) labial view of right incisors.

Dolichorhinus hyognathus is an intermediate-sized (table 2) brontothere that is most notable for its extremely dolichocephalic skull and dorsally arching cranium. Only Sphenocoelus uintensis has a similarly elongate cranium. D. hyognathus lacks conspicuous horns, although many specimens, such as AMNH 1851 possess small but distinct nasal protuberances. Others, such as FMNH P12175, have much weaker nasal protuberances. The nasal protuberance of AMNH 1851 is small, elliptical, projects laterally, and is positioned directly above the orbit. The frontonasal suture runs behind the nasal protuberance and recedes posteromedially, but it is acutely redirected anteriorly near the midline. D. hyognathus does not possess a forward projection of the frontal bone like that seen in Telmatherium validus. The hornlike protuberance rests entirely on the nasal bone. Thus, D. hyognathus is the only brontothere known to have a hornlike protuberance that is predominantly on the nasal bone rather than on both the frontal and nasal bone.

Table 2

Summary statistics for selected morphometric variables of Dolichorhinus hyognathus See Methods for measurement definitions

The nasal incision of Dolichorhinus hyognathus is shallow and long. It extends as far back as the posterior margin of the P4. From a lateral view the orbits are elliptical. The bottom of the orbit is positioned directly over the posterior half of M2 and anterior half of M3. The anterolateral root of M2 and the posterolateral root of M1 are positioned directly below the anterior orbital rim. The orbits do not protrude laterally as in Metarhinus.

The nasal bones tend to be poorly fused together. In FMNH P12175, the nasal process is slightly longer than the premaxillomaxillary rostrum, although in other specimens (e.g., AMNH 1851) the nasal process and the rostrum are about the same length. The nasal process is horizontal. The lateral margins of the nasal process form dorsoventrally deep and thin vertical walls. Typically, the depth of the lateral walls is nearly constant, but they become shallower near the distal end. The anterior margin of the nasal process is thin, roughened, and strongly deflected downward. From the dorsal view the nasal process is narrow and of nearly constant width throughout its length, although it is sometimes slightly constricted at its proximal end.

The premaxillomaxillary rostrum of Dolichorhinus hyognathus is long and slightly upturned. From a lateral view, the dorsal margin of the rostrum is horizontal. Normally, the rostral cavity, which houses the vomeronasal organ, is bordered laterally and ventrally by the maxillaries and is open dorsally, thus forming a continuous osteological space with the nasal cavity. However, in D. hyognathus, the rostral cavity is completely covered by bone dorsally and is separated from the nasal chamber. This condition is shared with Metarhinus. In most specimens, the elements of the rostrum are completely fused into a solid dorsal cover and a distinct dorsal ridge of bone (the apparent osteological marker for the cartilaginous nasal septum) runs the length of the rostrum and extends into the nasal cavity of the skull.

The skull of AMNH 13164 allows for a more precise description of the specialized rostrum. The premaxillae are joined at the midline for almost their entire length and they form a domelike roof above the incisors. The posterior tips of the premaxillae diverge posterolaterally forming a posterior notch between the premaxillae. A thin pair of bones emerges horizontally from the nasal cavity and covers the dorsal surface of the rostrum, inserting into the notch formed by the premaxillae. Laterally, these small bones are sutured to the maxillaries. The origin of this unusual bony cover is uncertain. It continues posteriorly into the skull, but most specimens are filled with sediment or they are too heavily damaged to trace this structure internally. It seems most likely that these bones represent extensions of the maxilloturbinates, which, in the cross-sectional view of AMNH 1851 appear to extend into the rostral cavity below the surface of the bony covering (see Osborn, 1929a: fig 254c).

The dorsal surface of the skull above the orbits is slightly concave. Behind the orbits, the dorsal surface is strongly convex because the posterior half of the skull is strongly arched dorsally. The parasagittal ridges do not join to form a sagittal crest; instead, the parasagittal ridges remain separate throughout their length, although they strongly constrict the dorsal surface of the skull posteriorly.

The zygomatic arches are thin, shallow, and slightly bowed laterally. The jugal portion of the zygomatic arch is horizontal, while the squamosal portion is angled posterodorsally, thus giving the zygomatic arch a weak curvature. A conspicuous rounded infraorbital process extends ventrolaterally from the jugal. This process is similar to those of Mesatirhinus and Sphenocoelus, though it is generally larger and more laterally projected. In most specimens of D. hyognathus the flange created by the infraorbital process extends onto the maxillary forming a distinct secondary process. The degree to which the jugal and maxillary infraorbital processes are separated varies; sometimes it appears as a single flange (e.g., CMNH 3117).

From a dorsal view of the skull the nuchal crest has a shallow but wide median notch. From the posterior view, the nuchal crest is strongly arched dorsally. The width of the upper portion of the occiput is similar to the width of the ventral portion in AMNH 1845, although in other specimens the dorsal portion of the occiput is narrower. The width of the occiput is not strongly constricted in the middle as in Mesatirhinus. Small occipital pillars are visible on the posterior surface of the occiput, although the middle of the occiput is not deeply recessed between these structures. The occipital condyles of Dolichorhinus hyognathus seem disproportionately large and are almost as wide as the entire occiput.

Peterson (1924) first noted many of the peculiar aspects of the posterior nares of Dolichorhinus hyognathus. A distinct rim of bone, seemingly the original position of the posterior nares, is positioned between the anterior margins of the M3s. The position of this rim fluctuates between the hypocones of the M2 (e.g., CMNH 3117) to between the M3 protocones (e.g., FMNH P12167). A raised palatal extension is seen behind the original margin of the posterior nares. In AMNH 13164, this palatal extension is about three cm long and shifts the position of the posterior nares to a point between the posterior margins of the M3s. However, this palatal extension is commonly longer and shifts the posterior nares to well behind the M3s (e.g., AMNH 1845). A short median process extends posteriorly from both the original border of the posterior nares and from the palatal extension. Behind the palatal extension, an elongate posterior narial canal is bisected lengthwise by a thin vomerine septum. Two elongate pouches of bone (choanal pouches) project beyond the palatal extension and fill roughly the anterior two thirds of the posterior narial canal. The bony choanal pouches that extend beyond the palatal extension are similar to those seen in Telmatherium validus and Metarhinus abbotti, although they are more elongate and positioned much farther posteriorly in D. hyognathus. The thin pouches appear to be extensions of the maxilloturbinates, as seen in a cross section of AMNH 1851 (see Osborn 1929a: fig 254c). Finally, the functional posterior nares are situated near the end of the posterior narial canal. In AMNH 13164, the posteriorly shifted right functional posterior naris can be clearly seen behind the bony choanal pouch. The left posterior naris of this specimen is still filled with sediment.

Dolichorhinus hyognathus lacks large paired ventral sphenoidal fossae as seen in Sphenocoelus uintensis, however in AMNH 13164 a pair of very small ovoid pits can be seen on the ventral surface of the basisphenoid. The pits are separated by a rod-shaped structure that joins the thin vomerine septum. However, most other specimens of D. hyognathus lack these small pits entirely and have a relatively normal basisphenoid.

The external auditory pseudomeatus enters the skull at a posteromedial angle, a condition shared with Sphenocoelus. Other aspects of the basicranium of Dolichorhinus hyognathus are more typical. For instance, the external auditory pseudomeatus is wide and unconstricted ventrally. The main basicranial foramina such as the foramen lacerum and the foramen ovale are widely spaced.

Upper Dentition

The following description of the upper dentition of Dolichorhinus hyognathus is based primarily on AMNH 13164 (fig. 14), although additional information from other specimens is provided. The three upper incisors are large and form an arched row that extends anterior to the canines. The incisors increase in size laterally. The crowns are subcaniniform: they are short, pointed, and lingually curved. Each crown is only slightly bilaterally asymmetrical from the lingual view. Each incisor has a thick lingual cingulum. No specimen of D. hyognathus exhibits labial incisor cingula. The precanine and postcanine diastemata are shorter than the P2. The canine of AMNH 13164 is small, as is typical of many specimens of this species.

The P1 crown is relatively simple with a single cusp and a narrow posterior heel. Other specimens (e.g., AMNH 1852) have slightly less elongate P1s. There is no P1–P2 diastema, although a p1–p2 diastema is found in the mandible (see below). The remaining premolars (P2–P4) become progressively larger and less oblique posteriorly. The anterior margin of the P2 crown is strongly posterolingually angled, although in other specimens the P2 is more nearly square (e.g., AMNH 1851). The parastyle and metastyle of P2 arch slightly lingually. The parastyle and metastyle of P3 are nearly straight, while those of P4 are angled labially. The labial paracone ribs of the P2–P4 become shorter and narrower in more posterior premolars. The metacone of P2 is shifted lingually in comparison to those of P3 and P4. Because of these differences, the labial wall of the P2 is rounder than those of P3 and P4. A small bulge of enamel can be found at the labial base of the metacone of P4. This small bulge is not present in every specimen, and in some others it is enlarged and forms a short mesostyle (CMNH 11081, FMNH P12182).

The lingual side of the P2 crown exhibits a large protocone, a distinct lingual crest extending posteriorly from the protocone, and a small preprotocrista. In P3, these crests are faint. P4 lacks these crests altogether, but there is a tiny beadlike paraconule. The lingual sides of the premolars of Dolichorhinus hyognathus are morphologically unstable. For instance, in FMNH PM3870 there are no lingual crests extending posteriorly from the protocones and each premolar (P2–P4) has a small but distinct preprotocrista. Other specimens have large lingual crests extending posteriorly from the protocones, or even distinct hypocones, but this last condition is very rare. For instance, CMNH 11081, a specimen with heavily worn dentition, has a small hypocone on P2. The shape of the wear facets on P3 and P4 of that specimen suggests that hypocones were present on those premolars as well, but were worn off.

The labial cingula of the premolars are thin but distinct. They tend to be discontinuous around the proximal base of the paracone in P2 and P3, but continuous around the base of the paracone of P4. The anterior and posterior premolar cingula are thick and stretch around the lingual sides of the crowns. They are not connected lingually in AMNH 13164, but in other specimens (e.g., AMNH 1850) they form continuous lingual cingula.

The molars of Dolichorhinus hyognathus exhibit numerous brontotheriine apomorphies such as tall, lingually angled ectolophs, weak labial ribs, thin lingual ectoloph enamel, and wedge-shaped lingual borders of the paracone and metacone visible in molars that are not heavily worn. The anterior cingulum is thin and passes proximally to the distal peak of the parastyle. D. hyognathus molars lack anterolingual cingular cusps. Shallow central molar fossae are invariably present. Each of the molars of AMNH 13164 retains a vestigial paraconule, however the size and presence of paraconules in the molars of D. hyognathus is variable (e.g., smaller in FMNH P12182, absent in FMNH P12175). All traces of a metaloph on M1 and M2 are lost. The M3 of AMNH 13164 has a small hypocone and small metalophlike ridge. However, the presence and size of these structures varies. For instance, in FMNH P12175 and FMNH P12188 the M3 hypocone is as large as that of the M2. On the other hand, FMNH PM3873 lacks a M3 hypocone altogether.

Mandible and Lower Dentition

The description of the mandible and lower dentition of Dolichorhinus hyognathus is based on two nearly perfectly preserved and essentially identical mandibles (AMNH 13164 and AMNH 1856) with complete and lightly worn teeth (fig. 15). The holotype mandible (YPM PU10273) is also pictured (fig. 16). The mandible of D. hyognathus is distinctive in the slenderness of the horizontal ramus. The ascending ramus is short. The coronoid process is long, curves posteriorly, and is taller than the mandibular condyle. The symphysis is long and relatively slender in comparison to most other brontotheriids. The inferior margin of the symphysis has a very shallow angle (much less than 45°), although the symphyses of AMNH 13164 and AMNH 1856 are somewhat steeper than that of the holotype. The symphysis extends posteriorly to the trigonid of the p3, although in some specimens it extends only to the p2 talonid (e.g., AMNH 1587).

Figure 15

Selected views of mandibles and lower dentitions of Dolichorhinus hyognathus. (A) Left view of AMNH 13164, (B) dorsal view of AMNH 1856, (C) left premolars of AMNH 1856, (D) lingual view of incisors and canines of AMNH 1856, (E) labial view of left incisors and canine of AMNH 1856.

Figure 16

Holotype of Dolichorhinus hyognathus (courtesy of Division of Vertebrate Paleontology, YPM PU10273. © 2005 Peabody Museum of Natural History, Yale University, New Haven, Connecticut, USA. All rights reserved.). (A) Left view, (B) dorsal view.

The three lower incisors are rather large and form a broad semicircular arch. The incisors are mesiodistally elongate and roughly of the same size. The crowns are short, lingually curved, slightly bilaterally asymmetrical, and with blunt distal points. From the lingual view, the incisors each have a prominent lingual rib and distinct lingual cingulid. It is apparent from specimens with more heavily worn incisors that these characters tend to fade with wear. No labial cingulids are seen on the lower incisors. There are no gaps between any of the incisors, although there is occasionally a very short precanine diastema. The lower postcanine diastema is typically longer than the upper postcanine diastema. In AMNH 1856 and AMNH 13164, the postcanine diastema is about the length of the p2. The lower canine is typically small.

The p1 is a small elongate tooth with a single cusp and a narrow bladelike talonid heel. A p1–p2 diastema is always present, although its length is variable. In AMNH 13164, for instance, it is visibly shorter than that of the holotype jaw, YPM PU10273. The trigonid of the p2 is much longer than the talonid, although the trigonid and talonid are of similar width. The paralophid of p2 is essentially straight, although in some specimens it can curve slightly lingually, creating a small lingual notch in the trigonid. The p2 protolophid is short, straight, and extends in a posterior direction from the protoconid. The p3 trigonid is slightly longer and narrower than the talonid. The paralophid of the p3 curves in a slightly lingual direction, creating a distinct lingual notch in the trigonid. The p3 protolophid extends about equally lingually and posteriorly from the protoconid. The p4 trigonid is clearly shorter and narrower than the talonid. The paralophid of p4 strongly arches lingually creating a very broad lingual notch in the trigonid. The p4 is the only premolar with a large lingually positioned metaconid. The p2 talonid has a short cristid obliqua and hypolophid with a shallow sloping notch on the lingual side of the crown. The talonids of p3 and p4 have longer and more well-developed cristids obliqua and hypolophids with broader and more nearly molariform talonid basins.

The lower molars of Dolichorhinus hyognathus have relatively thin lingual enamel, shallow talonid and trigonid basins, and an elongate m3. The cheek teeth of AMNH 13164 lack lingual cingulids and the labial cingulids are thin and discontinuous around the bases of the cusps. The distinctness of the labial cingulids is variable; this is at least partly related to the degree of dental wear.

Remarks

Dolichorhinus hyognathus (Osborn, 1889) was based on a mandible, YPM PU10273 (fig. 18). The holotype can be distinguished from Sphenocoelus, Mesatirhinus, Telmatherium, Sthenodectes, and Metarhinus by the combination of its more slender proportions, shallow angle of the ventral margin of the symphysis, and retention of a p1–p2 diastema. Osborn (1889) originally assigned this species to the genus Palaeosyops. Shortly thereafter, Earle (1892), who described YPM PU10273 in more detail, reassigned this species to the genus Telmatotherium (a variation on the spelling of Telmatherium).

Figure 18

Selected skulls referred to Sphenocoelus uintensis. (A) Left view of DMNH 479, the holotype of Tanyorhinus bridgeri, (B) ventral view of UCMP 81310, (C) dorsal view of DMNH 2830, (D) anterodorsal view of the nasal and rostrum of DMNH 2830.

In 1895, Osborn described a complete skull (AMNH 1851) that he referred to Telmatotherium cornutum; he also referred several skulls to the same species, including AMNH 1850, AMNH 1847, AMNH 1848, AMNH 1852, and AMNH 1837. These specimens represent a species that was “remarkable for its very long flat-topped cranium and its incipient knoblike osseous horns borne chiefly upon the nasals, but partly upon the frontals” (Osborn, 1895: 92). None of these skulls was associated with a jaw that would allow direct comparison with the type jaw of T. hyognathus. Nonetheless, Osborn conjecturally referred several mandibles (AMNH 1857, AMNH 1858, AMNH 1854, and AMNH 1855) to T. cornutum. Based on these jaws, he distinguished T. cornutum from T. hyognathus “by the presence of two incisors” (p. 92). However, this was clearly a mistake, because one of these mandibles (AMNH 1854) shows three unmistakable incisor alveoli. The other specimens are too damaged or the teeth are too worn to determine the number of incisors.

Shortly thereafter, Hatcher (1895) reassigned the species Telmatotherium cornutum to a new genus, Dolichorhinus, but made no mention of T. hyognathus. Subsequently, Osborn (1908a) named another species, D. intermedius. This species was based on a skull (AMNH 1837) that was essentially the same as those he had referred to D. cornutum. Douglass (1909) and Riggs (1912) continued the trend of erecting new species based on skulls similar to those that Osborn had referred to D. cornutum. Douglass (1909) erected D. heterodon (CMNH 2340) and D. longiceps (CMNH 2347), while Riggs (1912) erected D. fluminalis (FMNH P12205) and D. superior (FMNH P12168). (The holotype of D. superior was mistakenly reported by Riggs (1912) to be FMNH P12188. However, records at the FMNH indicate that the holotype specimen has always been catalogued as FMNH P12168 [William Simpson, personal commun., 2004]). When all of these supposed species were distinguished from others, differentiaion was based on (1) minor size differences or (2) taphonomic distortion and/or damage. However, all of these holotype skulls strongly resemble the holotype skull of Dolichorhinus cornutum, and I have not been able to find any compelling morphological differences between them. Peterson (1924) expressed doubt over the validity of the large number of Dolichorhinus species, although he did little to rectify this problem and continued to “provisionally” accept all of the species that had been named by Osborn, Douglass, and Riggs.

Osborn (1908a, 1929a) ultimately concluded that the neglected holotype jaw of Telmatotherium hyognathus (Osborn) (YPM PU10273) represented the same species as the holotype skull of Dolichorhinus cornutum (AMNH 1851). Osborn (1929a) reassigned T. hyognathus to the genus Dolichorhinus and considered D. cornutum to be a junior synonym of D. hyognathus. At that point, D. hyognathus became the type species of Dolichorhinus. Despite this one revision, Osborn (1929a) continued to recognize all of the other species of Dolichorhinus as valid, despite the similarities of the holotype skulls of each of these supposed species.

Mader (1989) expressed doubt about Osborn's (1908a) decision to synonomize D. cornutum with D. hyognathus; he considered D. cornutum to be the type species and considered D. hyognathus to be a nomen dubium. Subsequently, Mader (1998) accepted only two species of Dolichorhinus, D. hyognathus and D. intermedius, but he mistakenly reassigned these species to the genus Sphenocoelus. However, it can be shown that Dolichorhinus (sensu Hatcher and sensu Osborn) is clearly distinct from Sphenocoelus (sensu Osborn). There are at least six conspicuous differences between Sphenocoelus and Dolichorhinus: (1) Sphenocoelus lacks the specialized morphology of the premaxillomaxillary rostrum seen in Dolichorhinus. (2) Sphenocoelus has large ventral sphenoidal fossae, while Dolichorhinus does not. (3) Sphenocoelus has a sagittal crest, while Dolichorhinus does not. (4) The posterior half of the cranium of Dolichorhinus is more strongly arched than that of Sphenocoelus. (5) Dolichorhinus molars have distinct central fossae, whereas central molar fossae are apparently variable in Sphenocoelus. (6) Finally, there is a p1–p2 diastema in the mandibles of Dolichorhinus, whereas Sphenocoelus lacks a p1–p2 diastema.

Fortunately, there are at least two associated skulls and mandibles (e.g., AMNH 13164 and CMNH 11017) that allow one to confirm that the holotype mandible of Dolichorhinus hyognathus does belong to the same species as the highly elongate, distinctive skulls that represent other supposed species of Dolichorhinus (D. cornutum, D. intermedius, D. heterodon, D. longiceps, D. superior, and D. fluminalis). The characteristics found to be variable among these specimens is consistent with intraspecific variation found in most species of brontotheres and does not suggest multiple species. These variable characters include the presence or absence of M3 hypocones, the presence or absence of molar paraconules, the presence or absence of a minor P4 mesostyle, the variable lingual morphology of the premolars, and the size and distinctness of hornlike protuberances. Therefore, all other species of Dolichorhinus are considered junior synonyms of D. hyognathus (Osborn, 1889).

Sphenocoelus uintensis Osborn, 1895

Holotype

AMNH 1501, the posterior part of a skull.

Type Locality

Northeastern Utah, Wagonhound Member (Uinta B1) of the Uinta Formation, Wyoming.

Synonyms

Tanyorhinus blairi Cook, 1926 and Tanyorhinus bridgeri Cook, 1926.

Age

Middle Eocene (early Uintan land mammal “age”).

Referred Specimens

(From the Wagonhound Member of the Uinta Formation of Utah) CMNH 2963, a posterior part of a cranium (partially prepared); (from the Sand Wash Basin of Moffat County, Colorado) DMNH 479 (holotype of Tanyorhinus bridgeri), a skull with left P3–P4, M2–M3; DMNH 484, a right maxilla with C, P2–M2; DMNH 507, an anterior portion of a skull with left P1–M3; DMNH 509, an anterior portion of a skull with right P3–P4 (all partial) and left P2–P3; DMNH 517, a right premaxillomaxillary fragment with isolated incisors and isolated P2–M1; DMNH 541 (holotype of Tanyorhinus blairi, in part), a skull with P1–P4, M3, and left P1–M3; DMNH 542 (holotype of Tanyorhinus blairi, in part), a mandible with right i1–m3, left i1–p4, and m2–m3; DMNH 2830, a skull with right I2–I3, P1–M3, left P2–M3, and a complete mandible with complete dentition; DMNH 14219, a mandible with complete dentition; DMNH 29411, a skull with right P1–M3 and left C–M3; (from the Adobe Town Member [Washakie B of Granger, 1909] of the Washakie Formation, Wyoming) UCMP 81281, a partial mandible with right p3–p4 and left p2–m1; UCMP 81301, a skull that is missing the premaxilla and zygomatic arches with right P3–M3 and left M1–M3; UCMP 81443, a partial mandible with right i1–c, p2–p4, and left i1–p3.

Specimens Referred to cf. Sphenocoelus

The following specimens are consistent with Sphenocoelus uintensis and probably belong to that species, but they lack sufficiently diagnostic characters to definitively refer them to S. uintensis: (from the Sand Wash Basin of Moffat County, Colorado) DMNH 504, a mandible with right and left p1–m3; DMNH 505, a partial mandible with right m1–m2, left m2–m3, isolated canine, and right p4; DMNH 506, left M1–M3 (very worn); DMNH 506a, right M3; DMNH 556, a mandible fragment with right p4–m2 and m3 (partial); DMNH 2584, a partial mandible with right p2–m3; (from the Adobe Town Member [Washakie B of Granger, 1909] of the Washakie Formation, Wyoming) UCMP 81294, a partial mandible with right p3–m3; UCMP 81299, a partial mandible with right p3–m3; UCMP 81366, a mandible fragment with left p4–m2; UCMP 81369, a left M1 or M2; UCMP 81370, a mandible fragment with left p3–m1; UCMP 81402, a right maxilla fragment with P1–P4; UCMP 81412, a left P2; UCMP 81413, left p2; UCMP 81414, a left P3 or P4; UCMP 81416, a right P4; UCMP 81422, a right M2 (partial), and M3; UCMP 81438, a right maxilla fragment with P1 (roots) and P2–P3; UCMP 81440, a partial mandible with broken incisors, right p2–p4, left m2 (partial) and m3; UCMP 81448, a partial mandible with left p3–m3; UCMP 81451, a partial mandible with right p2 (partial), p3–m2, and m3 (partial); UCMP 81462, an isolated incisor.

Diagnosis

Sphenocoelus uintensis is an intermediate-sized hornless brontothere in which the frontal bone does not overlap or intrude into the nasal bone. The cranium of S. uintensis is highly dolichocephalic. The nasal incision extends as far back as the anterior margin of the M1. The nasal process is horizontal, unelevated, of relatively constant transverse width, narrow, with thin and relatively shallow lateral walls, and without a well-defined or strongly rounded distal margin. The orbits do not protrude laterally; they are positioned above the posterior part of M2 and the anterior part of M3 with the anterolateral root of M2 and the posterolateral root of M1 below the anterior orbital rim. There is a prominent infraorbital process on the jugal. The premaxillomaxillary rostrum deepens posteriorly and is not covered by bone dorsally. Other cranial characteristics include a short sagittal crest, thin and weakly curved zygomatic arches, a ventrally open and steeply posteriorly angled external auditory pseudomeatus, disproportionately wide occipital condyles, and large paired ventral sphenoidal fossae. The cranium is dorsally arched but more weakly so than Dolichorhinus.

Dentally, Sphenocoelus uintensis is characterized by large subcaniniform upper incisors, a postcanine diastema, a simple P1, a distinct P2 metacone, weak premolar preprotocristae, and with short crests extending posteriorly from the premolar protocones. Premolar hypocones are absent. The molars of S. uintensis have tall, lingually angled ectolophs with weak labial ribs, and thinned lingual ectoloph enamel with wedge-shaped paracones and metacones. A cingular parastyle shelf and an anterolingual cingular cusp are absent. Central molar fossae are present in some specimens but absent in others. S. uintensis molars occasionally retain vestigial paraconules, but all traces of a metaloph are lost. The lower dentition of S. uintensis includes large semispatulate incisors that are all of a similar size, a postcanine diastema, no p1–p2 diastema, an elongate p2 trigonid, a metaconid on p4 but not on p2 and p3, shallow molar basins, and a slender m3.

Sphenocoelus uintensis is most similar to Mesatirhinus junius, but it is clearly distinct from that species due to its larger size, more dolichocephalic proportions, and paired ventral sphenoidal fossae. Likewise, paired sphenoidal fossae, an unspecialized premaxillomaxillary rostrum, and a sagittal crest distinctly set it apart from Dolichorhinus hyognathus.

Description

Skull

The holotype of Sphenocoelus uintensis (AMNH 1501) is the posterior part of a skull from the Uinta Basin that has suffered minor shearing distortion but is otherwise in good condition (fig. 17). Other specimens referable to S. uintensis have been recovered from the Sand Wash Basin of Colorado (e.g., Cook, 1926) and from a collection of the Washakie Basin of Wyoming made by McKenna and Kent in 1954. Among these are several complete (or nearly complete) skulls and a variety of cranial fragments and mandibles. The following description of the skull and upper dentition is based upon the holotype and five other skulls, UCMP 81301 (fig. 18b) DMNH 479 (fig. 18a), DMNH 541 (not shown), DMNH 2830 (figs. 18c–d, 19a), and DMNH 29411 (not shown). Other specimens provide additional information on variation.

Figure 17

Holotype partial cranium of Sphenocoelus uintensis (AMNH 1501). (A) Ventral view, (B) dorsal view, (C) posterior view, (D) left view.

Figure 19

Views of the ventral skull surface and upper teeth of Sphenocoelus uintensis. (A) Ventral view of DMNH 2830, (B) left premolars of DMNH 29411, (C) left molars of DMNH 29411, (D) labial view and (E) occlusal view of isolated right incisors of DMNH 517.

Sphenocoelus uintensis is an intermediate-sized brontothere (table 3) whose skull is generally similar to Mesatirhinus junius but notably more dolichocephalic. Hornlike nasal or frontonasal protuberances are not seen in S. uintensis. The frontonasal suture, most clearly visible in UCMP 81310 (not shown), recedes posteromedially, but near the midline the direction of the suture is acutely redirected anteriorly. The frontal does not overlap the nasal or protrude into the nasal as in Telmatherium validus.

Table 3

Summary statistics for selected morphometric variables of Sphenocoelus uintensis See Methods for measurement definitions

The nasal incision is shallow and long and extends as far back as the anterior edge of M1, while the orbit is positioned above the posterior part of M2 and the anterior part of M3. The anterolateral root of the M2 and the posterolateral root of the M1 are situated directly below the anterior orbital rim.

The premaxillomaxillary rostrum lacks the specializations seen in Dolichorhinus. From a lateral view of the skull the dorsal edge of the premaxillomaxillary rostrum ascends posterodorsally. The posteriormost notch of the nasal incision is level with the top of the orbit. The nasal processes of the premaxillae diverge laterally and the premaxillomaxillary cavity is open dorsally. A distinct remnant of the right premaxillomaxillary suture in UCMP 81310 indicates that the premaxilla does not contact the nasal bone.

The bones of the nasal process are not strongly fused together. The nasal processes are slightly shorter and somewhat narrower than the premaxillomaxillary rostrum. The transverse width of the nasal process is nearly constant throughout its length. In the least distorted skulls, such as DMNH 479, the nasal process extends horizontally from the skull, or it bends just slightly downward so that the dorsal surface of the nasal process is slightly convex. The lateral walls of the nasal process are relatively thin and dorsoventrally shallow, but they are of constant depth throughout the length of the nasal process. The anterior border of the nasal process is thin, roughened, and angled downward moderately.

The dorsal surface of holotype skull fragment (AMNH 1501) is essentially flat, although complete skulls indicate a more dorsally arched cranium. In the least distorted skulls (DMNH 29411, DMNH 479, and UCMP 81301), the dorsal surface of the skull above the orbits is flat or slightly concave. However, the dorsal surface of the posterior half of the skull is strongly convex. The highly elongated posterior half of the cranium tends to be slightly dorsally arched from a lateral view, but not as strongly as the skull of Dolichorhinus hyognathus. In the holotype and in other specimens, the parasagittal ridges merge to form a short sagittal crest.

The zygomatic arches are relatively thin, slender, and slightly bowed laterally. The jugal portion of the zygomatic is essentially horizontal, while the squamosal portion rises posteriorly at a moderate angle, thus giving the zygomatic arch a weakly curved shape. The jugal of Sphenocoelus uintensis has a large rounded infraorbital process. The infraorbital process resembles those of Mesatirhinus and Dolichorhinus, but it is larger and more distinct than that of Metarhinus.

The occipital can be adequately described from the holotype skull (AMNH 1501). The occiput is moderately tilted backward. From a dorsal view of the skull the nuchal crest is thin and concave. From the posterior view the nuchal crest is arched. The center of the occiput is not deeply recessed between the small occipital pillars. The occiput is narrower dorsally than it is ventrally. The occipital condyles of Sphenocoelus uintensis are disproportionately very large and almost as wide as the entire occiput.

Many aspects of the ventral surface of the skull of Sphenocoelus uintensis are most clearly revealed in UCMP 81310 (fig. 18b). The anterior rim of the posterior nares is positioned at the anterior margin of the M3. The anterior edge of the posterior nares is abrupt; that is, there is no horseshoe-shaped emargination, nor is there a bony palatal extension like that of Dolichorhinus. A short median process protrudes posteriorly from the midline of the anterior edge of the posterior nares, marking the posteriormost contact point of the vomer and palatal bones. The posterior narial canal is extremely long and continues into a large cavity in the sphenoid bone. The vomer is missing in UCMP 81310, but it would have originally formed a long, thin plate of bone that bisected the elongate posterior narial canal. Remnants of the elongate vomer can be seen in the form of a thin ridge of bone running along the dorsal roof of the posterior narial canal in UCMP 81301.

The posterior narial canal continues into a large vacuity in the sphenoid bone. Deep vacuities in the sphenoid can also be seen in the holotype (AMNH 1501) and in all other skulls of Sphenocoelus uintensis. Osborn referred to these fossae as “pits” (Osborn, 1895) and “sphenoidal pits” (Osborn, 1929a). In some specimens, such as UCMP 81310, the cavities formed by the ventral sphenoidal fossae are large, although they seem to have been narrower in others such as AMNH 1501 and DMNH 2830. The basisphenoid is highly modified and forms a narrow septum that partitions the ventral sphenoidal fossae; this is most clearly seen in AMNH 1501 (fig. 17). The partitioning basisphenoid is so thin that it is commonly not preserved.

In most of the specimens the ventral sphenoidal fossae form a continuous channel with the posterior narial canal. The thin partitioning basisphenoid would have connected with the elongate vomer to form a continuous partition of the posterior narial canal and sphenoidal fossae. The holotype (AMNH 1501) is somewhat unusual in this respect because the ventral sphenoid fossae appear to extend deeper into the ventral surface of the cranium, thus forming distinct pits. Apparently the depth and diameter of the ventral sphenoidal fossae are intraspecifically variable.

The external auditory pseudomeatus, formed by the mastoid and postglenoid processes of the squamosal bone, enters the skull at a strongly posteromedial angle, a condition shared with Dolichorhinus. Other aspects of the basicranium of Sphenocoelus uintensis are more typical. For instance, the external auditory pseudomeatus is not enclosed ventrally and the foramen ovale is widely separated from the foramen lacerum. One final peculiar aspect of S. uintensis is a distinct fossa on the ventral surface of the zygomatic process of the squamosal just lateral to the glenoid fossa (this is best seen on the holotype skull fragment, AMNH 1501).

Upper Dentition

Unfortunately, none of the skulls of Sphenocoelus uintensis has an intact set of upper incisors. In DMNH 2830 there are six incisor alveoli that form a semicircular arch anterior to the canines and are separated from the canines by a short I3–C diastema (fig. 19a). The two preserved incisors (right I2, I3) are large and though they are worn, they appear to have been subcaniniform (conular and lingually curved). The I3 is larger than the I2. Three isolated incisors are associated with DMNH 517, a maxilla that appears to be S. uintensis (fig. 19d, e). Although there is no way to know which crown is which, judging by comparison with the partial set of incisors in the skull of DMNH 2830, they appear to form a right incisor row, with the smallest incisor I1 and the largest I3. The crowns are subcaniniform with lingually curved crowns and narrow lingual heels. The I1 is the shortest incisor, while I2 and I3 are progressively taller and larger in diameter.

The canines are not well preserved in any of the skulls. One specimen, DMNH 29411, retains a left canine with a tall crown and a rounded cross section. Another specimen, DMNH 2830, includes an isolated canine of similar morphology. Despite the rather dolichocephalic proportions of the skull of S. uintensis, the postcanine diastemata of S. uintensis is very short in all specimens and is usually only a few millimeters long.

The most adequately preserved sets of cheek teeth are of DMNH 2830, DMNH 29411, and UCMP 81310. Figured are the complete and relatively unworn cheek teeth of DMNH 29411 (fig. 19b, c). The P1 is simple, with a single cusp and a low posterior heel. The P2 is slightly oblique in outline due to a posterolingually angled anterior edge. The anterior and posterior sides of P3 and P4 are more nearly parallel. The labial side of P2 is strongly rounded, while those of P3 and P4 are progressively flatter. The parastyle of P2 arches slightly lingually, while the metastyle is nearly straight. The metacone of P2 is positioned slightly lingually with respect to the protocone. The parastyle and metastyle of P3 are straight, while those of P4 are angled slightly labially. Small labial paracone ribs are present on P2–P4 and become progressively shorter on more posterior premolars.

A very small but distinct preprotocrista connects the protocone with the lingual base of the paracone in P2. In P3 and P4 this crest becomes progressively smaller, so that it is barely perceptible in P4. A low crest descends the posterior slope of the protocone on P2–P4. This lingual crest is not present on more worn sets of dentition (e.g., DMNH 2830), although it is present on the P2, P3, and P4 of DMNH 29411. Yet in others (e.g., UCMP 81301) the lingual crest is not seen on P4. There are no hypocones on the premolars. The anterior and posterior cingula of the premolars are thick and most often stretch around the lingual side forming a continuous lingual cingulum, although it is occasionally discontinuous, as is the case in UCMP 81301. The labial premolar cingula of the P2 and P3 typically connect to the posterior ridge of the paracone rib, but the labial cingulum of P4 most often stretches across the base of the crown.

The molars are Sphenocoelus uintensis have tall, lingually angled ectolophs, weak labial ribs, thin lingual ectoloph enamel, and wedge-shaped lingual sides of the paracone and metacone in molars that are not heavily worn (e.g., M3). The anterior cingulum is thin and passes proximally to the distal peak of the parastyle. Sphenocoelus uintensis molars lack an anterolingual cingular cusp. Very small paraconules are occasionally retained on the molars, although no evidence of a metaloph was found on any specimen. Among the specimens referred to Sphenocoelus uintensis, some (DMNH 507, DMNH 484, and DMNH 29411) have shallow but distinct central fossae in the molars. However, in other specimens (UCMP 81301, DMNH 2830, DMNH 541, and DMNH 517) they appear to be absent or very shallow. A M3 hypocone is generally absent, although a thick cingulum traces around the distolingual corner of the M3 crown. Cook (1926) reported a small “postero-internal” cusp on the M3 of DMNH 479, but I was unable to examine the ventral surface of that skull because it is part of a composite skeleton encased in glass on exhibit at the DMNH. The lingual molar cingula are typically discontinuous around the protocone and hypocone. Likewise, the labial molar cingula are thin and tend to be discontinuous around the mesostyles.

Mandible and Lower Dentition

One skull of Sphenocoelus uintensis is associated with a mandible (DMNH 2830) that retains a complete set of well-worn lower teeth (fig. 20a–b). Additionally, several mandibles, some with less worn dentition, are morphologically consistent with the mandible of DMNH 2830, but are not directly associated with diagnosable skulls. The following description of the mandible and lower dentition of S. uintensis is based primarily on DMNH 2830 and other specimens with well-preserved incisors and premolars (UCMP 81281 and UCMP 81443) (fig. 20c–e).

Figure 20

Selected mandibles and lower dentitions of Sphenocoelus uintensis. (A) Right view of DMNH 2830, (B) dorsal view of DMNH 2830, (C) left p2, p3, and p4 of UCMP 81281, (D) dorsal view of UCMP 81443, (E) labial view of incisors and canines of UCMP 81443.

The horizontal ramus of S. uintensis is elongate and shallow, like that of Dolichorhinus. However, the ventral margin of the symphysis is steeper than Dolichorhinus hyognathus, and more similar to that of Mesatirhinus. The position of the posterior margin of the symphysis varies slightly but it usually extends to the posterior part of the p3. The three pairs of incisors are large and form an arched row that reaches anterior to the canines. The incisors increase in size laterally. There are small gaps between the incisors of DMNH 2830, but these gaps are probably a result of extensive wear. UCMP 81443 has a better set of incisors, though they are also heavily worn. In this specimen, there are no gaps between the incisors. Judging by the shape of the wear facets of UCMP 81443, the i1 and i2 were semispatulate, while the i3 was more subcaniniform. A distinct lingual cingulid can be seen on each incisor. Moreover, each incisor has a distinct labial cingulid.

The lower canines of DMNH 2830 are large although the canines of other specimens (UCMP 81443) are somewhat smaller. The canine is followed by a postcanine diastema that is generally shorter than the p2. However, the lower postcanine diastema is longer than the upper postcanine diastema. The p1 is a simple tooth with a single cusp and a short and narrow talonid. There is no p1–p2 diastema in DMNH 2839 or any other mandible referred to Sphenocoelus uintensis. This is an important diagnostic feature that helps to differentiate the mandibles of S. uintensis from those of Dolichorhinus hyognathus, a species with a distinct p1–p2 diastema.

The p2–p4 of DMNH 2830 and UCMP 81443 are moderately worn, but much of their morphology is still discernable. UCMP 81281 has unworn premolars. The p2 trigonid is much longer than the talonid, but the trigonid and talonid are of similar width. The talonid and trigonid of p3 are of comparable length but the trigonid is slightly narrower than the talonid. The trigonid of the p4 is both shorter and narrower than the talonid. The paralophid of the p2 is either straight (e.g., UCMP 81281) or slightly arched lingually (UCMP 81443), thus creating a small lingual notch in the trigonid. The p3 paralophid arches at a slightly more lingual angle than that of the p2. The p4 paralophid is strongly directed lingually, creating a broad lingual trigonid notch; that of p3 is angled somewhat lingually, and that of p4 arches fully lingually. A metaconid is present only on p4. The talonid of p2 has only a minor lingual notch and a short cristid obliqua and hypolophid. The p3 and p4 have longer cristids obliqua and hypolophids with much broader talonid basins.

The molars of Sphenocoelus uintensis are typical of brontotheriines, with relatively thin lingual enamel and an elongate m3. The m3 hypoconulid heel of DMNH 2830 is unusually narrow, but this is atypical. Other specimens have a broader m3 hypoconulid. Lingual cingulids are absent, while the labial cingulids of p2–m3 vary in distinctness (this is related to wear), and tend to be discontinuous around the protoconid. Occasionally the m3 cingulid traces around the distal end of the hypoconulid (e.g., DMNH 14219).

Remarks

Osborn (1895) recognized that AMNH 1501 (fig. 17) represented a new species, Sphenocoelus uintensis, due primarily to the peculiar pair of ventral sphenoidal fossae. Because the specimen lacked teeth, Osborn (1895) was initially unsure what family of perissodactyls S. uintensis belonged to, but he ultimately (Osborn, 1929a) concluded that S. uintensis was a brontothere largely because of the relative positions of the basicranial foramina, and the similarity of the glenoid facets to that of Dolichorhinus. Osborn (1929a) also noted similarities between AMNH 1501 and YPM PU10041, a braincase of a brontothere that is probably Mesatirhinus junius.

In retrospect, by the time Osborn's (1929a) argument for Sphenocoelus uintensis as brontothere was published, Cook (1926) had described several complete brontothere skulls from the Sand Wash Basin of Moffat County, Colorado. However, Cook (1926) made no comparison of this material with Osborn's holotype of S. uintensis, nor did he mention the conspicuous ventral sphenoidal fossae of these specimens. He erected a new genus and two new species, Tanyorhinus blairi (DMNH 541) and T. bridgeri (DMNH 479). These species were differentiated from each other by minor differences in size (6%), and a variety of other characters that can be attributed to taphonomic distortion (e.g., more curved zygomatic arches) and other aspects of variation that do not warrant species distinctions (e.g., variable presence of M3 hypocone). The apparent difference in head orientation between these two supposed species suggested by Cook (1926) is artificial and can mostly be attributed to deformation in the more poorly preserved specimen (DMNH 541).

Mader (1998) considered Tanyorhinus blairi and T. bridgeri to be junior synonyms of Sphenocoelus uintensis, a revision that is upheld here. Mader (1989) also considered the genera Dolichorhinus Hatcher (1895) and Dolichorhinoides Granger and Gregory (1943) to be junior synonyms of the genus Sphenocoelus. However, Dolichorhinoides is actually a synonym of Epimanteoceras Granger and Gregory (1943), a taxon that clearly differs from Sphenocoelus in numerous ways, including the absence of ventral sphenoidal fossae, the absence of a sagittal crest, the presence of small frontonasal protuberances, and significantly, more molarized premolars. Additionally, Dolichorhinus differs from Sphenocoelus in many significant ways, including the specialized rostrum, the bony palatal extension, the more strongly arched cranium, and the p1–p2 diastemata. Additionally, Sphenocoelus differs from Dolichorhinus in having large ventral sphenoidal fossa, a sagittal crest, and a steeper mandibular symphysis.

In overall appearance, the skulls of Sphenocoelus uintensis resemble those of Mesatirhinus junius, although M. junius lacks ventral sphenoidal fossae and is smaller than S. uintensis. Several brontotheres share large ventral sphenoidal fossae with Sphenocoelus including Protitan, Diplacodon, and Metatitan.

Microtitan mongoliensis (Osborn, 1925)

Neotype

AMNH 22099, a partial mandible with right p1 alveolus and p2–m3.

Type Locality

Ulan Shireh Formation, eight miles north of Tukhum Lamasery, Inner Mongolia, China.

Age

Middle Eocene (Irdinmanhan land mammal “age”).

Referred Specimen

(From the same locality as the neotype) AMNH 21611, a left maxilla with C–M3.

Diagnosis

Microtitan mongoliensis is one of several relatively small Asian brontotheres. The dorsal surface of the maxilla rises steeply posteriorly indicating that the rostrum lacks the specializations seen in Metarhinus or Dolichorhinus. Dentally, Microtitan mongoliensis is characterized by a simple P1, a distinct P2 metacone, exceedingly weak labial paracone ribs, and weak premolar preprotocristae. M. mongoliensis premolars lack lingual crests (although this trait could be intraspecifically variable). Premolar hypocones are absent. The molars of M. mongoliensis have tall, lingually angled ectolophs with weak labial ribs, and thinned lingual ectoloph enamel with wedge-shaped paracones and metacones. A cingular parastyle shelf is absent. M. mongoliensis molars lack paraconules or metalophs. Central molar fossae and anterolingual cingular cusps are absent. Lower dental characters of M. mongoliensis include the absence of a p1–p2 diastema, an elongate p2 trigonid, a large metaconid on p4 but not on p2 and p3, shallow molar basins, rounded molar crescents, and a very elongate m3.

Microtitan mongoliensis can be readily differentiated from the vast majority of brontotheres on the basis of size, although it is within the size range of Metarhinus. Microtitan can be readily differentiated from Metarhinus by the ascending dorsal surface of the rostrum and lack of central molar fossae.

Description

Skull

Microtitan mongoliensis is much smaller than the majority of brontotheres, yet it is distinctly larger than the three smallest species, Pygmaetitan panxianensis, Acrotitan ulanshirehensis, and Nanotitanops shanghuangensis. However, Microtitan mongoliensis is within the size range of Metarhinus. The best cranial fragment referred to M. mongoliensis is AMNH 21611, a left maxilla with a canine and a complete cheektooth series (fig. 21). Although the premaxilla of that specimen is not preserved, the remaining contact surface for the premaxilla can be seen on the inner side of the maxilla above the root of the canine. This surface ends above the P2. A small intact portion of the ventral edge of the nasal incision is visible on the maxilla above the P2. Although the nasal incision extended at least to this point, the specimen does not reveal the position of the posterior margin of the nasal incision. The anterodorsal surface of the preorbital portion of the maxilla rises posteriorly at a steep angle. This indicates that the premaxillomaxillary rostrum lacked the specializations seen in Metarhinus, in which the rostrum does not deepen posteriorly. The exact position of the orbit is indeterminate, although the surface of the orbital floor (formed by the maxilla) is intact and positioned above the M2 and M3. The orbit would have been positioned directly above the M2 or slightly behind it.

Figure 21

A cranial fragment referred to Microtitan mongoliensis (AMNH 21611). (A) Left view, (B) left molars, (C) left premolars.

Upper Dentition

The canine of AMNH 21611 is of moderate size, is slightly elliptical in cross section, and has no distinct cingulum. The length of the postcanine diastema is similar to the length of the P2. The P1 is a simple tooth with a single cusp and an elongate posterior heel (fig. 21c). Most of the enamel of P1 on the labial side of the cusp and on the posterior and lingual sides of the posterior heel is missing. Despite this damage, it is clear that P1 was a very narrow tooth. The P2 and P3 are obliquely shaped due to their posterolingually angled anterior margins, while P4 is more rectangular. The parastyle of P2 is arched lingually, while the parastyle of P3 is nearly straight, and the P4 parastyle is slightly angled labially. The P2 metastyle is angled slightly lingually while those of P3 and P4 are straight. Labial paracone ribs can be seen on the P3 and P4 but they are weak. The metacone of P2 is strongly shifted lingually while those of P3 and P4 are positioned directly behind the protocone. Because of these differences the ectoloph of P2 is more rounded than the ectolophs of P3 and P4. The lingual heels of P2–P3 have large protocones. Premolar hypocones are absent. A small preprotocrista can be seen on the P2. In P3, there is a faint preprotocrista, while P4 lacks this feature completely. None of the premolars exhibits a lingual crest. The labial premolar cingula are extremely weak. The anterior and posterior premolar cingula arch around the lingual sides of the crowns, but they do not join lingually to form continuous lingual cingula.

The upper molars of Microtitan mongoliensis exhibit typical brontotheriine apomorphies, including tall, lingually angled ectolophs, weak labial ribs, thin lingual ectoloph enamel, and wedge-shaped lingual sides of the paracone and metacone. The anterior cingulum is thin and passes proximally to the distal peak of the parastyle. Paraconules and metalophs are completely absent. M. mongoliensis molars lack both anterolingual cingular cusps and central molar fossae. The M3 of AMNH 21611 has a well-developed hypocone that is somewhat smaller than the hypocones of M1 and M2. Lingual and labial molar cingula are essentially absent. One peculiar aspect of the molars of this specimen is the deep notch formed at the labial side of the paracone of each molar.

Mandible and Lower Dentition

The neotype, AMNH 22099, is a partial right ramus with a p1 alveolus (p1 missing) and complete p2–m3 (fig. 22). The partial ramus is reconstructed from fragments with plaster filling a significant gap on the ventral border of the ramus. The dorsal view of the partial ramus suggests that the symphysis extended to the talonid of p2. There is no diastema between the p1 alveolus and the p2. The premolars are slender, particularly p2. The trigonid of p2 is elongate and at least twice as long as the talonid. The p3 trigonid is much longer than the talonid as well. Finally, the p4 trigonid and talonid are of similar length. The trigonids of all the premolars are somewhat narrower than their talonids. The paralophid of p2 curves slightly lingually, although there is almost no lingual trigonid notch. The p2 protolophid is lingually positioned but directed posteriorly. The p3 paralophid is strongly angled lingually, creating a more distinct lingual trigonid notch. The p3 protolophid is straight, but it is angled slightly lingually. The paralophid and protolophid of the p4 are more nearly molariform and arch fully lingually. Among the premolars, only the p4 exhibits a large, lingually positioned metaconid. However, a small metaconid-like feature can be seen at the junction of the protolophid and cristid obliqua on the p3. Give the rudimentary nature of this structure, it is possible that a p3 metaconid was variably present. The talonid of p2 has a short but well-developed cristid obliqua and a very short hypolophid. The lingual surface of the talonid is a sloped, slightly convex surface. On the other hand, the talonids of p3 and p4 are more nearly molariform with longer cristids obliqua, longer hypolophids, and much broader basins. Labial and lingual premolar cingulids are absent.

Figure 22

The proposed neotype jaw of Microtitan mongoliensis (AMNH 22099). (A) Right view, (B) dorsal view, (C) right p2, p3, and p4.

The lower molars of AMNH 22099 have relatively thin enamel, shallow talonid and trigonid basins, and very weak lingual ribs. The m3 is one of the most elongate among brontotheres. The molars of AMNH 22099 have unusually rounded crescents, as noted by Granger and Gregory (1943). The hypoconulids of m1 and m2, though small in all brontotheres, are quite pronounced in AMNH 22099. Labial molar cingulids are exceedingly faint, and lingual molar cingulids are absent. The m3 of AMNH 22099 has a groove near the labial base of its crown that is probably a hypoplasia.

Remarks

Osborn (1925, 1511929) described a new species, “Metarhinus?” mongoliensis, from a mandible fragment with a p4 and m1 (AMNH 20167) from the Irdin Manha Formation of Inner Mongolia (fig. 23). Despite the fragmentary nature of this specimen, it is clearly much smaller than any species of Asian brontothere that was known at the time. However, the size of AMNH 20167 is within the reach of the North American species Metarhinus fluviatilis. Therefore, Osborn (1925) questionably referred this new species to Metarhinus. Better material was available to Granger and Gregory (1943) and they erected a new genus, Microtitan, for this species.

Figure 23

The composite holotype specimen of Microtitan mongoliensis (AMNH 20167). (A) Right view of mandible fragment, (B) dorsal view of mandible fragment, (C) isolated left DP4.

Because the original holotype of Microtitan mongoliensis lacked any diagnostic feature other than size, Granger and Gregory (1943) designated a neotype, AMNH 22099, a partial right jaw with p2–m3. Due to the fragmentary nature of the holotype of Microtitan mongoliensis, a reconsideration of these specimens is warranted to determine which, if any, can be realistically referred to M. mongoliensis, or whether that species should even be considered valid. The holotype of M. mongoliensis (AMNH 20167) is a small fragment of mandible with right p4 and m1 (fig. 23a, b). Overall, it is a rather unremarkable fossil, but several peculiarities about AMNH 20167 suggest that it is actually made up of multiple individuals. For instance, AMNH 20167 includes a barely worn DP4 that is obviously from a very young individual (fig. 23c). On the other hand, the mandible fragment of AMNH 20167 contains an erupted p4 and m1. The m1 shows a significant amount of wear. The lower dentition obviously represents an older individual than the nearly unworn deciduous premolar. More evidence suggests that the mandible fragment itself is a composite specimen. The lower teeth are disproportionately small in comparison to the actual mandible fragment. Closer inspection reveals that the dorsal surface of the mandible fragment is actually a weathered surface. The true alveolar surface has been weathered or broken away. Because the teeth are resting directly upon the weathered surface of bone, the conclusion that the teeth are not in situ is inescapable. Various adhesive materials and plaster can be traced around the borders of the p4 and m1. Therefore, the mandible fragment is probably from a larger individual than those represented by the actual teeth. The two lower teeth are more consistent with a single individual in size and in their degree of wear. Therefore, the m1 and p4 are possibly from the same individual, but this is not certain. In terms of size, the teeth of AMNH 20167 are similar to the North American brontothere Metarhinus fluviatilis, but because the specimen is too fragmentary to readily distinguish it from this taxon, Microtitan mongoliensis could be considered a nomen dubium.

Granger's and Gregory's (1943) neotype (AMNH 22099) is a more sufficiently diagnostic specimen, particularly due to the relatively slender p2 and p3 with elongate trigonids, the rather rounded molar crescents, and the hyperelongate m3. I recommend that AMNH 22099 continue to be recognized as the neotype specimen for Microtitan. The alternative to designating a neotype, rejecting Microtitan outright, is not recommended. At present, it is well understood that Microtitan represents a small Irdinmanhan aged brontothere that, until this paper, was most adequately diagnosed and described by Granger and Gregory (1943).

The maxilla and upper cheektooth series (AMNH 21611) is not directly referable to Microtitan mongoliensis (due to a lack of associated skulls and jaws) but there is little doubt that it represents that same species because of its occurrence in the same locality, its nearly identical size, and relatively elongate M3. Therefore, I continue to include AMNH 21611 in M. mongoliensis.

No other specimens reported since Granger and Gregory (1943) can be assigned to Microtitan. The mandible (PIN 3107-25) assigned to Microtitan by Yanovskaya (1980) from the Khaichin Formation does not actually belong to a brontothere. “Microtitan?” elongatus Qi (1987) is presently a nomen dubium, although it is possibly a synonym of M. mongoliensis. Dental fragments from the southern Jiangsu Province of China referred to Microtitan sp. by Qi and Beard (1996) (herein referred to cf. Metarhinus sp.) belong to an unnamed new species of Microtitan-sized brontothere.

Fossendorhinus diploconus (Osborn, 1895) new genus

Holotype

AMNH 1863, a partial skull with right C–P1 (roots only), P2 (partial), P3–M3, left P1–P2 (roots only), and P3–M3.

Type Locality

Wagonhound Member (Uinta B) of the Uinta Formation, Uinta Basin, Utah.

Age

Middle Eocene (early Uintan land mammal “age”).

Etymology

Fossendorhinus combines the Latin term fossa (“ditch”) with Greek terms, endo (“inside”) and rhinus (“nose”). This combination refers to the internal fossae seen within the nasal cavity of this species.

Diagnosis

Fossendorhinus diploconus is an intermediate-sized hornless brontothere. The nasal incision extends posteriorly as far back as the anterior margin of the M1. The orbits are positioned above the M2 and protrude laterally, though not to the degree seen in Metarhinus. The premaxilla is robust and does not contact the nasal bone. The premaxillomaxillary rostral cavity is open dorsally and there are two large fossae inside the nasal chamber. The premaxillomaxillary rostrum is strongly upturned and is relatively constant in thickness throughout its length. Other cranial features include a well-developed sagittal crest, a strongly concave midcranial dorsal surface, a strongly convex posterior dorsal surface, thin and strongly curved zygomatic arches, and a ventrally open and mediolaterally directed external auditory pseudomeatus.

Dentally, Fossendorhinus diploconus has three large upper incisors, a distinct P2 metacone, weak premolar preprotocristae, short lingual crests extending posteriorly from the premolar protocones, and small hypocones on P2 and P3. The molars have tall, lingually angled ectolophs with weak labial ribs, and thinned lingual ectoloph enamel with wedge-shaped paracones and metacones. A cingular parastyle shelf and an anterolingual cingular cusp are absent. Central molar fossae are present. All traces of molar paraconules and metalophs are lost.

Fossendorhinus diploconus is most similar to Metarhinus, but can be clearly differentiated from Metarhinus by the autapomorphic fossae inside the nasal chamber, the less laterally protruding orbits, and the more strongly upturned rostrum.

Description

Skull

The holotype of Fossendorhinus diploconus (AMNH 1863) is a skull lacking the nasal process (fig. 24). The skull has been subjected to a minor amount of sheering distortion. Prior figures of the specimen in Osborn (1895: fig. 6) and in Osborn (1929a: figs. 362 and 363) misleadingly portray the skull as undistorted, and less damaged than it actually is. The shape of the right side of the skull is well preserved, except for a significant portion of the squamosal, which has been reconstructed with plaster. The left side of the skull is mostly complete, but it has been more severely crushed and the zygomatic arch has been forced inward.

Figure 24

The holotype of Fossendorhinus diploconus (AMNH 1863). (A) Right view, (B) left view, (C) dorsal view.

There is no hornlike protuberance on this specimen. Osborn's (1929a: fig. 363) figure of this specimen includes an unlabeled line that seems to portray the frontonasal suture. However, the line drawn in Osborn's figure actually corresponds to a large crack in the specimen, not a suture. There is no discernable frontonasal suture in this specimen.

The nasal incision of Fossendorhinus diploconus strongly constricts the face, but not to the degree seen in Metarhinus. The nasal incision extends as far back as the anterior margin of the M1. The orbit is positioned directly over the M2. The posterolateral root of M1 is positioned below the anterior rim of the orbit. The right orbit protrudes somewhat laterally from the skull, but not to the degree seen in Metarhinus. The proximal base of the nasal process is preserved on the right side. This remnant suggests a rather thin nasal process that was nearly flat or had very shallow lateral walls, similar to those of Metarhinus.

The most distinctive characteristics of Fossendorhinus diploconus are in the premaxillomaxillary rostrum and the nasal chamber. From the lateral view of the skull it can be seen that the premaxillomaxillary rostrum is a consistent depth from the proximal end to the distal end. The dorsal margin of the rostrum does not rise above the midlevel of the orbit. In comparison to Metarhinus, the premaxillomaxillary rostrum is more strongly curved upward. The pronounced upward curvature of the rostrum does not appear to be a result of taphonomic distortion. On the right side there are a few large cracks in the face, but the proportions appear to be essentially intact.

The premaxillae have become detached at the symphysis and the left premaxilla has been displaced laterally, ventrally, and anteriorly. A roughened groove of bone on the right side, probably representing the premaxillomaxillary suture suggests that the premaxilla did not extend to the posterior base of the nasal incision. The median symphyseal contact surface of the right premaxilla is flat and long (∼7.1 centimeters), although Osborn (1895) described it as short. The dorsal surface of the rostrum is not covered by bone as seen in Metarhinus or Dolichorhinus. However, the rostrum of F. diploconus shows a number of autapomorphic specializations (fig. 25a, b). The premaxillary symphysis arches dorsally. Behind the symphysis are two large internal ovoid fossae. The internal fossae are recessed below the dorsal surface of the rostrum and extend well behind the nasal incision. An additional smaller fossa appears in the upper corner, just beneath what was the proximal base of the nasal process. This small fossa is intact on the right side, but only the very bottom of the fossa is preserved on the left side. The bone that forms these internal fossae fills a large portion of the volume of the nasal chamber. Consequently, the internal nasal cavity is very narrow.

Figure 25

The holotype of Fossendorhinus diploconus (AMNH 1863). (A) Anterior view, (B) anterodorsal view of rostrum, (C) posterior view.

From a lateral view the midsection of the dorsal surface of the skull is strongly concave. Although two large cracks that run through the right side of the frontal area may exaggerate the concave superorbital profile, the individual bone fragments on this surface are concave as well. The dorsal surface of the posterior portion of the cranium is convex in lateral profile. The parasagittal ridges converge medially into a long sagittal crest.

The zygomatic arch is thin and bladelike in cross section. The right zygomatic is less damaged. It is only slightly bowed laterally. The jugal section of the zygomatic is dorsoventrally shallow and is more or less horizontal while the squamosal portion is deeper and angled posterodorsally, giving the zygomatic arch a moderately curved shape. Although Osborn (1908a) described this specimen as lacking an infraorbital jugal process, as seen in Metarhinus, the surface of the jugal is flaked off on the ventral surface of the inferior rim of the orbit. This damage seems superficial and it is doubtful that this specimen possessed a large infraorbital process, though it is possible that a small infraorbital process was present.

From the dorsal view, the nuchal crest is very narrow and swept backward, although this appears to be exaggerated by lateral crushing. From the posterior view (fig. 25c), the nuchal crest is strongly arched dorsally. The overall proportions of the occiput are distorted, although the occiput is narrower dorsally than it is ventrally, and it is slightly waisted. There are small but distinct occipital pillars, and a median depression in the occiput between the occipital pillars.

The ventral surface of the skull is poorly preserved, but several features are discernable (fig. 26a). Although the palate has been crushed laterally, the right side of the posterior nares is preserved and is positioned slightly anterior to the M3. The posterior narial canal appears to have been elongate, but it is severely damaged. Finally, the basicranium, though badly damaged, is typical with a foramen ovale well separated from the foramen lacerum. The opening for the external auditory pseudomeatus, preserved on the left side, is wide and unconstricted ventrally.

Figure 26

The holotype of Fossendorhinus diploconus (AMNH 1863). (A) Ventral view, (B) right molars, (C) right premolars.

Upper Dentition

No incisors or canines are preserved with AMNH 1863, however there are clearly three pairs of incisor alveoli. The size of the alveoli and shape of the premaxillae suggest relatively large incisors and an arched incisor row. The canine alveoli suggest relatively small canines that are comparable in size to those of Dolichorhinus and Metarhinus. There is both a short precanine diastema and postcanine diastemata.

The P1 is not preserved but the remnants of the roots indicate a small, double-rooted premolar (fig. 26c). The ectoloph of P2 is not preserved although the shape of the P2 appears to have been similar to those of P3 and P4. The parastyle and metastyle of P3 are relatively straight. The parastyle of P4 is labially angled, although the P4 metastyle is relatively straight. Prominent labial paracone ribs can be seen on P3 and P4; the P4 labial paracone rib is smaller. There is a large bulge at the proximal base of the P4 ectoloph near the metacone. This bulge is similar in position to the mesostyles to one that is occasionally seen on the P4s of some brontothere species. Each premolar (P2–P4) has a large centrally positioned protocone. Additional lingual features in the P2 include a rudimentary preprotocrista with a distinct paraconule and a distinct crest extending posteriorly from the protocone that is connected to a very small hypocone-like swelling. The lingual side of P3 is morphologically similar to that of P2. However, the preprotocrista, the lingual crest extending posteriorly from the protocone, and the hypocone are less distinct. Finally, the lingual side of the P4 crown is devoid of any features except for the large protocone. The labial premolar cingula of P3–P4 are very thin. The anterior and posterior premolar cingula wrap around the lingual side of the crown but do not join lingually.

The molars of Fossendorhinus diploconus exhibit numerous brontotheriine apomorphies including tall lingually angled ectolophs, weak labial ribs, thin lingual ectoloph enamel, and wedge-shaped lingual sides of the paracone and metacone in molars that are not heavily worn (M2 and M3) (fig. 26b). The anterior cingulum is thin and passes proximally to the distal peak of the parastyle. F. diploconus molars lack anterolingual cingular cusps but possess distinct central molar fossae. All traces of paraconules and metalophs are absent. The trivial name (diploconus) alludes to the two lingual cusps of the M3, a protocone, and a large hypocone that is similar in size to the hypocones of the more anterior molars. However, the size and/or presence of a M3 hypocone could vary in this species as it does in many other brontothere species. The labial molar cingula are distinct but weak and they are discontinuous around the mesostyles. The lingual molar cingula are distinct between the protocone and hypocone, but they are discontinuous around the protocone and hypocone.

Remarks

Fossendorhinus diploconus (Osborn, 1895) is based upon a single skull (AMNH 1863) from the Uinta Basin (Uinta B). Osborn (1895) originally referred this species to the genus “Telmatotherium” (a variant on the spelling of Telmatherium), but he subsequently reassigned it to the genus Metarhinus (Osborn, 1908a). Osborn (1908a) differentiated this species from other species of Metarhinus based on the lack of an infraorbital process. However, AMNH 1863 is damaged in the area where the infraorbital process would occur if it were present. It is unclear whether a small infraorbital process was present on the holotype skull.

Riggs (1912) and Osborn (1929a) reassigned “Metarhinus” diploconus to another genus, Rhadinorhinus (now considered a synonym of Metarhinus). More recently, Mader (1998) reassigned this species to the genus Metarhinus and synonomized it with Metarhinus abbotti (Riggs, 1912). The repeated revisions and associations of this species with Metarhinus are, nonetheless, all contradicted by the distinct characteristics of AMNH 1863, which seem to clearly differentiate it from other species of Metarhinus (M. fluviatilis and M. abbotti). These distinctions include the less prominently protruding orbits, the more strongly upturned rostrum, the more deeply concave dorsal midcranial surface, and the small premolar hypocones. More compellingly, the autapomorphic features found in the rostrum of AMNH 1863 are distinctly different from Metarhinus. Recent removal of the hard sandstone matrix from the nasal chamber of the holotype (AMNH 1863) revealed a pair of large fossae in the floor of the rostrum and inside the nasal chamber. These peculiar nasal fossae, not known to prior authors, warrant the designation to the new genus.

Currently, AMNH 1863 is the only specimen that clearly represents Fossendorhinus diploconus. However, AMNH 2055, a poorly preserved and incompletely prepared skull probably represents this species, but the diagnostic morphology of the rostrum is obscured by sandstone matrix.

Metarhinus fluviatilis Osborn, 1908a

Neotype

FMNH P12187, a complete skull missing only the incisors.

Type Locality

Wagonhound Member (Uinta B) of the Uinta Formation, Uinta Basin, Utah.

Age

Middle Eocene (early Uintan land mammal “age”).

Synonyms

Metarhinus riparius Riggs, 1912.

Referred Specimens

(From the Wagonhound Member of the Uinta Formation of Utah) FMNH P12186 (holotype of Metarhinus riparius), a skull with heavily worn dentition including right C–P1, P2–P3 (partial), M3, left C–P4, and M2–M3; FMNH P12201, a subadult skull with right I3–C (erupting), P1, P2–P3 (erupting), DP4, M1–M2, M3 (erupting), left P1, P2–P3 (erupting), DP4, M1–M2, and M3 (erupting); YPM 13125, a skull with heavily worn dentition including right C, P2, P3–M3, left C–M3, and a partial mandible with right i2 (?), and left p2–m3; FMNH P12173, a skull with right C, P3–P4, M2–M3 (all partial), left P2–P4, M2–M3; (from the Adobe Town Member of the Washakie Formation of Wyoming) FMNH PM1733, a palate with a partial nasal bone, right C–M3, and left P1–M3; FMNH PM3935, an anterior half of a skull with right P2–M3 and left P1–M3; FMNH PM44655, a fragmented skull with right M2–M3, left P2–P4, M1–M2 (partial), and M3; UCM 44939, a complete skull with right P2–M3 and left M3 (partial); UCMP 81278, a skull missing the posterior end with right and left C, and P2–M3.

Diagnosis

Metarhinus fluviatilis is a small hornless brontothere in which the frontal bone intrudes slightly into the surface of the nasal bone splitting off a small lateral nasal splint from the main body of the nasal. The posterior margin of the nasal incision is above the anterior margin of the M2. The nasal process broadens distally, is thin, horizontal, unelevated, with very shallow lateral walls, and without a strongly rounded distal margin. The orbits are positioned above the M2 and strongly protrude laterally. The premaxillomaxillary rostral cavity is enclosed by bone dorsally and its dorsal surface is nearly horizontal. Other cranial characteristics include a small infraorbital process, a sagittal crest, a dorsal cranial surface that is flat or slightly convex postorbitally, strongly curved and unbowed or weakly bowed zygomatic arches, and a ventrally open and mediolaterally angled external auditory pseudomeatus. Ventral sphenoidal fossae are absent.

Dentally, Metarhinus fluviatilis can be characterized as having large subcaniniform upper incisors, a simple P1, a distinct P2 metacone, weak premolar preprotocristae on P2 and P3, and with short lingual crests occasionally extending posteriorly from the premolar protocones. Premolar hypocones are absent. The molars of M. fluviatilis have tall, lingually angled ectolophs with weak labial ribs, and thinned lingual ectoloph enamel with wedge-shaped paracones and metacones. Central molar fossae and anterolingual cingular cusps are present. Cingular parastyle shelves, paraconules, and metalophs are absent. The mandibular symphysis extends to the p3. The p3 has a distinct metaconid (although this character may be variable). The lower molars of M. fluviatilis can be characterized as having shallow molar basins and a slender m3.

Metarhinus fluviatilis shares with Dolichorhinus hyognathus and Metarhinus abbotti a rostrum that is sealed dorsally by bone. The skull of M. fluviatilis most clearly differs from Dolichorhinus in its shorter proportions, prominent sagittal crest, and laterally protruding orbits. The skull of M. fluviatilis is undifferentiated from M. abbotti except for the distally broadening nasal process.

Description

Skull

The proposed neotype of Metarhinus fluviatilis (FMNH P12187) is an exceptionally complete and undistorted skull with lightly worn teeth (figs. 27, 28a). It is missing only the incisors. Riggs's original figure of FMNH P12187 (Riggs, 1912: pl. VIII) depicts a left I2 (?) although it is presently lost, or not with the specimen. In addition, several views of another well-preserved specimen, UCM 44939, are shown (fig. 29).

Figure 27

The neotype of Metarhinus fluviatilis (FMNH P12187). (A) Left view, (B) dorsal view, (C) anterior view, (D) posterior view.

Figure 28

The neotype of Metarhinus fluviatilis (FMNH P12187). (A) Ventral view, (B) left molars, (C) left premolars.

Figure 29

A skull (UCM 44939) of Metarhinus fluviatilis. (A) Right view of face showing frontonasal suture, nasomaxillary suture, and nasal-lacrimal contact, (B) right view of premaxillomaxillary rostrum, (C) ventral view.

Metarhinus fluviatilis is a rather small (table 4), hornless brontothere that is most similar in size to the contemporaneous early Uintan brontotheres M. abbotti and Fossendorhinus diploconus. In most specimens the sutures of the facial bones are not discernable, although they can be seen in UCM 44939. The posterolateral portion of the nasal bone is split by a short triangular process of frontal bone. The lateral portion of the frontal bone forms a short but distinct lateral nasal splint that extends in a posteroventral direction between the frontal and maxilla and makes a narrow contact with the lacrimal bone. This configuration is also seen in several other hornless brontotheres (e.g., Telmatherium), but in M. fluviatilis, the triangular process and lateral nasal splint are relatively short. Riggs (1912) and Osborn (1929a) described rudimentary horns in specimens of M. fluviatilis (FMNH P12186, FMNH P12187, FMNH P12194), but no such structures can be discerned on the actual specimens. The lateral views of some skulls, such as FMNH P12187, are deceptive because the nasal process is angled slightly downward; this gives the appearance of a raised area between the orbits and the nasal incision. Often, the nasal bone is forced further downward taphonomically, exaggerating the effect. However, there are no distinct hornlike protuberances on any specimen of M. fluviatilis.

Table 4

Summary statistics for selected morphometric variables of Metarhinus fluviatilis See Methods for measurement definitions

The face of Metarhinus fluviatilis is highly constricted by the long nasal incision. The nasal incision extends to the anterior rim of the orbit and to the anterior margin of the M2. The orbit is situated over the M2, with the posterolateral root of M1 and anterolateral root of M2 situated below the anterior orbital rim. The orbits of this species have been described as prominent, or laterally protruding (Osborn, 1908a, 1929a). The appearance of prominent orbits is manifested by a broad and relatively flattened frontal between the orbits, a transverse constriction of the dorsal surface of the skull anterior to the orbits, and a recessed surface of bone between the deep nasal incision and the anterior rim of the orbit that forms a narrow wall of bone facing anteriorly just in front of the orbit.

The nasal process of Metarhinus fluviatilis is long. In FMNH P12187 it is as long as the premaxillomaxillary rostrum. The nasal process is angled slightly downward and it is somewhat bowed downward in the center so that the dorsal surface is concave. The lateral walls of the nasal process are thin, shallow, and nearly constant in dorsoventral depth from the proximal end to about the midpoint of the nasal process. On the distal half of the nasal process the lateral walls become shallower distally. The anterior edge of the nasal process is thin, roughened, and deflected downward. From the dorsal view the width of the nasal process is constricted proximally, broadens distally, and has an imperfectly rounded anterior edge. Other specimens with complete (or nearly complete) nasal processes are consistent with this description except that the nasal process is slightly shorter than the premaxillomaxillary rostrum most of the time (e.g., FMNH P12173, FMNH PM3935, UCM 44939, UCMP 81278, YPM 13145).

From the lateral view, the premaxillomaxillary rostrum is long, slightly upturned, and of relatively constant thickness throughout its length. The dorsal margin of the rostrum is not raised above the lower half of the orbit. However, the posterior margin of the nasal incision extends to the level of the upper rim of the orbit. The rostrum of Metarhinus fluviatilis is highly specialized in the same manner as Dolichorhinus and M. abbotti where the rostral cavity is entirely enclosed by bone. In the available specimens of M. fluviatilis the dorsal covering is solid and no distinct sutures can be seen. The premaxillary symphysis is elongate and extends the entire length of the rostrum. This can be best seen in UCM 44939, where the symphysis is unossified. From an anterior view of FMNH P12187 the premaxillae form a solid dome of bone with a thick median ridge of bone (the osteological marker for the nasal septum) running anteroposteriorly along the superior margin of the premaxillomaxillary process. This ridge continues along the dorsal surface of the rostrum into the skull.

From a lateral profile of FMNH P12187, the dorsal surface of the skull is essentially flat over the orbits, and convex over the posterior half of the skull. From the dorsal view of the skull, the parasagittal ridges converge into a short sagittal crest. The zygomatic arches are thin and not strongly bowed. The jugal portion of the zygomatic arch is dorsoventrally shallow and horizontal, while the squamosal portion is deeper and rises posteriorly, giving the zygomatic arch a strong curvature. Some of the specimens are more gracile than the holotype with longer and thinner sagittal crests and thinner zygomatic arches (e.g., FMNH P12186, UCM 44939). Despite this variation, there are no discrete broad-skulled and narrow-skulled groups as hypothesized by Riggs (1912). There is a small infraorbital process on the jugal. The infraorbital process is much smaller than those of Dolichorhinus, Sphenocoelus, or Mesatirhinus, but is similar to that of Metarhinus abbotti. In the center of the zygomatic arch the ventral margin of the jugal is extended ventrally forming a small flange, although this structure is less pronounced in other specimens of Metarhinus fluviatilis and is always much less conspicuous than the large ventral zygomatic flange seen in Metatelmatherium ultimum.

From a dorsal view the nuchal crest is strongly notched medially. From a lateral view the occiput is moderately tilted backward. From a posterior view the dorsal border of the occiput is arched. The dorsal half of the occiput is narrower than the ventral portion and it is somewhat constricted in the middle. The center of the occiput is deeply recessed between two prominent occipital pillars.

The ventral surface of the skull of Metarhinus fluviatilis is well preserved in both FMNH P12187 and UCM 44939. The anterior rim of the posterior nares varies in its position from between the M2 hypocones (FMNH P12187) to between the anterior edges of the M3s (UCM 44939). A narrow horseshoe-shaped rim emarginates the anterior and lateral margins of the posterior nares. The anterior rim of bone and the emargination each have a small posteriorly projecting median process. Remnants of posteriorly shifted turbinates can be seen in the posterior narial canal of UCM 44939, where the matrix inside the skull has been removed. Though the matrix has not been removed from inside the skulls of other specimens, thin slivers of bone suspended in the matrix filling the posterior narial choanae can be seen in some skulls (e.g., FMNH P12201) that also appear to represent posteriorly extended turbinates. The posterior narial canal is elongate and tends to extend onto the anterior part of the sphenoid, but large ventral sphenoidal fossae are absent. In other ways the basicrania of these specimens are otherwise typical with a widely separated foramen ovale and foramen lacerum. The mastoid process does not contact the postglenoid process ventrally; therefore, the external auditory pseudomeatus is open ventrally.

Upper Dentition

No known skull of Metarhinus fluviatilis contains upper incisors. Earlier figures of the neotype (Riggs, 1912: pl. VIII) depict a large, subcaniniform I2(?), but that incisor is no longer attached to the skull and it is apparently lost. The incisor alveoli and/or partial roots indicate an unreduced number of incisors (three pairs) that form an arched row anterior to the canines. The canines are generally small, but tend to vary in size more than the other teeth. Short precanine and postcanine diastemata are consistently present.

The description of the cheek teeth of Metarhinus fluviatilis is primarily based on FMNH P12187 (fig. 28b, c), but is supplemented by information from other specimens. The P1 crown is a small and simple tooth with a single cusp and an elongate posterior heel. In FMNH P12201 (not shown), the P1 is unworn; the cusp of that tooth curves lingually and a thin lingual cingulum is present. The P2 of FMNH P12187 is slightly more oblique in outline than P3 and P4 due to a more posterolingually angled anterior margin. In other specimens the P2 can be less oblique (e.g., FMNH PM3935). The parastyle of P2 is straight, while those of P3 and P4 are somewhat labially directed. The metastyles of P2 and P3 are essentially straight, while the P4 metastyle is angled slightly labially. The labial walls of P2–P4 have distinct labial paracone ribs that become shorter and narrower in more posterior premolars. The lingual heel of the P2 is as developed as those of P3 and P4. There is only a single large lingual cusp (protocone) on the P2–P4; premolar hypocones are absent. On P2 and P3 there is both a small but distinct preprotocrista and a short lingual crest. In P4, the preprotocrista is extremely faint, and there is no lingual crest. The labial cingula are discontinuous around the paracone of P2 and P3, but there tends to be continuous cingulum around the labial base of the crown of P4. The anterior and posterior premolar cingula extend around the lingual sides of the crowns but do not join in FMNH P12187. Occasionally (e.g., FMNH P12173), there are continuous lingual premolar cingula.

Molar apomorphies seen in Metarhinus fluviatilis include tall, lingually angled ectolophs, weak labial ribs, thin lingual ectoloph enamel, and wedge-shaped lingual sides of the paracone and metacone in molars that are not heavily worn. The anterior cingulum is thin and passes proximally to the distal peak of the parastyle. The molars invariably have shallow central molar fossae. Very small anterolingual cingular peaks can be seen on FMNH P12187, although in other specimens (e.g., UCM 44939 and UCMP 81287), small anterolingual cingular cusps are usually more distinct. There are no traces of paraconules or metalophs. The M3 of FMNH P12187 lacks a hypocone, but the posterolingual cingulum is thick and rough. However, an M3 hypocone is occasionally present (e.g., FMNH P12186), though it is never as large as that of the M1 or M2 hypocone. The labial molar cingula are thin but distinct and discontinuous around the labial bases of the mesostyles; lingual molar cingula are essentially absent.

Mandible and Lower Dentition

Only one identifiable skull of Metarhinus fluviatilis (YPM 13125) is associated with a partial mandible. The p3 of this mandible has a large metaconid, a character that is variable among the Metarhinus sp. mandibles that are further described below.

Metarhinus abbotti (Riggs, 1912)

Holotype

FMNH P12179, a complete skull, somewhat crushed dorsoventrally, with complete dentition.

Type Locality

Wagonhound Member (Uinta B) of the Uinta Formation, Uinta Basin, Northeast Utah.

Age

Middle Eocene (Early Uintan land mammal “age”).

Referred Specimens

(From the Wagonhound Member of the Uinta Basin, Utah) CMNH 2866, a complete skull with heavily worn teeth including right P2–M3 and left P1–M3; CMNH 3510, an anterior portion of a skull (partially prepared) with heavily worn teeth including right I2?–I3?, P1 (partial), P2–M1, left I3?, P1–M3.

Diagnosis

Metarhinus abbotti is a small hornless brontothere. The nasal incision extends to the anterior margin of M2. The nasal process tapers distally. It is thin, horizontal, unelevated, and with very shallow lateral walls. The orbits are positioned above the M2 and strongly protrude laterally as in Metarhinus fluviatilis. The premaxillomaxillary rostral cavity is enclosed by bone dorsally and its dorsal surface is nearly horizontal. Other cranial characteristics include a small infraorbital process, a sagittal crest, a dorsal cranial surface that is flat or slightly convex postorbitally, strongly curved and unbowed or weakly bowed zygomatic arches, and a ventrally open and mediolaterally angled external auditory pseudomeatus. Ventral sphenoidal fossae are absent among the known specimens.

Dentally, Metarhinus abbotti has large subcaniniform upper incisors, a simple P1, a distinct P2 metacone, occasional weak premolar preprotocristae, and occasional short lingual crests. Premolar hypocones are absent. The molars of M. abbotti have tall, lingually angled ectolophs with weak labial ribs, and thinned lingual ectoloph enamel with wedge-shaped paracones and metacones. Central molar fossae and anterolingual cingular cusps are present. Cingular parastyle shelves, paraconules, and metalophs are absent.

Metarhinus abbotti shares with Dolichorhinus hyognathus and M. fluviatilis a rostrum that is sealed dorsally by bone. The skull of M. abbotti most clearly differs from D. hyognathus in its shorter proportions, prominent sagittal crest, and laterally protruding orbits. The skull of M. abbotti is undifferentiated from M. fluviatilis except for the distally tapered nasal process.

Description

Skull

The holotype of Metarhinus abbotti (FMNH P12179) is a complete skull although it is somewhat crushed dorsoventrally (fig. 30). Small portions of the right premaxilla and right zygomatic arch are reconstructed with plaster. The dentition of FMNH P12179 is complete; however, substantial portions of the canine crowns are plaster so that measurements of the canine crowns are probably not reliable. There are no visible sutures on FMNH P12179. Two additional skulls, CMNH 2866 and CMNH 3510, are identified as Metarhinus abbotti. Although it is more poorly preserved than the holotype, CMNH 2866 (fig. 31) is undistorted and more faithfully reveals the general shape of the cranium from a lateral view.

Figure 30

The holotype of Metarhinus abbotti (FMNH P12179). (A) Left view, (B) dorsal view, (C) anterior view.

Figure 31

Left view of a skull referred to Metarhinus abbotti (CMNH 2866).

Metarhinus abbotti is a small (table 5) hornless brontothere most similar in size to M. fluviatilis and Fossendorhinus diploconus. The nasal incision extends as far back as the posterolateral root of M1. The orbit is positioned directly above M2, while the posterolateral root of M1 is positioned below the anterior rim of the orbit. The face is greatly constricted in FMNH P12179 but somewhat less than that seen in some specimens of M. fluviatilis. However, CMNH 2866 shows greater facial constriction. The orbits of M. abbotti protrude laterally to a degree similar to that of M. fluviatilis (see description of M. fluviatilis for further explanation).

Table 5

Summary statistics for selected morphometric variables of Metarhinus abbotti See Methods for measurement definitions

The nasal process of the holotype is very thin, narrow, shorter than the premaxillomaxillary rostrum, and it projects in a slightly upward direction. The upward orientation of the nasal process of the holotype appears to be a taphonomic artifact related to the dorsoventral crushing above and behind the orbits. The nasal process of CMNH 2866 is horizontal. The side of the nasal bone forms a very shallow lateral wall. The lateral wall is truncated at the midpoint of the nasal process and the distal half of the nasal process is nearly flat. In FMNH P12179 the nasal process tapers continuously from the proximal end to the distal end and the distal margin is rounded. However, the shape of the nasal bone appears to be variable. In CMNH 2866 and CMNH 3510 the proximal two thirds of the nasal process has a constant width while the distal third tapers.

The premaxillomaxillary rostrum is undifferentiated from that of Metarhinus fluviatilis. From a lateral view, the premaxillomaxillary rostrum is long, of relatively constant dorsoventral depth, and slightly curved upward. The dorsal surface of the rostrum is horizontal and the rostrum does not deepen proximally. The premaxillary symphysis is very long and it extends the entire length of the rostrum. Consequently, the dorsal surface of the rostrum is completely covered by a solid layer of bone. From the anterior view the premaxillae form a dome with a tall ridge of bone running mesially along the full length of the rostrum.

From a lateral view the dorsal surface of FMNH P12179 is flat above and behind the orbits. However, in the uncrushed specimen, CMNH 2866, the postorbital dorsal surface is more convex. The sagittal crest is very thin. Likewise, the zygomatic arches are very thin, dorsoventrally shallow, and are not laterally bowed. From a lateral view, the zygomatic arches are strongly curved. The infraorbital jugal process of Metarhinus abbotti can be seen most clearly from the anterior view (fig. 30c). It is small like that of M. fluviatilis.

From a lateral view the occiput is moderately tiled backward. From a dorsal view the nuchal crest is deeply notched medially. From a posterior view (not shown) the dorsal margin of the occiput is arched and the center of the occiput is deeply recessed between two prominent occipital pillars.

In comparison to specimens of Metarhinus fluviatilis (particularly FMNH P12187, seen in fig. 28), the holotype of M. abbotti is gracile. However, other specimens such as CMNH 2866 are more robust with somewhat deeper zygomatic arches and less slender proportions. In particular, CMNH 2866 resembles some of the more gracile specimens of Metarhinus fluviatilis. Therefore, it is difficult to separate these two species based on the depth of the zygomatic arches or general robusticity of the skull.

The ventral view of FMNH P12179 (fig. 32a) does not notably differ from Metarhinus fluviatilis. The anterior rim of the posterior nares is positioned between the M2s, the emargination of the posterior nares is narrow, and there are no ventral sphenoidal fossae, although the posterior narial canal extends onto the anterior part of the sphenoid. The foramen oval is widely separated from the foramen lacerum and the external auditory pseudomeatus is open ventrally. A pair of thin bony choanal pouches can clearly be seen in the anterior portion of the posterior narial canal of FMNH P12179.

Figure 32

The holotype of Metarhinus abbotti (FMNH P12179). (A) Ventral view, (B) right molars, (C) left premolars, (D) lingual view of left incisors and canine.

Upper Dentition

The holotype of Metarhinus abbotti (FMNH P12179) is the only specimen with well-preserved teeth (fig. 32), although some information on variation in the premolars can be gleaned from other specimens. M. abbotti has an unreduced upper dental formula (3-1-4-3). The incisors are large and form an arched row that extends anterior to the canines. The incisors are subcaniniform with short, lingually curved crowns, and distinct lingual cingula. The incisors increase in size laterally, with i3 being the largest and most caniniform incisor. There is both a short precanine diastema and a postcanine diastema. The crowns of the canines are fragmented and reconstructed with plaster. The remnants of real enamel embedded in the plaster reconstruction suggest that the canines were rather small.

The P1 is a small and simple tooth with a single cusp and an elongate posterior heel. There is a distinct lingual cingulum on the P1. The P2 is more oblique in outline than the P3 and P4 because of a slightly more posterolingually angled anterior margin and a slightly more lingually shifted metacone. The P2 parastyle is straight, while those of P3 and P4 are angled slightly labially. The metastyles of P2–P4 are nearly straight. The labial paracone ribs of P2–P4 are distinct and become smaller in more posterior premolars. The protocones of P2–P4 are relatively tall. In P2, the protocone is slightly ovoid, but those of P3 and P4 are progressively more cone-shaped. There are no additional lingual cusps or crests on the premolars of FMNH P12179; however, the premolars of CMNH 3510 exhibit vestigial yet distinct preprotocrista on P2 and P3. The anterior and posterior premolar cingula typically join lingually, forming continuous lingual cingula, but this is not always the case, as in the P2 of FMNH P12179, where the lingual cingula are discontinuous.

The molars of Metarhinus abbotti show typical brontotheriine apomorphies including tall, lingually angled ectolophs, weak labial ribs, thin lingual ectoloph enamel, and wedge-shaped lingual sides of the paracone and metacone in molars that are not heavily worn (M2 and M3). The anterior cingulum is thin and passes proximally to the distal peak of the parastyle. Each molar has a shallow central molar fossa. There are no traces of paraconules or metalophs. There are strong anterolingual cingular peaks on the M2 and M3. It is weaker on M1, but this is because it is more heavily worn. A M3 hypocone is present, but it is not bilaterally symmetrical; on the right side it is a small but distinct cusp and the distal cingulum ascends to the apex of the right M3 hypocone; on the left side the hypocone is smaller and the cingulum appears to wrap around the mesiodistal corner of the M3. The labial molar cingula are thin and discontinuous around the labial bases of the mesostyles. The lingual molar cingula are discontinuous around the protocone and hypocone.

Metarhinus sp.

Includes nomina dubia: Metarhinus fluviatilis Osborn, 1908a; Metarhinus earlei Osborn, 1908a; Metarhinus cristatus Riggs, 1912; Heterotitanops parvus Peterson, 1914a; Telmatherium accola Cook, 1926; Telmatherium advocata Cook, 1926; Metarhinus pater Stock, 1937

Referred Specimens

(From the Wagonhound Member of the Uinta Formation, Uinta Basin, Utah) AMNH 1500 (holotype of Metarhinus fluviatilis), a skull with right I1–M3 and left I2–M3; AMNH 1859, a mandible with right c–m3 and left i3–m3; AMNH 1864, a left maxilla fragment with P3–M1, M2 (partial); AMNH 1865, a partial mandible with left p3–m3; AMNH 1877, a crushed skull; AMNH 1946, right P3–M3, left P2–M3, and a partial mandible with right c–m3; AMNH 2059, a partial mandible with right i1, c, p2, and left i1–m3; CMNH 2909, a juvenile skull and skeleton (holotype of Heterotitanops parvus); CMNH 3098, a skull right P2–M3, left C–M3, and isolated incisors and canine fragment; CMNH 3125, a partial mandible with right p3–m3; CMNH 3133, a mandible with right c–m1, left i2, and c–m3; CMNH 3142, a maxilla fragment with partial molars and a partial mandible with right i1–c, p2–m3, and left i1–c; CMNH 3371, a partial mandible with left p1–m3; CMNH 3842, a partial mandible with right i1–m2, m3 (partial), left i1–m3; CMNH 11924, a mandible with right i1–i2, p3–m3, and left i1–m3; CMNH 16736, a palate with right P4–M3 and left P1–M3; FMNH P12169, a skull with right P1–M3 and left I3–M3; FMNH P12174, a crushed skull with right and left M1–M3; FMNH P12178, a mandible with right p3–p4 and left p2–m3; FMNH P12178, a mandible with right p2–m3, left i1, and p1–m3; FMNH P12183, a partially prepared skull and mandible with exposed left upper and lower cheektooth rows; FMNH P12190, a mandible with right and left m1–m3; FMNH P12194 (holotype of M. cristatus), the posterior portion of a skull with right and left M2–M3; FMNH P12195, a mandible with right c, p2–m3, left c, and p2–m3; YPM 11284, a partial mandible with right p4 and m1–m3 (partial); (from the Adobe Town Member of the Washakie Formation of Wyoming) AMNH 13166 (holotype of Metarhinus earlei), a skull with right P4, M3, left P2–P4, and M2–M3; AMNH 13179, a mandible with right p1–m3 and left c–m3; CMNH 9409, a crushed skull with right P2–M3 and left P4–M3; FMNH PM1456, a partial mandible with left p3–m3 (all partial); FMNH PM1511, a partial mandible with right i2–c, p2–p4, and left i3–m3; FMNH PM1517 (in part), a right maxilla fragment with P2–P4; FMNH PM1519, a palate with right P2–P4, and left P3–M3; FMNH PM1675, a partial mandible with right and left dp2–dp4, m1, and m2 (unerupted); FMNH PM1680, a rostrum with right P1–P2 (partial), and P3–M1; FMNH PM1715, a mandible with right i1 (?), p2–p4, m3, left i1 (?) p1–p4, and m3; FMNH PM1730, a rostrum with right P2–M1; FMNH PM1731, the posterior portion of a skull with right and left M1–M2, and isolated right P3–P4; FMNH PM1732, a left maxilla and jugal with P2–M3; FMNH PM1734, a partial mandible with right p2–m3; FMNH PM39947, a mandible with right p2–m3; UCMP 81273, a mandible with right c–m3 and left p1–m3; UCMP 81285, a right maxilla with P4–M2; YPM 16834, a partial mandible with right i1–i2, c, left i1–c, and p2–p4; (from the “Metarhinus Quarry” of the Adobe Town Member of the Washakie Basin, Washakie Formation, Wyoming) FMNH PM28001, a mandible with right i2, c, p2–m3, and left i1–m3; FMNH PM28002, a partial mandible with left i2–m3; FMNH PM28003, a partial mandible with right p1–m3; FMNH PM28004, a partial mandible with right i2–m3; FMNH PM28006, a mandible with right c, p2–m3, left c, and p2–m3; FMNH PM28014, a partial mandible with right and left p1, dp2–dp4, and m1 (unerupted); FMNH PM28342, a partial mandible with right p2–m3, left p1–p3, and m2–m3; FMNH PM28343, a partial mandible with right p2–m3 and left p2–m2; FMNH PM28344, a partial mandible with right p2–m2, left c, p2, p3–p4 (partial), m2–m3 (partial); FMNH P28345, a skull fragment with right P2–M3 and left P2–M3; FMNH PM28348, a palate with right I2 (?), C, P1–M3, left C, P1, and P3–M3; FMNH PM28359, a right p2–p4; FMNH PM30388, a right maxilla with P2–M3; FMNH PM30422, a mandible with right i2 (?), p1–m3, left i3, and p1–m3; FMNH PM30432, a crushed skull with right and left P2–P4; FMNH PM30434, a partial mandible with right and left p1, dp2–dp4, m1 (unerupted); FMNH PM30435, a mandible with right i2–i3, p1–m3, left i3, and p1–m3; FMNH PM35932, a skull with right and left P1–M3; FMNH PM35933, a mandible with right p3–m3 and left p2–m3; FMNH PM35970, a mandible with right p2–m3, left canine, and p2–m3; FMNH PM35996, a mandible with right i2–i3, p2–m3, left i2 (?), and p2–m3; FMNH PM36053, a skull fragment with left P3–M1 (partial); FMNH PM36054, a skull fragment with right P3–M3 (partial), left P1–M1, and M2 (partial); (from the Sand Wash Basin of Moffat County, Colorado) DMNH 543, a partial mandible with left p2–m3; DMNH 544 (holotype of Telmatherium accola), a mandible with right p2–m3 and left c–m3; DMNH 550 (holotype of T. advocata), a mandible with right i3, left i1, c, p2–m2, and m3 (erupting); DMNH 2611, a right maxilla with M1–M3; (from the Friars Formation of San Diego County, California) LACM/CIT 2037 (holotype of Metarhinus pater), a right maxilla with C–M3; UCMP 95774, a right maxilla with M1–M3; UCMP 95808, a right maxilla with DP4 and M1; UCMP 95809, a left maxilla fragment with P3–P4; UCMP 95831, a left M1 or M2; UCMP 95780, a partial mandible with left m1–m3; UCMP 95813, a partial mandible with right m2 (partial), and m3; UCMP 95841, a mandible fragment with right p2–p3; UCMP 106011, right p3, p4, and m1; UCMP 113182, fragments of a skull, jaw and some isolated lower teeth; UCMP 113189, a right p2; UCMP 113194, a mandible fragment with right m2; UCMP 113201, a partial mandible with right dp3–dp4, and m1; UCMP 113203, a mandible fragment with left m3.

Description

Because the only clear distinction between Metarhinus fluviatilis and M. abbotti is the shape of the nasal bone, the vast majority of Metarhinus specimens cannot be assigned to either species. These specimens include jaws as well as skulls that lack preserved nasal bones. Among the fossil collections studied for this revision, specimens that could belong to either species of Metarhinus are known from the early Uintan Wagonhound Member of the Uinta Formation of Utah, the middle Adobe Town Member of the Washakie Formation of Wyoming, the Sand Wash Basin of Colorado, and from the Poway and Murray Canyon local faunas of the Friars Formation of San Diego County, California. These include 30 skulls and skull fragments. Four of these have associated mandibles and/or lower dental elements. Also included are an additional 50 complete and partial mandibles.

Among these specimens are the holotype skulls of Metarhinus fluviatilis Osborn (1908a) (AMNH 1500), as well as the holotypes of M. earlei Osborn (1908a) (AMNH 13166), M. cristatus Riggs (1912) (FMNH P12194), and M. pater Stock (1937) (CIT 2037). This large group of specimens provides further information on the morphology of Metarhinus. Therefore, they are described below, particularly as they pertain to intraspecific variation and missing phylogenetic data for M. fluviatilis and M. abbotti.

Skull and Upper Dentition

None of the Metarhinus sp. skulls have well-preserved upper incisors, however many specimens have preserved alveoli or partial sets of well-worn incisors. These all suggest large incisors, consistent in morphology with those of the holotype of Metarhinus abbotti. The incisor morphology of M. fluviatilis was probably not different from that of M. abbotti. A significant number of the skulls show variable premolar morphologies. Numerous specimens (e.g., AMNH 1864, FMNH PM1517, FMNH PM1730, FMNH PM1732, and FMNH P12169) have a small crest extending posteriorly from the protocone of P2 and P3. Occasionally a very small lingual crest can be seen on the P4 protocone. A premolar hypocone is never present. One atypical specimen (FMNH PM36054) has a P1 with an unusual lingual heel with a large protocone. The M3s of these specimens consistently have hypocones that are variable in size but they are always smaller than the hypocones of the M1 and M2.

Mandible and Lower Dentition

The following description of the mandible of Metarhinus sp. is primarily based on AMNH 2059 (fig. 33), although additional information from other specimens is given. The inferior margin of the symphysis is angled about 45° or somewhat less than that. The posterior margin of the symphysis is between the talonids of the p3 in AMNH 2059. However, its position fluctuates between the anterior margin of p3 and the posterior margin of p3. The dental formula is unreduced (3-1-4-3). The incisors are large, form a semicircular arch anterior to the canines, and are positioned closely together. The incisors are subcaniniform with short lingually curved crowns and blunt points. Each incisor has a strong lingual cingulid. The canines are somewhat variable in size, but generally, they are small and slender with lingual cingulids that are much weaker than those of the incisors. There is no precanine diastema in AMNH 2059 although a short precanine diastema is occasionally present in other specimens. The postcanine diastema of AMNH 2059 is shorter than the p2 although some specimens have a slightly longer postcanine diastema. AMNH 2059 lacks a p1–p2 diastema while other specimens (e.g., UCMP 81273) have a minor p1–p2 diastema.

Figure 33

A mandible (AMNH 2059) referred to Metarhinus sp. (A) Left view, (B) left premolars, (C) dorsal view, (D) left incisors and canine, lingual view, (E) left incisors and canine, labial view.

The p1 is a simple tooth with a single cusp and a short talonid heel. The p2 and p3 trigonids are longer than the talonid, while the p4 trigonid is shorter than the talonid. The trigonids of p2–p4 are narrower than their respective talonids. The paralophid of p2 arches slightly lingually, creating a small lingual trigonid notch. The p2 protolophid is lingually positioned and posteriorly directed. The p2 lacks a metaconid. The p3 and p4 trigonids are more molariform with strongly lingually arched paralophids and protolophids, a broad lingual notch, and a large lingually positioned metaconid. The talonid of the p2 has a well-developed cristid obliqua, a small lingual notch, and a short hypolophid. The talonids of p3 and p4 are more developed with more basinlike depressions and longer hypolophids. The lower molars are typical with thin enamel, shallow trigonid and talonid basins, and an elongate m3.

Although the majority of the unidentified mandibles have a large metaconid on p3, it is occasionally absent (e.g., AMNH 1859, CMNH 3125, FMNH P12178) or very small (e.g., CMNH 3371). The presence or absence of a p3 metaconid is known to be variable in Lambdotherium (Bonillas, 1936) and among some brontotheriids (e.g., Rhinotitan). In this instance, it is possible that either this character is intraspecifically variable in Metarhinus fluviatilis and M. abbotti, or the m3 metaconid is present in M. fluviatilis but absent in M. abbotti. The former scenario is more likely based on the monospecific death assemblage of Metarhinus sp. from the Washakie Formation where the p3 metaconid is both absent and present. If this assemblage is truly monospecific, as it has been presumed to be (Turnbull and Martill, 1988; see below), one must conclude that the p3 metaconid is intraspecifically variable. It is reasonably safe to infer that the jaws and lower dentition of M. fluviatilis and M. abbotti were not significantly differentiated. Therefore, these mandibles were used in coding the mandible and lower dental characters of M. fluviatilis and M. abbotti for the phylogenetic analysis.

Remarks

In 1908, Osborn (1908a) named a new genus and species, Metarhinus fluviatilis, from a skull (AMNH 1500) from the Uinta Basin with worn dentition and missing the nasal process. The specimen shows a unique combination of characters that clearly sets it apart from other species that were known at the time. These characters include prominent orbits, a long nasal incision, and a small infraorbital process. In the same paper, Osborn (1908a) named another species, Metarhinus earlei, based on AMNH 13166, a skull also missing the nasal bones that is slightly larger than AMNH 1500 but similar to it in other respects. In 1912, Riggs (1912) erected two more species of Metarhinus. These include M. riparius, based on a complete skull (FMNH P12186) with nasal bones, and M. cristatus, based on the posterior portion of a skull (FMNH P12194). Riggs (1912) thought that he saw two lineages of Metarhinus evolving (presumably) from a small primitive M. fluviatilis. M. earlei and M. cristatus represented a broad-headed form, and M. riparius represented a narrow-headed form. Osborn (1929a) generally agreed with this interpretation. However, most of the differences in these “lineages” can be attributed to taphonomic distortion. For instance, the sagittal crest is exaggerated in skulls with crushed braincases. Other reported differences are simply unsubstantiated. For instance, the expansion of the nasal bones was one of the distinctions of these lineages, however, only one of the four holotypes has preserved nasals. Curiously, Osborn (1929a) recognized the influence of taphonomic distortion on the differences in the proportions of the various holotypes but he did not reject any of the supposed Metarhinus species. Finally, Stock (1937) named a fifth species, Metarhinus pater, based on CIT 2037, a right premaxilla and maxilla from the sandstones of the Poway Conglomerate of San Diego County, California. Although Stock (1937) questionably referred this species to Metarhinus, the holotype is consistent with other species that had been referred to Metarhinus.

Riggs (1912) based a new genus and species, Rhadinorhinus abbotti, on a complete skull (FMNH P12179). Riggs' (1912) decision to erect a new genus for FMNH P12179 is curious because it closely resembles the other supposed species of Metarhinus. However, the nasal process of FMNH P12179 tapers distally, whereas some of the skulls that have been attributed to Metarhinus have a distally widening nasal process. However, Riggs' (1912) and Osborn's (1929a) referrals of species with the distally widening nasals to the genus Metarhinus and referrals of species with distally tapering nasals to the genus Rhadinorhinus is arbitrary because the morphology of the nasal processes of the type species of Metarhinus, M. fluviatilis, is unknown.

Despite the numerous species named by Osborn and Riggs, there appear to only be two diagnosable species, one with distally tapering nasals and one with distally broadening nasals. These species appear to be undifferentiated in every other respect. Unfortunately, the holotype of the type species of Metarhinus (M. fluviatilis) lacks a nasal bone and its specific identity is uncertain. Therefore the validity of the genus Metarhinus is questionable. On the other hand, the genus Rhadinorhinus is less problematic because the holotype of R. abbotti includes a complete (distally tapering) nasal bone. Nonetheless, in more recent revisions, Mader (1989) continued the dubious practice of using Metarhinus fluviatilis for the species with distally widening nasals, and Rhadinorhinus abbotti for the species with distally tapering nasals. In a later revision (Mader, 1998), Rhadinorhinus abbotti was mistakenly considered a junior synonym of Metarhinus diploconus (now Fossendorhinus diploconus). However, Fossendorhinus diploconus is clearly different from both M. fluviatilis and M. abbotti and can no longer be considered a potential synonym of either species.

Despite the fact that Metarhinus is problematic due to unknown nasal morphology of the type species, M. fluviatilis, Metarhinus is a well-known taxon with biostratigraphic importance (e.g., Prothero 1996; McCarroll et al. 1996b; Robinson et al. 2004). To reject Metarhinus outright would generate serious inconsistencies in the literature on North American mammal biostratigraphy. It has been long assumed that M. fluviatilis represents the particular species with the distally widening nasals (Osborn 1929a; Mader 1989, 1998). Therefore, to continue the use of Metarhinus, I suggest that FMNH P12187, a complete skull with distally broadening nasals, be designated as the neotype for Metarhinus fluviatilis. Secondly, Rhadinorhinus abbotti is valid and represents the species with the distally tapering nasal bone. However, it is convenient to consider Rhadinorhinus a junior synonym of Metarhinus because the large number of museum specimens that pertain to either fluviatilis or abbotti but cannot be identified to either due to nonpreservation of the nasal bone are identified on specimen labels and museum catalogs as Metarhinus sp. Grouping fluviatilis and abbotti into one genus, Metarhinus, maintains accuracy in the identification of these specimens and published biostratigraphic data that are based on these specimens. The remaining species of Metarhinus are invalid. M. riparius is the only other Metarhinus species represented by a holotype with a preserved nasal bone. The nasal bone of that specimen indicates that M. riparius is a junior synonym of M. fluviatilis. Other species attributed to Metarhinus whose holotypes lack nasal bones, M. earlei, M. cristatus, M. pater, are nomina dubia.

In addition to the above nomina dubia, Peterson (1914a) described a skull and skeleton of a very young, possibly fetal brontothere from the early Uintan (Uinta B1) that he gave a new name, Heterotitanops parvus (CMNH 2909). Given the very young ontogenetic age of the animal, this species is not valid and it probably represents a known species of similar age although this specimen cannot be clearly assigned to any particular species. Osborn (1929a) noted that it resembled Dolichorhinus in the absence of a sagittal crest but this similarity was dismissed as an artifact of ontogeny. Ultimately Osborn (1929a) considered it most likely to be Metarhinus fluviatilis, primarily based on comparisons of the deciduous dentition with the adult dentition of that species. However, the assignment to M. fluviatilis is uncertain since the neonate specimen lacks a nasal process. At any rate, the specimen is simply too young to clearly refer to any particular species of brontothere; juvenile brontothere materials are uncommon and too poorly documented to make an appropriate comparison of CMNH 2909 with other species. Heterotitanops parvus is a nomen dubium, but it could be a synonym of Metarhinus sp.

The summary statistics of those few specimens that are directly identifiable as Metarhinus abbotti and M. fluviatilis can be found in tables 4 and 5. M. abbotti appears to be larger than M. fluviatilis in premolar length and ventral skull length. These species slightly overlap in total cheektooth row length and completely overlap in molar row length. Although the few specimens identifiable as M. abbotti appear to be larger in some respects than those specimens identifiable as M. fluviatilis, plots of any of these variables (not shown) for the entire Metarhinus sp. group do no reveal any clear bimodal distributions that might suggest two size groups.

A catastrophic death assemblage found in overbank deposits associated with sandstone channels from the Adobe Town Member of the Washakie Formation provides a rare glimpse at the population variability of a brontothere species. This sample was initially identified as Mesatirhinus sp. (Turnbull and Martill, 1988), but was demonstrated by McCarroll et al. (1996a) to be Metarhinus sp. No specimen with intact nasal bones is known from the quarry. Therefore, the quarry sample could represent either Metarhinus fluviatilis or M. abbotti. It is often presumed that large quarry samples, particularly catastrophic death assemblages, are monospecific. Turnbull and Martill (1988) assumed this sample to represent a monospecific herd. Modern ungulate catastrophic death assemblages are monospecific (Berger et al., 2001), although they do not necessarily indicate herds (Mihlbachler, 2003b). It is probable that this sample, though not necessarily representing a herd, is a monospecific death assemblage. (Indeed, I have assumed this as outlined in the explanation of the methodology used in this study). The summary statistics for the Metarhinus sp. quarry sample are given in table 6. The coefficients of variation are compatible with a monospecific population, although they do no falsify the possibility that two similarly sized taxa are present. The average values of the quarry sample seem slightly closer to the average values of M. fluviatilis. However, the size ranges of two of the variables (P3 length, P2–P4 length) virtually span the size range of both species. Additionally, M1–M3 length and P2–M3 length span the entire size range of M. abbotti. Clearly, size is not a realistic means for assigning specimens to either species of Metarhinus.

Table 6

Summary statistics for selected morphometric variables of Metarhinus sp. from the “Metarhinus quarry” See Methods for measurement definitions

Sthenodectes incisivum (Douglass, 1909)

Holotype

CMNH 2398, a skull lacking the nasal process with right I1, I2–C (crowns broken off), P1–P4 (partial), M1–M3, left I1–C, and P2–M3.

Type Locality

Wagonhound Member (Uinta B) of the Uinta Formation, Northeast of Well #2 Uinta County, Utah.

Synonyms

Sthenodectes priscus Peterson, 1934.

Age

Middle Eocene (early Uintan land mammal “age”).

Referred Specimens

(From the Wagonhound Member of the Uinta Formation of Utah) CMNH 9928, right and left maxillary fragments with right P2–P3 (partial) and left P3–P4 (partial); CMNH 11437 (holotype of Sthenodectes priscus), a partial skull with right C–M3, left P3–M3 (all partial), and an associated mandible with right i1–i2, c–m3, and left i1–m3; FMNH P12166, a mandible with complete dentition; FMNH P12165 (mistakenly referred to P12168 by Osborn [1929a]), a partial skull with complete dentition; (from the Washakie Formation of Wyoming) YPM 16883, a partial mandible with right i2–p2, p3 (partial), p4, m1(partial), left i1–i2, and c–m3.

Diagnosis

Sthenodectes incisivum is an intermediate-sized hornless brontothere in which the frontal bone intrudes into the nasal bone, thus splitting off a small lateral splint of nasal bone from the main body of the nasal. The nasal splint and overlapping frontal process are small and the suture is not often discernable. The nasal incision extends as far back as the P1. The nasal process is very short, horizontal, unelevated, of relatively constant transverse width, narrow, and with thin and moderately deep lateral walls. The orbits are positioned above M2 and protrude laterally, but as strongly as in Metarhinus. The lateral margin of the premaxillomaxillary rostrum deepens posteriorly and the rostral cavity is not sealed dorsally. Other cranial characteristics include a strongly concave midcranial dorsal surface, a flatter posterior cranial surface, a sagittal crest, strongly curved zygomatic arches, and a ventrally open and mediolaterally angled external auditory pseudomeatus. Ventral sphenoidal fossae, infraorbital processes, and a ventral zygomatic flange are all absent.

Dentally, Sthenodectes incisivum has oversized subcaniniform incisors, simple P1 that is surrounded on all sides by cingula, and a distinct P2 metacone. The lingual features of the P2–P4 are tall, and the lingual premolar cingula are thick and continuous. Other dental characteristics include weak premolar preprotocristae, and short lingual crests extending posteriorly from the premolar protocones. Premolar hypocones are absent. The upper molars of S. incisivum have tall, lingually angled ectolophs with weak labial ribs and thinned lingual ectoloph enamel with wedge-shaped paracones and metacones. A cingular parastyle shelf is absent. Central molar fossae and anterolingual cingular cusps are present, while paraconules and metalophs are absent. The lower dentition of S. incisivum is characterized by oversized subcaniniform incisors of uniform size, no p1–p2 diastema, a metaconid on p4 but not on p2 or p3, shallow molar basins, and a slender m3.

Only two brontotheres have greatly enlarged incisors, Sthenodectes incisivum and Pygmaetitan panxianensis. P. panxianensis is significantly smaller than Sthenodectes incisivum and it has a more complex p1.

Description

Skull

The holotype of Sthenodectes incisivum (CMNH 2398) is a nearly complete skull (fig. 34). The only significant missing portion is the nasal process. The skull has been dorsoventrally flattened, thus distorting the shape of the skull from a lateral view. Earlier figures in Douglass (1909) misleadingly depict a seemingly undistorted skull. The shape of the skull is less distorted from a dorsal view (fig. 34b) although the braincase has collapsed inwardly, exaggerating the height of the sagittal crest. Two other skulls of S. incisivum are known. These include CMNH 11437 (holotype of S. priscus) (fig. 35) and FMNH P12165 (fig. 36a). The skull of CMNH 11437, though less complete, is not dorsoventrally crushed and it offers a more faithful representation of the shape of the skull from a lateral view. The dorsal surface of FMNH P12165 is not preserved, but that skull is uncrushed as well, and its ventral surface is the best of the three skulls.

Figure 34

The holotype of Sthenodectes incisivum (CMNH 2398). (A) Left view, (B) dorsal view, (C) anterior view, (D) anterolateral view.

Figure 35

Right view of a skull referred to Sthenodectes incisivum (CM 11437).

Figure 36

A skull referred to Sthenodectes incisivum (FMNH P12165). (A) Ventral view, (B) left premolars, (C) lingual view of incisors and canine, (D) labial view of left incisors and canine.

Sthenodectes incisivum is a medium-sized hornless brontothere. Osborn (1929a: fig. 301) labeled a “horn” and Peterson (1934) discussed “horn swellings”, however, there are no hornlike protuberances in the holotype or any other specimen of S. incisivum. In CMNH 2398 there appears to be a thickened ridge of bone in front of the orbit that arches over the maxilla. This feature is not seen in the undistorted specimen, CMNH 11437. Portions of the facial sutures can be seen on CMNH 2398 and, although they are indistinct, they reveal the specialized facial configuration where the frontal bone intrudes into the nasal bone. Part of the right frontonasal suture can be seen on the dorsal surface. The frontal bone projects anteriorly and splits off the posterolateral portion of the nasal bone from the main body of the nasal. From the lateral view of the right side of the skull (not shown) a short lateral nasal splint can be seen dividing the frontal bone and maxilla, although it is indistinct and difficult to see without carefully examining the actual specimen. Note that Douglass' (1909: fig. 1) original figure of the holotype does not accurately portray the shape of the frontonasal contact.

The nasal incision is very short and most similar to those of Metatelmatherium ultimum and Wickia brevirhinus. This can most easily be seen on CMNH 11437, in which the nasal process is complete. In that specimen the nasal incision extends as far back as the P1. The orbit is situated directly above the M2 and the posterior portion of the M1. The anterolateral root of M1 is positioned below the anterior rim of the orbit. The region of maxillary bone between the orbits and nasal incision forms a shallow concavity.

In the holotype (CMNH 2398), there is a triangular vacuity on either side of the skull just in front of the orbit. Douglass (1909) described these as a feature of Sthenodectes incisivum. However, Gregory (1912) and Osborn (1929a) concluded they were an artifactual result of damage, a conclusion that is borne out by other specimens that are undistorted and lack these vacuities.

CMNH 11437 has a complete nasal process. The nasal process is horizontal, relatively narrow, and it is slightly shorter than the premaxillomaxillary rostrum. The sides form shallow lateral walls. The anterior edge of the nasal process is not preserved.

The premaxillomaxillary rostrum of Sthenodectes incisivum is large and thick. The robust condition of the rostrum relates to the enlarged anterior dentition of this species. The premaxillae are completely fused at the midline. There are no discernable premaxillomaxillary sutures in the holotype. In FMNH P12165 an indistinct premaxillomaxillary suture indicates that the premaxilla terminates about at the posterior notch of the nasal incision. The premaxilla does not contact the nasal. From the lateral view the dorsal margin of the rostrum slopes posterodorsally. From the anterior view it can be seen that the sides of the rostrum diverge posterolaterally and the rostral cavity is dorsally open.

From a lateral view the dorsal surface of the skull is concave midcranially, but it becomes flattened or even slightly convex posteriorly. From a dorsal view the skull is broad between the orbits. The orbits protrude laterally to a degree similar to Fossendorhinus, but not to the extreme degree seen in Metarhinus. The parasagittal ridges converge and form a true sagittal crest, although the height of the sagittal crest of the holotype is exaggerated by the crushed braincase. The zygomatic arches are broad and bowed from the dorsal view. From a lateral view the jugal portion of the zygomatic arch is shallow and horizontal. The squamosal portion is deeper and angled posterodorsally, giving the zygomatic arch a strong curvature. Sthenodectes incisivum lacks an infraorbital process like that seen in Metarhinus. It also lacks an enlarged ventral zygomatic flange as seen in Metatelmatherium.

The nuchal crest is thin and from a dorsal view it is angled anteromedially. The occiput is only slightly tilted backward. From a posterior view, the occiput is not well preserved. However, it seems that the dorsal portion of the occiput is somewhat narrower than the ventral portion. The dorsal margin of the occiput is arched moderately. There are distinct occipital pillars on the posterior surface of the occiput; however, the deeply recessed pit in the center of the occiput between the occipital pillars is exaggerated by distortion.

The ventral surface of the skull of Sthenodectes incisivum is best preserved in FMNH P12165 (fig. 36a). The anterior rim of the posterior nares is positioned anterior to the M3. There is a narrow emargination around the anterior and lateral sides of the posterior nares. The posterior narial canal is elongate and extends posterior to the pterygoid processes, but it does not continue into the sphenoid. A remnant of the thin vomerine septum is preserved at the posterior end of the posterior narial canal of the holotype, but it is not preserved in FMNH P12165. The posterior narial canal of FMNH P12165 is partially filled with sediment. In the holotype (CMNH 2398), the posterior narial canal is mostly cleaned of matrix. In that specimen, the anterior half of the posterior narial canal is lined with a surface of wavy bone with anteroposteriorly directed grooves. It is possible that this wavy layer of bone represents maxilloturbinates continuing into the posterior narial canal, similar to that seen in Dolichorhinus; however, the dorsoventral crushing of the holotype limits interpretation.

In the basicranium of the holotype skull (CMNH 2398) the mastoid process contacts the posterior wall of the postglenoid process, thus forming a tubelike external auditory pseudomeatus. Gregory (1912) and Osborn (1929a) noted this characteristic; however, the ventrally closed external auditory pseudomeatus of the holotype appears to be an artifact of distortion. In the less distorted specimen, FMNH P12165 (fig. 36a), the postglenoid and mastoid processes do not contact each other, thus leaving a ventrally open external auditory pseudomeatus. In other respects, the basicranium of Sthenodectes incisivum is typical of brontotheres with a widely separated foramen ovale and foramen lacerum.

Upper Dentition

The upper dentition of the holotype (CMNH 2398) and that of FMNH P12165 are essentially the same although those of the later are more complete and less worn (fig. 36). Dentally, Sthenodectes incisivum is most easily distinguished by its large teeth, particularly the oversized incisors and canine that rival the premolars in size. The incisor row forms a semicircular arch that is positioned anterior to the canines. The anterior teeth are positioned closely together and there are no diastemata in the entire dental battery. The crowns of the I1s converge medially and make contact at the midline. The incisors progressively increase in size laterally. The apices of I1 and I2 are worn slightly, but all three incisors appear to have been tall and essentially subcaniniform in shape. All three upper incisors display a broad lingual cingulum. The lingual heel of I1 is thick and tall, forming a large fossa in the center of the crown between the main cusp and the lingual heel. In more distal incisors the central incisor fossa is less prominent. The canine is extremely tall and pointed and it has a distinct posterolingual cingulum. The canine of CMNH 11437 is smaller than the canines of FMNH P12165 or CMNH 2398.

There is no postcanine diastema in FMNH P12165 or in CMNH 2398 although the P1s are placed somewhat medially to the canines. The lateral view of CMNH 11437 suggests a short postcanine diastema. This is partially related to the medial placement of the P1. Additionally, the canine appears to be displaced slightly in this specimen.

The P1 is poorly preserved in the holotype, but that of FMNH P12165 is intact. The P1 is a small tooth with a single cusp and a posterior heel. However, the P1 is unusual in that the crown is completely encircled by a cingulum. The same characteristic morphology is seen in CMNH 11437. The P2–P4 are essentially rectangular in outline although the lingual margin of P2 is rounded. The parastyle of P2 is directed anteriorly while the parastyles of P3 and P4 are directed slightly labially. The metacone of P2 is positioned directly posteriorly from the paracone. The metastyle of P2 arches somewhat lingually while those of P3 and P4 are nearly straight. There are distinct labial paracone ribs on the P2–P4 that become progressively smaller in more posterior premolars.

The lingual features of the P2–P4 are unusually tall, although photos of the occlusal surfaces do not clearly reveal this character. On each premolar there is a tall protocone, but hypocones are not present. On the P2 there is both a small preprotocrista as well as a short lingual crest extending posteriorly from the protocone. On the P3 these crests are less distinct and in P4 they are essentially absent. This lingual premolar morphology of the holotype differs slightly. In that specimen the lingual side of the P2 has a single ovoid loph that arches around the lingual side of the crown. On the P3 there is a similar structure, but it could best be described as an ovoid protocone. The maxillary fragments of CMNH 9928 offer a glimpse at the unworn lingual morphology of the P3 and P4. The lingual morphology is similar to that of the other specimens, but a very short thin lingual crest can be seen extending posteriorly from the P4 protocone. This crest would be quickly obliterated by wear and could have originally been present in premolars of specimens with more dental wear. The labial cingula of P2–P4 tend to be continuous or slightly discontinuous. The lingual cingula, on the other hand, are always thick and continuous.

The molars of Sthenodectes incisivum exhibit numerous brontotheriine apomorphies, including tall, lingually angled ectolophs, weak labial ribs, thin lingual ectoloph enamel, and wedge-shaped lingual sides of the paracone and metacone in molars that are not heavily worn. The anterior cingulum is thin and passes proximally to the distal peak of the parastyle. S. incisivum molars exhibit both small anterolingual cingular peaks and shallow central fossae. There are no traces of paraconules of metalophs, nor is there a hypocone on the M3, although the posterolingual cingulum of the M3 can be very thick. The labial cingulum is distinct, but it is discontinuous around the mesostyles. Unlike the lingual premolar cingula, the lingual molar cingula are weak.

Mandible and Lower Dentition

Fortunately, one skull of Sthenodectes incisivum (CMNH 11437) is associated with a mandible (fig. 37a, b). That specimen reveals that the lower incisors, like the uppers incisors, are greatly enlarged. Other mandibles are referable to S. incisivum due to their large incisors, including YPM 16883, a partial mandible with very lightly worn lower dentition (fig. 37c–f).

Figure 37

Selected views of mandibles referred to Sthenodectes incisivum. (A) Left view and (B) dorsal view of CMNH 11437. (C) Left molars, (D) left premolars, (E) labial view of right incisors and canine, and (F) lingual view of right incisors and canine of YPM 16883. (© 2005 Peabody Museum of Natural History, Yale University, New Haven, Connecticut, USA. All rights reserved.)

The inferior margin of the symphysis is steep (≥ 45°). The position of the posterior margin of the symphysis fluctuates between the midpoint of the p3 and the posterior margin of p3. The lower incisors form a semicircular arch anterior to the canines. The crowns are distinctly tall and pointed with very thick lingual cingulids. The incisors are all of similar size although the i3 is shorter and more mesiodistally elongate than the other incisors. The canine is very tall with a thick lingual cingulid.

There are no diastemata throughout the entire lower dentition of YPM 16883. However, in other specimens, FMNH P12166 and CMNH 11437l, there are short postcanine diastemata. This is perhaps related to the fact that the canines of these specimens are smaller that those of YPM 16883.

The p1 is a small single-cusped tooth with a slightly elongate talonid heel. The trigonid of p2 is nearly twice as long as the talonid. The p3 trigonid is somewhat longer than the talonid, while the p4 trigonid is similar in length to the talonid. The talonid and trigonid are of similar width in p2, while in p3 and p4 the talonid is broader than the trigonid. The paralophid of p2 is incomplete, but other specimens indicate that it arches slightly lingually, resulting in a small lingual trigonid notch. The protolophid is straight but positioned lingually. The trigonid of p3 is similar to that of the p2, but the lingual notch is somewhat larger and the protolophid is more lingually directed. The paralophid and protolophid of p4 arch 90° lingually. The protolophid of p4 ends in a large metaconid, while metaconids are absent on p2 and p3. The talonid of p2 has a well-developed cristid obliqua, a short hypolophid, and a weakly developed lingual notch. The talonids of p3 and p4 are more strongly developed with longer hypolophids, longer cristids obliqua, and more molariform basins. The p1 and p2 have distinct lingual and labial cingulids. The p3 and p4 lack lingual cingulids but retain labial cingulids.

The molars are typical advanced brontothere molars with thin lingual enamel, weak lingual ribs, an elongate m3, and a widened m3 hypoconulid. The labial molar cingulid is thick but variably continuous and discontinuous around the paracone and hypocone.

Remarks

When Douglass (1909) named Sthenodectes incisivum he questionably referred it to the genus Telmatherium, although he commented that it probably was a different genus. Douglass' (1909) description of “Telmatherium?” incisivum was brief and insufficient. Gregory (1912) made a more thorough comparison of the holotype with other brontotheres and concluded that it represented a new genus, and, thus, revised the name to S. incisivum. Gregory (1912) distinguished S. incisivum from Metatelmatherium ultimum by (1) the enlarged incisors, (2) the absence of a postcanine diastema, (3) thickened premolar cingula, (4) the ventrally enclosed external auditory pseudomeatus, (5) tapering nasals, and a few other less significant details. Gregory's (1912) observations generally distinguished S. incisivum from all known brontotheres, except for character four, an error related to taphonomic distortion of the holotype skull. In the same year, Riggs (1912) referred two additional specimens to S. incisivum, a skull (FMNH P12165, mistakenly referred to as P 12168 by Osborn [1929a]) and a mandible (FMNH P12166).

Peterson (1934) named another species, Sthenodectes priscus, based on CMNH 11437. Peterson distinguished this species from S. incisivum based on the rounder shape and thinner cingulum of the canine, less hypsodont dentition, less developed cingula, and a short diastema. Each of these characteristics is refutable. The rounder shape of the canine and its thinner cingulum are related to the smaller size of the canine. Canine size is variable within many brontothere species and, thus, does not warrant a distinct species. Although the premolar cingula of S. incisivum tend to be thick in comparison to other species, the minor differences in the thickness of the lingual premolar cingulum between CMNH 11437 and CMNH 2389 are consistent with a pattern of intraspecific variability in cingula thickness seen in most brontothere species. Finally, the apparently less hypsodont dentition has to do with the fact that the teeth of CMNH 11437 are more worn than are those of CMNH 2398. Therefore, S. priscus is considered a junior synonym of S. incisivum.

Telmatherium validus Marsh 1872

Lectotype

YPM 11120, a partial skull with right I3, C, P1, P3–M2 and left I1, I2, C, P1–M3.

Type Locality

Near Henry's Fork of the Green River, Wyoming, Twin Buttes Member of the Bridger Formation (Bridger C or D).

Synonyms

Telmatherium cultridens Osborn, Scott, and Speir, 1878; Manteoceras manteoceras Hay, 1901; Manteoceras washakiensis Osborn, 1908a.

Age

Middle Eocene (late Bridgerian land mammal “age”).

Referred Specimens

(From the Twin Buttes Member [Bridger C–D] of the Bridger Basin, Wyoming) AMNH 1511, a skull with right I3–M3, and left P2–M3; AMNH 1532, a palate with right P2–M3, left C (partial), and P2–M3; AMNH 1545 (associated with USNM 6700), a mandible with right p4–m3, left p4–m1, m2 (partial), and m3; AMNH 1560, a mandible with right i3–c, p3, p4, m2, m3 (partial), left i2, c, and p2–m3; AMNH 1563, a mandible with right p2–m3 and left p3–m3; AMNH 1566, a mandible with right i1–m3 and left i3–m3; AMNH 1569 (holotype of Manteoceras manteoceras), a skull with no complete teeth; AMNH 1587, a crushed skull with right M1–M3 and left P1–M3, and a right mandibular ramus with p4 (partial) and m1–m3; AMNH 12193, a mandible with right m2, left c, and p4–m3; AMNH 12194, a left maxilla with P4–M3; AMNH 12204, a crushed skull with right P4–M3, left P3–M3, and a crushed mandible with right c, p3, p4, left c, p4, and m1–m3; AMNH 12210, a mandible with p4–m3; AMNH 12214, a right M3; AMNH 12678, a skull with right I1–P4, M2, M3, left I2–C (partial), P1–M1 (severely damaged), M2, and M3; AMNH 12687, a right mandibular ramus with p2–m3; AMNH 12681, a mandible with left p3 (partial), and p4–m3; AMNH 12683, an anterior portion of a skull (in two pieces) with right C–M1, M2–M3 (all partial), and left I1–M3; UCM 69355, a mandible with right and left i1–13, c (erupting), and p1–m2; UCM 73736, a crushed skull fragment with left P1–P3; UCMP 31344, a right mandibular ramus with p3–m3; USNM 6700 (associated with AMNH 1545), a crushed skull with right P1–M3, left C, P2–M3; USNM 12836, a mandible with right p3–m3, left p4–m3, and an isolated canine; USNM 13455, a partial mandible with right p3–m3, left p2, and p4–m3; USNM 13456, a partial skull with right and left maxillae with C–M3; USNM 26113, a right maxilla with P4–M3; USNM 26114, a left maxilla fragment with M1–M3; USNM 26119, a crushed skull with right P2–M3, left C, and P2–M3; USNM 26121, a right mandibular ramus with i2–m3; USNM 26140, a skull with right P3 (partial), P4–M1, and left P3–M1; USNM 26162, a partial skull with right and left P2–M3, and a mandible with right p2–m3; USNM 26302, a mandible with right p4–m3 and left i2–m3; YPM 16415, a left maxilla fragment with M1, M2, and M3 (partial); YPM 16729, a partial skull (in two pieces) with left C, and P2–M3; YPM PU10027 (holotype of Telmatherium cultridens), a right maxilla with I1–M3 and a left mandible fragment with p2–m3; YPM PU10361, a mandible with m1–m3; (from the lower Adobe Town Member [Washakie A of Granger, 1909] of the Washakie Formation of Wyoming) AMNH 1570, an anterior portion of a skull with right and left P1–M3; AMNH 2353, a skull with right P2, M1–M3, and left P1–M3; AMNH 2354, a skull fragment with right P1, M2–M3, and left P2–M3; AMNH 2356, a right mandibular ramus with dp3, dp4, m1, and m2; AMNH 13165 (holotype of Manteoceras washakiensis), a partial skull with right M2, M3, and left C–M3; AMNH 13176, a crushed mandible with left p1–m3; FMNH PM2328, a mandible with right i3–m3 and left p1–m3; FMNH PM55576, a partial mandible with right i1–m2, and left i1–p4; YPM PU16103, a mandible with right p1–m3 and left c–m3; (from the Aycross Formation, Absaroka Range, Wyoming) AMNH 105429, associated upper dentition including right P2–M3 and left P2–M2; (no locality data) YPM 47288, associated dentition with right and left P2–M3; YPM PU25022, a palate with right P2–M3 and left P1, P2, P4–M3.

Diagnosis

Telmatherium validus is an intermediate-sized hornless brontothere. The frontal bone intrudes into the surface of the nasal bone splitting off a small lateral nasal splint from the main body of the nasal. The nasal splint is larger, more strongly arched, and more distinct than those of Wickia brevirhinus or Metatelmatherium ultimum. The nasal incision extends to the P2. The nasal process is horizontal, unelevated, of relatively constant transverse width, narrow, with thin and relatively deep lateral walls, and without a well-defined or strongly rounded distal margin. The orbits are positioned above the M2 and do not project laterally. The premaxillomaxillary rostral cavity is not enclosed by bone dorsally. Other cranial characteristics include a dorsal cranial surface that is either flat or strongly elevated and convex posteriorly, strongly convergent parasagittal ridges that are separated posteriorly by a narrow pit, strongly curved and moderately bowed zygomatic arches, and a ventrally open and mediolaterally angled external auditory pseudomeatus. Ventral sphenoidal fossae are absent.

Telmatherium validus has large subcaniniform upper incisors, a postcanine diastema, a simple P1, a distinct P2 metacone, weak premolar preprotocristae, and short lingual crests occasionally extending posteriorly from the premolar protocones. Premolar hypocones are absent. The molars of T. validus have tall, lingually angled ectolophs with weak labial ribs, and thinned lingual ectoloph enamel with wedge-shaped paracones and metacones. Central molar fossae, a cingular parastyle shelf, and well-developed anterolingual cingular cusps are absent. Metalophs are absent, but vestigial paraconules are occasionally present. The lower dentition of T. validus includes large subcaniniform incisors that are all of similar size, a p1–p2 diastema (variably present), a postcanine diastema, an elongate p2 trigonid, a metaconid on p4 but not on p2 or p3, shallow molar basins, and a slender m3.

Telmatherium validus is most similar to Metatelmatherium ultimum and Wickia brevirhinus. It differs from Metatelmatherium ultimum in the more posteriorly extended nasal incision, thicker parasagittal ridges, broader forehead, and by lacking a ventral zygomatic flange. Telmatherium validus is distinct from Wickia brevirhinus in the more posteriorly extended nasal incision and the less robust premaxillae.

Description

Skull

Although numerous skulls of Telmatherium validus are known, no single specimen allows for a complete description of the entire skull. The following description of the skull of the T. validus is based primarily on the lectotype YPM 11120 (fig. 38), AMNH 12678 (fig. 39), AMNH 1511 (fig. 40), USNM 26140 (fig. 41), AMNH 13165 (fig. 42), and AMNH 1570 (fig. 43), but additional information from other specimens is provided. T. validus is an intermediate-sized (table 7) hornless brontothere whose skull is most similar to Wickia brevirhinus and, to a lesser extent, Metatelmatherium ultimum. The frontonasal suture is plainly visible on AMNH 12678, USNM 26140, and most other skulls. The frontal bone forms a pair of short triangular processes just above and anterior to the orbits. On the surface this frontal process appears to protrude into the posterolateral portion of the nasal bone. However, in specimens where the frontal and nasal bones have become detached (e.g., YPM PU10027) it can be seen that the frontal process actually onlaps the nasal bone and sits within a grooved triangular depression in the nasal bone. On the surface, the frontal splits the posterolateral corner of the nasal bone, forming a lateral nasal splint that separates off from the main body of the nasal bone. The nasal splint is strongly arched and broadly contacts the maxillary. The nasomaxillary suture forms a tall arch. The overlapping triangular frontal process and lateral nasal splint of T. validus is shared with many other brontotheres including Metarhinus, Sthenodectes, Qufutitan, Wickia, Metatelmatherium, as well as those brontotheres with frontonasal protuberances (horns). Among hornless brontotheres that are characterized by this particular frontonasal configuration, with the possible exception of Qufutitan, the lateral nasal splint appears to be longer, wider, and more distinctly curved in T. validus, and its contact with the frontal and maxilla is more distinct on the surface of the skull.

Figure 38

Lectotype of Telmatherium validus (YPM 11120). (A) Ventral view, (B) right view, (C) anterior view. (© 2005 Peabody Museum of Natural History, Yale University, New Haven, Connecticut, USA. All rights reserved.)

Figure 39

A skull (AMNH 12678) of Telmatherium validus. (A) Left view, (B) dorsal view, (C) anterior view, (D) posterior view.

Figure 40

Right view of a skull (AMNH 1511) of Telmatherium validus.

Figure 41

Anterolateral view of the left side of the face of Telmatherium validus (USNM 26140).

Figure 42

The dorsal view of a skull of Telmatherium validus (AMNH 13165) showing intact parasagittal ridges.

Figure 43

Ventral view of a skull of Telmatherium validus (AMNH 1570).

Table 7

Summary statistics for selected morphometric variables of Telmatherium validus See Methods for measurement definitions

Telmatherium validus has been described as having a rudimentary horn or frontonasal swelling (Osborn 1895, 1929a; Mader, 1989, 1998). Figures of T. validus, particularly those published by Osborn (e.g., fig. 7 in Osborn, 1895) greatly exaggerate the relief of the overlapping portions of the frontal and nasal bone. I can find no distinct frontonasal protuberances on any of the specimens. The overlapping triangular frontal process is flush with the surface of the nasal bone. However, some specimens present an illusion of small frontonasal protuberances from a lateral viewpoint. For instance, in AMNH 12678 the region above the preorbital concavity appears to be raised. This effect in AMNH 12678 is partly an illusion related to a large crack running transversely across the dorsal surface of the skull. The skull roof above the orbits has been forced downward. Additionally, the shallow preorbital fossa in the maxilla gives the region above it a prominent appearance. However, the triangular frontal process is flush with the dorsal surface of the skull. An actual frontonasal protuberance is not present in USNM 26140, where the facial area is undistorted.

The overlapping frontal and nasal bones of Telmatherium validus have long been thought of as the morphological precursor of the frontonasal horns seen in other brontotheres, thus suggesting that Telmatherium is the ancestor or sister taxon of horned brontotheres (Osborn, 1929a; Mader, 1989, 1998). It is, no doubt, very likely that the configuration of facial bones of T. validus does represent the ancestral state of brontothere horns, but that does not suggest any particularly close relationship of T. validus with horned brontotheres because the same facial configuration exists among numerous hornless brontotheres.

The nasal incision of Telmatherium validus is longer than those of Wickia brevirhinus and Metatelmatherium ultimum. In AMNH 12678, the nasal incision extends approximately to the protocone of the P2. In all specimens of T. validus the nasal incision extends posteriorly past the anterior margin of the P2. The longest nasal incision seen in a skull of T. validus is found in AMNH 1511, where the nasal incision extends to the posterior margin of P2. The orbits do not protrude laterally as in Metarhinus. The orbit of AMNH 12678 is positioned directly over the M2 and over the alveolus of the M1. The anterior rim of the orbit is positioned directly above the alveolus of the anterior lateral root of the M1. However, the position of the orbit varies slightly from individual to individual, and in specimens with less extremely worn dentition (e.g., AMNH 1511), the posterior lateral root of the M1 is positioned below the orbit.

The nasal bones tend to be poorly fused together. The nasal process is shorter than the premaxillomaxillary rostrum. In AMNH 12678, the nasal process is angled slightly downward, although the downward angle is somewhat exaggerated. In specimens with less dorsoventral crushing, such as USNM 26140, the nasal process is more horizontal, but its dorsal surface is slightly convex from its lateral profile. The lateral walls of the nasal process are vertical and dorsoventrally deep. The sides of the nasal process form deep and essentially vertical lateral walls that are of relatively constant depth throughout the length of the nasal process. The distal edge of the nasal process is thin, rough, and deflected downward (fig. 39c). From the dorsal view, the nasal process is narrower than the premaxillomaxillary rostrum, and is more or less constant in width throughout its length, although it is sometimes slightly constricted proximally (e.g., USNM 26140).

From a lateral view the posterodorsal slope of the dorsal margin of the premaxillomaxillary rostrum is relatively shallow. The premaxillomaxillary sutures of the lectotype YPM 11120 and many other specimens (e.g., AMNH 12678, AMNH 1569, AMNH 1570, USNM 26140) are plainly visible. The premaxilla truncates before reaching the posterior notch of the nasal incision. Thus, the premaxilla does not contact the nasal bone. The premaxillae are generally thick and well developed although they tend not to be as robust as those of Wickia brevirhinus. In AMNH 12678 the premaxillae are strongly fused at the symphysis. However, in other specimens, including the lectotype (YPM 11120), the premaxillae are not strongly fused. The degree of ossification of the symphysis seems to covary inversely with the size of the canines and the corresponding robustness of the maxillae. In AMNH 12678, the canines are relatively small, the maxillaries are comparatively slender and narrow, and the premaxillae are fully coossified. In contrast, the canines of YPM 11120 and AMNH 1570 are much larger in diameter, the maxillaries are more robust, and the premaxillary symphysis is not strongly coossified. Posterior to the premaxillary symphysis, the nasal processes of the premaxillae diverge laterally. The premaxillomaxillary rostral cavity is not dorsally enclosed by bone.

In comparison to Wickia brevirhinus and Metatelmatherium ultimum, the dorsal roof of the skull of Telmatherium validus is wider and less transversely arched. From a lateral profile, the dorsal surface of the skull above the orbits is slightly concave or it can be flat. Behind the orbits the dorsal surface is flat in AMNH 12678. In some specimens (e.g., AMNH 1511), the postorbital portion of the dorsal cranial surface is strongly convex with an elevated posterior end. The dorsal cranial profile of AMNH 1511 bears a strong resemblance to Palaeosyops, with a steep slope between the elevated posterior portion and the lowered midcranial frontal area.

The parasagittal ridges of Telmatherium validus tend to be prominent, although there is variation in the thickness of the parasagittal ridges. For instance, AMNH 12678 and AMNH 13165 are among the more gracile specimens. One of the most robust specimens is AMNH 1569 (not shown) with extremely thickened parasagittal ridges. The parasagittal ridges of all T. validus specimens strongly constrict the dorsal surface posteriorly, but they never completely join to form a sagittal crest. Instead, they remain separated by several millimeters and a deep narrow pit is formed in the midline of the dorsal surface between the medially constricted parasagittal ridges.

The depth and thickness of the zygomatic arches of Telmatherium validus are variable. Those of AMNH 12678 are among the more gracile. From a lateral view, the jugal portion of the zygomatic arch is essentially horizontal, while the squamosal portion rises posteriorly, giving the zygomatic arch a distinct curvature. The zygomatic arches are moderately bowed laterally. Telmatherium validus lacks a prominent ventral flange on the jugal below the contact with the squamosal as seen in Metatelmatherium. Likewise, there is no infraorbital jugal process as seen in Sphenocoelus.

The nuchal crest is of moderate thickness, and from the dorsal view of the skull of AMNH 12678 it is concave. From a lateral view the occiput of T. validus is moderately tilted backward. From the posterior view, the nuchal crest is dorsally arched; the upper portion of the occiput is narrower than the posterior portion and the occiput is constricted in the middle. The occipital pillars are distinct but not massive and the central depression of the occiput is shallow.

The palate and posterior nares of Telmatherium validus are best preserved in AMNH 1570. The posterior nares are rimmed anteriorly and laterally by a horseshoe-shaped emargination. The anterior edge of the posterior nares is positioned about at the posterior margin of the M2, although in other specimens the position of the posterior nares fluctuates from between the M2 hypocones to slightly behind the anterior margin of M3. In AMNH 1570 two thin unbroken choanal pouches (maxilloturbinates) protrude beyond the anterior rim of the posterior nares. These choanal pouches are similar in position to those of Metarhinus abbotti, and they are not as posteriorly shifted as those of Dolichorhinus. There are clear remnants of bony choanal pouches in others specimens of T. validus as well (e.g., AMNH, 12678, AMNH 1569; UCMP 6700).

The posterior narial canal is elongate, but it does not extend significantly into the basisphenoid. As in other brontotheres, a thin vomerine septum bisects the posterior narial canal, although it is usually obscured by sediment filling the posterior narial canal, or it was destroyed when the sediment filling the posterior narial canal was removed. Although there are no large ventral sphenoidal fossae in Telmatherium validus, in at least one specimen (USNM 6700), the posterior narial canal extends slightly into the sphenoid, and the anterior end of the sphenoid body narrows anteriorly and forms part of the part of the partitioning septum, along with the vomer. The basicranial foramina of T. validus are typical of brontotheres, with a widely separated foramen ovale and foramen lacerum. The mastoid process does not contact the postglenoid process and the external auditory pseudomeatus is, therefore, unconstricted ventrally.

Upper Dentition

The description of the upper dentition of Telmatherium validus is primarily based on the lectotype (YPM 11120) (figs. 38, 44), but other specimens are noted to document variation. The lectotype includes a complete set of lightly worn upper teeth indicating an unreduced dental formula (3-1-4-3). The incisors are large and increase in size laterally. The crowns of I1 and I2 are short, round in outline, pointed at the apex, and slightly lingually curved. Each incisor is bordered lingually by a distinct cingulum. The I3 crown is taller but otherwise similar to I1 and I2. The incisor row strongly arches anterior to the canines and it is separated from the canines by a short precanine diastema.

Figure 44

The lectotype of Telmatherium validus (YPM 11120) (A) Occlusal view of left premolars, (B) lateral view of left premolars, (C) occlusal view of left molars, (D) occlusal view of incisors and canine, (E), labial view of left I2, I3, C. (© 2005 Peabody Museum of Natural History, Yale University, New Haven, Connecticut, USA. All rights reserved.)

The lectotype canine is large and tall with two distinct ridges that extend down the sides of the crown. The canines of Telmatherium validus are variable in size, but they could never be characterized as being very small. The length of the postcanine diastema is very short in YMP 11120, but in other specimens it can be slightly longer than the P2.

The P1 is small and simple with a single cusp, a posterior heel, and a thin but distinct lingual cingulum. A very short P1–P2 diastema is not seen in the lectotype. In the remaining skulls a P1–P2 diastema is most often absent, but it is occasionally present (e.g., AMNH 1569). The anterior margin of the P2 of YPM 11120 is steeply angled posterolingually. There is a marked discontinuity in shape from the rather oblique P2 crown to the more rectangular crowns of P3 and P4.

The parastyle of P2 arches slightly lingually, while the parastyle of P3 is nearly straight, and the P4 parastyle is angled labially. The P2 metastyle is slightly angled lingually, while those of P3 and P4 are more or less straight. The labial ribs of the paracones are prominent, although they follow the typical pattern of becoming narrower in progressively posterior premolars.

The lingual heel of the P2 is small and posteriorly shifted. A small protocone is seen on the lingual shelf. In the lectotype the small preprotocrista is angled anteriorly and forms part of the anterior cingulum. The lingual heels of P3 and P4 are much wider so that the anterior and posterior sides of the crowns are nearly parallel and the protocones are more centrally positioned between the paracone and metacone. Tiny paraconules can be seen on the P3 and P4 of the lectotype, although they are most often absent in other specimens. More typically, a very low preprotocrista connects the protocone to the lingual side of the ectoloph of P3 and P4 (e.g., YPM PU10027, AMNH 13165).

The premolars (P2–P4) of the lectotype lack distinct lingual crests; however, on other specimens, such as YPM PU10027, there is a minor lingual crest on P3 and a barely discernable lingual crest on P4. Occasionally, P2 lacks a distinct protocone; instead, a tall crest of enamel arches around the lingual side of the crown (e.g., YPM 47288). In one specimen (YPM PU10027) P2 has a small hypocone; however, a hypocone does not occur on the premolars of any other specimen of Telmatherium validus.

The labial cingula of the P2–P4 are thin and incomplete. In the lectotype the lingual cingulum of P2 is continuous around the base of the protocone. The anterior and posterior cingula of the P3 and P4 arch around the lingual side of the crown, but they are not connected lingually. However, in other specimens continuous cingula wrap around the anterior, lingual, and posterior sides of the crowns of P3 and P4.

The molars of Telmatherium validus exhibit the derived characteristics typical of brontotheriines including tall, lingually angled ectolophs, thin labial ribs, thin lingual ectoloph enamel, and wedge-shaped lingual sides of the paracone and metacone. The anterior cingulum is thin and passes proximally to the distal peak of the parastyle. T. validus molars lack an anterolingual cingula cusp or central molar fossae. All traces of a metaloph are lost, although each of the molars of the lectotype retains a tiny paraconule. In other specimens tiny paraconules are variably present. In YPM PU10027, for instance, a paraconule is present on the M1, but this feature is not detectible in the M2. Other specimens lack distinct paraconules, but they have a very low preprotocrista-like crest at the lingual base of the paracone (e.g., AMNH 2353). Yet others show no evidence of paraconules or preprotocristae (e.g., AMNH 12194), but this is the least frequent condition. None of the specimens of T. validus has an M3 hypocone, although the posterolingual cingulum of the M3 is rather thick and raised in the lectotype. The labial cingula of the molars tend to be thin and discontinuous around the mesostyles. The anterior cingulum of each molar terminates at the lingual base of the protocone and does not continue around the lingual side of the crown.

Mandible and Lower Dentition

A number of Telmatherium validus skulls are associated with mandibles: AMNH 1587, AMNH 2353, AMNH 12204, and YPM PU10027. There are additional mandibles without associated skulls from the upper Bridger C–D and the lower Washakie formations whose molars are far too slender for Palaeosyops, fall well above the upper size range of Mesatirhinus junius, and are within the size range of T. validus. The most complete mandibles that are referred to T. validus are AMNH 1560, AMNH 1566 (fig. 45b, d), FMNH PM2328 (fig. 45a), FMNH PM55576 (fig. 45c), and USNM 26121. The general shape of these mandibles does not differ notably from the contemporaneous taxon, M. junius. The inferior margin of the symphysis is typically steep (≥ 45°). The symphysis extends to a point between the talonid of the p2 (e.g., AMNH 1560) and the trigonid of the p3 (e.g., AMNH 1566).

Figure 45

Selected views of mandibles referred to Telmatherium validus. (A) Right view of FMNH PM2328, (B) dorsal view of AMNH 1566, (C) lingual view of incisors and canine of FMNH PM55576, (D) right premolars of AMNH 1566.

The lower dentition of T. validus is not differentiated from the contemporaneous species, M. junius in any way other than size. FMNH PM55576 includes a complete set of lower incisors. The incisors are large, positioned closely together, and form a row that arches anterior to the canines. The apex of i1 is worn off. The apex of i2 is lightly worn but the crown is still rather pointed. Finally, the i3 is essentially unworn and has a pointed apex. All of the incisors are roughly of the same size; however the i3 is more mesiodistally elongate than the other incisors. Each incisor has a thin but distinct lingual cingulid. A precanine diastema is generally absent. The postcanine diastema is generally about the same length as p2 or slightly shorter. The p1–p2 diastema is typically very short (e.g., AMNH 1566) or absent (e.g., FMNH PM2328); however one mandible (AMNH 1563) has a p1–p2 diastema that exceeds the anteroposterior length of the p2.

The p1 is a small and simple tooth with a single cusp and a short talonid heel. (Note that the posterior end of the p1 is broken in AMNH 1566 [fig. 45d]). The p2 trigonid is nearly twice the length of the talonid. The p3 trigonid is also longer than the talonid, while the p4 trigonid is slightly shorter than the talonid. The p2 trigonid and talonid are of similar width. The p3 trigonid is about the same width as the talonid or slightly narrower. The p4 talonid is slightly wider than the trigonid. The paralophids of p2 and p3 are slightly angled lingually, creating a very small lingual notch in the trigonids. The protolophids of the p2 and p3 are essentially straight. However, in comparison to that of p2, the p3 protolophid is positioned more lingually with respect to the protoconid. Both p2 and p3 lack metaconids. The trigonid of p4 is significantly more molariform. The paralophid arches fully lingually. The p4 protolophid arches 90° lingually and is connected to a large, lingually positioned metaconid. The lingual orientation of the p4 paralophid and protolophid results in a very broad lingual trigonid notch. The talonid of the p2 is narrow with a weakly developed cristid obliqua and a very short hypolophid. The lingual side of the p2 trigonid forms a slightly concave sloped surface. The p3 and p4 have more developed cristids obliqua and longer hypolophids. The lingual-talonid notches of p3 and p4 are broader than that of p2, although they do not quite form molariform basins. Lingual premolar cingulids are absent, while the labial premolar cingulids tend to be either weak or absent.

The lower molars of Telmatherium validus are typical with relatively thin enamel, shallow talonid and trigonid basins, and an elongate m3. Labial molar cingulids are generally weak while lingual molar cingulids are absent.

Remarks

The taxonomic history of Telmatherium validus and its various synonyms is remarkably convoluted. Marsh (1872) named T. validus from “the greater portion of a skull with teeth, and portions of several other skeletons” that were collected by the Yale expedition of 1871 to the Green River Basin. Later, Marsh (1880) amended the genus name to Telmatotherium and this spelling was adopted by numerous authors (e.g., Earle, 1892; Hatcher, 1895; Osborn, 1895), although Osborn (1929a) eventually reverted to the original spelling, Telmatherium.

Marsh (1872) did not specify which of the specimens was to be the holotype of Telmatherium validus, although his description depended heavily on a partial skull. Osborn (1929a) considered the holotype to be YPM 11120. No other specimen in the Yale Peabody Museum matches the description and measurements provided by Marsh (1872). Therefore, Osborn's assignment of YPM 11120 as the holotype specimen is undoubtedly accurate. However, the specimen is a lectotype because “portions of several other skeletons” were initially reported along with YPM 11120.

Osborn et al. (1878) named Leurocephalus cultridens from a partial skull and mandible from the Bridger Basin. No number was given to the specimen, but their figure of L. cultridens clearly matches YPM PU10027. Earle (1891, 1892) subsequently reassigned this species to the genus Telmatherium, but he continued to recognize it is as distinct from the type species, T. validus.

Eventually, Earle (1892) and Osborn (1895) began to use the genus Telmatherium as a dumping ground for middle Eocene brontotheres. Several species that eventually were assigned to other genera (e.g., Metatelmatherium, Dolichorhinus, Mesatirhinus) were grouped into Telmatherium at one time or another. Among these was Palaeosyops vallidens Cope (1872), a taxon that was reassigned to Telmatherium by Osborn (1895). Telmatherium vallidens (Cope) (not to be confused with Telmatherium validus Marsh) is represented by a partial mandible (AMNH 5098, a lectotype assigned by Osborn [1929a]). This species is presently considered a nomen dubium. However, in 1895, Osborn considered Telmatherium vallidens (Cope) to be valid and went so far as to conjecturally refer two additional skulls to this species (AMNH 1569 and AMNH 1570).

Osborn's (1895) referral of these two skulls to Telmatherium vallidens (Cope) was a grave mistake because it initiated a string of taxonomic blunders that was continued by Hatcher, Osborn, Matthew, and Hay, thus generating a confusing series of invalid taxonomic revisions related to these two skulls. Going on a recommendation made by J. L. Wortman, Hatcher (1895) reassigned T. vallidens (Cope) to a new genus, Manteoceras; however, Hatcher's diagnosis of Manteoceras vallidens (Cope) does not actually refer to the lectotype jaw (AMNH 5098). Instead, Hatcher's (1895) diagnosis is exclusively made up of cranial characters, and his description and figure (Hatcher, 1895: pl. 39: fig. 2) of Manteoceras vallidens were clearly based on the skulls (AMNH 1569 and AMNH 1570) that Osborn had conjecturally referred to that species.

Another species was reported by Matthew (1899), Palaeosyops manteoceras Osborn ex ms., although no holotype was reported. Hay (1901) designated a holotype for this species, AMNH 1569, one of the skulls that had been conjecturally referred to Manteoceras vallidens by Osborn (1895). Hay (1901) also reassigned Palaeosyops manteoceras to Hatcher's genus Manteoceras. A third species, Manteoceras washakiensis, was eventually named by Osborn (Osborn, 1908a) with yet another skull (AMNH 13165).

Ultimately, in contrast to his opinion of 1895, Osborn (1929a) concluded that the skulls he had referred to Manteoceras vallidens (Cope) (AMNH 1569 and AMNH 1570) did not really belong to that species. He removed those specimens from M. vallidens (Cope), reassigned it to the genus Dolichorhinus, and redefined Dolichorhinus vallidens (Cope) based only on the lectotype jaw (AMNH 5098).

At that point, it would have been more reasonable of Osborn to consider Manteoceras the senior synonym of Dolichorhinus. However, because Hatcher's (1895) earlier diagnosis of Manteoceras was based entirely on skulls (AMNH 1569 and AMNH 1570) rather than the lectotype jaw (AMNH 5098) of Manteoceras vallidens (Cope), Osborn (1929a) rationalized that Manteoceras should continue to be applied to the species to which these skulls were now thought to have belonged, that is, Manteoceras manteoceras Hay.

Regardless of how Manteoceras was ultimately used by Osborn (1929a), that genus name is invalid because Hatcher's (1895) original diagnosis of Manteoceras is based on an invalid species concept, Manteoceras vallidens sensu Osborn (1895) (i.e., it was based on two skulls that were incorrectly referred to the type species of Manteoceras). Despite the confusing series of taxonomic errors associated with these skulls, it was never recognized by any of the involved authors that AMNH 1569 and AMNH 1570 actually belong to Telmatherium validus Marsh (1872).

Three previously named species can be considered junior synonyms of Telmatherium validus Marsh. These include Telmatherium (Leurocephalus) cultridens (Osborn et al., 1878), Manteoceras manteoceras Hay, and M. washakiensis Osborn. Osborn (1929a) believed Manteoceras and Telmatherium to be two distinct “lineages”. However, most of the differences noted by Osborn (1929a) between the Telmatherium and Manteoceras “lineages” were are actually based on comparisons of specimens belonging to Telmatherium validus (but assigned by Osborn to M. manteoceras and M. washakiensis) with another species altogether, T. ultimum Osborn (1908a), a species that Osborn had believed was the terminal member of the Telmatherium “lineage”. Telmatherium ultimum (described below) is clearly distinct from T. validus or any of its synonyms, including T. cultridens, M. manteoceras, and M. washakiensis; it was finally removed from Telmatherium by Granger and Gregory (1943) and assigned to Metatelmatherium. Excluding observations involving Metatelmatherium ultimum, the actual differences noted by Osborn (1929a) between Manteoceras and Telmatherium are minor and can easily be attributed to dental wear, damage, and/or intraspecific variation. The similarities between Manteoceras and Telmatherium (sans Metatelmatherium ultimum) were attributed by Osborn (1929a) to parallel evolution, although without justification. Mader (1989) concluded that Manteoceras was synonymous with Telmatherium. In a later revision, Mader (1998) considered T. cultridens, M. manteoceras, and M. washakiensis to be junior synonyms of T. validus Marsh; these revisions are all upheld here, although justification of these revisions (none were provided by Mader, 1998) are needed.

Telmatherium cultridens was thought by Osborn (1929a) to represent a less “progressive” stage of Telmatherium, although the minor differences in size and in the premolar morphology of its holotype (YPM PU10027) could just as easily represent individual variation. The holotype of Manteoceras manteoceras (AMNH 1569) is a skull that lacks teeth. It is therefore not possible to directly compare it to the lectotype of T. validus (YPM 11120), which mostly consists of upper teeth. However, numerous skulls similar to that of AMNH 1569 have teeth that are consistent in size and morphology with those of the lectotype of T. validus. The holotype of M. washakiensis (AMNH 13165) is one of the more gracile skulls among this group, but nonetheless one that is consistent cranially and dentally with other specimens of T. validus.

Finally, Telmatherium validus Marsh can be considered valid for the following reasons. (1) The dentition of the lectotype of T. validus (YPM 11120) is similar to that of Mesatirhinus junius (a contemporaneous Bridgerian species), but it is larger than the upper size limit of M. junius. (2) Additionally, its P2 is somewhat different from that of Mesatirhinus junius, with a less lingually positioned metacone. (3) The lectotype is morphologically consistent with and falls within the size range of a group of more complete specimens (including the holotype skulls of Manteoceras manteoceras and Manteoceras washakiensis) from the upper Bridger (C–D) and lower part of the Washakie Basin (Lower Adobe Town Member of McCarroll et al., 1996b) that represents a species of hornless brontothere. (3) Telmatherium validus is most similar to Wickia brevirhinus and Metatelmatherium ultimum, although it can be distinguished from both by a deeper nasal incision; additionally, the splint of nasal bone that separates the triangular frontal process from the maxilla on the surface of the skull is larger and more prominent in T. validus.

Qufutitan zhoui Wang and Wang, 1997

Holotype

IVPP V8067, the anterior portion of a skull with right I2, C, P2–M2, M3 (partial), and left I3–M1, M2 (partial).

Type Locality

Dong Huangzhuang, Qufu City, Shandong Province; Huangzhuang Formation, China.

Age

Late middle Eocene (Sharamurunian land mammal “age” (Wang and Wang, 1997).

Diagnosis

Qufutitan zhoui is the largest known brontothere that lacks conspicuous frontonasal protuberances. The frontal bone forms an anteriorly projecting triangular process that overlaps the nasal bone, thus splitting off a lateral nasal splint from the main body of the nasal. The lateral nasal incision does not extend posterior to the anterior margin of the P1. The nasal process slightly broadens distally; it is nearly horizontal, unelevated, narrow, with thin lateral walls, and without a well-defined or strongly rounded distal edge. The orbits are positioned directly above the posterior portion of M1 and the M2. The orbits do not project laterally. The premaxillomaxillary rostrum deepens posteriorly and it is not enclosed by bone dorsally. The premaxillary symphysis is vertical with a prominent dorsal ridge and a convex anterior margin. The dorsal surface of the skull is broad and nearly flat above the orbits, and the skull was probably not saddle-shaped.

Dentally, Qufutitan zhoui has small, globular upper incisors that form an arched incisor row. Premolar characteristics include a distinct metacone and hypocone on the P1, a distinct P2 metaconid, and weak premolar preprotocristae. Weak premolar hypocones are present on P2, P3, and P4. The lingual margins of the premolar metacones are strongly angled anterolingually. The molars of Qufutitan zhoui have tall, lingually angled ectolophs with weak labial ribs, and thinned lingual ectoloph enamel with wedge-shaped paracones and metacones. A cingular parastyle shelf is absent. Central molar fossae, anterolingual cingular cusps, paraconules, and metalophs are absent.

Qufutitan zhoui closely resembles Telmatherium validus, Metatelmatherium ultimum, Wickia brevirhinus, and Epimanteoceras formosus, but it retains the following unique combination of characters: a long face with a relatively short nasal incision, a broad and flat forehead, a complex P1, small premolar hypocones, and the lack of central fossae and anterolingual cingular cusps on the upper molars. Moreover, Q. zhoui is the only brontothere that lacks conspicuous frontonasal protuberance but at the same time possesses small globular upper incisors and a vertical premaxillary symphysis.

Description

Skull

The holotype of Qufutitan zhoui (IVPP V8067) consists of the anterior portion of a very large skull (figs. 46–Figure 4748). The antorbital region is essentially complete and in good condition, although it suffers from numerous cracks and a minor amount of distortion. The portion of the skull above the orbits is crushed downward, a large transverse crack runs across the proximal portion of the nasal process, and the nasal process appears to be artificially deflected downward. Additionally, the left nasal bone is pressed downward.

Figure 46

The holotype of Qufutitan zhoui (IVPP V8067). (A) Left view, (B) dorsal view, (C) right view. In A and B, the black arrows are resting on the frontal bone. In A, the white arrows are resting on the maxilla. All arrows point to the frontonasal and nasomaxillary sutures.

Figure 47

Anterior view of the holotype skull of Qufutitan zhoui (IVPP V8067).

Figure 48

Ventral view of the holotype skull of Qufutitan zhoui (IVPP V8067).

Judging by the holotype, Qufutitan zhoui is the largest brontothere to lack conspicuous frontonasal swellings. In terms of the overall shape of the skull, Q. zhoui closely resembles Telmatherium validus, Metatelmatherium ultimum, Wickia brevirhinus, and Epimanteoceras formosus. Q. zhoui is significantly larger than Telmatherium validus, Metatelmatherium ultimum, and Wickia brevirhinus, but Epimanteoceras formosus is of a similar size.

A portion of the large anteriorly projecting triangular protrusion of bone on the left dorsal surface of IVPP V8067 is partly an artifact of damage to the fossil. However, the right margin of the triangular protrusion on IVPP V8067 is a crack. The actual frontonasal sutures of IVPP V8067 can be distinguished from cracks upon close inspection (fig. 46). From the dorsal view of the skull, a triangular process of the frontal bone overrides the nasal bone. From the left lateral view, an arched splint of the nasal bone can be seen running between the maxilla and the overriding frontal process, thus maintaining contact with the lacrimal.

The lateral antorbital surface of the maxilla of IVPP V8067 forms a shallow facial concavity. The relative distance between the orbit and lateral nasal incision is greater than that of comparable brontotheres, and that distance contibutes to a longer face in Qufutitan zhoui. The nasal incision of Qufutitan is similar in length to Telmatherium validus and Epimanteoceras formosus. However, the faces of these taxa are relatively shorter and the nasal incisions of Telmatherium and Epimanteoceras extend to a point above the P2. However, the lateral nasal incision of Qufutitan zhoui does not extend posteriorly beyond the anterior margin of P1.

The orbit of Qufutitan zhoui is positioned directly above the posterior portion of M1 and M2, while the anterior orbital rim is situated above the anterior root of M1. This orbital position is slightly more anterior than that of Epimanteoceras formosus, Metatelmatherium ultimum, and Wickia brevirhinus, but similar to Telmatherium validus.

The moderate downward angle of the nasal bone of IVPP V8067 is probably a taphonomic artifact and, originally, it would have extended more horizontally from the skull. The nasal process is slightly longer than the premaxillomaxillary rostrum. The sides of the nasal process form deep and thin lateral walls that extend to the distal end of the nasal process. The nasal bones are incompletely co-ossified. The left and right nasal bones have become detached at the midline and the left side has been artificially depressed. From the dorsal view it can be seen that the nasal process is narrower than the premaxillomaxillary rostrum. The nasal process is of nearly constant width throughout its length, although it broadens slightly near the distal end. The anterior margin of the nasal process is mostly broken away although intact remnants of the distal edge remain on the anterolateral corners, indicating a thin and roughened distal nasal margin. The slight distal flaring of the nasal process resembles that in Epimanteoceras formosus in particular. However, the overall the morphology of the nasal process of Qufutitan zhoui does not differ significantly from that of Telmatherium, Metatelmatherium, or Wickia.

Some aspects of the rostrum of Qufutitan zhoui are unspecialized and typical. For instance, from the lateral view of IVPP V8067, the rostrum deepens posteriorly; the dorsolateral margin slopes posterodorsally and rises to about the midlevel of the orbit. The rostrum lacks the highly specialized dorsal bony cover seen in Dolichorhinus and Metarhinus. Despite these regularities, the rostrum of Q. zhoui has several peculiar features that distinctly differ from most other brontotheres. The elongate premaxillomaxillary rostrum of Q. zhoui contributes to the appearance of the relatively long face. Although there is a slightly upward curvature to the rostra of most brontotheres, the rostrum of IVPP V8067 seems to curve slightly downward. The premaxillomaxillary suture can be seen on the left side of the specimen, running along the dorsolateral border of the rostrum. The premaxilla does not contact the nasal bone; however, the nasal process of the premaxilla is long, slender, and extends nearly to the posterior notch of the lateral nasal incision. The premaxillary symphysis of Q. zhoui is nearly vertical, thus giving the incisors a nearly vertical orientation. In lateral profile the anterior margin of the symphysis is concave. Finally, there is a prominent U-shaped ridge of bone surrounding the notch on the dorsal surface of the premaxillary symphysis that is formed by the posterolaterally divergent premaxillary nasal processes. In contrast to Qufutitan zhoui, the premaxillae of nearly all other brontotheres have a more posterodorsally reclined premaxillary symphysis with more anteriorly angled incisors, a more convex or flat anterodorsal surface, and a less prominent (or absent) dorsal ridge.

Little can be said about the postorbital portion of the skull of Qufutitan zhoui; however, the skull does not appear to have been saddle-shaped. The forehead is realtively broad and flat. In this respect Q. zhoui resembles both Telmatherium and Epimanteoceras. The foreheads of Metatelmatherium and Wickia are narrower and more transversely arched.

From the ventral view of IVPP V8067 the anterior rim of the posterior nares is positioned just anterior to the M3. There is a relatively wide emargination along the anterior margin that is marked by a thin ridge of bone. The lateral margins of the posterior nares are not preserved; however, the emargination probably continued around the lateral margins of the posterior nares, thus forming a horseshoe-shaped emargination similar to that seen in most other brontotheres.

Upper Dentition

The holotype of Qufutitan zhoui (IVPP V8067) retains a nearly complete set of upper dentition (figs. 48, 49, 50). Only two incisors are preserved, although the alveoli are reasonably well preserved and indicate a dental formula of 3-1-4-3. The incisors form an arched row anterior to the canines. A small median notch on the alveolar surface of the premaxilla suggests a minor diastema between the central incisors. The right I2 and left I3 are in a minimal state of wear. Both incisors are similar in size and are relatively small and globular in shape. There are no incisor cingula. The remaining unworn surfaces of the enamel crowns are crenulated. Each incisor bears one or more irregularly shaped and irregularly positioned enamel protuberances. The relatively small globular incisors, a condition similar to the incisors of more advanced horned brontotheres (e.g., Duchesneodus, Dianotitan), are unusual for a hornless brontothere. There is a short diastema between the I3–C and a longer postcanine diastema. The canines of IVPP V8067 are very large and posteriorly curved.

Figure 49

Incisors and incisor alveoli of the holotype skull of Qufutitan zhoui (IVPP V8067). (A) Ventral view, (B) labial view.

Figure 50

Cheek teeth of the holotype skull of Qufutitan zhoui (IVPP V8067). (A) Right molars, (B) right premolars (P2–P4), (C) right premolars rotated slightly (P2–P4), (D) left premolars (P1–P4).

The left premolar row (P1–P4) of IVPP V8067 is complete with a particularly well-preserved and minimally worn P1 and P2. However, the surfaces of the left P3 and P4 are somewhat weathered. The right P1 is missing although the remaining premolars are complete, minimally worn, and unweathered. The P1 of IVPP V8067 is substantially smaller than the other premolars, although it exhibits relatively advanced crown morphology. The P1 crown has both a well-developed ectoloph and lingual heel. The crown is nearly rectangular, although the anterior edge is strongly angled posterolingually, thus giving the tooth an oblique appearance. There is both a distinct paracone and metacone on the P1. The labial swelling of the paracone is much larger than the metacone. The parastyle is long and slightly lingually arched, while the metastyle is shorter and uncurved. The lingual side of the crown bears a well-developed and somewhat posteriorly shifted protocone, as well as a strongly developed preprotocrista that connects the protocone to the lingual base of the paracone. A much less distinct and shorter lingual crest projects straight posteriorly from the protocone. A continuous cingulum stretches from the lingual edge of the parastyle, wraps around the lingual margin of the crown, and meets the lingual edge of the metastyle.

The P2, P3, and P4 are more nearly rectangular in outline than the P1, with parallel anterior and posterior sides. Distinctly pinched paracone ribs can be seen on the labial surfaces of P2–P4. The labial swellings of the metacone are broader and rounder that the labial paracone ribs. The metacone of P2 is only slightly more lingually positioned than those of P3 and P4. The P2 parastyle projects straight anteriorly, while the P3 and P4 parastyles are strongly angled labially. The metastyles of P2 and P3 project straight posteriorly, while the P4 metastyle is angled somewhat labially. The lingual band of enamel on the ectolophs of P2, P3, and P4 is thinner than the labial band of enamel. A small but distinct preprotocrista connects the paracone and metacone on P2. P3 and P4 have progressively smaller and less distinct preprotocrista. No paraconules or metaconules can be seen on any of the premolars. The lingual margins of the metacone of P2, P3, and P4 form distinct vertical wedges that are strongly angled anterolingually.

A lingual crest of relatively low relief extends posteriorly from the protocone of the P2 and connects it to a very small hypocone that can now be identified only by a small exposure of dentine on the posterolingual corner of the crown. The lingual morphologies of the P3 and P4 are similar, although the hypocones are more distinct than that of the P2. Each of these premolars (P2–P4) has a strong continuous cingulum that wraps around the anterior, lingual, and posterior margins of the tooth. Likewise, each of these premolars has a distinct and continuous labial cingulum.

The premolars of Qufutitan zhoui distinctly differ from those of Telmatherium, Metatelmatherium, and Wickia. These taxa have simpler single-cusped P1s, and hypocones are not seen on any of the premolars. Moreover, Telmatherium occasionally has premolar paraconules and a diastema between P1 and P2. The premolars of Epimanteoceras formosus compare well with Q. zhoui in most respects. In particular, both taxa tend to have poorly developed hypocones. However, the strongly wedged and anterolingually angled margin of the premolar metacones of Q. zhoui is an autapomorphic condition.

The right molar row of IVPP V8067 is moderately worn and the posterior portion of M3 is not preserved. Typical brontotheriine molar apomorphies seen in Qufutitan zhoui include a narrow anterolabial cingulum that passes below the apex of the parastyle, a relatively tall and lingually angled ectoloph, weak labial paracone and metacone ribs, thin lingual ectoloph enamel, and wedge-shaped lingual margins of the paracone and metacone. Anterolingual cingular cusps and central molar fossae do not occur on any of the molars. Q. zhoui molars lack vestigial paraconules. Although the posterolingual corner of the M3 is lost, enough of the specimen is preserved to suggest that a hypocone either was not present or was poorly developed. Lingual molar cingula in IVPP V8067 are thin and discontinuous around the margins of the protocone.

Remarks

Wang and Wang (1997) named a new genus and species of very large hornless brontothere, Qufutitan zhoui, based on the anterior portion of a skull (IVPP V8067). Wang and Wang (1997) diagnosed Q. zhoui by its large size, straight nasals, shallow nasal incision, weak hornlike swelling, well-developed premaxillary suture projection, upper incisors reduced with “global-shaped” crowns, large upper canine, developed C1–P1 diastema, nonmolariform premolars, weak hypocone on P2–P4, wide upper molars with no preconule ( =  paraconule as used in this paper) or metaconule. Not all of these characters are particularly diagnostic. For instance, the molars of Qufutitan, described as wide by Wang and Wang (1997), are not different in overall proportions than other brontotheres. Moreover, the premolars of all brontotheres can roughly be described as “nonmolariform”.

I was not able to confirm that IVPP V8067 has a “weak horn-like swelling” (contra Wang and Wang, 1997). As noted in the above description, there is a large triangular projection of bone on the left dorsal surface of the skull that overlaps the nasal bone, although in Qufutitan zhoui the surface of the overlapping frontal process is flush or nearly flush with the dorsal surface of the skull. A number of other middle Eocene brontothere taxa are known to exhibit the same frontonasal configuration, but like Qufutitan they lack conspicuous frontonasal protuberances. These taxa include Telmatherium validus, Metarhinus fluviatilis, Sthenodectes incisivum, Wickia brevirhinus, Metatelmatherium ultimum, and Epimanteoceras formosus. Among these taxa, only Epimanteoceras formosus has a minor frontonasal swelling, and even in that species the swelling occurs in only one of the two known specimens.

Wang and Wang (1997) briefly compared Qufutitan zhoui to Metatelmatherium; a more extensive comparison of Qufutitan zhoui to other brontotheres is warranted. As noted above, Qufutitan zhoui closely resembles Telmatherium validus, Wickia brevirhinus, Metatelmatherium ultimum, and Epimanteoceras formosus. All express the same derived frontonasal configuration, but they otherwise have relatively unspecialized skulls. Closer examination reveals that in addition to its unusually large size Qufutitan zhoui expresses a unique combination of characters that strongly differentiates it from Telmatherium, Wickia, Metatelmatherium, and Epimanteoceras; these characters include a long face with a relatively short nasal incision, a broad and flat forehead, a complex P1, small premolar hypocones, and the lack of a central fossa and anterolingual cingular cusps on the upper molars. Additionally, the strongly posterolingually angled premolar metacones is an autapomorphic condition that is not seen in other brontotheres. However, it is the unusual morphology of the premaxilla and the small globular incisors that most conspicuously distinguishes Qufutitan zhoui from other hornless brontotheres. With the exception of Qufutitan zhoui, small globular incisors are only seen among the most advanced horned brontotheres. All hornless brontotheres other than Qufutitan zhoui have larger and more subcaniniform incisors.

Strangely, several aspects of the incisor and rostral morphology of Qufutitan zhoui resemble the giant late Eocene Asian brontothere, Embolotherium. For instance, E. andrewsi, E. grangeri, and Q. zhoui are the only brontotheres to have globular upper incisors but retain a plesiomorphically arched incisor row. Other brontotheres with globular incisors exhibit a straight incisor row. Secondly, the vertical orientation of the premaxillary symphysis of Q. zhoui, with its concave anterior margin and pronounced dorsal ridge, are conditions that are otherwise only seen in the premaxillae of E. andrewsi and E. grangeri. Aside from these similarities Q. zhoui is obviously very different from Embolotherium.

Qufutitan zhoui is known only from a partial skull. However, there are a variety of brontothere species, known only from mandibles, to which Qufutitan cannot be directly compared. Nonetheless, it is worth considering the possibility that one of these “mandible taxa” is synonymous with Q. zhoui. Based on the long rostrum and long postcanine diastema of Q. zhoui, the mandible probably had a relatively elongate symphysis with a rather long lower postcanine diastema and small globular or semispatulate incisors. Pollyosbornia altidens and Hyotitan thomsoni closely resemble the anticipated mandible of Q. zhoui. Both are relatively large brontotheres exhibiting an arched incisor row, a long symphysis, and a long postcanine diastema. Hyotitan thomsoni, in particular, seems a likely synonym of Q. zhoui. The lower incisors of Hyotitan are rather smallish and form a slightly arched row. Further congruence between the skull of Q. zhoui with the mandible of H. thomsoni is represented by the relatively advanced upper and lower premolar morphologies, the very large canines, and the unusually pronounced labial premolar cingula/cingulids. H. thomsoni is more or less compatible with the anticipated mandible and lower dentition of Q. zhoui and it is possible that these species are synonyms. Therefore one of these species must be considered dubious. H. thomsoni Granger and Gregory, 1943, is the senior name. However, as a matter of convenience, I presently consider Q. zhoui to be the valid species; the character information derived from its holotype skull is superior to the more limited character data retrievable from the holotype jaw (AMNH 26401) of H. thomsoni. If further discoveries indicate that these species are indeed synonymous, Q. zhoui should be considered a junior synonym of H. thomsoni.

Wickia brevirhinus, new genus and species

Holotype

CMNH 11380, a skull with right I3, C, P2–M3, left C, P2 (partial), and P3–M3.

Type Locality

Sand Wash Basin, Moffat County Colorado.

Age

Middle Eocene (early Uintan land mammal “age”).

Etymology

“Wick”, nickname of paleontologist William Berryman Scott (Scott, 1939). The trivial name, brevirhinus, refers to the short nasal incision and short nasal process of this species.

Referred Specimen

(From the Sand Wash Basin, Moffat County Colorado) CMNH 11382, an anterior portion of a cranium with right I3–M3, left I1–M3, and a mandible with right p1–m3, left i2, and p1–m3; (from the Adobe Town Member of the Washakie Formation, Sweetwater County, Wyoming) UCMP 81300, a skull missing the right zygomatic arch, with right I3, C (partial), P1–M3, and left I3–M3.

Diagnosis

Wickia brevirhinus is a relatively large hornless brontothere, one of several in which the frontal bone intrudes into the nasal bone, thus splitting off a small lateral splint of nasal bone from the main body of the nasal. The size and shape of the nasal splint is similar to that of Metatelmatherium ultimum and less pronounced than that of Telmatherium validus. The nasal incision is short and extends posteriorly to the P1. The nasal process is very short, horizontal, unelevated, of relatively constant transverse width, narrow, with thin and relatively deep lateral walls, and without a well-defined or strongly rounded distal margin. The orbits do not project laterally and they are positioned above M2 with the roots of M1 below the anterior orbital rim. Neither an infraorbital process nor a ventral zygomatic flange is present on the jugal. The premaxilla is extremely robust and does not contact the nasal bone. The premaxillomaxillary rostral cavity is not enclosed by bone dorsally. Other cranial characteristics include a relatively flat dorsal cranial surface, an elevated occiput, deep, strongly curved, and laterally bowed zygomatic arches, and a ventrally open and mediolaterally angled external auditory pseudomeatus. Large ventral sphenoidal fossae are absent. A sagittal crest is absent, but the dorsal surface of the cranium is very narrow and the parasagittal ridges strongly constrict the dorsal surface of the cranium.

Dentally, Wickia brevirhinus has large subcaniniform upper incisors, a simple P1, a distinct P2 metacone, weak premolar preprotocristae, and short lingual crests extending posteriorly from the premolar protocones. Premolar hypocones are absent. The upper molars of W. brevirhinus have tall, lingually angled ectolophs with weak labial ribs, and thinned lingual ectoloph enamel with wedge-shaped paracones and metacones. Small anterolingual cingular cusps are present, but central molar fossae and cingular parastyle shelves are absent in the molars. W. brevirhinus molars appear to retain vestigial paraconules. Anterolingual cingular cusps are present. The upper molars lack paraconules or metalophs. The lower dentition of W. brevirhinus includes a postcanine diastema, a p1–p2 diastema (variably present), an elongate p2 trigonid, a poorly developed p2 talonid, a metaconid on p4 but not on p2 or p3, shallow molar basins, and a slender m3.

Wickia brevirhinus is most similar to Telmatherium validus and Metatelmatherium ultimum, but it is distinct from these species primarily by the combination of a very short nasal incision, a short nasal process, lack of a sagittal crest, lack of a ventral zygomatic flange, and possibly a poorly developed p2 talonid.

Description

Skull

The following description of Wickia brevirhinus is based upon the holotype skull (CMNH 11380) (figs. 51, 54a) as well as the referred specimens (UCMP 81300) (figs. 52 and 54c) and CMNH 11382 (figs. 53, 54b). W. brevirhinus is a large hornless brontothere that is most similar in size to Metatelmatherium ultimum. All skulls reveal a frontonasal configuration that resembles those of Metatelmatherium and Telmatherium, but the sutures can most clearly be seen on CMNH 11382. A thin, arched splint of the nasal bone can be seen branching in a posteroventral direction from the main body of the nasal. A triangular process of the frontal bone enters the notch between the main body of the nasal bone and the posteroventrally directed nasal splint. From the dorsal views of all skulls, parts of the frontonasal suture can be seen receding in a posteromedial direction. The shape of the nasal splint and the intruding frontal bone is more like that of Metatelmatherium where the nasal splint is relatively short, narrow, and only slightly arched. Likewise, the triangular process that intrudes into the nasal bone is relatively short and inconspicuous. In Telmatherium validus the nasal splint and frontal process are larger and more pronounced.

Figure 51

Holotype skull of Wickia brevirhinus (CMNH 11380). (A) Left view, (B) dorsal view, (C) anterior view, (D) posterior view.

Figure 54

Views of the ventral surface of skull and upper dentition of Wickia brevirhinus. (A) Ventral view of CMNH 11380, (B) right premolars of CMNH 11382, (C) ventral view of UCMP 81300.

Figure 52

A skull referred to Wickia brevirhinus (UCMP 81300). (A) Left view, (B) dorsal view, (C) anterior view.

Figure 53

A partial skull referred to Wickia brevirhinus (CMNH 11382). (A) Left view, (B) close up of right face showing frontonasal and nasomaxillary sutures. White arrows on the maxilla point to the nasomaxillary suture. Black arrows on the frontal point to the frontonasal suture.

The face of Wickia brevirhinus is tall. The maxilla extends upward to the level of the upper rim of the orbit, where it meets the ventral edge of the nasal splint and forms an arched nasomaxillary suture. The maxilla forms a very shallow preorbital fossa. The nasal incision of W. brevirhinus is shorter than that of Telmatherium but similar to Metatelmatherium; the posterior notch of the nasal incision is positioned between the canine and the P2 in CMNH 11380. The nasal incision extends as far back as the P1 in CMNH 1382 and UCMP 11380. The orbit of W. brevirhinus is positioned directly above M2. The posterolateral root of M1 is rooted below the anteriormost edge of the orbital floor. The anterolateral root of M1 is situated directly below the anterior orbital rim.

The nasal process of the holotype is not preserved, but it is complete in the referred specimens. The nasal bones are poorly fused together. The nasal process is shorter than the premaxillomaxillary rostrum in UCMP 81300, but it is of similar length to the rostrum in CMNH 11382. The nasal process is slightly curved downward, resulting in a convex dorsal surface. The lateral margins of the nasal process form dorsoventrally deep and thin vertical walls that are of relatively constant depth throughout most of their length, but become shallower near the distal end. From a dorsal view the nasal process is narrower than the premaxillomaxillary rostrum and it tapirs slightly distally with a somewhat convex distal edge. The anterior edge of the nasal process is thin, roughened, and deflected downward.

From the lateral view the dorsal margin of the premaxillomaxillary rostrum rises steeply posterodorsally so that the posterior notch of the nasal incision is at the level of the upper half of the orbit. The rostral cavity is open dorsally and continuous with the nasal cavity. The premaxillomaxillary sutures are clearly visible in CMNH 11380. The nasal processes of the premaxillae diverge laterally in a posterior direction and they are truncated before reaching the posterior notch of the nasal incision. The premaxillae of CMNH 11380 are massive and they do not contact each other medially, thus the premaxillary symphysis is not ossified. The premaxillae of UCMP 81300 and CMNH 11382 are somewhat less massive than those of the holotype and they contact each other mesially, although they are not strongly co-ossified. Overall, the premaxillae of Wickia brevirhinus seem more massive than those of Metatelmatherium or Telmatherium. The degree of premaxillary robusticity is somewhat variable in these species and relates to variation in canine size.

The skull of Wickia brevirhinus is rather tall and the posterior portion of the cranium is elevated. From a lateral view the dorsal surface of the skull above the orbits is flat. Behind the orbits the dorsal surface of the skull is slightly convex in lateral profile. From a dorsal view the dorsal surface of the skull is postorbitally very narrow, although the parasagittal ridges do not meet to form a true sagittal crest. An elongate shallow pit resides in the constriction of the parasagittal ridges at the posterior end of the skull.

The zygomatic arches of Wickia brevirhinus are deep and very strongly bowed laterally. The jugal portion of the zygomatic is essentially horizontal while the zygomatic process of the squamosal is steeply angled posterodorsally, giving the zygomatic arch a strongly curved appearance. The large ventral flange that extends from the inferior margin of the zygomatic process of the jugal in Metatelmatherium is not seen among the skulls of W. brevirhinus. W. brevirhinus also lacks an infraorbital jugal process like that seen in Sphenocoelus uintensis.

The nuchal crest of CMNH 11380 is concave from the dorsal view. From the lateral view the occiput is moderately tilted backward. From the posterior view the nuchal crest is dorsally arched, the upper half of the occiput is narrower than the lower portion and the middle is slightly constricted. The occipital pillars are prominent like in Telmatherium validus and the central occipital pit between them is shallow. The nuchal crest of CMNH 11380 is very rugose on its posterior surface, while the remainder of the occipital surface is much smoother. Wickia brevirhinus lacks a distinctive notch with facetlike margins above the foramen magnum that is seen in Metatelmatherium ultimum.

From the ventral views of CMNH 11380 and UCMP 81300 the posterior nares are rimmed by a distinct horseshoe-shaped emargination. The anterior rim of the posterior nares is positioned between the hypocones of the M3. A distinct anteroposteriorly oriented ridge emerges from the midline of the palate and extends posteriorly beyond the anterior rim of the posterior nares. Neither specimen bears evidence of posteriorly extended maxilloturbinates, although they could have been present but were not preserved.

The vomer forms a thin septum that bisects the elongate posterior narial canal. The posterior narial canal of CMNH 11380 extends into the sphenoid bone, but it does not extend past the foramen ovale. The morphology of the posterior narial canal of the referred specimen, UCMP 81300, is presently obscured by a large mass of wax, although the posterior narial canal does not appear to extend quite as far posteriorly in this specimen. Other aspects of the basicranium of Wickia brevirhinus are typical. For instance, the foramen ovale is widely separate from the foramen lacerum. The external auditory pseudomeatus enters the skull in a strictly mediolateral direction and is not constricted ventrally by the mastoid process.

Upper Dentition

All three specimens of Wickia brevirhinus have partial sets of upper incisors (fig. 54). The incisor row forms a short arch anterior to the canines. Though the incisor rows are incompletely preserved, the premaxilla shows room for three large incisors. Little else is noteworthy about the incisors of CMNH 11380. In UCMP 81300, only the left and right I3s are preserved. The nearly intact I3s are large with pointed and lingually curved subcaniniform crowns. A thin cingulum can be seen tracing around the lingual base of the right I3. The I3s are separated from the canines by a distinct diastema on both specimens. The canines themselves are relatively large. The left I1 and I2 of CMNH 11382 are preserved, but they are heavily worn. These incisors are large, yet they are smaller than the large caniniform I3.

The P1s of CMNH 11380 are lost and their alveoli are mostly reabsorbed, nearly erasing all evidence of P1. Due to the absence of P1 the length of the postcanine diastema cannot easily be judged from this specimen, although the gap between the canine and P2 is quite long. The P1s are still present in CMNH 11382 (fig. 54b) and UCMP 81300 (fig. 54c). In these specimens the postcanine diastema is relatively short. P1 is small with a simple crown with a single cusp, a posterior heel, and a thin lingual cingulum. There is no P1–P2 diastema.

The crowns of the P2–P4 of UCMP 11380 are heavily worn and reveal only basic details, while those of UCMP 81300 and CMNH 11382 are substantially less worn. The anterior margin of P2 is angled posterolingually, giving the crown of that tooth a somewhat more oblique outline. P3 is less oblique in outline while the anterior and posterior sides of P4 are essentially parallel. The parastyle and metastyle of the P2 arch slightly lingually, while those of P3 are approximately straight. The parastyle of P4 is strongly angled anterolabially, while the metastyle of P4 is straight. The labial sides of the paracones of P3 and P4 are more strongly convex than those of the metacones.

In CMNH 11380 and CMNH 11382, the lingual side of the P2 is heavily worn. However, on the P2 of UCMP 81300 a short crest extends anterolingually from the protocone and continues along the anterior side of the crown. The lingual side of P2 exhibits a short preprotocrista as well as a short lingual crest extending posteriorly from the protocone. In P3 there is a small but distinct preprotocrista and a short lingual crest. Finally P4 has a very indistinct preprotocrista and there is no lingual crest. However, a lingual crest is present on the P4 of CMNH 11382. None of the premolars has a hypocone.

The molars of CMNH 11380 are extremely worn, and their proportions have been significantly altered by interstitial wear. The molars of UCMP 81300 and CMNH 11832 serve as better sources for the molar morphology of this species. The molars of Wickia brevirhinus exhibit typical brontotheriine traits including tall, lingually angled ectolophs, weak labial ribs, thin lingual ectoloph enamel, and wedge-shaped lingual sides of the paracone and metacone in molars that are not heavily worn (e.g., M3). The anterior cingulum is thin and passes proximally to the distal peak of the parastyle. W. brevirhinus molars exhibit small anterolingual cingular peaks, but they lack central molar fossae. The molars retain vestigial paraconules. The M3 lacks a hypocone.

Mandible and Lower Dentition

The holotype of Wickia brevirhinus lacks an associated mandible, however, a referred specimen, CMNH 11382, includes a complete mandible with partially preserved incisors and complete cheektooth rows (fig. 55). The ramus of CMNH 11382 is similar in proportion to those of Telmatherium validus and Metatelmatherium ultimum, including a tall coronoid process and relatively steeply angled (∼45°) ventral margin of the symphysis. The symphysis extends posteriorly to the p3 metaconid.

Figure 55

Mandible of Wickia brevirhinus (CMNH 11382). (A) Right view, (B) left premolars, (C) dorsal view.

The remnants of the three incisors of CMNH 11382 are relatively large and form a short arch anterior to the canines, although they are too damaged to describe other details of their morphology. There is no precanine diastema, but the postcanine diastema is somewhat longer than the p2.

The p1 of CMNH 11382 is small and simple with a single cusp and a short talonid heel. There is no p1–p2 diastema. The p2 of CMNH 11382 is slender and ovoid in outline. The trigonid of the p2 is nearly twice the length of the talonid, the p3 trigonid is somewhat longer than the talonid, and the p4 trigonid is slightly shorter than the talonid. The p2 trigonid and talonid are of similar width, while the trigonids are narrower than the talonids in p3 and p4. The p2 paralophid projects anteriorly from the protoconid and there is only a slight lingual notch in the p2 trigonid. The p2 protolophid is short and posteriorly directed. The paralophid of p3 is angled about 45° lingually, creating a distinct lingual notch in the trigonid. The p3 protolophid is straight but positioned lingually. Finally, the p4 trigonid is essentially molariform with a lingually arching paralophid and protolophid. A distinct metaconid is seen only on the p4. The p2 talonid of CMNH 11382 is very simple; the cristid obliqua is not well developed and there is essentially no talonid basin or hypolophid. The p3 and p4 of CMNH 11382 have more well-developed cristids obliqua, elongate hypolophids, and broader basins.

The molars of Wickia brevirhinus are typical, with thinner lingual enamel, shallow talonid and trigonid basins, and an elongated m3. The labial cingulid is discontinuous around the bases of the cusps, but it is continuous between them. The m3 cingulum does not wrap around the distal end of the m3.

Remarks

Wickia brevirhinus is based on a skull (CMNH 11380) from the Sand Wash Basin of Colorado. In addition, a partial skull and mandible (CMNH 11380) from the Sand Wash Basin and a partial skull (UCMP 81300) from the middle Adobe Town Member of the Washakie Formation of Wyoming are referable to W. brevirhinus. West and Dawson (1975) originally assigned two of these specimens, CMNH 11380 and CMNH 11382, to Manteoceras pratensis Cook (1926). However, M. pratensis is based upon a poorly preserved partial mandible with deciduous teeth and is considered a nomen dubium in this paper.

Nonetheless, West and Dawson (1975) noted that CMNH 11380 and CMNH 11382 share attributes with Manteoceras manteoceras ( =  Telmatherium validus) and Metatelmatherium ultimum. However, Wickia brevirhinus distinctly differs from both of these species due to a unique combination of characters. The skulls of W. brevirhinus differ most conspicuously from Telmatherium validus in the shorter nasal incision although there are other minor differences such as the narrower and shorter lateral nasal splint, less prominent parasagittal ridges, and minor anterolingual cingular cusps on the molars. W. brevirhinus differs from Metatelmatherium ultimum most conspicuously in the lack of a true sagittal crest, the lack of a large ventral flange on the zygomatic arch, and in having more prominent occipital pillars. Moreover, W. brevirhinus lacks the notch on the anterior rim of the foramen magnum that is seen in the type skull of Metatelmatherium ultimum. Finally, the p2 talonid of W. brevirhinus is poorly developed in comparison to Telmatherium validus or Metatelmatherium ultimum.

In many ways Wickia brevirhinus is morphologically intermediate between Telmatherium validus and Metatelmatherium ultimum. Temporally, it also bridges the gap between the late Bridgerian T. validus and the late Uintan M. ultimum. W. brevirhinus specimens from the Sand Wash Basin of Colorado are either latest Bridgerian or Early Uintan in age. The Washakie Basin skull, UCMP 81300, from above level 17 of Granger (1909), which occurs within the middle Adobe Town Member of the Washakie Formation (TWKA2), suggests an earliest Uintan age for W. brevirhinus (McCarroll et al., 1996b).

Metatelmatherium ultimum (Osborn, 1908a)

Holotype

AMNH 2060, a complete skull with right I2, C (partial), P1–M3, left I2–I3, C (partial), P1–M3, and a partial mandible with symphysis and right ramus with c (partial), p3 (partial), p4–m3.

Type Locality

Myton Member (lower part of Uinta C) of the Uinta Formation of Utah.

Synonyms

Manteoceras uintensis Douglass, 1909; Metatelmatherium cristatum Granger and Gregory, 1938.

Age

Middle Eocene (Uintan and Irdinmanhan land mammal “ages”).

Referred Specimens

(From the Myton Member of the Uinta Basin of Utah) AMNH 2004, an anterior portion of a skull with right I2 (partial), I3–M3, left I2 (partial), I3, and P3–M3; AMNH 2029, a crushed ventral surface of a skull with right P1, P2, P4, M1–M3 (partial), left P1–P4, and M1–M3 (all partial); AMNH 2033, a mandible with partial symphysis and right ramus with p2–m3; CMNH 2339, a partial skull with no teeth; CMNH 2388 (holotype of Manteoceras uintensis), a crushed anterior portion of a skull with right I1–M3, left I1, I3, C (partial), P1–P3, P4–M3 (all partial); CMNH 2354, a skull fragment with right P2–M1 and left P2–P4 (partial); (from the Adobe Town Member of the Washakie Formation of Wyoming) UCMP 81447, a skull with right I1–M3 (all broken) and left I3–M3; (from the Wind River Basin, Hot Springs County, Wyoming) YPM 14158, a skull with no complete teeth; (from the “Irdin Manha” Formation [sensu Radinsky, 1964] Camp Margetts area, Inner Mongolia, China) AMNH 26411 (holotype of Metatelmatherium cristatum), a complete skull with right and left I2–M3, and a mandible with right i1, i3–p1, p2 (partial), p3–m3, and left i1–m3.

Diagnosis

Metatelmatherium ultimum is a large hornless brontothere. The frontal bone intrudes into the nasal bone and splits off a small lateral splint of nasal bone from the main body of the nasal. The size and shape of the nasal splint is similar to that of Wickia brevirhinus and less pronounced than that of Telmatherium validus. The nasal incision is short and does not extend posteriorly beyond the P1. The nasal process is very short, horizontal, unelevated, of relatively constant transverse width, narrow, with thin and relatively deep lateral walls, and without a well-defined or strongly rounded distal margin. The orbits do not protrude laterally. The orbits are positioned directly above the M2 with the posterolateral root of M1 directly below the anterior rim of the orbit. The premaxillomaxillary rostrum deepens proximally and the rostral cavity is dorsally sealed by bone. Other cranial characteristics include a relatively flat dorsal cranial surface (from a lateral view), a short sagittal crest, an elevated occiput, deep and strongly curved zygomatic arches that are very strongly bowed laterally, a large ventral zygomatic flange on the jugal, and a ventrally open and mediolaterally angled external auditory pseudomeatus. Ventral sphenoidal fossae are absent.

Dentally, Metatelmatherium ultimum has large subcaniniform upper incisors, a simple P1, a distinct P2 metacone, a postcanine diastema, weak premolar preprotocristae, and short lingual crests extending posteriorly from the premolar protocones. Premolar hypocones are absent. The molars of M. ultimum have tall, lingually angled ectolophs with weak labial ribs, and thinned lingual ectoloph enamel with wedge-shaped paracones and metacones. Central molar fossae and cingular parastyle shelves are absent, but small anterolingual cingular cusps are present. The upper molars lack paraconules and metalophs. The lower dentition of M. ultimum includes large subcaniniform incisors that are all of similar size, a postcanine diastema, a p1–p2 diastema (variably present), an elongate p2 trigonid, a metaconid on p4 but not on p2 or p3, shallow molar basins, and a slender m3.

Metatelmatherium ultimum is most similar to Telmatherium validus and Wickia brevirhinus, but it is clearly distinct from these species primarily due to the unique combination of a very short nasal incision, a true sagittal crest, absence of strong occipital pillars, a deep notch in the dorsal rim of the foramen magnum, and a large ventral zygomatic flange.

Description

Skull

The holotype skull of Metatelmatherium ultimum (AMNH 2060) is nearly whole and undistorted (fig. 56). Only the nasal process is fragmentary, but it is nearly complete and has been reassembled and reattached to the skull. Seven other skulls from North America are referable to M. ultimum, as well as an associated skull and mandible from Asia (AMNH 26411) (fig. 57). The following description of the skull of M. ultimum is based primarily on the holotype but is supplemented by other specimens where noted.

Figure 56

Holotype skull of Metatelmatherium ultimum (AMNH 2060). (A) Left view, (B) dorsal view, (C) anterior view, (D) posterior view.

Figure 57

Skull of Metatelmatherium ultimum (AMNH 26411) from Asia. (A) Left view, (B) right view.

Metatelmatherium ultimum is a relatively large (table 8) hornless brontothere whose skull most closely resembles those of Telmatherium validus and Wickia brevirhinus. Upon close inspection the sutures in the facial area are discernable in AMNH 2060 although they are indistinct. The sutures of the facial bones are more distinct in a referred specimen (YPM 14158) (fig. 58). The frontonasal suture recedes posteriorly toward the midline, but near the midline it is acutely redirected anteriorly. The nasal bone is split by a small triangular process of the frontal just above and anterior to the orbits. The small splint of nasal bone that branches off the main body of the nasal arches posteroventrally between the frontal and the maxilla. Below this splint, the nasomaxillary suture is strongly dorsally arched. The face of M. ultimum is tall and the maxilla forms a shallow preorbital fossa. The shape and configuration of the facial bones of M. ultimum most closely resembles that of Wickia brevirhinus. The frontonasal suture in Telmatherium validus is more pronounced, the triangular frontal process is longer, and the nasal splint is broader, longer, and more arched. Previous figures of AMNH 2060 incorrectly depicted the configuration of the facial sutures of M. ultimum (Osborn, 1908a: fig. 17; 1929a: figs. 126, 294–296, pl. 51a). In these earlier figures, the frontal bone is depicted as having a broad contact with the maxilla rather than being separated by the lateral nasal splint.

Table 8

Summary statistics for selected morphometric variables of Metatelmatherium ultimum See Methods for measurement definitions

Figure 58

Close-up view of the right face of a skull of Metatelmatherium ultimum showing the configuration of the maxilla, nasal, and frontal bones. (Courtesy of Division of Vertebrate Paleontology, YPM 14158. © 2005 Peabody Museum of Natural History, Yale University, New Haven, Connecticut, USA. All rights reserved.)

Earlier descriptions of Metatelmatherium ultimum are misleading in other respects. For instance, Osborn (1929a: 346) described “horn rudiments” in male skulls. He considered the holotype (AMNH 2060) to be a female skull but regarded AMNH 2004 as male. In reexamining AMNH 2004, the triangular frontal process does not form a thickened protuberance; nor can I find a structure that could be interpreted as a “horn rudiment” in any other specimen of M. ultimum.

The nasal incision of Metatelmatherium ultimum is very short; in AMNH 2060 it extends only to the anterior margin of the P1. The position of the posterior margin of the nasal incision is similar to that of Wickia, but much shorter than Telmatherium. Because of the short nasal incision the preorbital region of the face appears to be quite long. The orbits are positioned directly above M2 while the posterior lateral root of M1 is positioned directly below the anterior rim of the orbit.

The nasal bones are poorly fused. The nasal process is thin, projects horizontally from the skull, and is shorter than the premaxillomaxillary rostrum. In the holotype the nasal process is straight with a flat dorsal margin. The sides of the nasal process form deep, thin, and vertical lateral walls. The depth of the lateral walls is relatively constant throughout the length of the nasal process, but they become shallower at the very distal end. From a dorsal view the nasal process is narrower than the rostrum. Other specimens with more intact nasal processes (e.g., AMNH 2004, CMNH 2339) are morphologically consistent with that of the holotype. These specimens also more clearly reveal that the anterior edge of the nasal process is thin, roughened, and very slightly downwardly deflected.

From the lateral view the dorsal border of the premaxillomaxillary process is steeply sloped posterodorsally so that the posterior margin of the nasal incision is elevated to the level of the upper rim of the orbit. The rostral cavity is open dorsally. The premaxillomaxillary suture is clearly visible on the right side of AMNH 2060 and the premaxilla terminates near the posterior notch of the nasal incision. There is no contact of the nasal and premaxillary bones. The premaxillae are not co-ossified at the symphysis. The premaxillae are not nearly as massive as those of Wickia.

Like Wickia, the skull of Metatelmatherium ultimum is tall and the occiput is elevated. This condition is most extreme in CMNH 2339 (fig. 59). From a lateral view, the dorsal surface of the skull is flat or slightly convex over the orbits. Postorbitally, the dorsal surface is flat, although it is angled posterodorsally. From the dorsal view of AMNH 2060 the skull is very narrow between the orbits. From the anterior view the dorsal cranial roof is vaulted. The appearance of a narrow arched skull roof seems to be slightly exaggerated by some lateral crushing in the holotype. In specimens that are less laterally crushed (e.g., CMNH 2399, YPM 14158, UCMP 81477), the dorsal roof of the cranium is wider and less vaulted than that of the holotype. Nonetheless, the superorbital dorsal skull roof is narrow in M. ultimum. The parasagittal ridges merge posteriorly to form a short but thin sagittal crest. Each parasagittal ridge forms a prominent ridge on the sagittal crest, although there is no distinct median pit between the parasagittal ridges as seen in Telmatherium.

Figure 59

Lateral right view of a skull (CMNH 2339) of Metatelmatherium ultimum.

From a dorsal view of AMNH 2060 the nuchal crest is very thin and concave. From a lateral view the occiput is slightly tilted backward. From a posterior view, the nuchal crest is strongly arched dorsally. The dorsal portion of the occiput is somewhat narrower than the ventral portion and the middle of the occiput is constricted. The occiput of AMNH 2060 is unusual in a number of respects. The entire surface of the occiput is roughened, particularly above the occipital condyles and below the nuchal crest. Two short crests originate near the top of the nuchal crest and converge ventromedially into a small process in the central depression of the occiput. The thick occipital pillars seen in most other brontothere skulls are not discernable in AMNH 2060. Finally, there is a tall and narrow median notch on the upper margin of the foramen magnum. On the lateral edges of this notch are two elongate, smooth, facetlike surfaces.

The zygomatics of Metatelmatherium ultimum are deep, relatively thin, and very strongly bowed laterally. The jugal portion of the zygomatic process is angled downward at a shallow angle, while the squamosal portion of the zygomatic is steeply angled posterodorsally, giving the zygomatic arch a strong curvature. The curvature of the zygomatic is exaggerated by a large flange on the ventral margin of the zygomatic process of the jugal just below the junction with the squamosal. A similar flange is occasionally seen on the skulls of other brontotheres (e.g., Epimanteoceras formosus), but it is always less conspicuous in comparison to M. ultimum, where the ventral zygomatic flange is pronounced and greatly influences the overall appearance of the zygomatic arch. The ventral flange is always conspicuous but its size varies; in CMNH 2339 the prominence of the ventral zygomatic flange is extreme (fig. 59). An infraorbital process of the jugal, such as that of Sphenocoelus, is not seen in M. ultimum.

In AMNH 2060, the posterior nares are rimmed by a narrow horseshoe-shaped emargination (fig. 60a). The anterior rim of the posterior nares is situated between the anterior edges of the M3s. In other specimens, the anterior rim of the posterior nares fluctuates in its position from between the anterior edge of M3 to between the protocones of M2. A thin ridge of bone runs along the midline of the palate and extends past the anterior margin of the posterior nares. The posterior narial canal does not extend into the basisphenoid bone. The thin vomerine septum that bisected the posterior narial canal is not preserved in AMNH 2060, although remnants of it can be seen in the roof of the posterior narial canal. The basicranium of AMNH 2060 is quite short. The configuration of the basicranial foramina is typical, with the foramina of the alar canal, the foramen ovale, and the foramen lacerum being widely separated. The external auditory pseudomeatus is narrow; however, the mastoid process does not make contact with the postglenoid process and the external auditory pseudomeatus is unconstricted ventrally.

Figure 60

Ventral view of the skull and upper dentition of Metatelmatherium ultimum. (A) Ventral view, (B) left molars, and (C) left premolars of AMNH 2060, (D) upper incisors and canines of CMNH 2388.

Upper Dentition

The following description of the upper dentition is primarily based on the holotype (AMNH 2060) (fig. 60a–c), although this specimen has an incomplete incisor row. The incisors of AMNH 2060 are large and arch anterior to the canines. Alveoli indicate three pairs of incisors. The I1 crowns of AMNH 2060 are not preserved, although the root of the left I1 is preserved. Judging by the root of I1 it was the smallest incisor. The complete I1s of CMNH 2388 (fig. 60d) confirm that the I1 is the smallest incisor. The I1 is subcaniniform and similar in shape to the I2 and I3. I2 and I3 are large and subcaniniform with a single lingually curved cusp and a very thin lingual cingulum. No labial cingula are seen on the upper incisors. The I3 is larger in diameter and taller than the I2. The canine is bordered by a short precanine diastema and a longer postcanine diastema. The crowns of both canines are fragmentary in AMNH 2060, but in other specimens the canines vary in size from moderate to relatively large, except in the Asian skull (AMNH 26411), which has very large canines.

P1 is relatively large. It is about the same length as P2, but it is narrower and with a simpler morphology, including a single cusp and a posterior heel. A thin cingulum can be seen in the lingual side of the posterior heel. The posterior heel of the P1 of the Asian specimen (AMNH 26411) is somewhat wider than those of the North American specimens. There is no P1–P2 diastema in any specimen. The P2 of AMNH 2060 is oblique in shape due to the steeply posterolingually angled anterior margin. P2 is less oblique in AMNH 2004, although the P2s of other specimens are more similar to the holotype in this respect. In AMNH 2060 and other specimens P3 is much less oblique and P4 is not oblique.

The P2 parastyle and metastyle arch slightly lingually, while the P3 parastyle and metastyle are nearly straight. The P4 parastyle is strongly angled lingually, while the metastyle is straight. Because of these shape differences the outer wall of the ectoloph of P2 is more curved than those of P3 and P4. In AMNH 2060 each premolar (P2–P4) has a small distinct labial paracone rib though the ectolophs are too worn to compare their relative sizes. The premolars of AMNH 2004 are less worn and reveal that the labial paracone ribs are distinct and become narrower and shorter in more posterior premolars.

On the P2 of AMNH 2060 there is a small protocone with a small but distinct preprotocrista and a short lingual crest extending posteriorly from the protocone. The lingual morphology of P2 varies in other skulls. For instance, on the P2 of AMNH 2004 there is no distinct protocone; instead, a large crest arches around the lingual side of the crown. The lingual heels of P3 and P4, though larger and wider than that of P2, have less distinct preprotocristae and less distinct lingual crests. There are no hypocones on any premolars of Metatelmatherium ultimum. The labial cingula of the premolars tend to be discontinuous around the paracones except on the P4. The lingual cingula of the P2–P4 vary from being completely continuous to slightly discontinuous around the protocone.

The molars of AMNH 2060 are heavily worn and M1 and M2 have been shortened by interstitial wear. The M3 is minimally worn and preserves a number of brontotheriine traits including tall, lingually angled ectolophs, weak labial ribs, thin lingual ectoloph enamel, and wedge-shaped lingual sides of the paracone and metacone. The anterior cingulum is thin and passes proximally to the distal peak of the parastyle. There is a small anterolingual cingular peak on each molar. Central molar fossae are absent. The molars have no visible remnants of paraconules or metalophs. There is no hypocone on the M3, but a thick cingulum traces around the distolingual corner of the M3 crown. The labial molar cingula tend to be thick at the bases of the cusps, but they are discontinuous around the mesostyles. The lingual cingulum of the molars is thick, but it is only continuous around the base of the protocone of the M3.

Mandible and Lower Dentition

The holotype of Metatelmatherium ultimum (AMNH 2060) includes an associated partial mandible with p4–m3. Another partial mandible, AMNH 2033, is consistent with the holotype mandible, and has a nearly complete symphysis with complete p2–m3 (fig. 61). The mandible associated with the Asian skull, AMNH 26411, is fully complete with a nearly complete set of lower dentition (fig. 62).

Figure 61

Mandible of Metatelmatherium ultimum (AMNH 2033). (A) Right view (B) dorsal view.

Figure 62

Mandible of Metatelmatherium ultimum (AMNH 26411) from Asia. (A) Left view, (B) dorsal view, (C) left premolars, (D) labial view of incisors and canines, (E) lingual view of incisors and canines.

In comparison to AMNH 26411, the partial mandible of AMNH 2033 has a slightly more slender coronoid process and a somewhat narrower symphysis with a steeper inferior margin. However, these differences could be attributed to distortion and damage in AMNH 26411. The coronoid process of AMNH 26411 is crushed and partially reconstructed with plaster. In addition, the symphysis has gaps that are filled with plaster and the seemingly wider symphysis and shallower inferior margin could relate to the way in which this specimen has been distorted and subsequently reconstructed. Additionally, the larger canines of the Asian specimen, with their rather bulbous roots, influence the size and shape of the symphysis. The posterior margin of the mandibular symphysis fluctuates in position from between the talonid of the p2 (AMNH 26411) to between the anterior margins of the p3 (AMNH 2033).

The lower incisors of AMNH 26411 are somewhat smaller than the upper incisors of the same specimen. The incisor row forms a short arch anterior to the canines. The apex of i1 is rounded although this seems due to wear. The i2 is larger and the apex is much more pointed. The i3 is about the same size as i2, although it is stouter and more subcaniniform. Lingual cingulids can still be discerned on these incisors, but labial cingulids are absent. The extremely procumbent condition of the incisors of AMNH 26411 is an artifact of distortion. In life the lower incisors would have been more nearly vertical. The canines of the Asian specimen are relatively large. There is no precanine diastema, but the postcanine diastema in these specimens is relatively longer.

The p1 of AMNH 26411 is small and simple with a single cusp and a short talonid heel. AMNH 2033 has a short p1–p2 diastema. However, AMNH 26411 lacks a p1–p2 diastema. The p2 trigonid is nearly twice as long at the talonid. The p3 trigonid is somewhat longer than the talonid. The talonid and trigonid of the p4 are of similar length. The trigonids of p2, p3, and p4 are narrower than the talonids. The p2 paralophid curves slightly lingually as it projects from the protoconid, creating a slight lingual notch in the trigonid. The p2 protolophid is straight but slightly angled lingually. The paralophid of p3 is angled about 45° lingually, creating a distinct lingual notch in the trigonid. The p3 protolophid is straight but weakly angled in the lingual direction. Finally, the p4 trigonid is essentially molariform with a lingually arching paralophid and protolophid. A distinct metaconid is seen only on the p4. The talonids of the p2–p4 are well developed with distinct cristids obliqua and hypolophids. The lingual surface of the p2 talonid forms a steeply sloped surface, although the p3 and p4 have nearly molariform talonid basins.

The molars of Metatelmatherium ultimum are typical with thinner lingual enamel, shallow talonid and trigonid basins, and an elongated m3. The labial cingulid is discontinuous around the bases of the cusps. The m3 cingulid does not wrap around the distal end of the m3.

Remarks

Osborn (1908a) originally referred Metatelmatherium ultimum to the genus Telmatherium. Although this species was based on a very well-preserved skull and a partial right mandible (AMNH 2060), Osborn's (1908a) initial description amounted to no more than four brief sentences. Douglass (1909) named another species, Manteoceras uintensis, on an anterior portion of a cranium with a complete set of upper dentition (CMNH 2388). Douglass' only stated reason for differentiating this species from Telmatherium ultimum is, at best, a dubious distinction, “The zygomatic arch is not very heavy…. It is not nearly so heavy as in Telmatherium ultimum” (Douglass, 1909: 307). The figure of CMNH 2388 in Douglass (1909: figs. 4, 5) shows this specimen to be more complete and better preserved than it actually is. Although the specimen is figured as having zygomatic arches, only the anteriormost portion of the right arch is actually preserved.

Osborn (1929a) continued to accept Telmatherium ultimum and Manteoceras uintensis as separate species, and he believed them to belong to different lineages. He distinguished M. uintensis from T. ultimum based on “the obliquely flattened form of the infraorbital portion of the malars” (Osborn, 1929a: 374). However, when one considers that the specimens are subtly distorted in different dimensions, one cannot realistically consider this difference as significant. Mader (1989) considered these two species to be synonymous and that revision is upheld here.

Granger and Gregory (1938) named a new genus and species, Metatelmatherium cristatum from a complete skull and mandible (AMNH 26411) from the Camp Margetts area of Inner Mongolia. Granger and Gregory (1943) noted that North American specimens referred to Telmatherium ultimum are nearly identical to M. cristatum. Therefore, they referred the North American species, T. ultimum, to their new genus, Metatelmatherium. They noted that the type skull of the North American species “so closely resembles the type of our Metatelmatherium cristatum that one can barely discover specific differences between them, while their congeneric relationship becomes more evident the longer they are studied” (Granger and Gregory, 1943: 356).

Granger and Gregory (1943) distinguished Metatelmatherium cristatum from M. ultimum by its apparently larger size, the shorter and broader coronoid process, and the relatively longer and more sloping mandibular symphysis. However, all of these distinctions are dubious. The Asian skull is larger than the holotype but not significantly. In fact, many of its dimensions fall well within the range of North American M. ultimum. Though the skull is uncrushed, it has experienced severe expanding matrix distortion (sensu White, 2003). The skull looks larger than it probably was when it was intact, and some aspects of its shape could be subtly distorted. In the jaw of the Asian specimen the coronoid process is fragmented, its distal tip is missing, and there are copious amounts of plaster filling gaps between the bone fragments. The symphysis seems wider that that of the North American specimen, AMNH 2033. However, there are large amounts of plaster in the mandibular symphysis of the Asian jaw, and the degree to which the shape of the symphysis has been influenced by distortion and reconstruction is uncertain. Given the variability in canine size of other brontotheres species, the influence that the unusually large canine of the Asian specimen had on the overall shape of the anterior portion of the mandible is probably not taxonomically significant. Another difference, the wider posterior heel of the P1, is not much more compelling particularly when one considers the high level of intraspecific variation in brontothere premolars. For instance, wide and narrow P1 posterior heels can be found in a single individual of Protitanotherium emarginatum (YPM PU11242).

In more recent revisions Mader (1989; 1998) continued to recognize Metatelmatherium cristatum and M. ultimum as distinct species. However, M. cristatum Granger and Gregory (1938) is here considered to be a junior synonym of Metatelmatherium ultimum (Osborn, 1908a). Thus, M. ultimum becomes the type species of Metatelmatherium.

Metatelmatherium ultimum as defined here primarily occurs in the late Uintan deposits of the Uinta Basin, but it is also known from the Adobe town member of the Washakie Formation (UCMP 81447), the Wind River Basin (YPM 14158), and the “Irdin Manha” (put in quotes to denote uncertain relationship with the type Irdin Manha beds [Radinsky, 1964]) of the Camp Margetts area of Inner Mongolia (AMNH 26411). These specimens are morphologically consistent with the M. ultimum material from the Uinta Formation, although the Washakie Basin specimen is notably smaller and could represent an earlier (early Uintan) species of Metatelmatherium, although, based on the present material, the only differentiating factor would be size. Presently, I include this specimen with M. ultimum.

Some authors have not accepted the generic distinction of Metatelmatherium from Telmatherium (Qi, 1987), while most others have accepted it (Russell and Zhai, 1987; Prothero, 1996; Mader, 1989, 1998). Metatelmatherium ultimum is similar to Telmatherium validus, but can be differentiated from it on the following features: the very large ventral flange on the zygomatic arch, the extremely shallow nasal incision, short nasal processes, thinner sagittal crest, the narrow vaulted cranial roof, and small anterolingual cingular peaks on the molars. M. ultimum is more similar to Wickia brevirhinus, but it can be distinguished from that species primarily by the large autapomorphic ventral zygomatic flange, thin sagittal crest and, possibly, the more developed p2 talonid. Several other species have been assigned to Metatelmatherium, although they are all now considered to be nomina dubia and are further discussed in the section dealing with miscellaneous dubious taxa.

Epimanteoceras formosus Granger and Gregory (1943)

Holotype

AMNH 21613, a complete skull with complete dentition.

Type Locality

Ulan Shireh Formation, four miles north of Tukhum Lamasery, Inner Mongolia, China.

Age

Middle Eocene (Irdinmanhan land mammal “age”).

Synonyms

Dolichorhinoides angustidens Granger and Gregory, 1943.

Referred Specimen

(From the Ulan Shireh Formation, eight miles north of Tukhum-in-Sumu, Inner Mongolia) AMNH 21607 (holotype of Dolichorhinoides angustidens), a partial skull with right P1–M3, left P4 (partial), and M1–M3.

Diagnosis

Epimanteoceras formosus is a large brontothere with poorly developed frontonasal horns. The frontal bone intrudes into the surface of the nasal bone, thus splitting off a lateral nasal splint from the main body of the nasal bone. Some specimens of Epimanteoceras formosus exhibit a distinct but weakly developed frontonasal protuberance. The nasal incision extends as far back as the P2. The nasal process broadens distally; it is nearly horizontal, unelevated, narrow, with thin lateral walls, and without a well-defined or strongly rounded distal margin. The orbits are positioned above the posterior portion of M2 and the anterior portion of M3. The orbits do not protrude laterally. The premaxillomaxillary rostrum deepens posteriorly and it is not enclosed by bone dorsally. Other cranial characteristics include a dorsal cranial surface that is convex but incompletely saddle-shaped, separated parasagittal ridges, distinctly curved zygomatic arches, and a ventrally open and mediolaterally angled external auditory pseudomeatus. Additionally, E. formosus is unique in having a very robust superorbital process and a distinct lateral temporal ridge.

Epimanteoceras formosus has subcaniniform upper incisors, a metacone on the P1, a distinct P2 metacone, and weak premolar preprotocristae. Premolar hypocones are present and connected to the protocones, but they are small and occasionally absent in some of the premolars. The molars of E. formosus have tall, lingually angled ectolophs with weak labial ribs, and thinned lingual ectoloph enamel with wedge-shaped paracones and metacones. A cingular parastyle shelf is absent. Central molar fossae are present, however, anterolingual cingular cusps are absent. Paraconules and metalophs are absent.

The skull of Epimanteoceras formosus is morphologically intermediate between Telmatherium and Protitan. It can be most easily distinguished from the former by its greater size, more widely separated parasagittal ridges, and distinct central molar fossae. It can be distinguished from the later by its more weakly developed frontonasal protuberance, incompletely saddled-shaped cranium, and lack of a postzygomatic process. The upper premolars are significantly more molarized than either Protitan or Telmatherium.

Description

Skull

Epimanteoceras formosus is a large brontothere that is most similar in size to Protitan grangeri and Protitanotherium emarginatum. This species is known from a nearly complete and undistorted holotype skull (AMNH 21613) (fig. 63). The only significantly reconstructed portion of the skull is the left posterior region. An additional partial skull, AMNH 21607, is referable to E. formosus (fig. 64). This skull is less complete, laterally compressed, and the dorsal surface is distinctly warped; however, this skull is superior to the holotype in some ways; it is ontogenetically younger than the holotype with less worn cheek teeth and more distinct cranial sutures. Additionally, this specimen was less extensively prepared and some aspects of the ventral surface of the skull are less damaged and not covered with plaster.

Figure 63

The holotype skull of Epimanteoceras formosus (AMNH 21613). (A) Right view, (B) dorsal view, (C) anterior view, (D) posterior view.

Figure 64

A skull (AMNH 21607) referred to Epimanteoceras formosus. (A) Right view with skull rotated slightly to see to dorsal surface, (B) ventral view, (C) left molars, (D) right premolars.

It can be seen on AMNH 21613 and even more clearly on AMNH 21607 that the distinctive configuration of facial bones seen in some hornless brontotheres, such as Telmatherium, is shared by Epimanteoceras formosus. A large triangular process of the frontal bone splits the posterolateral portion of the nasal bone. Dorsally, the frontonasal suture recedes posteromedially, but at the midline it is acutely reoriented anteriorly. Laterally, the frontonasal suture is directed sharply backward toward the orbit. The lateral nasal splint arcs posteroventrally and contacts the lacrimal bone, thus preventing contact of the maxilla and frontal bone. The nasomaxillary suture is less distinct than the frontonasal suture (and cannot clearly be seen in the photo), but it forms a dorsally arched contact between the lateral nasal splint and the maxilla.

In AMNH 21613 the overlapping frontal process forms a small but distinct protuberance above and directly anterior to the orbit. This frontonasal protuberance is not as well developed as those of Protitan species, but it is more distinctly more developed than that of Telmatherium in which the triangular frontal process is flush with the dorsal surface of the skull. In addition to the frontal nasal protuberance, AMNH 21613 has a very thick and prominent superorbital process that overhangs the orbits. This superorbital process is thickened with a roughened surface that is semicontinuous with that of the frontonasal protuberance. In AMNH 21607, the triangular process does not form a distinct frontonasal protuberance and it is smooth and nearly flush with the surface of the skull. Likewise, the superorbital processes seem less laterally prominent but this could be attributed to the fact that the skull is laterally compressed, and/or that the specimens is ontogenetically younger than the holotype skull.

The maxilla forms a shallow preorbital concavity. The face is only moderately constricted by the nasal incision, which extends to a point above the anterior margin of P2. The orbit is positioned over the anterior portion of M3 and the posterior portion of M2. The anterolateral root of M2 and posterolateral root of M1 rest directly below the anterior rim of the orbit.

The nasal process is the same length as the premaxillomaxillary rostrum. The nasal process of AMNH 21613 is nearly straight and the dorsal surface of the nasal process is slightly convex from the lateral view. The nasal process projects anteriorly from the skull in a slightly upward direction. In contrast, the nasal process of AMNH 21607 has a slight downward curvature. This subtle difference in nasal orientation may relate to taphonomic distortion in the latter specimen. The sides of the nasal process form deep and thin vertical walls that extend to the distal end of the nasal process. From the dorsal view it can be seen that the nasal bones are poorly ossified. The nasal process is narrower than the premaxillomaxillary rostrum, although the nasal process significantly broadens distally. The distal edge of the nasal process is thin, roughened, and nearly flat from a dorsal view. However, the anterior view reveals that the anterior edge is turned downward. The mesiodistal corners of each nasal bone form a single downturned distal tip.

The premaxillomaxillary rostrum is robust and from a lateral view it appears to deepen in a posterior direction. The dorsolateral surface of the rostrum rises posteriorly at a shallow angle so that the posterior notch of the nasal incision is level with the upper rim of the orbit. The premaxillomaxillary suture is distinct in AMNH 21613. The nasal processes of the premaxillae are truncated anterior to the posterior notch of the nasal incision and do not contact the nasal bone. The premaxillary symphysis of AMNH 21613 is not ossified. From the anterior view the lateral margins of the rostrum are posterolaterally divergent and the rostral cavity is open dorsally and continuous with the nasal cavity.

The dorsal surface of the skull of Epimanteoceras formosus forms a shallow concavity. However, the skull is incompletely saddle-shaped; the dorsal surface is flat or slightly convex posteriorly. From a dorsal view the parasagittal ridges remain separate throughout their length and only moderately constrict the dorsal surface posteriorly. The parasagittal ridges are prominent and they can be seen standing out in relief from the posterior view of AMNH 21613.

The zygomatic arches are rather thin. From a dorsal view the jugal portion of the zygomatic is straight and strongly angled posterolaterally. From a lateral view the jugal portion is shallow and horizontal, while the squamosal portion of the zygomatic arch is deeper and rises posteriorly at a shallow angle, thus giving the zygomatic arch a distinct curvature. A small ventral flange can be seen below the jugal-squamosal contact, but this flange is not nearly as conspicuous as that of Metatelmatherium. A postzygomatic process, as seen in Protitan, is not present. One peculiar aspect of the skull of Epimanteoceras formosus is the presence of a prominent temporal ridge that extends anteroposteriorly on the lateral surface of the skull behind the zygomatic arch.

The occipital region is the least well-preserved portion of the holotype specimen. From the right side it can be seen that the occiput is strongly tilted backward. From a dorsal view the nuchal crest is mildly concave. From the posterior view the dorsal margin is weakly dorsally arched. The upper half of the occiput is somewhat narrower than the lower half and it appears to be somewhat constricted in the middle. One can discern a weak occipital pillar on the right posterior surface of the occiput although few other details of the occiput are preserved.

The posterior nares of both specimens of Epimanteoceras formosus are rimmed by a U-shaped emargination (figs. 64b, 65a). The posterior narial canal is elongate, but it does not extend into the basisphenoid. The anterior margin of the posterior nares is positioned between the M3 protocones in AMNH 21613, although the exact edge is obscured by plaster that fills the posterior narial canal. In AMNH 21607 the anterior edge of the posterior nares is slightly anterior to the M3 protocones. Many details of the holotype specimen are obscured by plaster filling the posterior narial canal. However, in AMNH 21607 many of the thinner more fragile elements of the posterior narial canal can be seen (fig. 64b). A thin, raised, horizontal plate of bone extends posteriorly from the palatal margin of the posterior nares and covers the choanae to a point behind M3. A large crack can be seen running through this bony palatal extension and the posterior portion is crushed. Behind the palatal extension, the posterior narial canal is filled with sediment. The thin elongate vomer can be seen bisecting the posterior narial canal. Two thin and broken choanal pouches of bone suspended in the remaining sediment are exposed directly behind the secondary bony palate. These bony pouches most likely represent posteriorly shifted maxilloturbinates, as seen, for instance, in Dolichorhinus. Behind these choanal pouches, the sediment filling the posterior half of the right posterior naris is free of any bone fragments. The left posterior naris has been filled with white plaster and is clearly visible. This plaster filled gap fills the posteriorly shifted functional posterior nares. Thus, the functional posterior nares are posteriorly shifted in Epimanteoceras formosus, a condition closely resembling that of Dolichorhinus.

Figure 65

The holotype skull and upper dentition of Epimanteoceras formosus (AMNH 21613). (A) Ventral view of skull, (B) left premolars, (C) lingual view of incisors and canines, (D) labial view of left incisors.

The configuration of the basicranial foramina is typical; the foramina of the alar canal, the foramen ovale, and the foramen lacerum are widely separate. The external auditory pseudomeatus is wide, ventrally unconstricted, and enters the skull in a mediolateral direction.

Upper Dentition

The holotype (AMNH 21613) retains a complete set of upper dentition (fig. 65) indicating an unreduced dental formula (3-1-4-3) while the referred specimen, AMNH 21607, retains a set of less worn cheek teeth (fig. 64b, c, d). The incisors of AMNH 21613 are similar in size to those of Protitan. There are small diastemata between each incisor. The incisor apices are worn, but these teeth are essentially subcaniniform, with short, pointed, and slightly lingually curved crowns, and with distinct lingual cingula. The incisors become progressively larger from I1 to I3. The incisor row forms a broad arch anterior to the canines. The canines of AMNH 21613 are heavily worn and broken though they appear to have been quite large.

The premolars of the two skulls differ in several ways and there are notable bilateral asymmetries in right and left dentitions, suggesting that the premolars may have been morphologically unstable in Epimanteoceras formosus. The P1 of AMNH 21613 is heavily worn, although it appears that a paracone and a smaller metacone were present. The P1 of AMNH 21607 is incomplete, but it is less worn. That specimen suggests a more complex P1 with a lingual heel and possibly a small protocone. The P2 of both specimens is more oblique than P3 or P4. The parastyle of AMNH 21613 is straight, although the less worn P2 of AMNH 21607 suggests that the parastyle arches slightly lingually. The parastyle of P3 on AMNH 21613 is broken, but it is straight on AMNH 21607. The p4 parastyles of both specimens are angled slightly labially. The metastyles of P2 and P3 are straight while that of P4 is angled slightly labially. Distinct labial paracone ribs can be seen in AMNH 21607; these become smaller in more posterior premolars. Finally, the metacone of the P2 in both specimens is slightly lingually shifted.

The lingual features of the premolars (P2–P4) are higher in relief than those of Protitan and the lingual heels are relatively broader. Very faint preprotocristae can be seen on the P2–P4 of AMNH 21613, but only on the P2 of AMNH 21607. The P2s of AMNH 21613 each retain a prominent lingual crest extending posteriorly from the protocone, although there are no P2 hypocones. In contrast, hypocones on the P2s of AMNH 21607 are present and positioned very closely to the protocone. On AMNH 21613, the hypocones of the P3s are bilaterally asymmetrical; it is absent on the right P3 but on the left P3 there is a hypocone positioned close to the protocone and connected to it by a short lingual crest. The P3s of AMNH 21607 have distinct hypocones that are poorly separated from the protocones. The P4s of AMNH 21613 have distinct hypocones that are connected to the protocones by short lingual crests. Hypocones are absent on the P4s of AMNH 21607; nonetheless, the P4 crowns retain prominent lingual crests extending posteriorly from the protocones. The labial premolar cingula of both specimens are weak. On AMNH 21613 the lingual cingulum of P2 is continuous, while cingula of P3 and P4 are slightly discontinuous. On AMNH 21607 the lingual cingula of the P2–P3 are thicker and more continuous, but this final difference may relate to the fact that the premolars of this specimen are less worn.

The molars of Epimanteoceras formosus exhibit numerous brontotheriine apomorphies, including tall, lingually angled ectolophs, weak labial ribs, thin lingual ectoloph enamel, and wedge-shaped lingual sides of the paracone and metacone in those molars that are not heavily worn. The anterior cingulum is thin and passes proximally to the distal peak of the parastyle. Shallow central molar fossae are present, but anterolingual cingular cusps are absent. E. formosus molars lack vestigial paraconules, and all evidence of metalophs in M1 and M2 is lost. The M3 of AMNH 21613 has a small metalophlike ridge, while AMNH 21607 has a very tiny hypocone. Labial molar cingula in AMNH 21613 are thin and beaded, while lingual molar cingula are absent.

Remarks

Epimanteoceras formosus was first described from a single specimen (AMNH 21613), an essentially complete and remarkably undistorted skull with heavily worn teeth. Granger and Gregory (1943) remarked that E. formosus was “distinctly advanced beyond the stage of Manteoceras ( =  Telmatherium) and in the direction of Rhinotitan” (Granger and Gregory, 1943: 358). Seemingly, this remark stems from the fact that the holotype cranium of E. formosus is intermediate in size and morphology between Telmatherium validus and those of brontotheres with conspicuous frontonasal protuberances such as Rhinotitan. Although other brontothere species with the characteristic overlapping triangular frontal process, such as Telmatherium validus, have been misleadingly described as having rudimentary horns (e.g., Osborn, 1929a), E. formosus is the only brontothere known to have weakly developed frontonasal protuberances. Other brontotheres either lack a distinct frontonasal protuberance or have a more conspicuous one.

Granger and Gregory (1943) noted that another specimen, AMNH 21607, is remarkably similar to Epimanteoceras formosus, but they erected a new taxon, Dolichorhinoides angustidens, for it. It is curious that Granger and Gregory (1943) chose the name Dolichorhinoides because they state “this skull differs conspicuously from those of the American Dolichorhinus” (Granger and Gregory, 1943: 363). Mader (1998) considered both Dolichorhinoides and Dolichorhinus to be junior synonyms of Sphenocoelus. It was demonstrated earlier in this paper that Dolichorhinus is actually very distinct from Sphenocoelus (see remarks under Dolichorhinus hyognathus), and it is argued below that the holotype of Dolichorhinoides angustidens is likewise different from both Sphenocoelus and Dolichorhinus.

The bony palatal extension, the extremely posteriorly situated maxilloturbinates, and posteriorly shifted functional posterior nares seen in AMNH 21607 resemble Dolichorhinus hyognathus. However, because these elements are extremely fragile, they cannot be investigated in the majority of brontothere specimens, and hence these characters are unknown for many species. Additionally, AMNH 21607 is inconsistent with the following aspects of Mader's (1998) diagnosis of Sphenocoelus (which, in his concept, included Dolichorhinus). (1) Sphenocoelus (sensu Mader, 1998) lacks hypocones on the premolars (unmolarized premolars, sensu Mader, 1998) whereas the P2–P3 of AMNH 21607 have hypocones. (2) Small, vestigial paraconules or metaconules are variably present on molars of Sphenocoelus (sensu Mader, 1998), but they are absent on AMNH 21607. (3) The skulls of Sphenocoelus (sensu Mader, 1998) are highly elongate; however, the apparently elongate skull of AMNH 21607 is largely a product of moderate lateral crushing. (4) Sphenocoelus (sensu Mader, 1998) possesses a large infraorbital process, but AMNH 21607 lacks this character. Additionally, the triangular process of the frontal bone overlapping the nasal bone that is clearly discernable on AMNH 21607 is not seen in Sphenocoelus uintensis or Dolichorhinus hyognathus. Finally, the postorbital portions of the crania of Dolichorhinus and Sphenocoelus are arched dorsally while that of AMNH 21607 is not.

After further examination of the holotypes of Epimanteoceras formosus (AMNH 21613) and Dolichorhinoides angustidens (AMNH 21607), these specimens do not appear to vary in any way that is seemingly taxonomically significant, and these specimens almost certainly represent the same species. Variability in the size and distinctness of the frontonasal protuberance is found among nearly all species of horned brontotheres. Likewise, extensive variation in the lingual morphology of the premolars is the rule rather than an exception within brontothere species.

Granger and Gregory (1943) concluded that the only clear distinction between Epimanteoceras formosus (AMNH 21613) and Dolichorhinoides angustidens (AMNH 21607) was the more elongate molars of the later. However, the difference can be attributed to dental wear in AMNH 21613, and, less significantly, to taphonomic deformation of AMNH 21607. Wood (1938) documented extreme ontogenetic shortening of the molars of rhinoceroses. A similar phenomenon occurs among brontotheres. Brontothere upper molars are longer on the labial side of the tooth. In AMNH 21607, where the molars are less worn, they are longer on the labial side than on the lingual side. Consequently, adjacent molars contact each other on the labial side but do not contact each other on the lingual side (fig. 64c). In comparison, the dentition of AMNH 21613 is more extensively worn (fig. 65a). In the later specimen, the labial and lingual sides of M1 and M2 are nearly the same length and the adjacent molars contact each other both labially and lingually. However, the anterior and posterior enamel of the M1 and the anterior enamel of M2 are completely worn off. Clearly, the ectolophs of these teeth have been significantly shortened by interstitial wear, yet the sides of the molars are still in close contact. Apparently, the maxilla remodeled as the teeth shortened so that molars maintained contact despite having become significantly shorter on the labial side. Wear patterns in other brontothere molars exhibit the same pattern of ontogenetic tooth-row shortening. It is therefore evident that maximum lengths of the molars and even tooth-row lengths are influenced by dental wear.

In addition to differences in molar proportions due to interstitial wear, the molar proportions of AMNH 21607 have been taphonomically distorted. That specimen is laterally compressed and although the teeth appear to be minimally damaged, closer inspection reveals that the molars have been labiolingually compressed, thus exaggerating their relative length. A structural weakness exists in brontothere molars between the ectoloph and the lingual cusps. Those features contributing to the weak point include (1) increased ectoloph height without associated increase in height of the lingual side of the tooth, and (2) the appearance of the central molar fossa at the base of the ectoloph. Brontothere molars with these features are often cracked at this weak point and are susceptible to significant distortion from lateral crushing but without obvious damage to the tooth. Essentially, the tall ectoloph can be forced up over the lower lingual side of the tooth. This sort of damage is most evident in the right M3 of AMNH 21607 (fig. 64b) where the tooth has been considerably narrowed because the ectoloph has been forced over the lingual side of the tooth. The distortion is subtler on the left M3 where the shape of the tooth along the anterior margin is intact, but the posterior half of the ectoloph has been lingually displaced (fig. 64c). The damage to the M1 and M2 is less obvious, but tiny thrust faults in the enamel suggest a small amount of lingual displacement of the ectolophs.

A reliable comparison of the relative dimensions of the molars of AMNH 21613 and AMNH 21607 is clearly hampered by the extensive wear in the former and subtle distortion in the latter. However, measurements of a limited number of seemingly intact dimensions are possible. For instance, the left M3 of AMNH 21613 is less extensively worn. It is thus possible to compare M3 anteroposterior length in AMNH 21613 and 21607 if measured at the proximal base of the crown. At this location, the lengths are similar (AMNH 21613  =  70 mm; AMNH 21607  =  68 mm). Likewise, the left M3 of AMNH 21607 seems to be nearly intact along its anterior edge, thus allowing for a single comparison of the undistorted labiolingual widths of the M3s (AMNH 21613  =  58 mm; AMNH 21607  =  57 mm). Ratios calculated from the length of M3 divided by the width of the anterior margin yield nearly identical values (1.2 for AMNH 21613 and 1.19 for AMNH 21607). Therefore, it is relatively safe to conclude that whatever apparent differences exist in the dental dimensions of AMNH 21613 and 21607 that were observed by Granger and Gregory (1943) they can largely be attributed to wear and distortion. With little else to distinguish Dolichorhinoides angustidens as a valid taxon, it is here considered a junior synonym of Epimanteoceras formosus.

The only valid species of Epimanteoceras is E. formosus. Other species assigned to Epimanteoceras, including E. praecursor Yanovskaya (1953) from the upper Eocene of Kazakstan, and E. amplus Yanovskaya (1976) from the late Eocene Ergilin Dzo of Mongolia, are dubious species and are further discussed in the section dealing with nomina dubia and other problematic taxa. Another dubious taxon, “Protitan?” cingulatus, known only from a set of mandibles, is possibly a junior synonym of E. formosus.

Protitan grangeri (Osborn, 1925)

Holotype

AMNH 20103, a complete skull and mandible.

Type Locality

Irdin Manha Formation, one half mile south of the Kalgan Urga telegraph line, Inner Mongolia, China.

Age

Middle Eocene (Irdinmanhan land mammal “age”).

Synonyms

Dolichorhinus olseni Osborn, 1925; “Manteoceras?” irdinensis Osborn, 1925; Protitan robustus Granger and Gregory, 1943; Protitan bellus Granger and Gregory, 1943; Protitan obliquidens Granger and Gregory, 1943.

Referred Specimens

(From the Irdin Manha Formation, Inner Mongolia) AMNH 19179, a mandible fragment with left p2 (partial), p3–m2, and m3 (partial); AMNH 20104 (holotype of Protitan robustus), a partial mandible with right i1–c, p2–p3, and left i1–m3; AMNH 20108, a right maxilla with P1–M3; AMNH 20109 (holotype of Dolichorhinus olseni), a mandible with right i3, p1–m3, and left c–m3; AMNH 20111 (holotype of “Manteoceras?” irdinensis), a mandible fragment with right m1–m3; AMNH 20112, a partial mandible with i2–c and p2–m3; AMNH 20113, an anterior portion of a cranium with right C, P2–P4, and M1 (partial); AMNH 20114, an anterior portion of a cranium with right I2–M1, left C, and P2–M3; AMNH 20119, a mandible fragment with right p4–m2; AMNH 20120, a right maxilla fragment with M1–M3; AMNH 20123, a left maxilla fragment with P4–M2; AMNH 20125 (holotype of Protitan obliquidens), a left maxilla fragment with P1–P3; AMNH 20126, a mandible fragment with left p3–m2; (from the Ulan Shireh Formation of Inner Mongolia) AMNH 26104 (holotype of Protitan bellus), a ventral surface of a cranium with right and left I2–M3; (from the ?“Houldjin” beds at Camp Margetts) AMNH 26421, a mandible with some intact incisors, right p2–m3, and left p1–m3.

Diagnosis

Protitan grangeri is a large brontothere with small but distinct elliptical frontonasal horns positioned low on the skull and far in front of the orbit. The nasal incision is dorsoventrally shallow and extends posteriorly to the P3. The nasal process is slightly angled downward, unelevated, long and broad, with thickened sides, and with a thin and strongly rounded distal edge with a downturned distal tip. The orbits are positioned above the posterior portion of M2 and the anterior portion of M3. The premaxillomaxillary rostrum deepens posteriorly and it is not enclosed by bone dorsally. Other cranial characteristics include a saddle-shaped cranium, separate parasagittal ridges that strongly constrict the dorsal width of the cranium posteriorly, postzygomatic processes, large ventral sphenoidal fossae, nearly straight zygomatic arches, and a ventrally unconstricted and mediolaterally angled external auditory pseudomeatus.

Dentally, Protitan grangeri has three large to intermediate-sized subcaniniform upper incisors, a simple P1, and a distinct P2 metacone. Premolar hypocones are absent, although a short crest usually extends from the protocones. The molars of P. grangeri have tall, lingually angled ectolophs with weak labial ribs, and thinned lingual ectoloph enamel with wedge-shaped paracones and metacones. A cingular parastyle shelf is absent. Very shallow central molar fossae are present. Anterolingual cingular cusps are absent. Paraconules and metalophs are absent. The lower dentition of P. grangeri includes three large incisors, a semispatulate i1 and i2, and a more subcaniniform i3. Additional lower dental characters include, a distinct postcanine diastema, a metaconid on p4 but not on p2 or p3, shallow molar basins, and a slender m3.

Protitan grangeri closely resembles Protitanotherium emarginatum and Protitan minor. It can be most easily distinguished from Protitanotherium by the larger more subcaniniform incisors, a ventrally unconstricted external auditory pseudomeatus, and postzygomatic processes. Protitan grangeri can be distinguished from Protitan minor by its anteroposteriorly shorter nasal incision, thicker nasal, more anteriorly positioned horns, and mediolaterally angled auditory pseudomeatus.

Description

Skull

The holotype of Protitan grangeri (AMNH 20103) is a complete and relatively undistorted skull (figs. 66, 67) and jaw (fig. 68) with a complete set of dentition, but with rather poorly preserved cheek teeth. No other complete skulls of P. grangeri are known, although there are several additional partial crania, including AMNH 26104, a complete ventral surface, and several anterior portions of skulls including AMNH 20108, AMNH 20114, and AMNH 20113. The following description of P. grangeri is based primarily on AMNH 20103, but additional information is taken from these other specimens where noted.

Figure 66

The holotype skull of Protitan grangeri (AMNH 20103). (A) Right view, (B) dorsal view, (C) oblique view, (D) anterior view, (E) posterior view.

Figure 67

The ventral surface of the skull of Protitan grangeri and upper dentition. (A) Ventral view of AMNH 20103 (holotype), (B) left molars of AMNH 26104, (C) left premolars of AMNH 26104, (D) lingual view of right incisors and canine of AMNH 20103 (holotype), (E) labial view of the right incisors and canine of AMNH 20103 (holotype).

Figure 68

The holotype mandible of Protitan grangeri (AMNH 20103). (A) Right view, (B) dorsal view.

Protitan grangeri is a large (table 9) brontothere with an elongate cranium and small but conspicuous frontonasal protuberances that are similar in size to those of Protitan minor and Protitanotherium emarginatum. From the right side of the skull the frontal bone can clearly be seen overlapping the nasal bone and extending to the peak of the horn. The horns of AMNH 20103 are short and elliptical in shape with a longer anteroposterior axis. The peaks of the horns lack distinct rugosities, and the surfaces of the horns are no rougher than the remaining surface of the skull. The horns project dorsolaterally from the sides of the skull and they are widely separated by a broad and flat forehead. The horns are positioned low on the skull although they are much further anterior to the orbits in comparison to those of Protitan minor and Protitanotherium emarginatum.

Table 9

Summary statistics for selected morphometric variables of Protitan grangeri See Methods for measurement definitions

Like Protitan minor and Protitanotherium emarginatum, the nasal incision is dorsoventrally shallow. However, it is anteroposteriorly shorter than in other taxa. The nasal incision extends posteriorly to the P3 and the orbit is positioned over the anterior portion of M3 and the posterior portion of M2. The anterolateral root of M2 rests below the anterior rim of the orbit.

From the lateral view of the skull the nasal process of the holotype curves downward anteriorly and is the same length as the premaxillomaxillary rostrum. The sides of the nasal process are thickened, rounded, and angled downward. The sides are not upturned as are the nasals of Diplacodon elatus. Thin, vertical lateral walls, such as those seen in Epimanteoceras formosus and other hornless brontotheres, are not seen in Protitan grangeri. From a dorsal view of the skull it can be seen that the nasal bones are strongly fused together. The nasal process is wide, about the same width as the premaxillomaxillary rostrum. Throughout its length the nasal process is nearly constant in width. The distal edge of the nasal process is strongly rounded although there is a slight anterior notch at the midline. From the lateral and anterior views of the skull it can be seen that the anterior end of the nasal process is thinned and strongly curved downward. The anterior edge of the nasal process is thin and irregular.

From the lateral view of the holotype skull, the dorsolateral margin of the premaxillomaxillary rostrum rises posteriorly at a shallow angle so that the posterior notch of the nasal incision is at the level of the upper rim of the orbit. The premaxillae contact each other medially at the premaxillomaxillary symphysis although the symphysis is not completely fused. The premaxillomaxillary suture can be seen in AMNH 20103 most clearly from the oblique view of the skull. The premaxilla is truncated before reaching the posterior notch of the nasal incision, thus, the premaxilla does not contact the nasal bone. The dorsolateral margins of the premaxillae diverge posterolaterally and the rostral cavity is dorsally open.

From the lateral view the dorsal surface of the skull is fully concave from the orbits to the nuchal crest, forming a completely saddle-shaped cranium. The parasagittal ridges remain separate and do not form a sagittal crest although they strongly constrict the dorsal surface of the cranium posteriorly. The zygomatic arches are rather shallow dorsoventrally and have a narrow rectangular cross section. The zygomatic arches of AMNH 26104 are somewhat thicker and more robust than those of AMNH 20103, but they are otherwise similar. The zygomatic arch is essentially uncurved from a lateral view. From a dorsal view the zygomatic arches are bowed slightly inward. However, the zygomatic arches of AMNH 20103 are composed of several fragments that have been bonded together with plaster and it is probable that the inwardly bowed shape is an artifact of distortion and subsequent reconstruction of the fossil. Unlike the holotype, the zygomatic arches of AMNH 26104 are not inwardly bowed. Discounting the slight inward bowing of the zygomatics of the holotype, the zygomatic arches appear to have been straight and strongly angled posterolaterally. The posterodorsal end of each of the zygomatic arches exhibits a small dorsally projecting process that was referred to by Granger and Gregory (1943) as a “postzygomatic horn” (Granger and Gregory, 1943: 359). One of these postzygomatic processes can most easily be seen on the right side of the skull (fig. 66).

From the lateral view of the skull the occiput is moderately angled backward. From the dorsal view the nuchal crest is nearly flat. From the posterior view the dorsal rim of the occiput is dorsally arched. The dorsal portion of the occiput is slightly narrower than the ventral portion of the occiput and the occiput is strongly constricted in the middle. Weak occipital pillars are evident on the posterior surface of the occiput, although the central depression of the occiput is rather shallow.

From the ventral view of AMNH 20103 (fig. 67a) the anterior rim of the posterior nares is situated slightly behind the M3 protocones. The narial canal is rimmed by a rather wide U-shaped emargination. In both the holotype specimen and the other complete ventral skull surface (AMNH 26104), the elongate posterior canal is filled with plaster. In AMNH 20103 roughly the posterior half of the elongate posterior narial canal is free of plaster. In that specimen the canal extends significantly into the basisphenoid. The ventral sphenoidal fossae are rather large, but they are shallow. These were called presphenoid pits by Osborn (1925, 1929a) and were called basisphenoid pits by Granger and Gregory (1943). A remnant of the thin bony septum formed by the main body of the sphenoid that partitions the ventral sphenoidal fossae is still preserved in position.

Other aspects of the basicranium of AMNH 20103 and AMNH 26104 are rather typical, such as the widely separated foramen ovale and foramen lacerum. The occiput and basicranium are not disproportionately widened as those of Rhinotitan andrewsi or Metatitan. The external auditory pseudomeatus is directed mediolaterally. From the lateral view the external auditory pseudomeatus is wide and open ventrally.

Upper Dentition

The holotype of Protitan grangeri (AMNH 26103) retains a complete dentition (fig. 67a). Although the incisors of the holotype are in good condition (fig. 67d–e), the cheek teeth are poorly preserved. However, AMNH 26104 has a nearly complete set of upper dentition whose cheek teeth are significantly less worn and less damaged than those of AMNH 26103. Therefore, the cheek teeth of AMNH 26104 are shown in close-up (fig. 67b, c).

Protitan grangeri retains an unreduced dental formula (3-1-4-3). The incisors and canines of the holotype are relatively small, though by no means are they as reduced as those of Protitanotherium emarginatum. The anterior dentitions of other specimens referred to Protitan grangeri are larger. For instance, the incisors and canines of AMNH 26104 are distinctly larger than those of AMNH 20103. However, the premaxillae of both specimens are of similar size; in AMNH 20103 there are small diastemata between the incisors, whereas in AMNH 26104 there are no diastemata between the incisors (except for a median diastema). Osborn (1929a) attributed the smaller size of the incisors of AMNH 20103 to sexual dimorphism. Variation in incisor and canine size seen here and in other brontotheres such as Aktautitan hippopotamopus and Gnathotitan berkeyi probably represents sexual dimorphism.

The incisors of Protitan grangeri form an arched row anterior to the canines. They increase in size laterally and retain plesiomorphic subcaniniform morphology with short but conical and lingually curved crowns with distinct lingual cingula. Labial incisor cingula are absent. The canine of AMNH 20103 is small while that of AMNH 26104 is larger, even though it is otherwise morphologically similar. There is both a short precanine diastema and a longer postcanine diastema. The postcanine diastema of AMNH 26104 seems proportionally shorter but this difference can largely be attributed to the larger canine.

The P1 is a single-cusped tooth with an elongate posterior heel. The anterior margin of P2 is angled posterolingually, giving the tooth a somewhat oblique shape. The anterior and posterior margins of P3 and P4 are more nearly parallel. The parastyle and metastyle of P2 arch somewhat lingually. The parastyle and metastyle of P3 are nearly straight. The parastyle of P4 is slightly angled lingually while the metastyle is more nearly straight. The labial margin of the P2 protocone is very convex, while the paracones of P3 and P4 have distinct labial ribs. Finally, the metacone of P2 is lingually shifted, while those of P3 and P4 are more labially positioned. As a consequence, the ectoloph of P2 is rounder than those of P3 and P4.

The lingual features of the premolars have a relatively low level of topographic relief and they exhibit considerable intraspecific variation. On the P2 of the holotype specimen (AMNH 20103) there is a distinct protocone followed by a short lingual crest. However, in AMNH 26104 a single loph of enamel stretches around the anterolingual side of the P2 crown and although there are no distinct lingual cusps. The P2s of AMNH 20103 and AMNH 26104 have small preprotocristae. A less distinct preprotocrista is seen on the P3 of the holotype and it is absent on the P3 of AMNH 26104. The P3 protocones are followed by short lingual crests on both specimens. A preprotocrista is not seen on the P4 of either specimen. A lingual crest is not seen on the P4 of the holotype, but a short lingual crest is present on the P4 of AMNH 26104. A few specimens have small paraconules on the P2–P4 but these are most often absent. Premolar hypocones are nearly always absent in P. grangeri, though one specimen, AMNH 20123, has a very small hypocone positioned close to the protocone of the P4. Labial premolar cingula are weak. The anterior and posterior premolar cingula wrap around the lingual sides of the crowns and join to form continuous lingual cingula. Among all of the available specimens, the lingual premolar cingula range from continuous around the protocone to slightly discontinuous.

The upper molars of Protitan grangeri show numerous brontotheriine apomorphies, including tall, lingually angled ectolophs, very weak labial ribs, thin lingual ectoloph enamel, and wedge-shaped lingual sides of the paracone and metacone in molars that are not heavily worn. The anterior cingulum is thin and passes proximally to the distal peak of the parastyle. Shallow central molar fossae are present but anterolingual cingular cusps are absent. All evidence of paraconules and metalophs is lost. There is no trace of a hypocone on any M3 of P. grangeri. Labial molar cingula are weak and lingual molar cingula are absent.

Mandible and Lower Dentition

The holotype mandible (AMNH 20103) is missing only the condyles and coronoid processes (fig. 68). The lower dentition of the holotype is complete, but cheek teeth are poorly preserved. However, the lower dentition of a referred specimen (AMNH 20104) is complete, similar to that of the holotype, and in more pristine condition. Therefore, the lower teeth of this specimen are figured in close-up (fig. 69).

Figure 69

A mandible referred to Protitan grangeri (AMNH 20104). (A) Left view, (B) left premolars, (C) dorsal view, (D) lingual view of incisors and canines, (E) labial view of incisors and canines.

The coronoid process of AMNH 20104 is tall, slender, and moderately curved. The mandibular symphysis is long and narrow in AMNH 20103, although it is somewhat more robust in AMNH 20104. The angle of the ventral margin of the symphysis of AMNH 20103 is shallow (< 45°). However, as noted by Granger and Gregory (1943), the shallow orientation of the symphysis seems to have been influenced by taphonomic distortion. A segment of the right ramus below the p3 is reconstructed with plaster, suggesting that the anterior portion of the mandible has been forced or rotated downward. The anterior portion of AMNH 20104 is more intact and suggests a steeper mandibular symphysis. The exact posterior edge of the symphysis is obscured by plaster in AMNH 20103, but in AMNH 20104, the symphysis extends posteriorly to the talonid of the p3.

The lower incisors are relatively large and they form an arched row anterior to the canines. In the holotype the crowns of i1 and i2 are significantly worn, but those of AMNH 20104 are nearly unworn. The crowns of i1 and i2 are generally semispatulate with rounded apices, while i3 has a more sharply defined apex. There are distinct lingual cingulids on i1–i3. Labial cingulids are not seen. The i3 is mesiodistally elongate in comparison to i1 or i2. Each incisor is roughly of similar size. The lower canines of AMNH 20103 are notably smaller than those of AMNH 20104. The length of the lower postcanine diastema, the only diastema of the lower dentition, covaries with canine size. In AMNH 20103 the postcanine diastema is about double the length of the p2 whereas in AMNH 20104 the postcanine diastema is shorter, corresponding to the larger size of the canines.

The lower premolars are low crowned and relatively slender. The p1 is a very small tooth with a single cusp and a short talonid heel. The trigonid of p2 is much longer than the talonid. The p3 trigonid is somewhat longer than the talonid, while the p4 trigonid is similar in length to the talonid. The trigonid and talonid of p2 are of similar width, while the p3 and p4 trigonids are slightly narrower than their talonids. The p2 paralophid barely arches lingually and there is only a minor lingual trigonid notch. The p2 protolophid extends in a posterior direction, but it is positioned lingually. The p3 paralophid arches slightly lingually, creating a more distinct lingual trigonid notch. The p3 protolophid is moderately angled lingually. Metaconids are absent on p2 and p3. The trigonid of p4 is more molariform with a strongly lingually arching paralophid and protolophid and a large lingually positioned metaconid. The talonid of p2 is very small, with a very short cristid obliqua and hypolophid. The lingual face of the p2 talonid is a flat and sloped surface. The cristids obliqua and hypolophids of p3 and p4 are progressively longer although the lingual face of p3 forms only a broad concave surface. However, the talonid of p4 has a nearly molariform talonid basin. Labial and lingual premolar cingulids are essentially absent although a very faint labial cingulid can be seen on p4.

The lower molars of Protitan grangeri are typical with relatively thin lingual enamel, shallow trigonid and talonid basins, and an elongate m3. There are no lingual cingulids. Labial molar cingulids are distinct although they tend to be discontinuous around the paraconids and metaconids.

Remarks

Protitan grangeri (Osborn, 1925) is based on a complete skull and jaw (AMNH 20103). Osborn (1925) originally referred this species to the genus Protitanotherium, based on the “elongate horns”, the “broad shovel-shaped nasals”, the “saddle-shaped cranial top”, and several other minor details. Several other species have been referred to Protitan, although most of these turn out to be invalid synonyms and nomina dubia. Osborn named two other Asian brontothere species, Dolichorhinus olseni Osborn (1925) and “Manteoceras?” irdinensis Osborn (1925). In their revision of Mongolian brontotheres, Granger and Gregory (1943) erected a new genus, Protitan, with P. grangeri serving as the type species. In addition to Protitan grangeri, Granger and Gregory (1943) named five other species of Protitan, including P. minor, P. robustus, P. bellus, P. obliquidens, and “Protitan?” cingulatus. They considered Dolichorhinus olseni Osborn (1925) and “Manteoceras?” irdinensis Osborn (1925) to be synonyms of Protitan grangeri.

Among these species, Protitan minor is the only other species previously referred to Protitan that may actually belong to this genus, although even its membership is questionable (see remarks under Protitan minor). “Protitan?” cingulatus is a dubious species that may actually be synonymous with Epimanteoceras formosus. The remaining species, Dolichorhinus olseni, “Manteoceras?” irdinensis, P. robustus, P. bellus, and P. obliquidens are junior synonyms of P. grangeri.

Dolichorhinus olseni was based on a complete mandible (AMNH 20109) lacking most of its anterior dentition. It is not clear why Osborn (1925, 1929a) assigned this species to Dolichorhinus. Osborn did not attempt to differentiate Dolichorhinus olseni from Protitan grangeri, despite that fact that these taxa are based on very similar fossil material. Granger and Gregory (1943) considered D. olseni to be a junior synonym of P. grangeri, but they noted that the m3 was slightly shorter than on the holotype of P. grangeri (AMNH 20103).

“Manteoceras?” irdinensis is based on a partial ramus with m1–m3 that is reconstructed in such a way that suggests a very short space for a premolar row and a very short and flattened symphysis. All these features are relevant to Osborn's (1925) diagnosis, which includes an apparently reduced number of premolars (judging by the premolar alveoli), unusually procumbent incisors (judged by the angle of the symphysis), and a short symphysis. The short symphysis is artificial and related to the fact that the anterior end is missing. Additionally, the ramus has been incorrectly reconstructed; the symphysis is plastered directly to the horizontal ramus, although a segment of the horizontal ramus appears to be missing, thus giving the artificial appearance of a shortened jaw. Likewise, the shallow angle of the symphysis and, hence, the procumbent nature of the incisor alveoli are simply a result of how the specimen has been reconstructed. The molar measurements of this specimen are similar to Protitan grangeri. Thus, “Manteoceras?” irdinensis is considered a junior synonym of P. grangeri.

Granger and Gregory (1943) based Protitan robustus on a partial mandible with complete and lightly worn dentition (AMNH 20104). This species was diagnosed on the following criteria: (1) size large (presumably in comparison to P. grangeri), (2) incisors large, wide spreading, (3) canines very massive with recurved crowns, and (4) “postcanine diastema short”. However, none of these observations clearly differentiates P. robustus from Protitan grangeri. AMNH 20104 is larger than the holotype of P. grangeri (AMNH 20103), but the difference is not extreme and this specimen easily fits into an acceptable size range for P. grangeri (table 9). Granger and Gregory (1943) actually noted that the dentition of AMNH 20104 nearly fit onto AMNH 26104 (holotype of P. bellus, a synonym of P. grangeri; see next paragraph), but they even rejected the possibility that these two specimens are the same species because the match was not perfect! (Intraspecific variation and taphonomic distortion were largely ignored by Granger and Gregory [1943].) In comparing AMNH 20104 with the holotype mandible (AMNH 20103) of P. grangeri, the larger size of the incisors and canines of the former is consistent with sexual dimorphism. Likewise, the shorter postcanine diastema can be explained by the fact that the canine root of AMNH 20104 is much larger and takes up more space. Nothing else appears to differentiate this specimen from P. grangeri.

Protitan bellus was based on a ventral surface of a skull with a nearly complete set of upper dentition (AMNH 26104). Granger and Gregory (1943) distinguished P. bellus from P. grangeri with the following diagnosis: (1) upper molars anteroposteriorly longer than in P. grangeri, especially M1. (2) P3 with incipient tetartocone ( =  hypocone) swelling, (3) P4 more elongate, (4) incisors much larger than in the type of P. grangeri, and (5) width across opposite M3 distinctly larger. Observations 1, 3, and 5 were based on ratios in which AMNH 26104 differs slightly from AMNH 20103 (See table 5 of Granger and Gregory, 1943). However, either these differences can be attributed to the more damaged and more heavily worn state of the dentition of AMNH 20103 compared with that of AMNH 26104 (observations 1 and 3), or the differences are so small (4%) that they do not warrant a taxonomic distinction (observation 5). While I could not confirm the presence of an “incipient hypocone” on the P3 of AMNH 26104 (observation 2), there are small lingual crests on the P3 and P4. These are not present in the holotype of Protitan grangeri; however, this character is strongly variable in most brontothere species. Moreover, other specimens that Granger and Gregory (1943) referred to P. grangeri show similar variations in premolar morphology. Therefore, AMNH 26104 does not really stand out in this respect. Clear differences between AMNH 20103 and AMNH 26104 include the large incisors and canines (observation 4 of Granger and Gregory [1943]), larger size, and thicker zygomatic processes. These character differences are most consistent with sexual dimorphism. AMNH 26104 is marginally larger than the holotype of P. grangeri.

The obliqueness of the premolars (P2–P3) of a maxillary fragment (AMNH 20125) led Granger and Gregory (1943) to assign it to yet another a new species, Protitan obliquidens. Granger and Gregory (1943) remark in their diagnosis: crown pattern of P2, P3 very oblique both on the anterointernal and posteroexternal. However, this specimen does not really stand out in comparison to other specimens of Protitan. In particular, the P2 of all Protitan specimens is oblique in appearance, although the extent of obliqueness is somewhat variable. As Granger and Gregory (1943) note, the shape of the premolars closely resembles those of AMNH 26104 except in the much greater maximum oblique width of P2. However, the difference in the shape of the P2 of AMNH 20125 is subtle and the P2 and P3 are cracked in numerous places with bits of matrix and plaster between the spaces; the teeth of this specimen are clearly distorted to a minor degree. The minor differences in the apparent degrees of obliqueness of these teeth can largely be attributed to this distortion. Realistically, the molar dimensions of AMNH 20109 are similar to those of Protitan grangeri and this specimen is likely to represent the same species.

Subsequent to Granger and Gregory's (1943) publication, paleontologists working with Asian brontothere material have shown a mistaken tendency to assign large middle Eocene brontothere fossils to the genus Protitan, although their use of Protitan has been far too broad; none of the material outside of the original AMNH central Asiatic expedition collection resembles the species referred to Protitan by Granger and Gregory (1943). Two species named by Yanovskaya (1980), P. khaitshinus and P. reshetovi, are actually Metatitan. Yanovskaya (1980) incorrectly reassigned Protitan robustus to the genus Epimanteoceras and wrongly synonomized it with Epimanteoceras amplus (a dubious species that is, nonetheless, very different from Protitan and might be a synonym of Nasamplus progressus). None of the material described in Yanovskaya's (1980) monograph on Mongolian brontotheres appears to belong to any species of Protitan. Qi et al. (1992) referred another specimen (IVPP–V10104, presumed to represent a single individual) from the Tukhum beds of Erden Obo (Urtyn Obo) to Protitan sp. The material includes an elongate m3 and an assortment of postcranial elements. Although the size of the m3 is consistent with P. grangeri, it lacks diagnostic characters and could belong to one of a number of brontotheres including Epimanteoceras, Metatitan, or others. Preliminary reports of a brontothere from the Ily Basin of Kazakstan were referred to Protitan (Emry et al., 1997; Emry and Lucas, 2002, 2003; Lucas and Emry, 2001), although this brontothere turned out to be a new genus and species, Aktautitan hippopotamopus (Mihlbachler et al., 2004a). Most recently, Huang and Zheng (2004) named a new species, Protitan major, from the Lumeiyi Formation of China, although the material assigned to this species represents a brontothere with much more advanced dentition and it possibly belongs to Metatitan. Other material from the Lumeiyi formation assigned to Protitan cf. P. robustus by Zheng et al. (1978) and Huang and Zheng (2004) lack sufficient diagnostic characters for species or genus identification.

Protitan minor Granger and Gregory, 1943

Holotype

AMNH 26416, a skull missing the distal end of the nasal process. (Presently the canines are missing although they were originally recovered and can be seen in older photographs of the specimen).

Type Locality

“Probably top of Irdin Manha beds”, Camp Margetts, Inner Mongolia, China.

Age

Middle Eocene (Irdinmanhan land mammal “age”).

Referred Specimen

(From 1 mile west of Camp Margetts, ?Irdin Manha beds, Inner Mongolia) AMNH 26417, a left anterior quarter of a skull with I3 and P2–M3.

Diagnosis

Protitan minor is an intermediate-sized brontothere with small but distinct elliptical frontonasal horns that are positioned low on the skull and slightly in front of the orbits. The nasal incision extends posteriorly to the M1 and is dorsoventrally shallow. The nasal process is slightly angled downward, unelevated, long, broad, and with thin and shallow sidewalls. The orbits are positioned above the posterior portion of M2 and the anterior portion of M3. The premaxillomaxillary rostrum thickens posteriorly and it is not enclosed by bone dorsally. Other cranial characteristics include a saddle-shaped cranium, separate parasagittal ridges that strongly constrict the dorsal surface of the cranium posteriorly, postzygomatic processes, nearly straight zygomatic arches, and a ventrally unconstricted and posteromedially angled external auditory pseudomeatus.

Dentally, Protitan minor has three large subcaniniform upper incisors, a simple P1, and a distinct P2 metacone. Premolar hypocones are absent but short lingual crests are present in some premolars. The molars have tall lingually angled ectolophs with weak labial ribs, and thinned lingual ectoloph enamel with wedge-shaped paracones and metacones. A cingular parastyle shelf is absent. Very shallow central molar fossae are present. Anterolingual cingular cusps are absent. Paraconules and metalophs are absent.

Protitan minor most closely resembles Protitanotherium emarginatum and Protitan grangeri. It can be most easily distinguished from Protitanotherium by the larger and more subcaniniform incisors, postzygomatic processes, and the ventrally unconstricted external auditory pseudomeatus. Protitan minor can be distinguished from Protitan grangeri by a posteriorly deeper nasal incision, thinner nasal process, more posteriorly positioned horns, and a posteromedially angled auditory pseudomeatus.

Description

Skull

AMNH 26416 is a nearly complete skull with significant amounts of plaster filling in a large portion of the occiput and parts of the frontoparietal (fig. 70). The nasal process is mostly missing (although it is has been reconstructed as if very short), and the skull has suffered a minor amount of lateral sheering distortion. Otherwise the proportions of the skull appear to be relatively undistorted. A second specimen, AMNH 26417, a left anterior quarter of a cranium, has a more complete nasal process (fig. 71).

Figure 70

The holotype skull of “Protitan” minor (AMNH 26416). (A) Left view originally published by Granger and Gregory (1943: pl. 8a; photo from the American Museum of Natural History vertebrate paleontology archive) showing the left canine (now lost), (B) dorsal view, (C) anterior view, (D) posterior view.

Figure 71

Left lateral view of a partial skull (AMNH 26417) of “Protitan” minor.

Protitan minor is most similar to that of Protitan grangeri and Protitanotherium emarginatum although it is smaller than these species. The frontonasal protuberances are similar to those of Protitan grangeri and Protitanotherium emarginatum; they are short and elliptical with a longer anteroposterior axis. The frontonasal protuberances are positioned low on the skull and slightly in front of the orbits. This differs from Protitan grangeri whose horns are positioned far anterior to the orbits. The frontal bone can be seen overlapping the nasal bone and extending to the peak of the frontonasal protuberance.

The nasal incision is dorsoventrally shallow, but it is longer than that of Protitan grangeri and extends to a point above the M1 mesostyle. The anterior rim of the orbit rests over the posterior lateral root of the M1 and the anterolateral root of the M2. The orbit is positioned directly over the posterior half of the M2 and the anterior portion of the M3. In AMNH 26417 the nasal process is nearly horizontal, but it is angled downward slightly and is about as long at the premaxillomaxillary rostrum. The nasal process is much thinner than that of Protitan grangeri. In this respect it is similar to the nasal bones of Telmatherium validus and Epimanteoceras formosus. However, the lateral wall of the nasal process is much shallower than Telmatherium and most other hornless brontotheres.

The premaxillomaxillary rostrum of Protitan minor is typical in most respects. From the lateral view it deepens posteriorly and the dorsolateral margin rises posterodorsally so that the posterior notch of the nasal incision is at the level of the upper rim of the orbit. From the dorsal view of the skull the dorsolateral margins of the rostrum diverge posterolaterally and the rostral cavity (which is presently filled with plaster) is not dorsally sealed by bone. The premaxillae fully contact each other at the symphysis. The premaxillomaxillary suture is visible from the left side of the skull; the premaxilla truncates anterior to the posterior notch of the nasal incision and, therefore, does not contact the nasals.

The postorbital cranium is much longer than the facial region of the skull. The dorsal surface of the cranium of Protitan minor is essentially saddle-shaped. The posterior half of the frontoparietal is clearly concave. The anterior half is more flat, although this portion of the dorsal surface is somewhat warped, and from the dorsal view it can be seen that large portions of skull between the orbits are missing. Otherwise the general shape of the postorbital cranium of AMNH 26416 is similar to Protitan grangeri. The parasagittal ridges converge medially and constrict the dorsal surface of the skull posteriorly, but they do not form a sagittal crest. The zygomatic arches are relatively thin, shallow, and laterally unbowed. From a lateral view, the jugal portion of the zygomatic is horizontal. The squamosal portion of the zygomatic is angled upward posteriorly but at a very shallow angle, and there is essentially no curvature of the zygomatic arch. The posterodorsal end of each of the zygomatic arches exhibits a dorsally projecting postzygomatic process that is similar to that of Protitan grangeri.

Although poorly preserved, the overall shape of the occiput seems to resemble that of Protitan grangeri. The occiput appears to be tilted backward. From a dorsal view of the skull, the nuchal crest is concave. From the posterior view the dorsal half of the occiput is heavily reconstructed, but it appears to have been about as wide as the posterior portion. The occiput appears to have been constricted in the middle.

The posterior nares of AMNH 26416 are slightly more anteriorly positioned in comparison to Protitan grangeri (fig. 72a). The anterior rim of the posterior nares seems to be positioned slightly anterior to the M3. However, the elongate posterior narial canal is completely filled with plaster, thus obscuring other details of the posterior nares. Two large ventral sphenoidal fossae are present and are distinctly partitioned by a thin bony septum formed by the basisphenoid. Like Protitan grangeri, the external auditory pseudomeatus forms a broad ventrally unconstricted opening; however, it enters the skull in a posteromedial direction.

Figure 72

The holotype skull of “Protitan” minor (AMNH 26416). (A) Ventral view (note that the canines have been lost since Granger and Gregory [1943]), (B) right molars, (C) left premolars, (D) lingual view of incisors, (E) labial view of right incisors.

Upper Dentition

The teeth of the holotype skull (AMNH 26416) are complete (except for the lost canines), undamaged, and for the most part, they are only lightly worn (fig. 72). The incisors of Protitan minor are relatively large but with small diastemata between them. They form a distinct arch anterior to the canines. The tips of the incisors are worn, but each incisor appears to have been subcaniniform with a short, lingually curved crown and a distinct lingual cingulum. The I3 is distinctly larger and taller than the I1 or I2 and it is more caniniform. Relatively large canines were once present and can be seen in older photographs of AMNH 26416 (fig. 70a). However, the canines are now missing (fig. 72a). There are both short precanine and postcanine diastemata.

The cheek teeth are not differentiated from those of Protitan grangeri. The P1 is a single-cusped tooth with an elongate posterior heel. A small lingual cingulum is also visible on the P1. The anterior margin of P2 is angled distolingually, giving the tooth a strongly oblique shape. The anterior and posterior margins of P3 and P4 are closer to parallel. The parastyle of P2 is straight, although the ectoloph and the metastyle are directed in a posterolingual direction. The parastyle and metastyle of P3 are nearly straight, while the parastyle and metastyle of P4 are distinctly angled labially. The labial margin of the P2 paracone is very convex, while the paracones of P3 and P4 have distinct labial ribs. The metacone of P2 is lingually shifted, while those of P3 and P4 are more labially positioned. Because of these differences the outer surface of P2 is rounder than those of P3 and P4.

Like Protitan grangeri the lingual features of the premolars have low topographic relief. Only a single lingual cusp (protocone) is present on P2–P4. A small preprotocrista is seen on the P2. A faint preprotocrista can be discerned on the P3. Additionally, the protocones of P2 and P3 are each followed by a short lingual crest. The lingual heel of P4 is devoid of any crests. The anterior and posterior premolar cingula wrap around the lingual side of the crowns and nearly join together, but they do not form completely continuous lingual cingula.

The molars show typical brontotheriine apomorphies, including tall, lingually angled ectolophs, very weak labial ribs, thin lingual ectoloph enamel, and wedge-shaped lingual sides of the paracone and metacone in molars that are not heavily worn. The anterior cingulum is thin and passes proximally to the distal peak of the parastyle. Central molar fossae are present, but they are very shallow and almost indistinct. Anterolingual cingular cusps are absent. All traces of paraconules and metalophs are lost. M3 lacks a hypocone. Labial molar cingula are weak, and lingual molar cingula are absent.

Remarks

Protitan minor Granger and Gregory (1943) is based on a nearly complete skull (AMNH 26416) from the Camp Margetts area of Inner Mongolia. Along with P. grangeri, P. minor is one of only two brontotheres with conspicuous frontonasal horns that retains a ventrally unconstricted external auditory pseudomeatus. Granger and Gregory (1943) did not explicitly state how P. minor differed from P. grangeri although there are several differences. In addition to being distinctly smaller, P. minor can clearly be differentiated by its shorter more constricted face, more posteriorly positioned horns, and thinner nasal bone.

No mandibles or lower teeth can presently be referred to Protitan minor. However, considering that the upper dentitions of P. grangeri and P. minor are undifferentiated, it seems likely the lower dental morphologies of these species were similar. Granger and Gregory (1943) referred several mandibles from the Camp Margetts area of Inner Mongolia to P. minor; however, the large metaconid seen on the p3 of these specimens is inconsistent with what one would expect of the lower dentition of P. minor. Therefore, I have removed these specimens from P. minor and reassigned them to an unnamed taxon, informally referred to as Camp Margetts “taxon A” (see section on dubious and problematic taxa).

Yet another mandible (AMNH 26419) from the “Houldjin” beds of Camp Margetts, with extremely worn incisors, a complete and moderate-sized left canine, and right and left premolars, closely resembles what one would predict for Protitan minor in terms of size and overall morphology. Unlike the mandibles assigned to Camp Margetts “taxon A”, the p3 of this specimen lacks a metaconid. This specimen could represent P. minor. Additionally, Dong and Ai (2001) referred a left p1–p4 and right m1 from the lower part of the Tunggur Formation of Inner Mongolia to P. minor. These referrals were based primarily on size and there is no conclusive evidence that they actually belong to P. minor.

Protitanotherium emarginatum Hatcher, 1895

Holotype

YPM-PU 11242, the anterior portion of a cranium with incisors and canines, and a partial mandible with right i1–p2, left i1–m2, and m3 (partial).

Type Locality

Kennedy's Hole, Myton Member (Uinta C) of the Uinta Formation, Uinta Basin, Utah, eight miles north of White River and 25 miles east of Ouray Agency.

Age

Middle Eocene (late Uintan land mammal “age”).

Synonyms

Protitanotherium superbum Osborn, 1908a; Sthenodectes australis Wilson, 1977.

Referred Specimens

(From the Myton member of the Uinta Formation of Utah) AMNH 2501 (holotype of Protitanotherium superbum), a left maxilla fragment with M2–M3, and a partial mandible with right p1–m3 and left c–m3; YPM PU11213, a distal fragment of a nasal process; CMNH 2855, a dorsal surface of the anterior portion of a cranium; YPM 11146, an anterior portion of a mandible with right i1, i2 (partial), i3–c, p2–p3, p4 (partial), left i1, i2 (broken), and i3–p4; (from the Pruett Formation of the Agua Fria Area in Trans-Pecos Texas) TMM 41723-3 (holotype of Sthenodectes australis), a skull with right C–M3 and left P1–M3; TMM 41723-6, a skull with right and left P1–M3, TMM 41747-106, a partial skull with right C–M3, left P1–M3, and isolated right I3.

Diagnosis

Protitanotherium emarginatum is a large brontothere with small, elliptical frontonasal horns positioned low on the skull. The nasal incision is dorsoventrally shallow and extends as far back as the P4. The nasal process is nearly horizontal, unelevated, short and broad, with thickened sides, and with a thickened and imperfectly rounded distal edge with a distinct median notch. The lateral margins of the nasal process are not upturned. The orbits are positioned above M2. The premaxillomaxillary rostrum thickens posteriorly and it is not sealed by bone dorsally. Other cranial characteristics include a saddle-shaped cranium, separate parasagittal ridges that strongly constrict the dorsal surface of the cranium posteriorly, nearly straight zygomatic arches, and tube-shaped and mediolaterally angled external auditory pseudomeati. The posterior narial canal does not extend past the foramen ovale and large ventral sphenoidal fossae are absent.

Dentally, Protitanotherium emarginatum has three intermediate-sized subglobular upper incisors, and metacones on P1 and P2. Premolar hypocones are absent, although a short lingual crest can occasionally be seen extending from the protocones. The molars have tall lingually angled ectolophs with weak labial ribs, and thinned lingual ectoloph enamel with wedge-shaped paracones and metacones. A cingular parastyle shelf is absent. Central molar fossae are present, but anterolingual cingular cusps are absent. Paraconules and metalophs are absent. The lower dentition includes three intermediate-sized semispatulate incisors with an enlarged i2, a distinct postcanine diastema, a metaconid on p4 but not on p2 or p3, shallow molar basins, and a slender m3.

Protitanotherium emarginatum most closely resembles Protitan grangeri and Protitan minor, but it can be most easily distinguished from these species by the smaller, less subcaniniform incisors and tube-shaped external auditory pseudomeatus. Additionally, Protitanotherium emarginatum is distinct from Diplacodon elatus in its relatively smaller horns, dorsoventrally shallow nasal incision, less deeply saddle-shaped cranium, differently shaped nasal process, and less molarized premolars.

Description

Skull

The holotype of Protitanotherium emarginatum consists of the preorbital portion of a skull (fig. 73) and a partial mandible (fig. 77). The skull has been slightly sheared. None of the previously published figures of this specimen (Hatcher, 1895; Osborn, 1929a) clearly differentiates real bone from those parts that are reconstructed with plaster. A large fragment of maxilla and nasal bone is missing from the left face between the orbits and nasal incision. In the mandible the crowns of the left canine and left i3 are almost entirely reconstructed. The lingual half of the left m2 is plaster, as is the majority of the left m3.

Figure 73

The holotype skull of Protitanotherium emarginatum. (Division of Vertebrate Paleontology, YPM PU11242. © 2005 Peabody Museum of Natural History, Yale University, New Haven, Connecticut, USA. All rights reserved.) (A) Left view, (B) anterior view, (C) dorsal view, (D) lingual view of incisors and canines.

Figure 77

The holotype mandible of Protitanotherium emarginatum. (Division of Vertebrate Paleontology, YPM PU11242. © 2005 Peabody Museum of Natural History, Yale University, New Haven, Connecticut, USA. All rights reserved.). (A) Left view, (B) dorsal view, (C) right p1 and p2, (D) left premolars, (E) lingual view of incisors and canines, (F) labial view of incisors and canines.

In addition to the holotype, two nearly complete skulls from the Pruett Formation of Texas allow for a more thorough description of the skull of Protitanotherium emarginatum. TMM 41723-3 (fig. 74) is most significantly damaged in the region above the orbits; essentially the frontonasal horns appear to have been forced back and down over the orbits, significantly distorting this region of the skull including the frontonasal protuberances. Neither the original figures of this specimen nor the initial description of Wilson (1977) reveals that the specimen is damaged in this way. Additionally, the left zygomatic arch is incomplete although a large central segment of it is reconstructed. The second skull, TMM 41723-6 (fig. 75), is less distorted, although the nasal bone is split and the left zygomatic arch is incomplete.

Figure 74

A skull referred to Protitanotherium emarginatum (TMM 41723-3). (A) Dorsal view, (B) left view.

Figure 75

A skull referred to Protitanotherium emarginatum (TMM 41723-6). (A) Right view, (B) dorsal view, (C) ventral view.

The horns of the holotype (fig. 73) are oval, widely separated on the skull, and they project in a dorsolateral direction. They are larger and more prominent than those of Protitan grangeri or P. minor, but they are smaller than those seen on most specimens of Diplacodon elatus. The horns of TMM 41723-3 (fig. 74) are similar in size and shape although they seem to be directed somewhat more laterally. TMM 41723-6 almost completely lacks hornlike protuberances (fig. 75). The considerable size variation in the horns of P. emarginatum is not unlike other horned brontotheres (e.g., Diplacodon elatus).

From a lateral view of the holotype skull it can be seen that the short horns are formed by both the frontal and nasal bones. This configuration of facial bones resembles that of Telmatherium validus, which has an anteriorly projecting process of the frontal bone overlapping the nasal bone. The nasal bone forms the proximal base of the frontonasal protuberance. Two thin sheets of roughened bone appear to overlap the upper surface of the nasal swelling. The upper layer extends to the peak of the protuberance and represents the frontal bone. The lower layer extends over the entire protuberance. It is not clear whether the lower layer is part of the frontal bone or it represents secondary growth of the nasal bone. Below the frontonasal protuberance, the nasal contacts the maxillary in a distinct arch.

The horns of the holotype skull are positioned above the preorbital portion of the face and they are not highly elevated above the skull as in Diplacodon elatus. In TMM 41723-3 the horns are positioned directly above the orbits, although this portion of the cranium is clearly distorted in that area and the horns have been displaced posteriorly. An earlier reconstruction of this specimen (Wilson, 1977: fig. 2) places the horns anterior to the orbits.

Due to the fragmentary nature of the holotype specimen it is difficult to define the position of the posterior margin of the nasal or the position of the orbit with respect to the dentition. The length of the nasal incision is similar to Protitan minor. In TMM skulls, the nasal incision extends to the P4 and the orbits are positioned more or less above the M2.

Unlike Diplacodon elatus, the nasal incision of Protitanotherium emarginatum is dorsoventrally shallow. The nasal process of the holotype is unelevated, straight, angled slightly downward, and it is shorter than the premaxillomaxillary rostrum. The sides of the nasal process are thickened. The thickened lateral edges extend downward, but they do not form deep vertical walls. The sides of the nasal process are not upturned as in D. elatus. From a dorsal view the nasal bones are strongly fused but the nasal suture is still visible. The nasal process is as wide as the premaxillomaxillary rostrum although its width is slightly constricted proximally. The distal margin nasal process is nearly flat, although it is somewhat concave medially. The anterior edge of the nasal process of TMM 41723-3 is more rounded and more strongly notched at the midline in comparison to the holotype. From the anterior view of the holotype, the anterior edge of the nasal process is relatively flat, thick, and mildly roughened. The anterior margin is not strongly deflected downward at the midline as in Protitan grangeri or D. elatus.

The premaxillomaxillary rostrum is best preserved in the holotype (fig. 73). From the lateral view the dorsolateral margin of the rostrum rises posterodorsally at a very shallow angle. The premaxillae are strongly fused at the symphysis. A premaxillomaxillary suture can be seen although it is indistinct. The premaxillae end before reaching the posterior notch of the nasal incision and they do not contact the nasal bone. From the anterior view the dorsolateral margins of the rostrum are laterally divergent and the rostral cavity is not sealed dorsally by bone.

Although the holotype is merely an anterior cranial fragment, Hatcher (1895) described the skull of Protitanotherium emarginatum as “slightly concave anteroposteriorly? and further characterized by the absence of a sagittal crest” (Hatcher, 1895: 1085). Hatcher must have made these observations on parts of the holotype that have been lost. Hatcher (1895) indicates, “the posterior region had already weathered out and was badly injured, but many of the pieces have been fitted together and show some of the more important characters of this region of the skull” (Hatcher, 1895: 1084–1085). The TMM specimens are consistent with Hatcher's observations on these missing pieces. The dorsal surfaces of these skulls are concave from the orbits to the nuchal crest. The parasagittal ridges do not join to form a sagittal crest, although they strongly constrict the posterior portion of the dorsal surface.

The zygomatic arches of TMM 41723-3 are thick whereas those of TMM 41723-6 are thinner; this variation corresponds with the differing degrees of horn development and canine size in these specimens. In both specimens, the jugal portion of the zygomatic is straight from a dorsal view and from a lateral view it is horizontal. The squamosal portion of the zygomatic is very weakly angled posterodorsally and the zygomatic arch is almost completely straight from a lateral view. A posterior zygomatic processes is not seen in Protitanotherium emarginatum.

There is no discernable infraorbital process on the TMM specimens. Nonetheless, Wilson (1977) described the TMM specimens as “give[ing] the appearance of having an infraorbital protuberance, but if the effect of the crushing were removed, such a protuberance would probably not have been more prominent than in CMNH 2398” (Wilson, 1977: 5). (CMNH 2398 is the holotype of Sthenodectes incisivum.) It is not clear what Wilson (1977) meant by this statement, since the lower portion of the orbit and suborbital region do not appear to be significantly distorted.

The occiputs of the TMM specimens are wide and strongly tilted backward. The dorsal portion is essentially the same width as the ventral portion and the occiput is not constricted in its middle. The nuchal crest is thin and is somewhat concave at the midline.

The anterior rim of the posterior nares is positioned between the protocones of the M3s in TMM 41723-3 and at the anterior margin of M3 in TMM 41723-6. The posterior nares of both specimens are rimmed by a distinct U-shaped emargination. In TMM 41723-6 the posterior nares have been more fully cleared of sediment, revealing that the full width of the bony emargination that surrounds the posterior nares. In TMM 41723-6 the thin vomer can be seen bisecting the elongate posterior canal and joining with the sphenoid. The posterior narial canal, which is filled with sediment, extends somewhat into the sphenoid but not posterior to the foramen ovale. Large ventral sphenoidal fossae are not seen on this specimen. Other aspects of the basicranium, such as the widely separated foramen ovale and foramen lacerum, are typical for brontotheres. The mastoid process is short and curves anteroventrally, contacting the postglenoid process, thus, the external auditory pseudomeatus is closed ventrally and tube-shaped. This configuration differs from Protitan, but resembles all other horned brontotheres.

Upper Dentition

The upper incisors of the holotype are in good condition and lightly worn (fig. 73d). No other upper teeth are preserved on the holotype except for the poorly preserved canines. On the other hand, the TMM specimens lack incisors but have more complete cheek teeth, particularly TMM 41723-3 (fig. 76b, d).

Figure 76

Ventral views of skulls and upper dentition referred to Protitanotherium emarginatum. (A) ventral view of TMM 41723-3, (B) left premolars of TMM 41723-3, (C) M2 and M3 of AMNH 2501, (D) right P2–M3 of TMM 41723-3.

Overall, the incisors of YPM-PU 11242 are significantly smaller than those of Protitan. The incisors increase in size laterally. The crowns of I1 and I2 are very short and blunt, although distinct lingual cingula can be seen on these teeth. The I3, on the other hand, is much larger and more subcaniniform than I1 or I2. The incisor row forms a very shallow arch that is positioned anterior to the canines. The I3 and canine are separated by a large diastema. Though the crowns of both canines are incomplete, it is apparent from the roots and the base of the right canine crown that the canines of this specimen were very large. The canines of TMM 41723-3 are of a similar size, although the left canine alveolus of TMM 41723-6 (fig. 75c) indicates a somewhat smaller canine. A distinct postcanine diastema is consistently present in these specimens, although its length is variable.

No cheek teeth are attached to the holotype skull, however, Hatcher (1895: pl. 38) and Osborn (1929a: fig. 317) each figured and described a left P1. The tooth was lost sometime after the fossil was molded, because the tooth is present on plaster casts found in numerous North American museums. Hatcher (1895) described it as “a very simple tooth fixed in the jaw by two roots, and consisting of a single cone with a posterior heel” (Hatcher, 1895: 1086). Osborn described the P1 somewhat differently as a “bifanged tooth…with a simple protocone, a sessile or rudimentary posterior heel, and a posterointernal cingulum and concavity” (Osborn, 1929a: 378). Inspection of a cast (AMNH 10385) reveals that Osborn's (1929a) description is more accurate. There is both a paracone and a smaller metacone, although wear has virtually obliterated the metacone. However, Osborn's identification of a protocone on the P1 is misleading. There is a small, posteriorly positioned lingual heel with a minor undulation of enamel on it, but it does not bear a distinct protocone. The P1 of TMM 41723-3 is smaller than the P2, but it is morphologically very similar, with two labial cusps of similar size, a relatively straight ectoloph, and a well-developed lingual heel. The lingual heel appears to have a small protocone and a small anteroposteriorly oriented lingual crest.

The P2–P4 of TMM 41723-3 are in good condition (fig. 76b). These teeth are closely pressed together. The P2 crown is slightly oblique, while the P3 and P4 crowns are progressively more rectangular. The anterior margins of P3 and P4 are worn away because of interstitial wear resulting in the formation of concave anterior margins. It is interesting that the posterior margins of P2 and P3 fit perfectly into the concave worn anterior margins of P3 and P4, but curiously, the posterior margins of P2 and P3 seem not to have been experienced significant interstitial wear. The ectoloph of P2 is straight, although in P3 and P4, the parastyles are somewhat labially directed. Distinct labial paracone ribs can be seen on P2–P4, though they become progressively smaller in more posterior premolars. The lingual heels of the P2–P4 are broad with nearly flat lingual edges. There are no distinct lingual cusps on the P2. Rather, a short lingual crest extends anteroposteriorly along the lingual side of the crown and is joined anteriorly by a small preprotocrista. Both P3 and P4 have a distinct protocone. The preprotocrista of P3 is faint, while the P4 has no preprotocrista. The protocones of P3 and P4 are followed posteriorly by very weak lingual crests. There are no hypocones on any of the premolars, thus the premolars of Protitanotherium emarginatum are significantly less molarized than are those of the contemporary species, Diplacodon elatus. The labial premolar cingula are weak. The lingual P2 cingulum is strong and continuous. The lingual cingulum of P2 is slightly discontinuous and it is more discontinuous on P4.

In addition to the molars of TMM 41723-3 (fig. 76d), a close up view of the M2 and M3 of AMNH 2501 is provided (fig. 76c). The molars of Protitanotherium emarginatum exhibit typical brontotheriine apomorphies including tall, lingually angled ectolophs, weak labial ribs, thin lingual ectoloph enamel, and wedge-shaped lingual sides of the paracone and metacone in relatively unworn molars. The anterior cingulum is thin and passes proximally to the distal peak of the parastyle. The following traits can be more distinctly seen in AMNH 2501. Shallow but distinct central molar fossae are present. The lingual portion of the anterior cingulum is thickened, but it does not form a distinct anterolingual cingular peak. All remnants of paraconules or metalophs are lost. There is no hypocone on the M3, but a portion of the posterior cingulum is thickened. The labial molar cingulum tends to be weak and discontinuous around the mesostyles.

Mandible and Lower Dentition

In the mandible of the holotype (fig. 77) the inferior margin of the symphysis is angled slightly less than 45° and it extends to the anterior margin of p4. The ascending ramus is not preserved, although in another specimen, AMNH 2501 (fig. 78), the coronoid process is long, slender, moderately curved, stands higher than the condyle, and is not generally different from other brontotheres.

Figure 78

Left view of a mandible referred to Protitanotherium emarginatum (AMNH 2501).

The lower incisors are significantly smaller than those of Protitan grangeri. They are more similar in size to those of Diplacodon elatus, but they are not reduced to the essentially vestigial condition seen in more advanced horned brontotheres such as Duchesneodus uintensis. The lower incisors are only lightly worn. They are semispatulate in shape. The i2 is significantly larger than either the i1 or the i3 in height and in labiolingual width. There are no labial cingulids and the lingual cingulids are faint. The incisors form only a slight arch anterior to the canines. The lower canines of both YPM-PU 11242 and AMNH 2501 are rather large. There is no precanine diastema. The postcanine diastema is asymmetrical in YPM PU11242. On the right side it is slightly longer than the p2, but on the left side it is a little shorter.

The right and left p1s of YPM-PU 11242 are bilaterally asymmetrical. The right p1 is a narrow single cusped tooth with an elongate posterior talonid heel. The left p1 is notably wider, particularly the posterior portion of the crown. The p2–p4 have relatively low slender crowns. The trigonids of p2 and p3 are less than twice as long as the talonids. The trigonid and talonid of p4 are of similar length. The talonids of p2–p4 are slightly wider than the trigonids. On p2 and p3 the paralophids and protolophids extend from the protoconid in a slightly lingual direction. There is only a small lingual trigonid notch on the p2, although the lingual trigonid notch of the p3 is large. The paralophid and protolophid of p4 arch 90° lingually, forming a nearly molariform trigonid basin. Metaconids are absent on p2 and p3 but they are present on p4. The cristids obliqua and hypolophids of p2 and p3 are well developed, but they are short and the lingual-talonid notch does not form a molariform basin. However, the cristid obliqua and hypolophid of p4 are much longer and create a more molariform talonid basin. Lingual premolar cingulids are absent, while labial premolar cingulids are absent (p2, p3) or very weak (p4).

The molars are typical, with thin lingual enamel, weak lingual ribs, and shallow talonid basins. The m3 of the holotype is incomplete but that of AMNH 2501 is elongate. The labial cingulids of the molars are much stronger in comparison to the premolars.

Remarks

Hatcher (1895) named Protitanotherium emarginatum from a specimen (YPM-PU 11242) consisting of the anterior portion of a skull with short frontonasal horns and a partial mandible. Previously, Marsh (1875) had named another species of horned brontothere, Diplacodon elatus. When Hatcher (1895) described P. emarginatum the existence of a horn in D. elatus was in question. For this reason Hatcher (1895) cautiously referred his new species to the genus Diplacodon but suggested, “should future discoveries show that there are hornless forms with the same dental characters as Diplacodon, it will be necessary to establish for the present specimen (YPM-PU 11242) a new genus, which may be called Protitanotherium” (Hatcher, 1895: 1084). Although it turns out that Diplacodon elatus does have horns, Osborn (1929a) demonstrated several distinctions between Hatcher's (1895) D. emarginatum and Marsh's D. elatus. Therefore, he adopted Hatcher's recommended genus name, Protitanotherium. Lucas and Schoch (1989a) felt that Protitanotherium was a junior synonym of Diplacodon, however Mader (1989; 1998) concluded that Protitanotherium was sufficiently different to warrant its distinct genus name. Mader's conclusion is accepted here.

Several additional specimens from the Uinta Formation are referred to Protitanotherium emarginatum, including the holotype mandible (AMNH 2501) of P. superbum Osborn (1908a). Osborn's specific characters for P. superbum do not clearly diagnose a new species. These include: “Canines in males very robust; p1 double fanged (an erroneous description); post canine diastema abbreviated; premolar series relatively abbreviated; p2 with very large talonid and crescentic protoconid; p3, p4 with talonid heavy and prominent, i.e., submolariform, but no entoconid; m3 with hypoconulid sharply constricted off at base” (Osborn, 1908a). The mandible and cheek teeth of AMNH 2501 sufficiently resemble the holotype of P. emarginatum to refer it to that species. It distinctly differs from the contemporaneous horned species, Diplacodon elatus, particularly in the absence of a p3 metaconid. AMNH 2501 is somewhat larger than the holotype of P. emarginatum, but it falls within an acceptable body size range for that species. Other specimens, such as the nasal bone YPM PU11213 are larger than the holotype as well, suggesting that the type specimen is a rather small individual of the species. A pair of upper molars (left M2, M3) is also assigned to AMNH 2501 although Osborn (1908a) made no mention of these upper molars in the original description of P. superbum. Later, Osborn (1929a) assumed that these upper molars were part of the same specimen. Whether or not these upper molars are associated with the mandible (AMNH 2501) is questionable, but both, nonetheless, seem to belong to P. emarginatum.

A brontothere consistent with Protitanotherium emarginatum also occurs in the Whistler Squat local fauna of the Pruett Formation of the Big Bend–Trans-Pecos area, Texas, although until now that material has not formally been recognized as P. emarginatum. Wilson (1977) described this material as a new species, Sthenodectes australis. This species was based primarily on two nearly complete skulls. Wilson (1977) mistakenly referred this species to the genus Sthenodectes. According to Wilson (1977: 7) “the length of the tooth row and the approximate size of the teeth show that the partial skull is close to Sthenodectes.” And secondly “the very large molars in proportion to the skull length, the advanced condition of the premolars, and the very rudimentary horns…seems to best fit Sthenodectes” (Wilson, 1977: 8). Despite Wilson's conclusion, it is clear that these skulls are far closer to Protitanotherium than Sthenodectes. It seems that Wilson (1977) failed to compare the material to P. emarginatum, largely because his comparisons were based on upper tooth measurements, elements lacking largely lacking in the holotype of P. emarginatum. Moreover, as with most work done on brontotheres after 1929, Wilson (1977) drew heavily from Osborn (1929a). Osborn (1929a: fig. 301) erroneously figured Sthenodectes incisivum as having a small horn. Actually, none of the S. incisivum specimens has a horn. Mader (1998) suggested that the skulls described by Wilson (1977) might be referable to Protitanotherium, but he made no formal revision. Indeed, the skulls of TMM 41723-3 and TMM 41723-6 are consistent with P. emarginatum. Differences between these skulls and the holotype of P. emarginatum, such as the morphology of P1 and variation in horn and canine size, are characters that are intraspecifically variable in other brontotheres. S. australis appears to be a junior synonym of P. emarginatum. Wilson (1977) referred several other specimens to S. australis, including some partial mandibles. These specimens could belong to P. emarginatum, but they lack sufficiently diagnostic morphology to make conclusive identifications.

Rhinotitan kaiseni (Osborn, 1925)

Holotype

AMNH 20252, a skull and mandible with extremely worn teeth.

Type Locality

Shara Murun Formation, Ula Usu, Baron Sog Mesa, Inner Mongolia.

Age

Middle Eocene (Sharamurunian land mammal “age”).

Referred Specimens

(From the Shara Murun Formation, Ula Usu, Baron Sog Mesa, Inner Mongolia) AMNH 20257, a ventral surface of a skull with a complete set of upper dentition; FMNH P14048 (formerly AMNH 20260), a skull and mandible with complete sets of upper and lower dentition.

Diagnosis

Rhinotitan kaiseni is a large brontothere with small, elliptical frontonasal horns. The horns are positioned far in front of and high above the orbits. The nasal incision is dorsoventrally deep and extends posteriorly to P3. The orbit is positioned above M2 with posterolateral and anterolateral roots of M1 below the anterior orbital rim. The nasal process is unelevated, slightly angled upward, relatively narrow, not strongly rounded anteriorly, and with deep lateral walls that arch around the anterior end of the nasal process. Proximally, the lateral walls of the nasal process deepen and angle ventromedially. The premaxillomaxillary rostrum thickens posteriorly and it is not enclosed by bone dorsally. Other cranial characteristics include an incompletely saddle-shaped cranium, a dorsal cranial surface that is not constricted posteriorly by parasagittal ridges, a narrow emargination surrounding the posterior nares, nearly straight zygomatic arches, and a ventrally constricted and posteromedially angled external auditory pseudomeatus. Ventral sphenoidal fossae and postzygomatic processes are absent.

Dentally, Rhinotitan kaiseni has three large but short subcaniniform upper incisors. The P1 crown is complex and there is a distinct P2 metacone. Premolar hypocones are present although the protocones and hypocones of P2 and P3 are sometimes fused into a single lingual crest. The molars of R. kaiseni have tall, lingually angled ectolophs with weak labial ribs, and thinned lingual ectoloph enamel with wedge-shaped paracones and metacones. A cingular parastyle shelf is absent. Distinct central molar fossae are present but anterolingual cingular cusps are absent. Paraconules and metalophs are absent. The lower dentition of R. kaiseni includes three large incisors. The i1 and i2 are semispatulate, while the i3 is more subcaniniform. There is a distinct postcanine diastema and an occasional small p3 metaconid. The p2 trigonid is nearly twice the length of the talonid. The lower molars have shallow basins and the m3 is slender.

Rhinotitan kaiseni can be distinguished from most other horned brontotheres by the combination of large incisors and an incompletely saddle-shaped cranium. R. andrewsi shares these traits, but R. kaiseni can be further distinguished from R. andrewsi by the narrower basicranium, more anterior position of the frontonasal horns, downfolded anterior margin of the nasal process, and subcaniniform upper incisors.

Description

Skull

The holotype of Rhinotitan kaiseni (AMNH 20252) includes a complete skull (fig. 79) and an associated mandible. The skull is complete, but it has experienced some shearing distortion and the dorsal surface of the skull is warped. Additionally, there is another complete skull (fig. 80) and associated mandible (fig. 82) (FMNH P14048: originally AMNH 20260). The skull consists of hundreds of small bone fragments held together with plaster, yet the specimen retains its general shape although its overall size appears to be swollen due to expanding matrix distortion (sensu White, 2003). A third specimen (AMNH 20257) consists of an undistorted ventral surface of a skull (fig. 81a).

Figure 79

The holotype skull of Rhinotitan kaiseni (AMNH 20252). (A) Left view, (B) dorsal view, (C) anterior view, (D) posterior view.

Figure 80

A skull referred to Rhinotitan kaiseni (FMNH P14048). (A) Left view, (B) dorsal view.

Figure 82

Mandible (FMNH P14048) of Rhinotitan kaiseni, associated with the skull in Fig. 80. (A) Left view shown with a cast of the p1 (original p1 has been lost), (B) dorsal view, (C) left premolars shown with a cast of the p1 (now lost), (D) lingual view of incisors and canines, (E) labial view of left incisors and canines.

Figure 81

Ventral surface and upper dentitions of skulls referred to Rhinotitan kaiseni. (A) Ventral view of AMNH 20257, (B) left molars of AMNH 20257, (C) left premolars of AMNH 20257, (D) left premolars of FMNH P14048, (E) lingual view of incisors and canines of FMNH P14048.

Rhinotitan kaiseni is a large horned brontothere similar in size to Protitan grangeri and Protitanotherium emarginatum. There are no discernable sutures on the holotype skull (AMNH 20252), but judging by the extreme dental wear this animal died at a very old age. FMNH P14048 is an ontogenetically younger individual with minimally worn adult dentition. On that specimen the posterodorsal surfaces of the right and left horns are less damaged than the surrounding bone. Portions of the frontonasal suture are visible on both the lateral and dorsal views, indicating that the horn is composed of both frontal and nasal elements.

The horns of AMNH 20252 are small but prominent and project almost horizontally from the skull. They are proximally elliptical, but the distal peaks of the horns form round roughened knobs. In FMNH P14048 the horns do not project laterally. In both skulls, the horns are positioned far anterior to the orbits. This differs from Rhinotitan andrewsi where the horns are more posteriorly positioned. The horns of R. kaiseni are positioned well out onto the nasal process and are high above the orbits.

The nasal incision extends posteriorly to a point above the anterior margin of the P3. The nasal incision is dorsoventrally deep. The posterior notch of the nasal incision is positioned slightly higher than the orbit. The orbit is above M2 while the posterolateral and anterolateral roots of M1 rest below the anterior orbital rim.

From the lateral view the nasal process is slightly longer than the premaxillomaxillary rostrum. The nasal process of AMNH 20252 appears to be angled slightly upward. The sides of the nasal process form deep lateral walls; these are deeper and thinner than those of R. andrewsi. The lateral walls are shallower toward the distal end of the nasal process. From the dorsal view, the width of the nasal process is relatively constant throughout its length and the anterior margin is nearly flat. From the anterior view the distal end of the nasal process is downfolded. Toward the proximal end of the nasal process the lateral walls are weakly angled ventromedially and they constrict the space between them, but this condition is not as severe in R. kaiseni as it is in R. andrewsi.

From a lateral view the premaxillomaxillary rostrum is thin distally, but it deepens proximally. The dorsal margin of the rostrum is steeply sloped posterodorsally. From the anterior view the premaxillary symphysis is completely fused. The dorsolateral margins of the rostrum diverge laterally behind the symphysis and the rostral cavity is not sealed over by bone. From the anterior view the nasal opening of the skull is very tall and narrow.

The dorsal surface of the skull is artificially warped. However, unlike many horned brontotheres, the skull does not appear to have been saddle-shaped. Rather, the shape of the skull in lateral profile more closely resembles Rhinotitan andrewsi and more primitive hornless brontotheres (e.g., Telmatherium) in which the dorsal surface is concave over the middle of the skull, but the dorsal surface of the skull is more convex over the posterior end of the skull.

The parasagittal ridges are not prominent on AMNH 20252 and the dorsal surface of the skull is not laterally constricted posteriorly. From a lateral view, the zygomatic arches are relatively shallow. The jugal portion of the zygomatic is horizontal, while the squamosal portion rises posteriorly at a very shallow angle and the zygomatic arch is nearly straight. From the dorsal view, the zygomatic arches are thin, straight, and not strongly angled laterally. A postzygomatic process, as seen in Protitan grangeri and Metatitan relictus, is absent in Rhinotitan kaiseni.

From a lateral view the occiput is nearly vertical. The nuchal crest is damaged, but from a dorsal view, it appears to have been concave. From the posterior view, the nuchal crest is dorsally arched. The dorsal and ventral halves of the occiput are the same width, although the occiput seems to have been moderately constricted in the middle. The surface of the occiput has weak occipital pillars with a shallow pit in the center.

In the holotype specimen (AMNH 20252) the anterior rim of the posterior nares is positioned somewhere behind M2, although its exact position is obscured by plaster. In AMNH 20257 the anterior edge of the posterior nares is slightly anterior to the M3 protocones. The posterior nares are rimmed by a narrow emargination. The vomerine septum seen in AMNH 20257 bisecting the posterior narial canal is mostly made of plaster except at its posteriormost end. There are no sphenoidal fossae in any of the specimens of R. kaiseni. The external auditory pseudomeatus is tube-shaped and enters the skull at a strongly posteromedial angle. The configuration of the basicranial foramina is typical, with widely separated foramen ovale and foramen lacerum. The width of the basicranium, measured at the mastoid processes, is slightly narrower than the width across the M3s; this distinctly differs from skulls of Rhinotitan andrewsi whose basicranium is wider that the total width across the M3s.

Upper Dentition

The teeth of the holotype skull are too worn to describe, but the upper dentitions of AMNH 20257 and FMNH P14048 are complete and not heavily worn (fig. 81). The incisor rows of AMNH 20257 and FMNH P14048 are complete, although those of the latter are the least worn. The incisors number three per side, are large, and form an arched row that extends anterior to the canines. The incisors increase in size laterally. The incisors of Rhinotitan kaiseni are less globular than those of R. andrewsi. When the I1 and I2 are nearly unworn, as in FMNH P14048, the crowns retain a relatively subcaniniform shape, although they are very short and have smooth lingual cingula. The I3 has a similar shape, but it is taller and more subcaniniform. The canines of all R. kaiseni specimens tend to be relatively small. There is a short precanine diastema and a longer postcanine diastema.

The P1 crown of AMNH 20257 is not preserved. The P1 of FMNH P14048 is much smaller than more posterior molars, but its morphology is relatively advanced and the crown has a nearly round outline. A metacone is distinctly present and is similar in size to the paracone. There is a small lingual heel with a small protocone. In both specimens the crown of P2 seems more obliquely shaped than the crowns of P3 or P4 due to its more posterolingually angled anterior margin. The P3 and P4 are less oblique in shape and the anterior and posterior margins of these teeth are nearly parallel. The P2 metacone is not shifted lingually. The parastyle and metastyle of P2 are straight. The parastyles of P3 and P4 are directed somewhat labially, although the metastyles of these teeth are essentially straight. The labial margin of the P2 paracone is rounded, while the paracones of P3 and P4 have small but distinct labial ribs.

The lingual features of the crowns of P2–P4 have low relief and invariably include a protocone and a lingual crest that extends posteriorly from the protocone. Vestigial preprotocrista are seen on P2 and P3, but this trait is not evident on P4. In other respects the lingual premolar morphology is variable. For instance, there is notable bilateral asymmetry in the presence and distinctness of premolar hypocones in AMNH 20257. The left P2 has a hypocone that is incompletely separated from the protocone, while the P3 and P4 each have small, well-separated hypocones. The right P2 is similar to the left side, but unlike their left counterparts, the right P3 and P4 lack distinct hypocones. The premolars of FMNH P14048 are more bilaterally symmetrical; P2 has a distinct hypocone that is connected to the protocone, but P3 and P4 lack hypocones. The lingual premolar cingula are also variable. The lingual premolar cingula of AMNH 20257 are discontinuous, while the lingual premolar cingula of FMNH P14048 are continuous on the P2 and P3 but discontinuous on the P4. Likewise, the labial premolar cingula of FMNH P14048 tend to be weaker than those of AMNH 20257.

The upper molars of Rhinotitan kaiseni show typical brontotheriine apomorphies, including tall, lingually angled ectolophs, very weak labial ribs, thin lingual ectoloph enamel, and wedge-shaped lingual sides of the paracone and metacone in molars that are not heavily worn. The anterior cingulum is thin and passes proximally to the distal peak of the parastyle. Distinct shallow central molar fossae are present, but anterolingual cingular cusps are absent. All evidence of paraconules and metalophs is lost. There is no trace of a hypocone on any M3 of R. kaiseni, although the distolingual cingulum of the M3 is thickened. Labial molar cingula are weak and lingual molar cingula are absent.

Mandible and Lower Dentition

Although the holotype (AMNH 20252) includes an associated mandible, the lower teeth are extremely worn. However, a second skull, FMNH P14048, is associated with a complete and essentially undistorted mandible with a complete set of lightly worn dentition (fig. 82). The inferior margin of the symphysis has a shallow angle (< 45°). The symphysis extends to the talonid of the p3. The incisors number three per side and form an arch anterior to the canines. The lower incisors are large, while i2 is clearly the largest incisor. The crowns of i1 and i2 are tall and semispatulate with rounded apices. The i3 crown is shorter and mesiodistally more elongate. There are slight lingual incisor cingulids and no labial incisor cingulids. There are no diastemata between the incisors or the canines. The canines are rather small. The postcanine diastema of FMNH P14148 is similar in length to p2, although that of the holotype is slightly longer.

A left p1 was originally associated with FMNH P14048, but it has been lost. Fortunately, the left p1 is still present on a cast of the jaw (AMNH 20260). The p1 cast is shown in proportion to the original specimen in fig. 82a and 82c. The p1 is a small narrow tooth with a single cusp and a short talonid heel. The p2 trigonid is less than twice the length of the talonid, while the trigonid and talonid have similar widths. The talonid and trigonid of p3 and p4 are about equal in length, but the talonid is distinctly wider than the trigonid. The paralophid of p2 arches slightly lingually, creating a small lingual trigonid notch. The p2 protolophid is straight but lingually positioned. The p2 lacks a metaconid. The trigonid of the p3 is more molariform with a strongly lingually paralophid and metalophid that create a much broader lingual-trigonid notch. In addition, p3 has a small but distinct metaconid. The trigonid of p4 is essentially molariform with a fully lingually arched paralophid and protolophid, and a large lingually positioned metaconid. The talonid of p2 has a relatively short cristid obliqua and hypolophid; these features create a small lingual-talonid notch. The cristids obliqua and hypolophids of p3 and p4 are longer and the talonid basins are more molariform. There are no lingual cingulids on the premolars and the labial cingulids are weak.

The lower molars of Rhinotitan kaiseni are typical. They have relatively thin lingual enamel, shallow trigonid and talonid basins, and the m3 is elongate. There are no lingual cingulids. Labial molar cingulids are weak and they tend to be discontinuous around the labial cusps. A cingulid is not seen around the hypoconulid of the m3.

Rhinotitan andrewsi (Osborn, 1925)

Holotype

AMNH 20271, a skull missing the nasal, with right I2–M3, left I2–C, and P2–M3.

Type Locality

Shara Murun Formation, Ula Usu, Baron Sog Formation, Shara Murun Region, Inner Mongolia, China.

Age

Middle Eocene (Sharamurunian land mammal “age”).

Referred Specimens

(From the Shara Murun Formation, Ula Usu, Baron Sog Mesa, Shara Murun Region, Inner Mongolia) AMNH 20254, a skull with right I1–I2 (damaged), I3, C (damaged), P1–M3, left I2 (damaged), I3, C (damaged), and P1–M3; AMNH 20261, a laterally crushed skull with right C–M3 and left I3–M3; AMNH 20263, a palate with right and left I2–M3; IVPP V3254-1, a partial skull with right I1–C, P2–M3, left I1–I3, P2–M3, a mandible with right i1–c, p2–m3, left i1–c and p2–m3, and a skeleton; IVPP V3254-2, a complete skull with complete dentition except left I1, and a mandible with right p1–m3, left p2, and p4–m3; PIN 2198-3, a partial skull with right P2–M2, M3 (partial and unerupted) and left P1–M2; PIN 2198-5, a skull with right P2–M3 and left I2–C, P1–M3: PIN 7130-3 with right and left P2–M3.

Diagnosis

Rhinotitan andrewsi is a large horned brontothere with small, elliptical, and widely separated frontonasal horns. The horns are positioned slightly in front of the orbits and high above the orbits. The nasal incision is dorsoventrally deep and extends to the P4. The orbit is above the M2 with the posterolateral and anterolateral roots of M1 below the anterior orbital rim. The nasal process is either horizontal or slightly angled upward, relatively broad, not strongly rounded anteriorly, and with thick lateral walls that do not arch around the anterior margin. Proximally, the lateral walls of the nasal process deepen and angle ventromedially, greatly constricting the dorsal portion of the nasal cavity. The premaxillomaxillary rostrum thickens posteriorly and it is not enclosed by bone dorsally. Other cranial characteristics include a marginally saddle-shaped or incompletely saddle-shaped cranium, a dorsal cranial surface that is not constricted posteriorly by parasagittal ridges, a narrow emargination surrounding the posterior nares, nearly straight zygomatic arches, and a ventrally constricted and posteromedially angled external auditory pseudomeatus. Ventral sphenoidal fossae and postzygomatic processes are absent.

Dentally, Rhinotitan andrewsi has three large incisors. The I1 and I2 are subglobular, while I3 is more subcaniniform. P1 is complex, there is a distinct P2 metacone, and premolar hypocones are occasionally present, although in P2 and P3 the protocone and hypocone sometimes take the form of a single lingual crest. The molars have tall, lingually angled ectolophs with weak labial ribs, and thinned lingual ectoloph enamel with wedge-shaped paracones and metacones. A cingular parastyle shelf is absent. Distinct central molar fossae are present, but anterolingual cingular cusps are absent. Paraconules and metalophs are absent. The lower dentition of R. andrewsi includes three large incisors; the i1 and i2 are semispatulate, while the i3 is more subcaniniform. There is a distinct postcanine diastema and the p2 trigonid is nearly twice the length of the talonid. The lower molars have shallow basins and the m3 is slender.

Rhinotitan andrewsi can be distinguished from most other horned brontotheres by the combination of large incisors and an incompletely saddle-shaped cranium. R. andrewsi is most easily distinguished from R. kaiseni by the wider basicranium, the more posterior position of the frontonasal horns, the less downturned anterior margin of the nasal process, and the more subglobular I1 and I2.

Description

Skull

The holotype of Rhinotitan andrewsi (AMNH 20271) is a large skull missing the nasal bones (fig. 83). The ventral surface of the skull and the zygomatic arches are well preserved, although large portions of the cranium are incomplete and reconstructed with copious amounts of plaster. There are several additional skulls of R. andrewsi, all from the Shara Murun Formation, including material in the AMNH collection, the IVPP collection previously described by Wang (1982), and undescribed material in the PIN collection. Each skull is crushed, heavily reconstructed with plaster, and/or damaged in some way, rendering it difficult to precisely interpret the shape of the skull without considering several specimens. In addition to the holotype, I have figured lateral views of AMNH 20254, IVPP V3254-1, IVPPV 3254-2 (fig. 84), and a more extensive set angles of PIN 7130-3 (fig. 85).

Figure 83

Left view of the holotype specimen of Rhinotitan andrewsi (AMNH 20271).

Figure 84

Specimens referred to Rhinotitan andrewsi. (A) Left view of AMNH 20254, (B) right view of the skull of IVPP V3254-2, (C) right view of the mandible of IVPP V3254-1, (D) left view of the skull of IVPP V3245-1. D is from Wang (1982).

Figure 85

A skull referred to Rhinotitan andrewsi (PIN 7130-3). (A) Left view, (B) dorsal view, (C) anteroventral view, (D) posterior view. Note that the premaxilla and incisors seen in C are a plaster reconstruction. The reconstructed premaxilla has been cropped off in A.

Rhinotitan andrewsi is larger than R. kaiseni, but the size difference is probably not so extreme that there was no overlap in body size among members of these species. The horns of all R. andrewsi specimens are highly elliptical with the longer axis in the anteroposterior direction, and they tend to project strongly laterally. There is a moderate amount of size variation in the horns and some are rather smooth while others have more rugose surfaces. The horns are widely separated laterally and they are elevated above the orbits to a degree similar to Diplacodon elatus. From a lateral view the horns rest upon a superorbital pillar that rises from the orbit at about a 45° angle. The horns are more posteriorly positioned than those of R. kaiseni. They are positioned in front of the orbits and rest directly above the posterior margin of the nasal incision (IVPP V3254-2) or in front of it (PIN 7130-3). Such minor differences in the position and height of the horns may relate to taphonomic distortion.

The nasal incision is so dorsoventrally deep that its posterior margin is significantly higher than the orbit. The nasal incision of Rhinotitan andrewsi is longer than that of R. kaiseni. In the holotype (AMNH 20271), the incomplete posterior margin of the nasal incision appears to extend to the anterior edge of P4. The posterior margin of the nasal incision of PIN 7130-3 is also positioned about at the anterior margin of P4. There is some fluctuation in the exact position of the posterior margin of the posterior nares; in AMNH 20254 it is slightly more anterior than the P4, while those of IVPP V3254-1 and IVPP V3254-2 are slightly posterior to the anterior margin of P4.

The position of the orbit of Rhinotitan andrewsi is not significantly different from that of R. kaiseni. For instance in the holotype (AMNH 20271), the M2 is directly below the orbit and the lateral roots of M1 are below the anterior rim of the orbit. Among other specimens, the exact position of the orbit fluctuates slightly. For instance, in AMNH 20254, the posterolateral root of the M1 is clearly behind the anterior orbital rim.

In some of the skulls (e.g., IVPP V3245-1, AMNH 20254) the nasal process slants upward; this character has been used as evidence for drawing phylogenetic lines between Rhinotitan and embolotheres (Protembolotherium and Embolotherium) (Yanovskaya, 1980; Wang, 1982). However, for the most part, the upward slant of the nasal processes of these specimens seems related to taphonomic distortion. In IVPP V3254-1, the horns and the entire nasal process are strongly angled upward at about a 45° angle. However, there is an abrupt kink in the dorsal surface of the skull just behind the orbits, suggesting that the entire frontonasal region has been artificially rotated upward. Other specimens of R. andrewsi with upwardly angled nasals (e.g., AMNH 20245) also clearly suffer from heavy distortion in the anterior portion of the skull. IVPP 3254-2 is minimally distorted and suggests that the nasal process may have been slightly angled upward, similar to that seen in R. kaiseni. However, another skull, PIN 7130-3 has a more normal horizontal nasal process with a slightly downward bend to it. In conclusion, the nasal process was either horizontal or it was only slightly angled upward.

The nasal process is similar in length to that of the premaxillomaxillary rostrum in some specimens (e.g., IVPP V3254-1, and IVPP V3254-2), but in others (PIN 7130-3 and AMNH 20254) the nasal process clearly extends anterior to the rostrum. The most complete nasal processes (IVPP V3254-2 and PIN 7130-3) have distally tapering widths. The distal margin of the nasal process is flat or slightly rounded with a small median notch. From a strictly lateral view, the lateral wall of the nasal process appears to be relatively shallow. However, the view of the ventral side of the nasal process of PIN 7130-3 reveals that the lateral walls are strongly directed medially (fig. 85C). Proximally, the lateral walls become thicker, deeper, and they are strongly angled ventromedially, severely constricting the upper portion of the nasal cavity. The lateral walls of the nasal process are very shallow toward the distal end of the nasal and they do not continue around the anterior margin of the nasal as they do in Rhinotitan kaiseni.

The premaxillomaxillary rostrum is not clearly differentiated from Rhinotitan kaiseni. From the lateral view of the holotype (AMNH 20271), the rostrum is relatively shallow. At a point above the P3 the dorsal margin of the rostrum curves sharply upward, giving the dorsal surface of the rostrum a concave appearance. PIN 7131-3 has a similar appearance. On the other hand, the dorsal margins of the rostra of IVPP V3254-1 and IVPP V3254-2 are flatter and angled sharply posterodorsally. The premaxillary symphysis is completely ossified. Behind the symphysis, the dorsolateral margins of the rostrum diverge laterally and the rostral cavity is not sealed over by bone. Like in R. kaiseni, the nasal opening of the skull is very tall and narrow.

Interpretation of the shape of the dorsal surface of the skull of R. andrewsi is difficult. The dorsal surfaces of the holotype (AMNH 20271) and AMNH 20254 are nearly flat, but these specimens are unreliable due to the degree of plaster reconstruction. The dorsal surface of IVPP V3254-2 is barely concave, while that of PIN 7130-3 is somewhat more deeply concave. However, IVPP V3254-1 suggests a more bulbous cranium, with a somewhat convex postorbital dorsal surface similar to that of Metatitan. Like R. kaiseni, the dorsal surface of the skull broadens posteriorly and the parasagittal ridges do not greatly constrict the posterodorsal surface of the skull. (Note that the dorsal surface of PIN 7130-3 is artificially narrowed due to transverse buckling.) The zygomatic arches of R. andrewsi tend to be thicker and somewhat deeper than those of R. kaiseni, although in at least one specimen (AMNH 20261), the zygomatics are thin. From lateral views the entire zygomatic arch is straight and nearly horizontal. From a dorsal view the zygomatic arches are nearly straight are not strongly bowed laterally.

The occiput is nearly vertical like that of Rhinotitan kaiseni. The nuchal crest is slightly concave from a dorsal view. From the posterior view the nuchal crest is barely arched dorsally. Overall, the occiput is large, tall, and broad. The dorsal and ventral halves of the occiput are the about the same width, and the width of the occiput does not appear to have been constricted in the middle. The surface of the occiput has weak occipital pillars with a shallow pit in the center of the occiput.

The ventral surface of the holotype skull (AMNH 20271) is seen in fig. 86a. The anterior rim of the posterior nares is positioned slightly anterior to the M3 protocones. The posterior nares are rimmed by a very narrow emargination. In AMNH 20271 this emargination is partially covered by plaster, but it is completely exposed on AMNH 20254 (not shown) and slightly narrower that that of Rhinotitan kaiseni and more squarish (rather than horseshoe-shaped). The posterior narial canal does not extend into the sphenoid as it does, for instance, in Diplacodon elatus. The arrangement of the basicranial foramina is typical, with a widely separated foramen lacerum and foramen ovale. The mastoid process curves anteroventrally and contacts the postglenoid process forming a tube-shaped external auditory pseudomeatus. Like R. kaiseni, the external auditory pseudomeatus enters the skull in a strongly posteromedial direction.

Figure 86

Ventral view and upper dentition of Rhinotitan andrewsi. (A) Ventral view of the holotype skull (AMNH 20271), (B) left molars of AMNH 20254, (C) left premolars of AMNH 20254, (D) right premolars of the AMNH 20271 (holotype), (E) incisors and canines of the AMNH 20271 (holotype), (F) lingual view of left I2–C of AMNH 20263, (G) labial view of left I2–C of AMNH 20263.

One peculiar aspect of the basicranium of Rhinotitan andrewsi is its width. In specimens where the basicranium is complete and undistorted (AMNH 20271, AMNH 20254, IVPP V3254-2) the width of the basicranium (measured as the distance between the lateral sides of the mastoid processes) is consistently wider than the width measured between the lateral sides of the M3s. For instance, in the holotype (AMNH 20271), the distance across the mastoid processes (315 mm) exceeds the outside width of the M3s (260 mm). The condition seen in R. andrewsi is intermediate between most brontotheres, where the basicranium is narrower than the distance across the M3s, and Metatitan, where the basicranium is even more extremely widened. Nonetheless, R. andrewsi consistently differs in this regard from skulls of R. kaiseni where the outside width of the M3s is greater than the width of the basicranium.

Upper Dentition

Most of the specimens of Rhinotitan andrewsi have well-preserved and lightly worn teeth. In addition to the ventral surface of the holotype (AMNH 20271) (fig. 86a), pictured in close-up are the upper premolars and molars of AMNH 20254 (fig. 86b–c), the upper premolars, canines, and incisors of AMNH 20271 (fig. 86d–e), and the unworn incisors and canine of AMNH 20263 (fig. 86f–g).

The incisors are large and form a slightly arched row anterior to canines. Complete I1s are only seen with IVPP V2354-1 and IVPP V3254-2. The I1s of these specimens are worn, but they appear to be similar to the I2 although somewhat smaller. The I2 and I3 are intact in a number of specimens, including AMNH 20271. Another specimen, AMNH 20263, has incisors that are very similar to those of AMNH 20271, but they are essentially unworn. In these specimens, I2 is nearly a featureless crown, although in the least worn specimen the crown is subglobular in appearance with a very short cusplike apex. The I3 is slightly larger than the I2 and it is taller and with a more conical crown and a more distinct lingual cingulum. However, the incisors of AMNH 20271 and AMNH 20263 are more globular than those of Rhinotitan kaiseni. The size of the canines among specimens of R. andrewsi varies; they are relatively large in all specimens except AMNH 20261 (not shown). There are distinct precanine and postcanine diastemata in R. andrewsi. The postcanine diastema tends to be shorter than the P2.

The crown of P1 is nearly round in outline. There are two lingual cusps, a large paracone, and a smaller metacone that is positioned at the very posterior end of the P1 ectoloph. The lingual heel of P1 is well developed with a small protocone that is partially absorbed by a crest that runs along the lingual side of crown. P2–P4 are more rectangular; the anterior and posterior sides of these teeth are nearly parallel and the lingual sides are only slightly rounded. The parastyle and metastyle of P2 are straight and the metacone is not strongly shifted lingually, thus the ectoloph is nearly straight. On P3 and P4 the parastyles are strongly angled labially, while the metastyles are weakly angled labially. The outer swelling of the P2 paracone is minor, but there are small but distinct labial paracone ribs on P3 and P4. A mesostyle is not seen on any of the premolars of Rhinotitan andrewsi.

The lingual features of the premolar crowns have low relief. In the holotype, AMNH 20271, the lingual sides of the P2–P4 have two cusps, a protocone and hypocone that are connected by a lingual crest. A vestigial preprotocrista can also be seen on P2 and P3, but it is not detectable on P4. In P2, the protocone and hypocone are spaced apart, but these cusps are almost fully absorbed by the connecting crest. In P3, the cusps are also indistinct, but they are positioned more closely together. Finally, on the P4, the protocone and much smaller hypocone are widely separate and less strongly connected. Other specimens of Rhinotitan andrewsi show variable lingual premolar morphology. For instance, in AMNH 20261 (not show) there are distinct hypocones on P2–P4. On this specimen, the hypocones are only weakly connected to the protocone and the P3 hypocone is positioned more distantly behind the protocone. Yet in AMNH 20254 there are no distinct lingual cusps on P2 and P3; instead, there is a single lingual crest that arches along the lingual side of the crown. The P4 of AMNH 20254 has a protocone and a smaller hypocone that are distinctly separated. In AMNH 20263, the P2 and P3 are similar to those of AMNH 20254, but the P4 of that specimen has only a centrally positioned protocone with no hypocone or connecting crest. Labial premolar cingula are very weak; the lingual premolar cingula vary from continuous (e.g., AMNH 20271) to slightly discontinuous (e.g., AMNH 20254).

The upper molars of Rhinotitan andrewsi show numerous brontotheriine features, including tall, lingually angled ectolophs, very weak labial ribs, thin lingual ectoloph enamel, and wedge-shaped lingual sides of the paracone and metacone in molars that are not heavily worn. The anterior cingulum is thin and passes proximally to the distal peak of the parastyle. Distinct central molar fossae are present. However, anterolingual cingular cusps are not seen. No traces of paraconules or metalophs remain. One specimen, AMNH 20254, has a small hypocone on the M3, although other specimens show no trace of an M3 hypocone. Labial molar cingula are generally weak, and lingual molar cingula are very weak between the protocone and hypocone.

Mandible and Lower Dentition

The skulls, IVPP V3254-1 and IVPP V3254-2, are associated with mandibles. The mandible of IVPP V3245-1 is presently articulated with the skull of IVPP V3254-2 and is seen in fig. 84c. I was not able to find the mandible belonging to the skull of IVPP V3254-2, but a photo of it can be seen in Wang (1982: pl. 4). The horizontal ramus of IVPP V3254-1 is slender and deepens posteriorly, while that of IVPP V3254-2 is deeper overall. Additionally, the angle of the inferior margin of IVPP V3254-1 is very shallow (much less than 45°), while that of IVPP V3254-2 is angled about 45°. The symphyses of these mandibles extend to the trigonid of p4. The lower incisor row of IVPP V3254-1 is nearly intact and is not differentiated from Rhinotitan kaiseni. The incisors are large, tall, semispatulate, and form a slightly arched row anterior to the canines. The i2 is distinctly the largest incisor. Each of the incisors has a small lingual cingulid. The canines of that particular specimen are relatively large. The lower cheek teeth of R. andrewsi are not clearly differentiated from R. kaiseni. Both specimens lack evidence of a p3 metaconid. They fall within the morphological range of the specimens described below as Rhinotitan sp.

Rhinotitan sp.

Includes the nomen dubium, Rhinotitan mongoliensis (Osborn, 1923)

Referred Specimens

(From the Shara Murun Formation Ula Usu, Baron Sog Mesa, Shara Murun Region, Inner Mongolia) AMNH 18653 (holotype of Rhinotitan mongoliensis), a mandible fragment with right p2–m3; AMNH 20251, a left mandibular ramus with p2–m3; AMNH 20256, a palate with right C–M2, left P1–M2, M3 (unerupted), a mandible with right i1–i2, c (erupting), p1, p3–m2, m3 (erupting), left i1–i2, p1–m2, and m3 (erupting); AMNH 20262, a left mandibular ramus with p1–m2, m3 (erupting), four isolated incisors and a canine; AMNH 20269, a partial mandibular symphysis with right and left i1–i2, isolated canine and right p1; AMNH 20270, a right maxilla with P1–M2; AMNH 20272, mandible fragment with right i2?, c, p3–m3, and a left ramus with i2?, c, and p2–m3; AMNH 20273, right and left mandibular rami with p2–m3; AMNH 20280, a fragmented skull with right P2–P3, dp4, P4 (unerupted), M1, M2 unerupted, left P1–P3, dp4, P4 (unerupted), and M1; AMNH 21598, a palate with right and left P1–P3, dp4, P4 (unerupted), and M1–M2; (from the Shara Murun Formation, four miles north of Baron Sog Lamasery, Baron Sog Mesa, Shara Murun Region, Inner Mongolia) AMNH 21603, a partial mandible with right i2–p4 and left i1–m3; AMNH 21605, a partial mandible with poorly preserved incisors and right c, p2 (partial), and p3–m3; PIN 2198-2, a mandible with right and left p2–m3.

Description

Upper Dentition

Many specimens from the Shara Murun Formation with well-preserved upper and/or lower sets of cheek teeth clearly belong to a species of Rhinotitan, but because the two known species, R. kaiseni and R. andrewsi, do not have clearly differentiated cheek teeth, the majority of these specimens can be assigned only to the genus level (i.e., Rhinotitan sp.) Four palates with essentially unworn upper cheek teeth (AMNH 20256, AMNH 20270, AMNH 20280, AMNH 21598) demonstrate a degree of variability in the lingual features of the premolars (P2–P4) similar to both R. kaiseni and R. andrewsi. The lingual side of the P2 varies from having a single crest with no distinct cusps to having a slightly discernable protocone within the lingual crest. P3 varies in similar way, but occasionally there is a very small hypocone present. Finally, P4 has a large protocone, a very low crest is occasionally present, and a tiny hypocone is sometimes present.

Mandible and Lower Dentition

The mandibles referable to Rhinotitan sp. demonstrate variability in the presence of the p3 metaconid. In the majority of these specimens the p3 metaconid is absent. However, AMNH 18653 (holotype of Rhinotitan mongoliensis), AMNH 20251, and PIN 2198-2 exhibit small metaconids on the p3. The metaconid is smaller and not as lingually positioned as that of p4 or those of the molars. The canines of these mandibles are highly variable in size as well, with some specimens having large curved crowns and bulbous roots (AMNH 20269, 20272) and others with smaller less curved crowns and less bulbous roots (AMNH 21605).

Remarks

The first brontothere described from the Central Asiatic Expeditions of the American Museum of Natural History is Protitanotherium mongoliensis Osborn, 1923. Osborn (1923, 1925, 1929a) based this species on a partial lower cheektooth row (AMNH 18653). In 1925 Osborn named two more species. One of these, P. andrewsi, was based on a large skull (AMNH 20271) lacking the horns and nasal region. No mandible or lower teeth were associated with this skull but Osborn conjecturally assigned another mandible, AMNH 20251, as the paratype of P. andrewsi. Osborn then distinguished P. mongoliensis from P. andrewsi based on “progressive mutations and rectigradations warranting a specific separation” (Osborn, 1925: 6), but he did not indicate specifically what the distinguishing characters were.

The third species, Dolichorhinus kaiseni Osborn (1925), was based on an associated skull and mandible (AMNH 20252). Osborn assigned this species to Dolichorhinus because its elongate cranium and long nasal bones seemed similar to that of the North American genus Dolichorhinus. However, Osborn (1925) noted that it differed from North American Dolichorhinus primarily in its larger size and more distinct horns. Osborn (1925) did not compare D. kaiseni with Protitanotherium mongoliensis or P. andrewsi. The teeth of the holotype of D. kaiseni are so worn that any comparison to the dentition of other brontotheres is not possible. However, another specimen (FMNH P14048; formerly AMNH 20260) assigned to Dolichorhinus kaiseni by Osborn (1925) includes a nearly complete and minimally worn set of upper and lower teeth that Osborn could have compared to P. mongoliensis and P. andrewsi.

Granger and Gregory (1943) were the first to recognize the similarities among the numerous specimens that were referred Protitanotherium mongoliensis, R. andrewsi, and Dolichorhinus kaiseni. These were assigned to a new genus, Rhinotitan, to which R. kaiseni was considered the type species. Rhinotitan was characterized by Granger and Gregory (1943) as having long upturned nasals, small oval horns, a wide occiput, premolars with “tetartocone ridges”, and a metaconid variably present on p3. Granger and Gregory (1943) continued to accept all three species, but unfortunately they provided dubious species diagnoses. Rhinotitan kaiseni was diagnosed by its smaller dimensions and relatively wider P4. Even more dubious is their distinction of R. mongoliensis and R. andrewsi. The specific character of R. mongoliensis, “P2–P4 in neotype with tetartocones [hypocones] less distinct from the deuterocone [protocone] crest than in the type of R. andrewsi” (Granger and Gregory, 1943: 365) is based on a set of upper dentition (AMNH 20263) that they conjecturally referred to R. mongoliensis and considered to be the neotype.

Ironically, Granger and Gregory (1943) recognized that their specific distinctions of the three supposed species of Rhinotitan did not hold up when all the specimens in the AMNH collection were considered, and that the dentition of other specimens bridged the gaps between these species. Granger and Gregory (1943) considered the various holotypes, paratypes, and neotypes of the three species of Rhinotitan to represent arbitrary divisions between evolutionary stages for a continuous series represented by R. kaiseni–R. mongoliensis–R. andrewsi. Granger and Gregory (1943) did not consider the alternative possibility that the seemingly continuous variation, particularly that of the premolars, might represent intraspecific variation rather than progressive evolutionary stages. This alternative hypothesis is supported by the fact that nearly all of the Rhinotitan material is from the same locality and formation. Moreover, most other brontothere species exhibit similar levels of seemingly random intraspecific variation in the lingual morphology of the premolars.

Other attempts to revise the species-level taxonomy of Rhinotitan are problematic as well. Prior to Granger and Gregory's (1943) work, Takai (1939) recognized problems with Osborn's taxonomy and considered R. mongoliensis and R. andrewsi to represent opposite sexes of the same species. More recently Wang (1982) described two associated skulls and jaws (IVPP V3254-1 and IVPP V3254-2) and drew similar conclusions to those of Takai (1939). Wang (1982) considered R. andrewsi to be a junior synonym of R. mongoliensis, and considered the minor variations in the cheek teeth to represent individual variation. However, neither Takai (1939) nor Wang (1982) considered Rhinotitan kaiseni, although the cheek teeth of that species are indistinguishable from the other supposed Rhinotitan species.

My observations on the Rhinotitan material suggest two distinct species, R. kaiseni and R. andrewsi, that can be differentiated by the following characteristics: relative width of the basicranium, position of the frontonasal horns, shape of the nasal process, and the morphology of the upper incisors. Because the cheektooth morphologies of these species are not differentiated, much of the Rhinotitan material, consisting of partial jaws and palates, must be referred to Rhinotitan sp. For this reason, Rhinotitan mongoliensis (Osborn, 1923), whose holotype consists of a lower cheektooth row, must be considered a nomen dubium.

Nanotitanops shanghuangensis (Qi and Beard, 1996)

Holotype

IVPP V11032, a right premolar, most probably a P2 or P3.

Type Locality

IVPP locality 93006D, a fissure-filling located in the Shanghuang Limestone Quarry, near the village of Shanghuang, Liyang County, southern Jiangsu Province, China.

Age

Middle Eocene (Irdinmanhan or early Sharamurunian land mammal “age”).

Synonyms

Nanotitan Qi and Beard, 1996, not Nanotitan Sharov, 1968

Referred Specimens

(From the same locality as the holotype) IVPP V11002, a right P2 or P3; IVPP V11003, a left P2 or P3; IVPP V11004, a left P2 or P3; IVPP V11005, a left P1 or P2; IVPP V11006, a right P1; IVPP V11007, a left M1 or M2; IVPP V11008, a left M1 or M2; IVPP V11010, a left m1 or m2; IVPP V11011, a right m1 or m2; IVPP V11013, a right m1 or deciduous premolar; IVPP V11014: a right m1 or deciduous premolar.

Diagnosis

Nanotitanops shanghuangensis is the smallest known brontotheriid. It can be characterized as having a P1 with a weak metacone and small lingual heel, a distinct P2 metacone, weak premolar preprotocristae, and with short lingual crests extending posteriorly from the premolar protocones with an occasional premolar hypocone. The molars of N. shanghuangensis have tall, lingually angled ectolophs with weak labial ribs, and thinned lingual ectoloph enamel with wedge-shaped paracones and metacones. A cingular parastyle shelf is absent. Central molar fossae are present but anterolingual cingular cusps are absent. N. shanghuangensis molars retain vestigial paraconules, but all traces of metalophs are lost. The lower molars of N. shanghuangensis have shallow molar basins and weak lingual ribs.

Nanotitanops shanghuangensis can most easily be distinguished by its very small size. In addition to minuscule size, this species is the only brontothere to have a relatively complex P1 (with two labial cusps and a small lingual heel) but to retain vestigial paraconules in the upper molars.

Description

Upper and lower dentition

Nanotitanops shanghuangensis is known only from isolated teeth (fig. 87). In the original description, Qi and Beard (1996) identified all the specimens to specific dental homologues, but they did not explicitly state how or why these exact identifications were made. Upon reexamining the material, I am unconvinced that all of these isolated dental elements can be identified to their specific dental homologues. There are problems, in particular, with several of the original premolar identifications.

Figure 87

Specimens referred to Nanotitanops shanghuangensis by Qi and Beard, 1996. (A) IVPP V11032 (holotype), (B) IVPP V11003, (C) IVPP V11002, (D) IVPP V11004, (E) IVPP V11009 (does not belong to Nanotitanops shanghuangensis) (F) IVPP V11006, (G) IVPP V11005, (H) IVPP V11014, (I) IVPP V11013, (J) IVPP V11008, (K) IVPP V11011, (L) IVPP V11007, (M) IVPP V11010, (N) IVPP V11012. All photos from Qi and Beard (1996).

The original assignment of IVPP V11006 (fig. 87f) as a P1 seems accurate. It is significantly smaller than the premolars, IVPP V11004, IVPP V11032, IVPP V11003, and IVPP V11002, and it is semirounded in outline. There are two poorly differentiated labial cusps, there is no discernable labial rib on the paracone, the cingulum is thin, and the lingual shelf is very small with small protoconelike cusp and a short loph connecting the lingual cusp to the lingual base of the metacone.

IVPP V11005 (fig. 87g) was originally identified as a P2, although it might be a P1. The specimen is nearly the same size as IVPP V11006. Among brontotheres, P2s are usually much larger than P1s. IVPP V11005 differs from IVPP V11006 in some ways. The paracone and metacone are more distinct, and the small lingual shelf has a loph that arches around the anterolingual corner of the crown that connects to the lingual base of the paracone. This loph is absent in IVPP V11006. There is a short loph at the lingual base of the metacone. In IVPP V11006 this loph is continuous with the protocone, but in IVPP V11005 it is discontinuous from the protocone. The exact morphology of the lingual features of the premolars tend to vary in most brontothere species and it is therefore possible that morphological differences between IVPP V11005 and IVPP V11006 represent intraspecific variation of the P1.

Based on their larger size, labial paracone ribs, large lingual shelves with large protocone lingual crests, occasional small hypocones (IVPP V11032), rhomboidal or rectangular outlines, and thick cingula, IVPP V11032, V V11002, V11003, and V11004 most certainly represent premolar elements other than P1.

IVPP V11004 (fig. 87d) was originally identified as a P3 although this is questionable. A P4 can probably be ruled out because the parastyle arches lingually; in brontotheres, the P4 parastyle usually projects labially while the parastyles of P3 can be straight or slightly lingually angled. Therefore, this tooth could be either a P2 or P3. The size of the labial rib of the paracone suggests that a P2 is more likely, although the rectangular shape of the tooth is more consistent with a P3. The specific position of this tooth is basically indeterminate.

IVPP V11032, the holotype, was originally identified as a P4 (fig. 87a). This specimen has a very small hypocone that is placed distantly from the protocone, but is connected to it by the lingual crest. In species where a premolar hypocone is variably present (e.g., Epimanteoceras formosus), the occurrence of a hypocone on P2, P3, and P4 does not follow a clear pattern that might facilitate identification of the exact position of this tooth. However, the parastyle of IVPP V11032 arches slightly lingually and the labial paracone rib is wide, probably ruling out the P4. In comparison to IVPP V11004, this tooth is much more oblique, suggesting, rather, a P2.

IVPP V11002 (fig. 87c) was originally identified as a P4, but the lingually arched parastyle, oblique shape and wide labial paracone rib suggest a more anterior premolar. Finally, IVPP V11003 (fig. 87b), originally identified as a P3, could just as well be identified as a P2. To summarize, all of these premolars seem to represent P2s, P3s, or P4s, but none of them can be identified with any certainty.

IVPP 11007 and IVPP 11008 were originally identified as M1s, although they could just as easily represent M2s (fig. 87j, l); there are no obvious differences between the M1s and M2s of brontotheres other than relative size and overall degree of wear, but these differences are not helpful for identifying isolated teeth. The molars of Nanotitanops shanghuangensis exhibit increased ectoloph height, lingually arched labial walls, weak labial ribs, thin lingual ectoloph enamel, and wedge-shaped lingual sides of the paracone and metacone. The anterior cingulum is thin and passes proximally to the distal peak of the parastyle. N. shanghuangensis molars lack an anterolingual cingular cusp, although shallow central molar fossae are present. Each of these molars retains a vestigial paraconule, although there is no trace of a metaloph. Labial molar cingula are absent, although weak lingual molar cingula are present.

IVPP V11010 and IVPP V11011 were identified as m1s, although they could be m2s (fig. 87k, m). These lower molars exhibit morphology typical of advanced brontotheres with relatively shallow talonid and trigonid basins and very weak lingual ribs.

IVPP V11013 and IVPP V11014 were identified as a dp3 and dp4, respectively. These teeth are essentially molariform although IVPP V11014 is relatively narrower than the adult molars (fig. 87h). IVPP V11013, on the other hand, is more similar in proportions (fig. 87i). It is probable that these teeth are deciduous premolars, although there is no compelling reason to conclude that these teeth could not actually be adult molars of one or more smaller individuals.

Finally, IVPP V11009 was identified as a DP4, although this specimen is not from a brontotheriid (fig. 87e). There is no mesostyle, a feature seen in all brontothere upper molars and deciduous upper premolars (dp2–dp4) (Osborn, 1929a; Butler, 1952). The parastyle is much lower than the valley between the paracone and metacone, whereas in brontothere molars and deciduous premolars these two areas are the same height.

Remarks

Qi and Beard (1996) described Nanotitan shanghuangensis from numerous teeth from a single fissure quarry of what clearly represents the smallest known brontothere. Moreover, it is somewhat smaller than Lambdotherium, and with far more advanced dental morphology. The genus was later renamed Nanotitanops, because Nanotitan is preoccupied (Qi and Beard, 1998). Although the materials consist of isolated and unassociated teeth, their uniquely small size suggests that they can presently be referred to a single species. One of the specimens originally referred to this species (IVPP V11009) turns out not to represent a brontothere, but possibly some other perissodactyl.

Pollyosbornia altidens (Osborn 1908a) new genus

Holotype

AMNH 2025, a partial mandible with right p1–m3 and left c–m1.

Type Locality

Uinta Formation (Uinta C) of the Uinta Basin, Utah.

Age

Middle Eocene (Uintan land mammal “age”).

Etymology

“Polly” was the undergraduate nickname of paleontologist Henry Fairfield Osborn at Princeton, given to him because of his apparent “girlish appearance” (Scott, 1939).

Diagnosis

Pollyosbornia altidens is a large brontothere with an unreduced dental formula (3-1-4-3). The incisors are of moderate or large size and form an arched incisor row. The symphysis is relatively elongate and extends posteriorly to the p3 talonid. Other features include a relatively long postcanine diastema, a metaconid on p3 and p4 but not on p2, a very poorly developed p2 talonid, a p2 trigonid that is at least twice the length of the talonid, an elongate m3, and shallow trigonid and talonid molar basins.

From what can be discerned of the poorly preserved holotype mandible, Pollyosbornia altidens is more or less consistent with Diplacodon elatus except for its larger size, and more poorly developed p2 talonid.

Description

Mandible and lower dentition

The holotype and only known specimen of Pollyosbornia altidens (AMNH 2025) is nearly complete although it is poorly preserved (fig. 88). In particular, the symphysis appears to have been crushed transversely and a significant portion of the left ramus has been reconstructed. Therefore, some aspects of the shape of the jaw are unreliable. The incisors are missing, but the preserved alveoli indicate 3 pairs of incisors. Judging by the size of the alveoli, the incisors were large to moderate in size, but they were not reduced to a vestigial state. The alveolar surface of the incisor row arches anterior to the canines.

Figure 88

Holotype of Pollyosbornia altidens (AMNH 2025). (A) Right view, (B) dorsal view, (C) right premolars.

The mandibular symphysis is relatively long and the postcanine diastema is substantially longer than the p2, although the elongate nature of the symphysis could be exaggerated by lateral crushing. The symphysis extends to the talonid of the p3. The p1 has a simple single-rooted cusp with a short talonid heel. In comparison to those of p3 and p4, the trigonid of p2 is proportionately large, while the talonid is very small. The trigonid of p2 is more than twice as long as the talonid. On the p3, the trigonid is moderately longer than the talonid, while the p4 trigonid and p4 talonid are of similar length. The trigonid of p2 is slightly wider than the talonid, but on the p3 and p4, the trigonids are slightly narrower than the talonids. The paralophid of p2 is not angled lingually and there is no lingual trigonid notch on that tooth. The p2 protolophid is only barely angled lingually. The paralophids and protolophids of p3 and p4 strongly arch lingually. The lingual trigonid notch of p3 is long and narrow, while the trigonid of p4 is essentially molariform. The p2 lacks a distinct metaconid but large, lingually positioned metaconids are present on p3 and p4. The talonid of p2 is small and narrow with a poorly developed cristid obliqua and hypolophid and a very small lingual-talonid notch. However, the talonids of p3 and p4 are large with long cristids obliqua and hypolophids with molariform basins. The premolars lack lingual cingulids. Labial cingulids occur only in the notch between the protoconid and hypoconid of the p3 and p4. The molars of AMNH 2025 are indistinctive; the occlusal basins are shallow, and the m3 is elongate, lingual molar cingulids are absent, and labial cingulids are present, but they are discontinuous around the protoconid and hypoconid.

Remarks

Osborn's (1908a) original description of the only known specimen of “Telmatherium?” altidens (AMNH 2025) consists of three brief sentences that do not clearly justify a new species, nor do they justify his questionable assignment of this species to the genus Telmatherium. Later, Osborn (1929a) provided a more extensive description and included justifications for considering this species a member of Telmatherium (which, in Osborn's concept also included Metatelmatherium ultimum). These justifications include the large canine, the large lower incisors, the long postcanine diastema, the state of molarization of p2, and the size of the molars. However, all these arguments are refutable. For instance, the size of the canine tends to vary intraspecifically. Canine size is certainly not a diagnostic character of Telmatherium or Metatelmatherium. Moreover, the size of the only preserved canine of AMNH 2025 (left) is highly exaggerated by severe expanding-matrix distortion (sensu White, 2003). None of the incisors were recovered. The incisor alveoli are distorted to varying degrees, prohibiting a precise interpretation of the size of the incisors. The postcanine diastema of AMNH 2025 is actually longer than that of Telmatherium or Metatelmatherium. The p2 of AMNH 2025 does not appear to resemble Telmatherium or Metatelmatherium any more closely than it resembles a number of other brontotheres (e.g., Mesatirhinus junius). Finally, the molars of AMNH 2025 are much larger than Telmatherium and Metatelmatherium molars.

Judging by its size, long postcanine diastema, and large p3 metaconid, this specimen does not belong to Telmatherium or Metatelmatherium. Rather, AMNH 2025 is closer to horned brontotheres such as Protitanotherium emarginatum or Diplacodon elatus, although this mandible does not seem referable to these species either.

The premolars of AMNH 2025 exhibit a peculiar set of traits not shared by Protitanotherium or Diplacodon. P. emarginatum is most easily ruled out by its lack of a large p3 metaconid. On the other hand, D. elatus possesses a large p3 metaconid, as does AMNH 2025. The degree of molarization of the p3 and p4 of AMNH 2025 is similar to D. elatus, but the p2 is much less molariform. For instance, the paralophid is short and straight, the talonid is very small, and the cristid obliqua and hypolophid are poorly developed. Additionally, AMNH 2025 is larger than known specimens of D. elatus. The large size, relatively molariform p3 and p4, in combination with the relatively primitive p2, suggest that this mandible represents a unique species and undoubtedly belongs to a genus other than Telmatherium. Despite the poor nature of the holotype, and the lack of other referable specimens, there appears to be enough evidence to conclude that the species is valid, although I have removed it from Telmatherium and placed it in a new genus, Pollyosbornia. The long postcanine diastema, long symphysis, and relatively deep symphysis resemble Gnathotitan berkeyi to a degree, but it is not clear to what extent these traits are real or products of distortion.

Pygmaetitan panxianensis Miao, 1982

Holotype

IVPP V6521, a left upper tooth row with I3–M3.

Type Locality

Shinao Formation, Shinao Basin, Panxian County, Guizhou Province, China.

Age

Late Eocene? ( =  early Oligocene sensu Russell and Zhai [1987]).

Referred Specimens

(All from the same locality as the holotype) IVPP V6522, a left maxillary fragment with P4–M3; IVPP V6523, left P2 and a mandible fragment with right i3, m1 (partial), m2, m3 (unerupted), left i1–i3, and other incisor or canine fragments; IVPP V6524, a left mandible fragment with m2 (partial) and m3; IVPP V6525, a partial juvenile mandible with right dc?, dp2–dp4, m1 (unerupted), left di2?, di3?, dp2, dp3, and dp4 (partial); IVPP V6526-1, a fragment of a left juvenile maxilla with DC?, P1 (unerupted), DP2, DP3, and DP4 (partial); IVPP V6526-2, a left mandible fragment with dc (root only), p1 (unerupted), and dp3–dp4; IVPP V6527, a left maxillary and jugal fragment with DP3 and DP4; IVPP V6528, a right maxilla fragment with dp2–dp4 (all damaged); IVPP V6529, a left mandible fragment with dp2 and dp3 (partial).

Diagnosis

Pygmaetitan panxianensis is intermediate in size between Nanotitanops shanghuangensis and Microtitan mongoliensis. Dentally, Pygmaetitan panxianensis has an enlarged and subcaniniform incisor, a diminished postcanine diastema, a distinct P1 metacone, a distinct P2 metacone, and weak premolar preprotocristae. Hypocones are not seen on P2, P3, or P4. The molars of Pygmaetitan panxianensis have tall, lingually angled ectolophs with weak labial ribs, and thinned lingual ectoloph enamel with wedge-shaped paracones and metacones. A cingular parastyle shelf is absent. Central molar fossae and anterolingual cingular cusps are present, while paraconules and metalophs are absent. The upper molar mesostyles are labially expanded. The lower molars have deep valleylike basins and the m3 is elongate.

Pygmaetitan panxianensis shares with Sthenodectes incisivum the unique combination of an enlarged I3, diminished postcanine diastema, and no molar paraconules. However, P. panxianensis is clearly much smaller than S. incisivum and has a more complex P1 and taller lower molar crowns.

Description

Upper Dentition

The following description of the adult upper dentition of Pygmaetitan panxianensis is based on the holotype specimen, IVPP V6521 (fig. 89a), with additional observations made on IVPP V6522 (now shown), IVPP V6523 (fig. 89b), IVPP V6526-1 (fig. 92 a, b), and IVPP V6527 (fig. 92c). The holotype, consisting of a left maxillary fragment with a complete cheek-tooth row, represents a small but remarkably advanced brontothere. The maxillary fragment itself is too poorly preserved to describe any characteristics of the skull. The teeth are only moderately worn, thus facilitating a reasonably complete description of the upper dentition; however, most of the dental surface is corroded and pitted, thus obscuring some of the finer details.

Figure 89

Upper adult dentition of Pygmaetitan panxianensis. (A) The holotype (IVPP V6521), and (B) an additional left P2 (IVPP V6523).

Figure 92

Deciduous upper dentitions of Pygmaetitan panxianensis. (A) Left view of IVPP V6526-1, (B) ventral view of IVPP V6526-1, (C) ventral view of IVPP V6527.

The only preserved incisor on the holotype, presumably an I3, is greatly enlarged. This incisor is attached to a small fragment of bone that has been glued onto the main maxilla fragment, although these two fragments do not appear to occlude precisely. It is therefore not possible to definitively determine whether the disproportionately large incisor actually belongs to the same individual. However, this large upper incisor is consistent in size with the large lower incisors that characterize this species (see below). Pygmaetitan panxianensis is one of two brontotheres with greatly enlarged incisors. The other example is Sthenodectes incisivum. This upper incisor is plesiomorphic in other respects; the crown is subcaniniform with a strong lingual cingulum. It has developed a broad lingual wear facet the spans the entire height of the crown.

The upper canine is very large as well, with a diameter that exceeds that of the incisor by about a millimeter. The crown of the canine is either worn away or broken off. The specimen lacks a diastema between the incisor and canine, although this could be an artifact of the loss of the bony fragments that would have been positioned between the larger maxillary fragment and the smaller fragment holding the incisor. A minor gap of a few millimeters separates the canine from the P1.

The P1 of IVPP V6521 is severely worn. Discernable topographical features include a paracone, a straight parastyle, a part of a small lingual heel with a small protoconelike cusp (worn flat), and a small ridge situated anterolingually from the paracone. The posterior portion of the crown is broken off. An intact and unerupted P1 is preserved with IVPP V6526-1 (fig. 92a, b). The P1 of that specimen exhibits two distinct labial cusps, a large paracone and a somewhat smaller metacone. The parastyle is bent slightly lingually while the metastyle is essentially straight. The P1 does not show a well-developed lingual heel, although a thick cingulum extends along the entire lingual margin of the crown. A sharp ridge ascends the lingual side of the metacone and joins the lingual cingulum, forming a small triangular peak. A similar ridge climbs the lingual side of the paracone, although it is less distinct and forms a smaller peak at its junction with the cingulum.

The P2, P3, and P4 are successively larger. These premolars are strongly rectangular in outline with parallel anterior and posterior margins. The P2 of the holotype is incomplete; the labial sheet of enamel has been lost. However, IVPP V6523 includes a complete and unworn P2 (fig. 89b). This specimen exhibits a prominent labial paracone rib, and a much smaller labial metacone rib. The parastyle is straight. The metacone is fully developed, but it is positioned close to the paracone and is strongly shifted lingually. Finally, the metastyle is straight, but strongly angled lingually. These features give the ectoloph a rather arched or rounded appearance. The labial sheet of enamel has been lost on the P3 of the holotype. The P4 of the holotype has suffered breakage at the midline of the tooth and its width is probably exaggerated; it may have had a small labial rib, but damage to the surface of the enamel and a minor amount of extraneously concreted material obscures it. The ectolophs of the P3 and P4 are flexed slightly labially in the middle, giving them a sort of rudimentary mesostyle. This mesostyle is more apparent in the P4. The ectoloph of P3 is straight, while that of P4 is incipiently W-shaped due to an anterolabially angled parastyle in addition to the more distinct mesostyle. The lingual heels of P2, P3, and P4 are broad, although only a single lingual cusp (protocone) can be found on each premolar. The protocones of P2 and P3 are tall, slightly anteriorly positioned, and are followed by short lingual crests that arch posterolabially. On P2 this crest ends at the posterior cingulum. Additionally a faint preprotocrista can be seen on the P2, connecting the anterolabial base of the protocone to the lingual base of the paracone. The P2 of IVPP V6523 lacks a strong lingual crest, although there is a very small pinpoint of enamel in the hypocone position. The P3 of the holotype has a small but distinct lingual crest that ends at the junction of the metacone and posterior cingulum where there is a distinct swelling that could be interpreted either as a metaconule, or a labially positioned hypocone. The preprotocrista of P3 is exceedingly faint. A lingual crest can be seen on P4, although it is indistinct. The P4 lingual crest is more distinct on IVPP V6522, although it is much less developed than the lingual crest of the P3 of IVPP V6521. Given the rudimentary nature of the lingual crest, and the hypoconelike and metaconule-like structures on the lingual sides of the premolars, it is probable that these characters were intraspecifically variable, like many other brontotheres.

The labial cingula of P3 and P4 are absent or they were excessively weak. A strong and continuous cingulum stretches around the anterior, lingual, and posterior margins of P2, P3, and P4. On P3 and P4 the anterior cingulum has been partially worn away by interstitial wear.

The left M1 of IVPP V6521 is heavily worn and incomplete; fragments of the posterior side are missing and the tooth has experienced severe interstitial wear, thus distorting its length/width proportion. The M2 is heavily worn as well, although the specimen is complete and has suffered a lesser amount of interstitial wear. The M3 is complete, only lightly worn and most clearly exhibits a full suite of molar characters. Advanced molar characters, best seen in the moderately worn M3, include a narrow anterolabial cingulum that passes below the apex of the parastyle, a relatively tall and lingually angled ectoloph, thin lingual ectoloph enamel, and wedge-shaped lingual margins of the paracone and metacone. There are no discernable labial paracone or metacone ribs, although weak ribs might have been present. Other advanced features include well-developed anterolingual cingular cusps and a round central molar fossa. The molars lack any trace of vestigial paraconules. The mesostyle of M3 is labially expanded. M1 and M2 probably had similarly labially expanded mesostyles, although this structure has been more dramatically worn in these teeth. A small metalophlike ridge can be seen along the anterolabial slope of the hypocones of M1 and M2. The M3 hypocone is well developed but smaller than that of the M2, and it lacks the metalophlike ridge. Labial and lingual molar cingula are thin and discontinuous around the margins of the main cusps.

Lower dentition

IVPP V6523 includes a number of isolated lower incisors, including a complete left incisor row (i1–13) (fig. 90). The lower incisor crowns are rather large. They are all about the same size, although they differ in shape. The i1 and i2 have a semispatulate shape with rounded apices. The i3, on the other hand, is more subcaniniform with a pointed apex. From a lingual view, the i1 and i2 crowns are slightly asymmetrical, while the i3 is mesiodistally more elongate and more severely asymmetrical. Thin lingual cingulids are present in all three lower incisors. Labial cingulids are absent.

Figure 90

Lower left incisors of Pygmaetitan panxianensis (IVPP 6523). (A) Labial view, (B) lingual view.

IVPP V6524 includes a partial m2 and a complete m3 (fig. 91). IVPP V6525 contains an exposed but unerupted m1 (fig. 93). The lower molars of Pygmaetitan panxianensis have relatively thin enamel. The talonid and trigonid basins are relatively deep and valley-, or V-, shaped. Weak lingual protoconid and hypoconid ribs are often seen in the lower molars of brontotheres, but they are not evident in these specimens. The m3 is elongate. Labial molar cingulids are exceedingly faint, but this cingulid can be seen wrapping around the posterior margin of the hypoconulid. Lingual cingulids are absent.

Figure 91

A mandible fragment with a partial m2 and complete m3 referred to Pygmaetitan panxianensis (IVPP V6524). (A) Dorsal view, (B), left (labial) view, (C) medial (lingual) view.

Figure 93

Juvenile mandibles and deciduous lower dentition referred to Pygmaetitan panxianensis. (A) Dorsal view of IVPP V6525, (B) right view of IVPP V6525, (C) occlusal view of deciduous dentition of IVPP V6526-2, (D) medial (lingual) view of IVPP V6526-2, (E) right (labial) view of IVPP V6526-2.

Deciduous upper dentition

Most of the known specimens of Pygmaetitan panxianensis are juveniles. The best juvenile specimen with upper deciduous dentition is IVPP V6526-1 (fig. 92a, b). The anteriormost tooth of IVPP V6525-1 is probably a deciduous canine. The unerupted tooth just behind it is a P1 and is described above with the adult dentition. The remaining teeth are DP2, DP3, and a portion of DP4. An additional specimen, IVPP V6527 (fig. 92c), has a complete DP4. The deciduous premolars are more molarized than the adult premolars. The deciduous premolars show a trend of being progressively “derived” posteriorly. In other words, the anterior deciduous premolars are the least molariform and retain a number of plesiomorphic states. Moving posteriorly, the deciduous premolars increasingly show a number of apomorphic conditions that resemble the adult molars.

All of the deciduous premolars share with the molars the following conditions: relatively tall ectolophs, thin enamel on the lingual band of the ectoloph, angled lingual margins of the paracone and metacone, and relatively weak labial ribs. The DP2 is more asymmetrical than typical adult molars. The anterior portion of the ectoloph is more elongate than the posterior portion. The DP3 and DP4, on the other hand, are fully molariform in terms of their overall shape. The mesostyle of the DP2 is small and angled posterolabially. The parastyle is less strongly deflected lingually. Because of the smaller mesostyle and less labially directed parastyle, the labial margins of the cusps are not as deeply concave as those of the adult molars. The DP3 and DP4 mesostyles are well developed; however, the DP3 and DP4 mesostyles are less labially expanded that those of the adult molars. A small but distinct preprotocrista is present on the DP2, although a distinct paraconule is not embedded in this crest. DP3 lacks a preprotocrista, although it retains a small paraconule. DP4, like the adult molars, entirely lacks a preprotocrista and paraconule. A deep central molar fossa is not seen on DP2, although, like the molars, DP3 and DP4 express this condition. The lingual cingulum of DP2 climbs to the peak of the parastyle; this character resembles the most basal brontotheres such as Eotitanops and Palaeosyops, although the cingulum does not form a thickened parastylar shelf as seen in these taxa. In adult molars of more derived brontotheres, the anterior cingulum wraps around the anterolabial margin of the tooth in a position proximal to the peak of the parastyle; the DP3 and DP4 of Pygmaetitan panxianensis show this condition. DP4 shares with the adult molars a strong anterolingual cingular cusp, although this structure is absent on DP2 as well as DP3. Finally, the lingual cingulum of DP2 is thick and continuous with the anterior and posterior cingula; the lingual cingula of DP3 and DP4 are very weak and more closely match the adult condition.

Juvenile mandible and deciduous lower dentition

IVPP V6525 is a nearly complete juvenile mandible with deciduous dentition (fig. 93a, b). IVPP V6526-2 contains an additional set of deciduous premolars that are somewhat better preserved (fig. 92c, d, e). The mandibular symphysis of IVPP V6525 is short and very broad and the inferior angle of the symphysis is nearly vertical. Although there were undoubtedly ontogenetic changes in the overall shape of the mandible, it is likely that the adult mandible possessed similar peculiarities, even if not as extreme. Unfortunately, the crowns of the right di2, di3, and left dc are not preserved. A dp1 or p1 is not present in IVPP V6525, and there is a short gap between the dc and dp2. This gap is filled by an erupting P1 in IVPP V6526-2, thus indicating the absence of a postcanine diastema, at least at this ontogenetic stage. The morphology of the dp2 generally resembles a typical adult brontothere p2, although a large metaconid is present; metaconids are not typically found in the adult p2s of brontotheres. The dp2 metaconid is situated very close to the protoconid and is positioned posteriorly and slightly lingually from it. The dp2 contrasts with the molariform condition in the following ways: the trigonid is relatively longer, the paralophid and protolophid are straight rather than lingually arched, the protoconid is positioned medially rather than labially, and the talonid of dp2 is essentially molariform, with lingually arching cristid obliqua and hypolophid forming a U-shaped crest that is remarkably deep in comparison to typical brontothere adult premolars or molars. The dp3 and dp4 contrast with typical adult premolars in their possession of a pronounced entoconid, and they are essentially molariform, although they have somewhat more elongate proportions than typical adult molars, and also have remarkably deep U-shaped crests. As noted above, the lower adult molars are remarkably tall in comparison to typical brontotheres, with deep valley-shaped trigonid and talonid basins. The lateral views of IVPP V6525 and IVPP V6526-2 reveal that the deciduous premolars increase in crown height posteriorly. In IVPP V6525, the dp4 approaches the remarkable crown height of the unerupted m1.

Remarks

Pygmaetitan panxianensis Miao, 1982, is based on a maxillary fragment with a nearly complete set of adult upper dentition. Miao (1982) noted that P. panxianensis is easily differentiated from small North American brontotheres, and except for its enlarged incisors and canine, is similar in many ways to derived brontotheres from the late Eocene. Miao (1982) briefly compared P. panxianensis with Eotitanops and Microtitan and more detailed comparisons with small Asian and North American brontotheres are made here.

Pygmaetitan panxianensis is substantially larger than the minuscule Nanotitanops shanghuangensis, with cheek-tooth dimensions that are more than double. Additionally, N. shanghuangensis molars retain small paraconules, whereas P. panxianensis molars do not. The cheek teeth dimensions of the holotype specimen of P. panxianensis just slightly exceed the largest specimens of Eotitanops, and are similar in size to the smallest Palaeosyops specimens. Nonetheless, the molars of Pygmaetitan panxianensis are clearly more advanced than Eotitanops and Palaeosyops, and clearly establish it as a member of the Brontotheriinae. Other small brontotheres, Microtitan mongoliensis, Mesatirhinus junius, Metarhinus fluviatilis, Metarhinus abbotti, and Fossendorhinus diploconus exceed Pygmaetitan in size by a substantial margin. Moreover, none of these species has extremely enlarged incisors or a P1 metacone, nor do they lack a postcanine diastema. Size comparison with Acrotitan ulanshirehensis is complicated by the lack of comparable dental material, although Acrotitan seems somewhat larger. However, the narrow mandibular symphysis, small incisor alveoli, and long p1–p2 diastema of Acrotitan ulanshirehensis are obviously out of character with the enlarged upper incisor and reduced upper postcanine diastema of Pygmaetitan panxianensis.

The extremely enlarged incisor of Pygmaetitan panxianensis is a characteristic shared with Sthenodectes incisivum, a much larger North American hornless Uintan brontothere. The deep, valley-shaped trigonid and talonid basins seen in the adult lower molars closely resemble the gigantic late Eocene Asian brontothere, Embolotherium. The very labially expanded mesostyle that is best seen in the M3 of IVPP V6521 is perhaps an autapomorphic condition, although it is difficult to compare mesostyle proportions to species that are only known from specimens with heavily worn teeth.

Many juvenile specimens were found with adult specimens of Pygmaetitan panxianensis, and were assigned to this species by Miao (1982). As Butler (1952) noted, the upper deciduous premolars of brontotheres tend to be molariform, although the more anterior deciduous premolars tend to lag behind in the accumulation of typical molar apomorphies. This appears to be the case in Pygmaetitan. In terms of adult molar characters, the DP2 retains the greatest number of plesiomorphic character states; the DP3 has an intermediate number of plesiomorphic states, and the DP4 has the least. Despite the more “primitive” condition of the deciduous dentition in comparison to the adult molars, the deciduous premolars of Pygmaetitan are quite molariform in comparison to other brontothere species where the DP2 is substantially more elongated (Butler, 1952). However, because of the lack of comparable material for most other brontothere species, the phylogenetic or taxonomic value of the deciduous teeth of Pygmaetitan panxianensis is uncertain. Since Butler's (1952) analysis of the deciduous dentition of perissodactyls, deciduous dental material has been collected for several more brontothere species, although at present there are no descriptions of this material.

Gnathotitan berkeyi (Osborn, 1925)

Lectotype

AMNH 20106, a mandible with right c–m3, left c, and p2–m3.

Type Locality

Telegraph Line Camp, Irdin Manha Formation, Inner Mongolia, China.

Age

Middle Eocene (Irdinmanhan land mammal “age”).

Referred Specimens

(From the Irdin Manha Formation, Inner Mongolia) AMNH 20107, a partial right mandible with i1–m1, m2 (partial); AMNH 20115, a partial juvenile mandible with unerupted incisor, p1, dp2–dp4 (all damaged), and m1; AMNH 20121, a left maxilla fragment with canine alveolus and P1–M3; AMNH 141231 (formerly part of AMNH 20106), a right maxillary fragment with C–M3.

Two other specimens, AMNH 20124 (a mandible) and AMNH 20127 (a palate with premolars and molars), were referred to Gnathotitan berkeyi by Osborn (1925). Both of these specimens were sent to the Chinese Geological Survey in 1928 and are apparently lost.

Diagnosis

Gnathotitan berkeyi is a large brontothere with a semicomplex P1 (indistinct metacone and small lingual heel) and a distinct P2 metacone. Lingual crests extending from the protocones of the P2–P4 are always present. A premolar hypocone is occasionally present although it is never well separated from the protocone. The molars of G. berkeyi are elongate and have tall, lingually angled ectolophs with weak labial ribs and thinned lingual ectoloph enamel with wedge-shaped paracones and metacones. A cingular parastyle shelf is absent. Distinct central molar fossae and weakly developed anterolingual cingular cusps are present. Paraconules and metalophs are absent. The lower dentition of G. berkeyi includes three large subcaniniform incisors. There is a distinct postcanine diastema, a p2 trigonid that is more than twice as long as the talonid, a small metaconid on p3, and a large metaconid on p4. The lower molars have shallow basins and the m3 is slender and very elongate.

Gnathotitan berkeyi can be distinguished from all other brontotheres by its disproportionately large and deep mandible, which exceeds the mandibles of all other brontotheres in depth and size.

Description

Skull

The skull of Gnathotitan berkeyi is known only from two maxillary fragments, AMNH 20121 (fig. 94) and AMNH 141231 (fig. 95). The lower portion of the orbit is seen in AMNH 20121 and suggests that the anterior rim of the orbit was positioned over the posterior portion of M2. The dorsal margin of the maxilla is intact to a point between the P1 and P2 and indicates that the nasal incision extended at least to this point, but the true anteroposterior length of the nasal incision is unclear.

Figure 94

A maxilla referred to Gnathotitan berkeyi (AMNH 20121). (A) Left (labial) view, (B) left molars, (C) left premolars.

Figure 95

A maxilla referred to Gnathotitan berkeyi (AMNH 141231, formerly part of AMNH 20106). (A) Right (labial) view, (B) ventral view, (C) right premolars.

Upper Dentition

The upper incisors of Gnathotitan berkeyi are unknown, but the cheek-tooth rows are complete in both maxillae (figs. 94, 95). The canine of AMNH 20121 is not preserved, but the alveolus indicates a canine of very small diameter. In contrast, the canine of AMNH 141231 is large with a bulbous root. A postcanine diastema follows the canines of both specimens.

The P1 is a small ovoid tooth that consists of a large paracone, a much shorter and indistinct metacone, and a very small lingual heel with a small protocone. A short preprotocrista connects the protocone to the paracone. The size of the lingual heel is variable. On AMNH 20121 the P2 is more oblique than P3 or P4 due to the posterolingually angled anterior margin and the lingually shifted metacone. The parastyle of P2 is nearly straight, while the metastyle is angled lingually. The P3 and P4 are more rectangular because of nearly parallel anterior and posterior sides and a nearly flat lingual side. Likewise, the metacones are not as strongly shifted lingually in comparison to P2. The parastyles of P3 and P4 are directed somewhat labially, while the metastyles are nearly straight. The paracone of P2 is swelled buccally, while P3 and P4 have small but well-defined labial paracone ribs.

The lingual features of the P2–P4 have relatively high relief. On each of the molars, a very low and short preprotocrista can be seen; it is strongest on the P2 and smallest on the P4. A long lingual crest extends posteriorly from the protocone of P2. On the P3 of AMNH 20121 a tall but short lingual crest connects the protocone to a large hypoconelike swelling. The P3 is similar to P4 except that the protocone and hypocone are even more closely positioned, so that they form a single large ovoid cusp with a short lingual crest at its apex. The lingual premolar morphology of AMNH 141231 differs from that of AMNH 20121. In that specimen, the lingual sides of the P2, P3, and P4 have a single lingual cusp (protocone) and a long lingual crest extending posteriorly from the protocone. The labial premolar cingula are very weak. In AMNH 20121 the anterior and posterior premolar cingula do not join lingually. In AMNH 141231 there is a continuous lingual cingulum on P3 and P4 but not on P2.

The upper molars of Gnathotitan berkeyi are notably elongate. This is most evident in M2 and M3. M1 appears to have been shortened somewhat by interstitial wear. The molars show typical brontotheriine traits including tall, lingually angled ectolophs, very weak labial ribs, thin lingual ectoloph enamel, and wedge-shaped lingual sides of the paracone and metacone in molars that are not heavily worn. The anterior cingulum is thin and passes proximally to the distal peak of the parastyle. Shallow central molar fossae are seen on each of the molars. On the M1, there is a wear facet on the anterolingual cingulum; on M2 and M3 small but distinct anterolingual cingular peaks are evident. All evidence of paraconules and metalophs is lost. The M3 hypocone varies. In AMNH 20121, the M3 hypocone is nearly as large as those of the more anterior molars and is separated from the protocone by a deep valley. However, in AMNH 141231, there are two very small hypoconelike points of enamel. Labial molar cingula are weak, and lingual molar cingula are absent.

Mandible and Lower Dentition

The mandible and lower dentition of Gnathotitan berkeyi are known from the lectotype (AMNH 20106), which consists of a nearly complete mandible that is missing its incisors (fig. 96). (The left p1 is mistakenly glued onto the mandible as an incisor.) The lectotype is laterally crushed, thus its proportions are distorted, but from a lateral view the shape of the mandible seems reasonably intact. Additionally, a partial right ramus with well-preserved incisors, premolars, and a symphysis offer more details (fig. 97).

Figure 96

The lectotype mandible of Gnathotitan berkeyi (AMNH 20106). (A) Right view, (B) dorsal view, (C) left p2–p4, (D) left view.

Figure 97

A partial mandible referred to Gnathotitan berkeyi (AMNH 20107). (A) Right view, (B) medial view of right side, (C) right premolars, (D) lingual view of right incisors, (E) labial view of right incisors.

At 805 mm in length, AMNH 20106 is the largest known brontothere mandible. (The second largest brontothere mandible I have ever encountered (724 mm in length) is AMNH 1051, a specimen of Megacerops sp. (sensu Mihlbachler et al., 2004b). Despite the profoundly large mandible, Gnathotitan berkeyi was probably not the largest brontothere in terms of body mass. For instance, the length of the cheek-tooth row is exceeded by other large brontotheres (e.g., Aktautitan). In addition to its excessive size, the mandible is unique in the great depth of the horizontal ramus. Another peculiar aspect of its shape is the distinct bulge on the inferior margin of the horizontal ramus. The bulge is present on both sides, but it is somewhat asymmetrical. On the right side it is below m2 and m3, while on the left side it is below m3 and somewhat posterior to m3. The peculiar shape of the inferior margin was considered by Osborn (1925, 1929a) to be a species-level character. However, the bulge could be a neoplasia or tumor paleopathology (Bruce Rothschild, personal commun., 2003), although a thorough paleopathological analysis has not been undertaken. In addition to the large bulge in the center of the ramus, the mandible bulges directly below the symphysis. This second characteristic is repeated in AMNH 20107, suggesting that it is probably not related to a paleopathology.

Osborn (1925) described the coronoid process of AMNH 20106 as wide, though there is so much plaster involved in the reconstruction of that part of the mandible that this observation is doubtful. The symphysis of AMNH 20106 is long and its inferior margin is steeply angled slightly more than 45°. This great length and steep orientation of the symphysis is clearly related to the extreme depth of the ramus. However, lateral crushing has clearly exaggerated the narrow-elongate shape of the symphysis of the lectotype. The cross-sectional shape of the symphysis can be seen from the medial view of AMNH 20107. The symphysis extends to the talonid of the p3.

Although no incisors are preserved with AMNH 20106, there are clearly six alveoli that form an arched row anterior to the canines. A complete right lower incisor row is preserved in AMNH 20107. The incisors are large and positioned closely together with no diastemata. Each crown is composed of a dulled point with narrow mesial and distal margins. The i2 is the largest incisor, though barely so. It exceeds the other incisors in size primarily in crown height. The i3 is slightly more mesiodistally elongate than the other incisors, but otherwise it is similar to i1 and i2. Each incisor has a distinct lingual cingulid. There are no labial incisor cingulids in Gnathotitan berkeyi.

The canine of AMNH 20106 is very small, while the canine of AMNH 20107 is much larger. Likewise, the postcanine diastema of AMNH 20107 appears to be shorter than that of the lectotype.

The left p1 is a small narrow tooth with a single cusp and a small talonid heel. The talonid of p2 is more than twice the length of the talonid but about the same width. The p3 and p4 trigonids are about the same length as their talonids, but the talonids are somewhat wider. The paralophid of p2 arches slightly lingually, thus creating a small lingual-trigonid notch. The protolophid of p2 is straight and slightly deflected lingually. There is no metaconid on the p2. The p3 paralophid and protolophid are more strongly angled lingually, creating broader lingual-trigonid notches. The p3 has a small metaconid that is positioned about equally lingually and posteriorly from the protoconid. The p4 trigonid is essentially molariform with a fully lingually arched paralophid and protolophid, a molariform trigonid basin, and a large lingually positioned metaconid. The cristids obliqua and hypolophids of p2–p4 are well developed, although on the p2 the talonid basin is small. The talonid basins of p3 and p4 are broader and nearly molariform. Labial premolar cingulids of Gnathotitan berkeyi tend to be weak, while lingual premolar cingulids are absent.

The molars of Gnathotitan berkeyi are typical, with shallow basins, thin lingual enamel, and an elongate m3. However, the m3 is among the most elongate among brontotheres with an m3 length/width ratio (2.94) that rivals Aktautitan (Mihlbachler et al., 2004a). Labial molar cingulids are thin and lingual molar cingulids are absent. The m3 hypoconulid has a faint cingulid tracing around it.

Remarks

The Central Asiatic Expedition of the American Museum of Natural History recovered seven specimens (two are now lost) that represent a species of very large brontothere with elongate molars and a very deep mandible. Osborn (1925) assigned holotype status to a mandible and maxilla that had originally been given the same catalog number (AMNH 20106), although these specimens do not actually belong to a single individual (see below for clarification). Osborn (1925, 1929a) originally assigned Gnathotitan berkeyi to the genus Telmatherium. However, G. berkeyi is clearly different from any other material that Osborn (1929a) had referred to Telmatherium. Osborn's (1925) only indication as to why he though G. berkeyi belonged to Telmatherium was that “the canines resemble those of Telmatherium and Menodus” (Osborn, 1925: 9). However, no special similarities in the canines of these taxa were actually pointed out, nor could I find any. Granger and Gregory (1943) correctly recognized this species as distinct from all other brontotheres and assigned it to its own genus.

The holotype of Gnathotitan berkeyi (AMNH 20106), as previously described by Osborn (1925; 1929a) and Granger and Gregory (1943), includes a mandible and right maxilla that were both originally assigned to AMNH 20106. However, Walter Granger's field notes for the 1923 Central Asiatic Expedition in the AMNH vertebrate paleontology archives indicate for field number 163 ( =  AMNH 20106), “association between maxilla and lower jaw questionable”. This detail was either ignored or unnoticed by Osborn (1925, 1929a) and Granger and Gregory (1943), who considered the maxilla and mandible to be parts of a single holotype. Because the jaw and maxilla are not certainly associated (and evidence presented below further suggests they represent separate individuals), the jaw and mandible should be considered syntypes rather than a single holotype. The AMNH catalog has been modified to reflect this by assigning a new number to the maxilla (AMNH 141231), while the mandible retains the original number (AMNH 20106). Among the syntypes, the mandible (AMNH 20106) is designated as the lectotype.

Other evidence suggests that the lectotype mandible (AMNH 20106) and the maxilla (AMNH 141231) represent separate individuals of the same species. Most notably, the canine of the maxilla is much larger than the canine of the mandible. The diameter of the maxillary canine (36.4 mm) at the base of the crown is 71% greater than that of the mandibular canine (21.2 mm). Canine size is commonly variable in brontothere species. However, the upper and lower canines of a single individual are usually of a similar size. For comparison, the upper canine of KAN N2/875 (the holotype of Aktautitan hippopotamopus) is only 5% larger in diameter than the lower canine (Mihlbachler et al., 2004a). The disparity in canine size between the maxilla (AMNH 141231) and mandible (AMNH 20106) of the G. berkeyi syntypes strongly suggests that these specimens represent two individuals, possibly of different sexes. The other specimens referred to G. berkeyi also indicate canine size dimorphism. The additional mandible, AMNH 20107, has canines that are much larger (diameter  =  29 mm) than those of the lectotype mandible. Likewise, the additional maxilla, AMNH 20121, has a canine alveolus that is much smaller in diameter than that bulbous canine root of the syntype maxilla.

Aktautitan hippopotamopus Mihlbachler, Lucas, and Emry, 2004a

Holotype

KAN N2/875, a complete skull, mandible, and skeleton lacking only parts of the right tarsus and pes.

Type Locality

Kyzylbulak Formation, Kyzyl Murun near Aktau Mountain, Ily Basin, Kazakstan.

Age

Middle Eocene (Irdinmanhan land mammal “age”).

Referred Specimens

(From the same locality as the holotype) KAN N2/873, a complete skull and articulated mandible, complete dentition, and fully articulated right forelimb with radius, ulna, and manus, and a partial left manus; KAN N/2 639, anterior portion of a cranium.

Diagnosis

Aktautitan hippopotamopus is a large brontothere with small frontonasal horns that are elevated high above the orbits. The nasal process and horns are elevated to the peak of a tall frontonasal process that rises anterodorsally from above the orbit at an angle of about 45°. The nasal incision is dorsoventrally deep and it extends posteriorly to the P4. The orbit is positioned over the M2. The elevated nasal process is angled downward, not strongly rounded anteriorly, and with lateral walls that are deep proximally and shallow distally. The premaxillomaxillary rostrum deepens posteriorly and is not enclosed by bone dorsally. Other cranial characteristics include a saddle-shaped cranium, separate parasagittal ridges, nearly straight zygomatic arches, and a ventrally unconstricted and mediolaterally angled external auditory pseudomeatus. Postzygomatic processes, as seen in Metatitan, are absent.

Dentally, Aktautitan hippopotamopus has three large upper incisors including a subglobular I1, and more subcaniniform I2 and I3, a distinct postcanine diastema, a complex P1, a distinct P2 metacone, and weak preprotocristae on the P2 and P3. Premolar hypocones are absent. The upper molars have tall, lingually angled ectolophs with weak labial ribs, and thinned lingual ectoloph enamel with wedge-shaped paracones and metacones. A cingular parastyle shelf is absent. Distinct central molar fossae and anterolingual cingular cusps are present. Paraconules and metalophs are absent. The three lower incisors are large and subcaniniform. The i2 is the largest incisor. There is a distinct postcanine diastema and a small p3 metaconid. The p2 trigonid is nearly twice the length of the talonid. The lower molars have shallow basins and m3 is long and slender.

Aktautitan hippopotamopus shares with Metatitan the unique combination of having closely positioned horns and a nasal process that are elevated to the peak of a superorbital frontonasal process. However, it differs from Metatitan in its more anteriorly positioned orbits, more poorly developed p3 metaconid, postcanine diastemata, and larger and more subcaniniform incisors.

Remarks

Aktautitan hippopotamopus is a large brontothere from the middle Eocene Kyzylbulak Formation of Kazakstan that resembles Metatitan in its elevated horns and nasal process, but differs from Metatitan in a number of respects. Preliminary reports of this new species mistakenly attributed it to the genus Protitan (Emry et al., 1997; Emry and Lucas, 2002, 2003; Lucas and Emry, 2001). The formal description of A. hippopotamopus represents a part of this study and follows methods and a format similar to that used here but was published elsewhere (Mihlbachler et al., 2004a).

Brachydiastematherium transylvanicum Böckh and Maty, 1876

Holotype

Anterior part of a lower jaw with several incisors, right c, p2–p3 and left c–m1.

Type Locality

Andrásháza (Seibenbürgen), Hungary (Transylvania), eastern Hungary, about 150 miles northeast of Belgrade.

Age

Middle Eocene (?).

Diagnosis

Brachydiastematherium transylvanicum is a large brontothere with a mandible very similar to Metatitan. Features of the holotype jaw are a robust mandibular symphysis that extends behind the p4, intermediately sized incisors that are semispatulate and subcaniniform that form an arched row anterior to the canines. A postcanine diastema is absent. A metaconid is absent on p2, but present on p3 and p4. The p2 trigonid and talonid have similar lengths. The labial notch in the crown of p2 is broad.

Brachydiastematherium transylvanicum can be distinguished from Metatitan relictus and M. primus by its larger more spatulate incisors with distinct lingual cingulids and a much broader labial p2 notch. B. resembles M. khaitshinus in these respects; however, the p4 of B. transylvanicum has an autapomorphic crest extending posteriorly from the middle of the cristid obliqua.

Description

Mandible and Lower Dentition

The holotype of Brachydiastematherium transylvanicum consists of the anterior portion of a mandible with an incomplete set of lower teeth. The following description is based primarily on cast material in the AMNH (108188) and the figure provided by Osborn (1929a), reprinted here as fig. 98. The angle of the inferior margin of the symphysis was relatively shallow. The posterior margin of the symphysis is incomplete; however, what is left indicates that it originally extended behind the talonid of p4. The alveolar incisor surface is incomplete and all but one of the incisors (left i3) is completely detached. However, the preserved portion of the symphysis seems to indicate that the incisor row arched slightly anterior to the canines. The incisors are of moderate size. One of these incisors is semispatulate, while the other is slightly more conical. Each of the incisors has a distinct beaded lingual cingulid. The canines are somewhat large and have prominent lingual cingulids.

Figure 98

The holotype of Brachydiastematherium transylvanicum. (A) Left view, (B) medial (lingual) view of left cheek teeth, (C) dorsal view. Illustrations from Osborn (1929a).

Brachydiastematherium transylvanicum lacks a postcanine diastema. The p1 has a single cusp and a broad talonid heel. The trigonid of the p2 is only slightly longer than the talonid; the p3 trigonid is about the same length as the talonid, and the p4 trigonid is shorter than the talonid. The paralophid of p2 is slightly angled lingually, creating a small but distinct lingual notch. The p2 protolophid is straight but deflected slightly lingually. The trigonids of the p3 and p4 are more molariform with paralophids and protolophids that arch fully lingually, creating nearly molariform basins. The talonids of p2, p3, and p4 are well developed with long cristids obliqua, long hypolophids, and broad, valleylike talonid basins. Lingual premolar cingulids are absent. The labial cingulid of the p2 is relatively faint and discontinuous around the base of the protoconid and hypoconid. However, the labial cingulids of p3 and p4 are thin, but they are distinct and continuous. Finally, the p4 has an unusual crest of enamel extending posteriorly from the middle of the cristid obliqua. This crest is also present on p3, but it is shorter and less distinct. The left m1 is the only preserved lower molar. It is heavily worn and lacks any features worthy of mention.

Remarks

Böckh (1876) described the first undoubted brontothere from eastern Europe, Brachydiastematherium transylvanicum. Presently, no other material is directly referable to this species. This fossil is considered to be middle Eocene (Uintan equivalent) by Lucas and Schoch (1989a), although this assessment is based on the opinion, expressed by Lucas (1983a) and Lucas and Schoch (1989a), that B. transylvanicum is a senior synonym of North American Uintan (Uinta C) genus, Diplacodon. However, B. transylvanicum differs from Diplacodon elatus in significant ways. For instance, there is no postcanine diastema; the p3 metaconid is essentially molariform, and the talonids of the premolars are generally much broader. Brachydiastematherium is similar to Metatitan in these respects.

In an analysis of Asian brontothere relationships, Mihlbachler et al. (2004a) found Brachydiastematherium to share close phylogenetic affinity with Metatitan and considered the possibility that Metatitan is actually a junior synonym of Brachydiastematherium. In particular, B. transylvanicum is very similar to M. khaitshinus. Both species share moderately sized semispatulate incisors and a broad labial notch on the p2. M. relictus and M. primus have smaller more globular incisors and a much narrower labial notch on the p2. However, the unusual enamel crest extending posteriorly from the cristid obliqua of the p4 of B. transylvanicum is not seen in M. khaitshinus. The taxonomic significance of this unusual crest is uncertain. It could be an autapomorphic character of B. transylvanicum, although it is also possible that it is an anomalous characteristic of the individual represented by the only known specimen. Presently I consider it to be an autapomorphy of B. transylvanicum. If my interpretation of this character is correct, it validates B. transylvanicum, although better material is clearly needed.

When considering the taxonomic identity of Brachydiastematherium transylvanicum, one must also consider other brontothere fossils found in Europe. Nikolov and Heissig (1985) described several brontothere teeth from the Black Sea Coast of Bulgaria, representing the second find of European brontotheres. The incomplete nature of European brontothere fossils limits comparison of the material described by Nikolov and Heissig (1985) with the holotype of B. transylvanicum. It is nonetheless possible that they represent more material of the same species. However, the upper premolars of these specimens are most consistent with Sivatitanops birmanicum, a species otherwise known only from Pondaung sandstones of Myanmar, southeast Asia. A partial skull from Pondaung deposits that is possibly Sivatitanops birmanicum is also similar to the jaw holotype of B. transylvanicum in lacking a postcanine diastema, but having incisors that are somewhat larger and less globular than other brontotheres (e.g., Metatitan) that lack postcanine diastemata. It is therefore possible that S. birmanicum is a junior synonym of B. transylvanicum or that these species are closely allied (along with Metatitan?). (For more discussion of the possible synonymy of these taxa see remarks under S. birmanicum.)

Metatitan primus Granger and Gregory, 1943

Holotype

AMNH 26101, a partial skull with intact left side, left C–M3, partial mandible with complete incisor row, canines, and left p2–m3.

Type Locality

?Ulan Gochu Formation, Chimney Butte, North Mesa, Shara Murun Region, Inner Mongolia, China.

Age

?Late Eocene (?Ulangochuian land mammal “age”).

Referred Specimen

(From the same locality as the holotype) AMNH 26102, a partial mandible with left c and m1–m3.

Diagnosis

Metatitan primus is a large brontothere with small closely spaced frontonasal horns that are positioned high above the orbits. The nasal process and horns are elevated to the peak of a tall frontonasal process that rises anterodorsally from above the orbits at an angle greater than 45°. The nasal incision is deep and extends to the anterior margin of the M1. The orbit is positioned over the posterior portion of M2 and the anterior portion of M3. The elevated nasal process is horizontal, relatively broad, not strongly rounded anteriorly, and with lateral walls that are deep proximally and shallow distally. The premaxillomaxillary rostrum deepens posteriorly and it is not enclosed by bone dorsally. Other cranial characteristics include a bulbous and incompletely saddle-shaped cranium. The posterior end of the cranium is extremely widened. The parasagittal ridges are not prominent and they do not constrict the dorsal surface posteriorly. There is a small dome on the dorsal surface of the skull. The zygomatic blades are nearly straight and they extend nearly to the posteriormost end of the skull where they form a 90° angle with the lateral zygomatic wing of the squamosal. The external auditory pseudomeatus enters the skull in a mediolateral direction and is ventrally constricted. Finally, a broad postzygomatic process is present.

Dentally, Metatitan primus is characterized by a complex P1, a distinct P2 metacone, and hypocones on P2–P4 that are positioned close to the protocones and not well separated from them. Upper and lower postcanine diastemata are absent. The molars of M. primus have tall lingually angled ectolophs with weak labial ribs, and thinned lingual ectoloph enamel with wedge-shaped paracones and metacones. A cingular parastyle shelf is absent. Distinct central molar fossae and small anterolingual cingular cusps are present. Paraconules and metalophs are absent. The lower dentition of Metatitan includes three very small incisors that form a nearly straight row. The i1 and i2 are short and wedge-shaped, while the i3 is somewhat more conical. A metaconid is present on the p3 and p4. The p2 trigonid is not much longer than the talonid. The lower molars have shallow basins and the m3 is elongate.

Metatitan relictus shares with M. primus and M. khaitshinus the unique combination of the following traits: horns and a nasal process elevated to the peak of a superorbital frontonasal process, posteriorly positioned orbits, an extremely widened skull, anteroposteriorly shortened basicranium, and lack of postcanine diastemata. M. primus differs from these species most obviously by its remarkably bulbous cranium.

Description

Skull

The holotype of Metatitan primus (AMNH 26101) is complete on its left side and does not appear to be significantly distorted. The right side of the skull is severely weathered and essentially featureless (fig. 99). The facial area of M. primus is similar to that of Aktautitan hippopotamopus Mihlbachler et al. (2004a) and other species of Metatitan. The small horns are positioned closely together and are highly elevated above the orbits. The protuberances rest on a tall superorbital pillar that rises from above the orbits at an angle steeper than 45°. The nasal process is elevated to the peak of this structure. Consequently, the nasal cavity is dorsoventrally deep, and the posterior margin of the nasal incision rises much higher than the orbits. The nasal and frontal bones are completely co-ossified, so it is not possible to determine the exact position of the frontonasal suture. However, the morphology of the face of M. primus is so similar to A. hippopotamopus, M. relictus, and M. khaitshinus (where the frontonasal contact is visible) there is little doubt that this large process is composed of the nasal bone anteroventrally and the frontal bone posterodorsally. The frontonasal protuberances are small, rounded, and slightly rugose at the surface and are perched at the lateral edges of the peak of the singular frontonasal process.

Figure 99

The holotype skull of Metatitan primus (AMNH 26101). (A) Left view, (B) dorsal view, (C) anterior view, (D) posterior view.

The nasal incision extends posteriorly to the anterior margin of the M1. The orbit is positioned over the posterior portion of M2 and the anterior portion of M3. The anterolateral root of M2 and the posterolateral root of M1 are positioned below the anterior rim of the orbit. The nasal process of AMNH 26101 extends nearly horizontally from the peak of the frontonasal process, although the flat dorsal surface is angled slightly downward. The nasal process is broad and slightly shorter than the premaxillomaxillary rostrum. The lateral walls of the nasal process are dorsoventrally deep and thin. The lateral walls are deepest proximally. The lateral walls extend to the end of the nasal process, but they shallow distally. The anterior edge of the nasal process is slightly turned downward. From the dorsal view the anterior edge of the nasal looks rounded and the nasal process appears to taper distally, but these are artifacts of how the specimen has been distorted. Direct inspection of the specimen reveals that the nasal process was not tapered distally, and the anterior margin was nearly flat with a distinct median notch.

From a lateral view it can be seen that the premaxillomaxillary rostrum is mildly curved upward. The rostrum deepens proximally. From an anterior view, the rostrum is broad, although the right side has been smashed inward. The dorsolateral margin of the rostrum (intact on the left side) angles posterolaterally and the rostral cavity is not sealed over dorsally. The maxilla and premaxilla appear to be completely fused; thus, a premaxillomaxillary suture is not discernable in this specimen. The opening of the nasal cavity is large and taller than it is broad.

The overall shape of the cranium of Metatitan primus is distinctive and somewhat bizarre. In addition to the elevated horns and nasals, the postorbital cranium is remarkably bulbous. Nonetheless, the general proportions of the facial and cranial regions are similar to those of other brontotheres, with a short face and an elongate cranium. The dorsal surface in the midsection of the skull is strongly concave and forms a deep transverse channellike depression. The dorsal surface of the posterior half of the skull is convex, thus resulting in an incompletely saddle-shaped skull. Directly behind the midcranial concavity is a centrally positioned convex dome. Curiously, this dome is similar in size and position to the dorsal dome seen in Duchesneodus uintensis. Between the dome and the occiput is a distinct pit. The dorsal surface of the skull continues to broaden posteriorly and the parasagittal ridges do not constrict the skull posteriorly. The parasagittal ridges are essentially absorbed into the swollen cranium and exist only as faint ridges running from the postorbital processes of the frontal to the occiput.

Overall, the zygomatic arch is relatively deep and thin. From a lateral view the zygomatic arch is peculiar in that it extends almost to the very end of the skull. The entire zygomatic arch is straight except for the posterior end where is it deflected upward. From the dorsal view the zygomatic arch is thin and very straight and angled posterolaterally. If the right side of the skull were intact, the zygomatic arches would have given the skull a wedge-shaped appearance. A peculiar feature of the zygomatics of Metatitan primus, shared with M. relictus, is the fact that the winged-shaped extension of the squamosal that connects the cranium to the zygomatic arch is very short and extends almost straight laterally from the side of the skull and forms an abrupt 90° angle with the actual zygomatic blade. M. primus also has a posterior zygomatic process similar to that of Protitan grangeri, although from a lateral view, the posterior zygomatic process is much broader than that of P. grangeri.

From a dorsal view the nuchal crest is slightly concave. From a posterior view, the dorsal margin of the occiput is slightly arched dorsally. The occiput itself is vertical and extremely broad. The dorsal half is similar in width to the ventral half and the occiput is not constricted in the middle. The surface of the occiput has distinct occipital pillars with a shallow triangular depression between them.

The ventral surface of the skull of Metatitan primus is incomplete (fig. 100a), although the following details can be discerned. The posterior nares are not preserved, however a remnant of a wide lateral emargination of the posterior nares can be seen posteromedial to the left M3. This wide emargination resembles those seen in more complete specimens of M. relictus and M. khaitshinus. The basicranium of Metatitan primus is very short. The external auditory pseudomeatus is positioned at the very end of the skull and can be seen just behind the zygomatic arch from the lateral view, forming a tube-shaped opening. Although it is not possible to measure the width of the basicranium due to the damaged right side, the posterior end of the skull of Metatitan primus is very broad and the basicranium would have been much wider the outside distance across the M3s.

Figure 100

The holotype skull of Metatitan primus (AMNH 26101). (A) Ventral view, (B) left molars, (C) left premolars.

Upper Dentition

The upper incisors of Metatitan primus are unknown, but the left canine and cheek teeth are well preserved in the holotype specimen (fig. 100). The canine is of moderate size. The P1 crown is rounded in outline, but its exact morphology is obscured by damage and wear. However, its overall shape suggests a relatively advanced P1 with a distinct paracone, metacone, and well-developed lingual heel.

The crowns of P2–P4 are nearly rectangular in outline due to the nearly parallel anterior and posterior margins and the nearly flat lingual margins. The parastyle and metastyle of P2 are nearly straight. The parastyle of P3 is barely directed labially, while the P4 parastyle is more strongly angled labially. The metastyle of P3 is straight. The P4 metastyle could have been straight or slightly labially directed, but it has been worn away by the adjacent M1. There are distinct labial ribs on the paracones of P2–P4; it is broadest on P2 and is progressively narrower on more posterior premolars. There are no mesostyles on any of the premolars.

Small preprotocristae are seen on P2 and P3 but not on P4. The lingual sides of P2–P4 have distinct protocones and hypocones that are separated to varying degrees. On P2 the hypocone is smaller than the protocone, although the enamel of the hypocone is worn or broken off, thus giving it a slightly smaller apparent size. On P3 the hypocone is smaller and lower than the protocone and these cusps are positioned very closely together and are almost completely conjoined into a single ovoid cusp. Finally, the P4 hypocone is smaller and shorter than that of the protocone and sits on the shallow posterior slope of the protocone. Labial premolar cingula are distinct but thin. The lingual cingulum of P2 is broken off, but the lingual cingula of P3 and P4 are thick and continuous.

The upper molars of Metatitan primus are elongate, although the M1 has been significantly shortened by interstitial wear. The upper molars show typical brontotheriine traits, including tall, lingually angled ectolophs, very weak labial ribs, thin lingual ectoloph enamel, and wedge-shaped lingual sides of the paracone and metacone in molars that are not heavily worn (e.g., M3). The anterior cingulum is thin and passes proximally to the distal peak of the parastyle. Distinct shallow central molar fossae and small anterolingual cingular cusps are present. All evidence of paraconules and metalophs is lost. There is no hypocone on M3, although there is a distinct peak of enamel on the cingulum of the distolingual corner of M3. Labial molar cingula are thin but distinct, while lingual molar cingula are absent.

Mandible and Lower Dentition

The holotype mandible consists of a symphysis and a left ramus that is missing the coronoid process and mandibular condyle (fig. 101). The inferior margin of the symphysis is angled about 45°. From a dorsal view the symphysis is broad and extends to the anterior margin of the p4. The three incisors are very small and form a nearly straight row that is positioned only slightly anterior to the canines. The crown of i2 (most complete on the right side) is the largest incisor especially in crown height and labiolingual width. The apices of the i1 and i2 are worn off, but these incisors appear to have had very short, wedge-shaped crowns. The i3 is more rounded in outline, with a short conular crown. The i1 and i2 have a distinct but thin lingual cingulid. The lingual cingulid is faint on i3. There are no diastemata between the incisors or canines. The p1 is missing, but the p1 alveolus fills the gap between the canine and p2, indicating the absence of a postcanine diastema.

Figure 101

The holotype mandible of Metatitan primus (AMNH 26101). (A) Left view, (B) dorsal view, (C) left p2–p4, (D) lingual view of incisors and canines, (E) labial view of incisors and canines.

Overall, the p2–p4 are rather broad. The p2 trigonid is only marginally longer and is distinctly narrower than the talonid. The trigonids of p3 and p4 are both shorter and narrower that the talonids. The very anterior end of the of p2 is not preserved and the morphology of this area is obscure, but it seems that the paralophid is strongly curved lingually, thus creating a relatively broad lingual-trigonid notch. The p2 protolophid arches lingually nearly 90°. The p2 protoconid is positioned labially, and the protolophid and cristid obliqua intersect at a point lingual from the protoconid. This configuration is essentially molariform. Finally, there is a deep labial notch on the p2 between the trigonid and talonid that is directed anterolingually. In more posterior cheek teeth this labial notch is directed more lingually.

The paralophids and metalophids of p3 and p4 arch fully lingually, thus creating nearly molariform trigonid basins. Both of these premolars have a large lingually positioned metaconid. The morphology of the p2 is also consistent with a metaconid, although the crown is too damaged to determine if a metaconid was present on that tooth. The talonids of p2–p4 have well-developed cristids obliqua and hypolophids with molariform talonid basins. These structures are significantly broader in p3 and p4. Labial premolar cingulids are weak and lingual premolar cingulids are absent.

The lower molars of Metatitan primus have relatively thin lingual enamel and shallow trigonid and talonid basins. The m3 is elongate. There are no lingual cingulids. Labial molar cingulids are thin, but they are distinct and continuous around the paraconids and metaconids. A thin beaded cingulid can be seen tracing around the hypoconulid of the m3.

Metatitan relictus Granger and Gregory, 1943

Holotype

AMNH 26391, a skull heavily reconstructed with plaster, with right I2–M3, left I2–C, P2–M3, and a mandible with right i2–c, p3–m3, left i1–i3, and p4–m3.

Type Locality

“Houldjin” beds, one mile west of Camp Margetts, Inner Mongolia, China.

Age

Middle Eocene (Irdinmanhan land mammal “age”).

Referred Specimens

(From the “Houldjin” beds near Camp Margetts, Inner Mongolia) AMNH 26395, a skull missing the right horn, nasal, and occiput, with right I1, I3, C, M1–M3, left I1–I3, and canine (fragmentary); AMNH 26396, fragments of a facial portion of a cranium, palate, and basicranium with right P2–M (damaged); AMNH 26397, a partial skull with right I1–C, P4–M3, left I3, P4 (fragmentary), and M2–M3; AMNH 26398, a complete ventral surface of a skull with nasal, horn fragment, parietal fragment, right I1–M3, left I2–M1 (partial), and M2–M3; AMNH 26399, a complete left side of a skull with right C, P1–P4, left C (partial), and P4–M3; AMNH 26406, a left premaxilla and maxilla fragment with I3–M2; AMNH 26402, a mandibular symphysis and right ramus with right i1–m3 and left i2–c; AMNH 26404, a mandibular symphysis with complete incisor row, canines, right p2, and left p2–p3; AMNH 26405, a fragmentary mandibular symphysis and left partial ramus with right left canine, and a right partial ramus with p4–m1, and m2 (partial); AMNH 26407, a right mandibular ramus with c, and p2–m3; AMNH 26420, a left partial ramus with p3–m2, and m3 (partial); AMNH 26427, a partial mandible with right i3–c, p4–m3, and left c–p4; AMNH 26429, a right partial ramus with p4 (partial) and m1–m3.

Diagnosis

Metatitan relictus is a large brontothere with small frontonasal horns that are positioned high above the orbits. The nasal process and horns are elevated to the peak of a tall frontonasal process that rises from the orbits at about 45°. The nasal incision is dorsoventrally deep and extends to the anterior margin of the M2. The orbit is positioned over the posterior portion of M2 and the anterior portion of M3. The elevated nasal process is horizontal, relatively broad, not strongly rounded distally, and with lateral walls that are deep proximally and shallow distally. The premaxillomaxillary rostrum deepens posteriorly and it is not enclosed by bone dorsally. Other cranial characteristics include an incompletely saddle-shaped cranium. The posterior end of the cranium is extremely widened. The parasagittal ridges are prominent, but they do not constrict the dorsal surface posteriorly. The zygomatic blades are nearly straight and they extend nearly to the posteriormost end of the skull where they form a 90° angle with the lateral zygomatic wing of the squamosal. The external auditory pseudomeatus enters the skull in a mediolateral direction and it is ventrally constricted. A broad postzygomatic process is present. The emargination of the posterior nares is wide, and the posterior nares are completely behind the M3. Large ventral sphenoidal fossae are present.

Dentally, Metatitan relictus is characterized by three small incisors that form a nearly straight row, a globular I1 and I2, and a subcaniniform I3. The P1 is complex and the P2 has a distinct metacone. Premolar hypocones are uncommon, but there is usually a lingual crest extending posteriorly from the premolar protocones. Upper and lower postcanine diastemata are absent. The molars of M. relictus have tall, lingually angled ectolophs with weak labial ribs, and thinned lingual ectoloph enamel with wedge-shaped paracones and metacones. A cingular parastyle shelf is absent. Distinct central molar fossae and small anterolingual cingular cusps are present. Paraconules and metalophs are absent. The lower dentition of Metatitan relictus includes three very small incisors that form nearly a straight row. The i1 and i2 are short and wedge-shaped, while the i3 is somewhat more conical. Metaconids are present on p3, p4, and occasionally on p2. The p2 trigonid is not much longer than the talonid. The lower molars have shallow basins and the m3 is elongate.

Metatitan relictus shares with M. primus and M. khaitshinus the unique combination of the following traits: horns and a nasal process elevated to the peak of a superorbital frontonasal process, posteriorly positioned orbits, an extremely widened skull, anteroposteriorly shortened basicranium, and lack of postcanine diastemata. M. relictus differs from M. khaitshinus in having more posteriorly positioned posterior nares, small lower incisors forming a straight row, and a narrow labial notch on p2. M. relictus is most easily differentiated from M. primus by the extremely swollen appearance of the cranium of the latter.

Description

Skull

The badly damaged but nearly complete holotype skull of Metatitan relictus (AMNH 26391) is missing large portions, although the ventral half is essentially complete and undistorted (fig. 102). None of the numerous additional skulls of M. relictus is complete (fig. 103). Those figured are AMNH 26399 (fig. 103a), AMNH 26395 (fig. 103b, c), AMNH 26397 (fig. 103d) and a part of AMNH 26398 (fig. 103e).

Figure 102

The holotype skull of Metatitan relictus (AMNH 26391). (A) Right view, (B) dorsal view, (C) anterior view.

Figure 103

Specimens referred to Metatitan relictus. (A) Left view of AMNH 26399, (B) left view of AMNH 26395, (C) dorsal view of AMNH 26395, (D) posterior view of AMNH 26397, (E) dorsal view of occipital-parietal fragment of AMNH 26398.

The horns and nasal process of the holotype are not preserved, but this area can be described from other specimens. The facial area of Metatitan relictus is similar to that of M. primus and M. khaitshinus. The small horns are highly elevated above the orbits and rest on a tall frontonasal process that rises from above the orbits at approximately an angle of 45°. The horizontal nasal process is elevated to the peak of this tall process. A distinct line representing the frontonasal contact can be seen from the lateral views of AMNH 26399 and AMNH 26395. The frontonasal process is composed of the nasal bone anteroventrally and the frontal bone posterodorsally; the peak of the horn is formed by the frontal bone. The frontonasal process of M. relictus is somewhat shorter and less steeply angled than those of M. primus and M. khaitshinus; consequently, the nasal chamber is not as deep. Nonetheless, the posterior margin of the nasal incision is dorsoventrally relatively deep in comparison to brontotheres whose nasal processes are not elevated (e.g., Protitan). In specimens where the face is complete (AMNH 26399), the posterior margin of the nasal incision rises higher than the orbits.

The horns are variable in size, although they are always small. For instance, in AMNH 26399 they do not rise higher than the dorsal surface of the nasal process. In AMNH 26395 the horns are larger with more significant relief. Those with relief form small, rounded knobs that typically have rugose surfaces. The horns vary in their orientation; the majority of specimens have horns that do not project laterally, however, one specimen, AMNH 26396, has horns that project almost completely laterally.

The nasal incision of the holotype (AMNH 26391) extended at least to P4, but the nasal incision is incomplete. Other specimens with complete faces indicate that the nasal incision extended behind P4. In AMNH 26399 the nasal incision extends to the anterior margin of M2. The position of the orbit is the same as that of Metatitan primus where the posterior part of M2 and the anterior part of M3 are directly beneath the orbit with the anterior lateral root of M2 positioned below the anterior rim of the orbit.

The nasal process, most complete in AMNH 26395 and AMNH 26399, does not differ substantially from that of Metatitan primus. The nasal process tends to be shorter than the premaxillomaxillary rostrum. It projects horizontally from the skull from its elevated position, although the flat dorsal surface tends to be angled downward. The lateral margins of the nasal process are unthickened and they are deep proximally and shallow anteriorly. From the dorsal views the nasal processes of these specimens are not tapered or flared distally. The anterior edge of the nasal process is turned downward slightly and from the dorsal view it is nearly flat, although some, such as AMNH 26395, show a deep medial notch at the distal end.

The premaxillomaxillary rostrum deepens posteriorly and its dorsal margin is steeply sloped posterodorsally. A premaxillomaxillary suture is not visible in any specimen of Metatitan relictus. The rostrum is relatively wide. The dorsolateral margins of the rostrum are laterally divergent behind the premaxillary symphysis and the rostral cavity is not enclosed by bone dorsally.

The cranium of Metatitan relictus is not extremely swollen like that of M. primus. The dorsal surface of the holotype (AMNH 26391) is incomplete, but a deep concavity runs transversely just behind the orbits. Other skulls of M. relictus are also deeply concave midcranially just behind the orbits. The dorsal surface of the posterior part of the cranium is slightly convex, thus the skull of M. relictus is incompletely saddle-shaped. M. relictus lacks the central parietal dome seen in M. primus. The parasagittal ridges of M. relictus are prominent in contrast to M. primus and overhang the sides of the cranium somewhat. The dorsal view of AMNH 26395 best illustrates the very broad dorsal surface of the cranium.

The jugal portion of the zygomatic arch is shallow in comparison to the much deeper squamosal portion. From a lateral view the zygomatic blade is straight, although it is deflected upward at the posteriormost end, where there is a tall posterior zygomatic process. From the dorsal view of AMNH 26391 the zygomatic blades are thin and straight, and they diverge posterolaterally, creating a wedge-shaped skull. The zygomatic blade of AMNH 26395 is strongly bowed inward, but this seems to be an artifact of distortion. The wedge-shape of the zygomatic arch resembles Metatitan primus and M. khaitshinus where the lateral wings of the squamosals are positioned at the posteriormost end of the skull and they project laterally to form an abrupt 90° angle with the zygomatic blades.

The occiputs of most skulls of Metatitan relictus are poorly preserved, but all suggest a very broad occiput that is mildly tilted backward, unlike the more vertical occiputs of M. primus and M. khaitshinus. The skull of AMNH 26398 includes an unbroken fragment with the greater part of an intact nuchal crest (fig. 103e). This piece is consistent with a very wide occiput with a nuchal crest that is slightly concave from the dorsal view and dorsally arched from the posterior view. The occiput of AMNH 26397 is reasonably complete on the right side. That specimen suggests that the dorsal and ventral halves of the occiput are of a similar width and that it was not constricted in the middle. The surface does not suggest distinct occipital pillars, but this could relate to the poor condition of the surface of the occiput.

It can be seen from the ventral view of AMNH 26391 that the posterior nares are constricted by a wide horseshoe-shaped emargination (fig. 104a). The anterior rim of the posterior nares is shifted behind the M3s, so that the posterior nares are more posteriorly positioned than those of M. khaitshinus. The posterior narial canal appears to extend well into the sphenoid bone, although other aspects of the posterior narial canal are difficult to discern due to poor preservation and plaster. The basicranium is very wide and anteroposteriorly compressed. The width of the skull across the mastoid processes is much greater than the width across the M3s. A ratio calculated from the width of the basicranium of AMNH 26391 divided by the width across the M3s (ratio  =  1.63) is greater than that of any other brontothere for which this ratio can be calculated except for M. khaitshinus. The external auditory pseudomeatus, which is positioned at the posteriormost end of the cranium, is tube-shaped and it enters the skull in a mediolateral direction.

Figure 104

Ventral view of the skull and upper dentitions of Metatitan relictus. (A) Ventral view of AMNH 26391, (B) right molars of AMNH 26391, (C) right premolars of AMNH 26391, (D) left I3–P4 of AMNH 26406, (E) labial view of right canine and incisors of AMNH 26398, (F) lingual view of right canine and incisors of AMNH 26398.

Upper Dentition

In the holotype (AMNH 26391) half of the incisors (right I1, left I1, I2) are missing and have been sculpted with plaster (fig. 104a). However, the skull of AMNH 26398 includes a complete set of right upper incisors (fig. 104e, f). The incisors are positioned slightly anterior to the canines and form a nearly straight row. The number of incisors (three pairs) is unreduced, although they are very small and vestigial in appearance. The I1 and I2 have an amorphous globular appearance. The I3 is larger and has a taller, more conical crown. The apex is blunt, but this could relate to wear. There is a distinct lingual cingulum on the I3, but cingula are absent on I1 and I2. The canines of Metatitan relictus tend to be rather small. There seems to be a short precanine diastema in most specimens, but postcanine diastemata are absent. In AMNH 26406, a young individual, the erupting canine is crowded by I3 and P1 (fig. 104d). There is no postcanine diastema, not even when P1 and P2 are missing. For instance, on the left side of AMNH 26391, P1 and its alveolus are missing and P2 is closely pressed to the canine root, indicating that some remodeling of the left maxillary must have occurred in this individual subsequent to the loss of the left P1.

In addition to the relatively worn premolars of the holotype (AMNH 26391) (fig. 104c), the nearly unworn premolars of AMNH 26406 are shown in close-up (fig. 104d). The crown morphology of P1 is obliterated in AMNH 26391. In AMNH 26406 the P1 is nearly round in outline. It has a large paracone, a distinct but smaller metacone, and a small posteriorly shifted lingual heel. The P2 and P3 of M. relictus are slightly oblique due to a weakly distolingually angled anterior margin. P4, on the other hand, is more nearly rectangular, with parallel anterior and posterior margins. The parastyle of P2 is straight, but those of P3 and P4 are deflected anterolabially. The metastyle of P2 is straight, while those of P3 and P4 are slightly deflected posterolabially. The P2 ectoloph is essentially straight. Small labial paracone ribs can be seen on P2–P4; these become smaller in more posterior premolars. Mesostyles are absent on all premolars, although there is a distinct labial bulge at the base of the crown near the P4 metacone of AMNH 26391.

The lingual features of P2 are obliterated in AMNH 26391. Though heavily worn, the lingual features of P3 and P4 are still discernable. On both P3 and P4, there is an oval area of dentine exposed behind the protocone, suggesting that a small hypocone was present and well separated from the protocone. The lingual premolar morphologies of other specimens differ notably from that of AMNH 26391. In AMNH 26406 there is a large P2 protocone and a slightly smaller hypocone that is positioned closely to the protocone. The hypocone is connected to the protocone by a short lingual crest that is almost as tall as the cusp apices. On the P2 of AMNH 26399, a large ovoid cusp rests on the lingual side of the crown that could represent two conjoined cusps, but the hypocone is not distinct from the protocone. The P3 of AMNH 26406 has a large protocone and a distinct lingual crest, but there is no hypocone. The P4 of that specimen retains a single, large protocone with no crest or hypocone. The protocones of AMNH 26406 and AMNH 26391 are anteriorly positioned, but the protocones of P3 and P4 are sometimes more centrally positioned by comparison. Moreover, a lingual crest is variably present (e.g., AMNH 26397) and absent (e.g., AMNH 26396) on P3 and P4. Apparently, P3 and P4 hypocones were infrequently present; no other AMNH specimen of Metatitan relictus other than the holotype (AMNH 26391) has distinct P3–P4 hypocones. A small preprotocrista is usually visible on P2, but it is exceedingly faint on P3 and absent on P4. Labial premolar cingula are distinct but thin. The lingual premolar cingula are most often continuous around the lingual sides of the crowns.

The upper molars of Metatitan relictus are elongate, although the lengths of M1 and M2 on the holotype (AMNH 26391) have been significantly shorted by interstitial wear (fig. 104b). The molars of this specimen are well worn, but along with molars of other specimens, they indicate typical brontotheriine apomorphies, including tall, lingually angled ectolophs, very weak labial ribs, thin lingual ectoloph enamel, and wedge-shaped lingual sides of the paracone and metacone in molars that are not heavily worn. The anterior cingulum is thin and passes proximally to the distal peak of the parastyle. Distinct shallow central molar fossae and small anterolingual cingular cusps are present. All evidence of paraconules and metalophs is lost. There is no trace of a hypocone on M3, although the distolingual cingulum of the M3 is thickened and raised. Labial molar cingula are weak and lingual molar cingula are absent.

Mandible and Lower Dentition

The holotype mandible (AMNH 26391) is complete except for the left condyle and a large portion of the symphysis, which has been reconstructed with plaster (fig. 105a, c) The right coronoid process is tall, narrow, and slightly curved. The ventral margin of the symphysis is slightly steeper than 45°. (Other specimens with less plaster in the symphysis confirm that the reconstruction of AMNH 26391 is accurate). The symphysis is broad and extends posterior to the m1 metaconid.

Figure 105

Mandible and lower dentition of Metatitan relictus. (A) Right view of AMNH 26391, (B) right premolars of AMNH 26402, (C) dorsal view of AMNH 26391, (D) lingual and (E) labial views of the right incisors and canine of AMNH 26402.

The incisors are very small and form a nearly straight row between the canines. The i2 is somewhat larger than the other incisors. The incisor crowns of AMNH 26391 are heavily worn, but they are clearly short and have small lingual cingulids. The incisors of AMNH 26402 are less worn (fig. 105d, e). The crown of i1 is essentially featureless. The crown of i2 is somewhat larger and more wedge-shaped with a small lingual cingulid. The i3 is smaller than the i2, and it is more rounded in outline with a blunt apex. The lower canines of Metatitan relictus are variable in size, but they are generally small. There are no precanine or postcanine diastemata. In the holotype (AMNH 26391) p1 and p2 have fallen out. On the right side there is small space between the canine and the p2 alveolus, but on left side the p2 diastema is right next to alveolus of the canine. Other specimens with more complete sets of premolars lack postcanine diastema altogether, even when p1 is absent.

The more complete and less worn premolar row of AMNH 26402 is shown in close-up (fig. 105b). The p1 crown consists of a small cusp with a broad talonid heel. The trigonids of p2–p4 are similar in length to the talonids. The trigonid and talonid of p2 are of similar width, but on p3 and p4 the talonid is wider than the trigonid. The paralophid of p2 arches strongly lingually, although the lingual trigonid notch is small. The lingually positioned protolophid extends from the protoconid in a posterior direction. There is no p2 metaconid on AMNH 26402. Another specimen, AMNH 26427, exhibits a small p2 metaconid and a more lingually oriented protolophid. Like Metatitan primus, a deep and narrow labial groove extends between the p2 trigonid and talonid. This groove is directed anterolingually, while those of p3 and p4 are more lingually directed. In p3 and p4 of AMNH 26402 the paralophids and protolophids arch fully lingually, creating nearly molariform trigonid basins. The p3 and p4 also have large lingually positioned metaconids. The talonids are well developed in p2–p4 with long hypolophids, long cristids obliqua, and broad talonid basins. Labial premolar cingulids are weak and lingual premolar cingulids are absent.

The molars of Metatitan relictus are typical with shallow trigonid and talonid basins and thin lingual enamel. The m3 is very elongate. Labial molar cingulids are distinct, but they vary in thickness and are strongest on the talonid and usually discontinuous on the trigonid. Lingual molar cingulids are absent. The m3 cingulid of all specimens of M. relictus ends on the labial side and does not wrap around the hypoconulid as it does in M. primus.

Metatitan khaitshinus (Yanovskaya, 1954)

Holotype

PIN 3745-1, a skull with a large block of plaster obscuring its dorsal surface, with right and left P1–M3.

Type Locality

Khaichin Formation, Khaichin-Ula V, Southwestern Mongolia.

Age

Middle Eocene (Irdinmanhan land mammal “age”).

Synonyms

Metatitan reshetovi (Yanovskaya, 1980).

Referred Specimens

(From the Khaichin Formation, Khaichin-Ula V, Southwestern Mongolia) PIN 3745-2, a mandible with right i2 (?), c–m3 and left i2–m3; PIN 3745-3, a skull with right P2–M3 and left C–M3; PIN 3754-4, a skull with right P1–P3, P4 (partial), M2–M3 (partial) and left P1–M2; PIN 3745-8, a partial edentulous skull; PIN 3745-9, a partial edentulous skull; PIN 3745-11 (holotype of Metatitan reshetovi), a skull missing the nasal process with right C, P2–M3 and left C, P2–M3; PIN 3745-12, the ventral surface of a skull with right and left P2–M3; PIN 3745-28, a mandible with right i2–m3, left i3–c, and p2–m3; PIN 3745-31, a mandible with right p2–m3, left c, and p2–m3.

The specimens listed above are those that I was able to relocate in the PIN collection. Other specimens including two more partial skulls (PIN 3745-5, 6), and many mandibles (PIN 3745 14–17, 19, 22, 23, 26, 27, 30) were referred to M. khaitshinus by Yanovskaya (1980) but could not be located in the PIN collection.

Diagnosis

Metatitan khaitshinus is a large brontothere with small frontonasal horns that are elevated high above the orbits. The nasal process and horns are elevated to the peak of a tall frontonasal process that rises anterodorsally from above the orbit at an angle greater than 45°. The nasal incision is dorsoventrally deep and its posterior margin extends posteriorly to the anterior margin of M2. The orbit is positioned over the posterior portion of M2 and the anterior portion of M3. The elevated nasal process is horizontal, relatively broad, not strongly rounded distally, and with lateral walls that are deeper proximally and shallower distally. The premaxillomaxillary rostrum deepens posteriorly and it is not enclosed by bone dorsally. The cranium is incompletely saddle-shaped. The posterior end of the cranium is extremely widened. The parasagittal ridges are prominent, but they do not constrict the dorsal surface posteriorly. The zygomatic blades are nearly straight and they extend nearly to the posteriormost end of the skull where they form a 90° angle with the lateral zygomatic wing of the squamosal. The external auditory pseudomeatus enters the skull in a mediolateral direction and it is ventrally constricted. A broad postzygomatic process is present. The emargination of the posterior nares is wide and the anterior margin of the posterior nares is positioned between the M3s. Large ventral sphenoidal fossae are present.

Metatitan khaitshinus has a complex P1 and a distinct P2 metacone. Premolar hypocones range from poorly developed to well developed and completely separated from the protocone. Upper and lower postcanine diastemata are absent. The molars have tall, lingually angled ectolophs with weak labial ribs, and thinned lingual ectoloph enamel with wedge-shaped paracones and metacones. A cingular parastyle shelf is absent. Distinct central molar fossae and small anterolingual cingular cusps are present. Paraconules and metalophs are absent. The mandible has a robust symphysis that extends to p4. The lower dentition of M. khaitshinus includes three moderately sized incisors that form a slightly arched incisor row. The i1 and i2 are semispatulate, while the i3 is somewhat more conical. Metaconids are present on p3 and p4, but not on p2. The p2 trigonid is not much longer than the talonid. The lower molars have shallow basins and m3 is elongate.

Metatitan khaitshinus shares with M. relictus and M. primus the combination of the following traits: horns and a nasal process elevated to the peak of a superorbital frontonasal process, posteriorly positioned orbits, an extremely widened skull, anteroposteriorly shortened basicranium, and lack of postcanine diastemata. However, M. khaitshinus differs from M. relictus and M. primus in having more anteriorly positioned posterior nares, larger lower incisors forming an arched row, and a broad labial notch on p2. M. khaitshinus has a more vertical occiput than M. relictus. It has a less constricted face than M. primus and its lacks the conspicuously bulbous cranium of that species.

Description

Skull

The following description of the skull of Metatitan khaitshinus refers primarily to the holotype (PIN 3745-1) but additional observations on other specimens are noted. The holotype of M. khaitshinus (PIN 3745-1) is a nearly complete skull (figs. 106 and 107). A large block of plaster is presently fixed to the dorsal surface of the skull, suggesting that it is incomplete or damaged. However, the specimen is not significantly distorted or damaged in any other way. This species is also known from several other reasonably complete specimens; two of them are figured here. PIN 3745-3 is a complete skull that is slightly crushed dorsoventrally (fig. 108). The anterior rim of the orbit is broken and the dorsal portion is displaced medially and ventrally with respect to the ventral portion of the anterior margin; consequently the orbit has become ovalized in contrast to the very rounded orbit of the holotype. Another skull, PIN 3745-11 (fig. 109), is dentally immature with an unerupted M3 crown. This skull is undistorted, but its nasal process and horns are unpreserved. This skull differs from the others in several ways although the differences are probably related to its subadult age.

Figure 106

The holotype skull of Metatitan khaitshinus (PIN 3745-1). (A) Right view, (B) dorsal view, (C) ventral view.

Figure 107

The holotype skull of Metatitan khaitshinus (PIN 3745-1). (A) Anterior view, (B) dorsal view of rostrum, (C) posterior view.

Figure 108

A skull referred to Metatitan khaitshinus (PIN 3745-3). (A) Left view, (B) dorsal view, (C) ventral view.

Figure 109

Left view of the holotype skull of Metatitan reshetovi, now referred to Metatitan khaitshinus (PIN 3745-11).

The frontonasal region of Metatitan khaitshinus resembles M. primus and M. relictus in having closely positioned horns that are elevated above the orbits along with the nasal process. The horns and nasal process are elevated to a degree similar to that seen in M. primus, but to a degree that is greater than that of M. relictus. The horns of the holotype just barely protrude out from under the front edge of the plaster covering and are positioned over the nasal incision. In PIN 3745-3 the horns and nasal process are not as highly elevated, but the difference is related to the dorsoventrally crushed condition of that specimen. The horns of that specimen form small, rounded dorsolaterally projecting knobs. The contact of the frontal and nasal bone can be clearly discerned on the left horn of PIN 3745-3. The frontal bone forms the posterodorsal surface of the frontonasal process and extends to the peak of the horn. The nasals bone forms the anteroventral portion of the frontonasal process and makes up the base and anterior portion of the horn.

The nasal cavity of Metatitan khaitshinus is voluminous due to the elevated nasal process. In the uncrushed holotype the posterior margin of the nasal incision rises much higher than the orbits. The posterior margin of the nasal incision extends to a point above the M1. Like other species of Metatitan the orbit is positioned rather posteriorly. The posterior part of M2 and the anterior part of M3 are directly beneath the orbit with the anterolateral root of M2 positioned beneath the anterior rim of the orbit. The nasal process is much shorter than the premaxillomaxillary rostrum and it projects from the skull in a horizontal direction. The lateral walls of the nasal process are unthickened and relatively shallow although they slightly deepen proximally. The width of the nasal process tapers slightly distally. The anterior margin of the nasal process is not strongly rounded because of a deep median notch.

The premaxillomaxillary rostrum is wide; it deepens posteriorly and its dorsal margin is sloped posterodorsally. A premaxillomaxillary suture is not visible on any of the specimens. The dorsolateral margins of the rostrum diverge laterally behind the premaxillary symphysis. The rostral cavity is open to the nasal cavity and not dorsally enclosed by bone.

The skull of Metatitan khaitshinus is similar to other species of Metatitan in its exceptional width. The postorbital portion of the cranium most closely resembles M. relictus and does not take on the extremely swollen appearance seen in the cranium of M. primus. While most horned brontotheres have saddle-shaped skulls, the crania of M. relictus and M. primus are incompletely saddle-shaped. However, this character is difficult to determine for M. khaitshinus. The dorsal surface of the holotype is obscured. The dorsal surface of PIN 3745-3 is concave from the horns to the occiput although this may be an artifact of crushing. However, several other specimens with less distorted crania, such as PIN 3745-4, PIN 3745-8, PIN 3745-9, and PIN 3745-11, indicate an incompletely saddle-shaped skull with a deep convexity in the center of the skull, but with a flat or more concave surface over the posterior portion of the cranium. The central parietal dome seen in M. primus is distinctly absent in skulls of M. khaitshinus.

The parasagittal ridges of M. khaitshinus are prominent. In the holotype (PIN 3745-1) they can be seen originating from the postorbital process of the frontal and sloping in a posterodorsal direction toward the occiput. From the dorsal view of PIN 3745-3 the parasagittal ridges appear to strongly overhang the sides of the skull. The parasagittal ridges are widely separated and they do not constrict the posterodorsal surface of the skull.

From lateral views of the skulls of Metatitan khaitshinus the zygomatic blade is straight, although it is deflected strongly upward at the posterior end where there is a tall posterior zygomatic process similar to that of M. relictus. From dorsal views of the skulls the zygomatic blades are thin, straight, and they diverge strongly posterolaterally, thus creating a strongly wedge-shaped skull. Like other species of Metatitan, the zygomatics extend nearly to the posterior end of the cranium, and the lateral wings of the squamosals are positioned at the posteriormost end of the skull and form an abrupt, nearly 90° angle with the zygomatic blades.

The nuchal crest of the holotype is obscured by plaster. From a dorsal view of PIN 3745-3 the nuchal crest is strongly concave. From a posterior view the dorsal margin of the occiput is arched dorsally. The occiput itself is extremely broad and like M. primus it is vertical. The dorsal half of the occiput is similar in width to the ventral half and the occiput is not strongly constricted in the middle. The surface of the occiput has narrow occipital pillars with a shallow triangular depression between them.

The posterior nares of the holotype are partially damaged while those of PIN 3745-3 are intact. However, in both specimens the posterior nares are constricted by a wide horseshoe-shaped emargination similar to that of Metatitan relictus. However, the posterior nares are more anteriorly situated than those of M. relictus; the anterior margin of the posterior nares of M. khaitshinus is consistently positioned roughly between the protocones of M3 while the posterior nares of M. relictus are entirely behind M3. This is one of the major features distinguishing these two species. Other aspects of the ventral surface of M. khaitshinus closely resemble M. relictus. The posterior narial canal is elongate and extends well into the sphenoid bone where it broadens notably, forming a rounded depression. The elongate vomer is not preserved in the holotype, but remnants of it are seen in PIN 3745-3 bisecting the elongate posterior narial canal and ventral sphenoidal fossae. In PIN 3745-3 the left sphenoidal fossa is clear of sediment while the right sphenoidal fossa is still filled with matrix. The basicranium is very wide and anteroposteriorly compressed. The width of the skull, measured from the left and right mastoid processes, is much greater than the width across the M3s. Ratios calculated from the width of the basicrania of PIN 3745-1 and PIN 3745-3 divided by the width across the M3s yield values (1.73 and 1.47, respectively) that are similar to those calculated for M. relictus. The external auditory pseudomeatus is tube-shaped, enters the skull in a mediolateral direction, and it is positioned at the posteriormost end of the cranium.

Upper Dentition

The upper incisors of Metatitan khaitshinus are unknown. The incisor alveolar border of PIN 3745-1 suggests a broad and somewhat arched incisor row (fig. 106) This interpretation agrees with the lower incisors (see below). The juvenile skull, PIN 3745-11, has a pair of small canines (fig. 109). Another specimen, PIN 3745-12 (not shown), has a pair of similarly sized canines that are very heavily worn. A postcanine diastema is not present in any specimen. The short postcanine gap in PIN 3745-11 is the result of P1 having fallen out.

The premolars of Protitan khaitshinus show a remarkable amount of intraspecific variability that is, nonetheless, a pattern typical of brontothere species. The labial aspects of the premolars are morphologically fairly stable, while the lingual sides are morphologically variable. To demonstrate this pattern of intraspecific variability close-ups of the premolars of the holotype (PIN 3745-1) (fig. 110a) and three other specimens (PIN 3745-3, PIN 3745-11, PIN 3745-12) (fig. 110b–d) are shown.

Figure 110

Close-ups of the upper dentition of Metatitan khaitshinus. (A) Left premolars of PIN 3745-1 (holotype), (B) left premolars of PIN 3745-3, (C) reflection of right P2–P4 of PIN 3745-11, (D) left premolars of PIN 3745-12, (E) left upper molars of PIN 3745-11.

The P1 of the holotype is heavily worn. PIN 3745-3 has a less worn P1 and has a morphology that is typical of many advanced brontotheres. The crown is rounded in outline with a distinct paracone, a metacone of roughly similar size, and a small lingual heel with a small lingual crest. P2–P4 have parallel anterior and posterior sides. P2 is slightly oblique in outline, while P3–P4 are more nearly rectangular. The parastyle of P2 is straight, but those of P3 and P4 are deflected anterolabially. The metastyles of P2–P4 are nearly straight. A broad labial paracone rib can be seen on the P2. The labial paracone ribs of P3 and P4 are much narrower and progressively smaller. Mesostyles are absent on all premolars.

As noted above, the lingual heels of the P2–P4 are variable. The holotype specimen is the least molariform, while the premolars of the remaining specimens are more molariform. The P2 of PIN 3745-1 has only one distinct lingual cusp, consisting of a large central ovoid cusp with a very short lingual crest at its peak. A small preprotocrista arches anterolabially from the peak of the lingual cusp, but it does not reach the paracone. The P2 of PIN 3745-3 is similar, but it appears to have two lingual cusps that are strongly fused together. PIN 3745-11 has two well-separated lingual cusps that are connected by a small lingual crest. The lingual crest continues posterior to the apex of the hypocone and ends at the posterior cingulum. Finally, the P2 of PIN 3745-12 has two equally sized and fully separated lingual cusps.

The P3 shows a similar degree of variability in relative molarization. The least molarized P3s have partially separated protocones and hypocones (PIN 3745-1, PIN 3745-3). Other specimens have two fully separated lingual cusps of similar size. One of these, PIN 3745-11 has a small lingual crest, while the other, PIN 3745-12, does not. The lingual heel of the P4 shows a lesser amount of variation and a lesser degree of overall molarization. Three of the figured upper premolar rows have a single lingual cusp without a preprotocrista or lingual crest on the P4. Only one of these specimens (PIN 3745-11) has a small lingual crest. The posterior portion of this crest thickens into an elongate hypoconelike structure. Labial premolar cingula are distinct but thin. The lingual premolar cingula are thick and continuous around the lingual sides of the crowns.

PIN 3745-11 (fig. 110e) has the least worn and best-preserved molars available for Metatitan khaitshinus. Other than the fact that they are less worn, they do not differ in any notable way from those of the holotype or other specimens belonging to this species. The upper molars are elongate. The molars of other specimens tend to be shorter, but this is primarily related to wear. Typical brontotheriine apomorphies are present, including tall, lingually angled ectolophs, very weak labial ribs, thin lingual ectoloph enamel, and wedge-shaped lingual sides of the paracone and metacone in molars that are not heavily worn. The anterior cingulum is thin and passes proximally to the distal peak of the parastyle. Distinct, shallow central molar fossae and small anterolingual cingular cusps are present. There is no evidence of paraconules or metalophs. There is no hypocone on the M3 although the distolingual cingulum of the M3 is thickened and raised. Labial molar cingula are weak and lingual molar cingula are absent or extremely weak.

Mandible and Lower Dentition

Yanovskaya (1980) referred numerous mandibles to Metatitan khaitshinus although I was able to locate only one of these, PIN 3745-31 (fig. 111). This specimen is nearly complete but it lacks the condyles. The angle of the ventral margin of the symphysis is near 45°. The symphysis is broad and extends to the P4. The dental formula is unreduced (3-1-4-3).

Figure 111

Mandible referred to Metatitan khaitshinus (PIN 3745-31). (A) Right view, (B) dorsal view, (C) right p2–p4, (D) lingual view of incisors and canines, (E) labial view of incisors and canines.

The incisors are of moderate size and they form a slight arch anterior to the canines. This differs from other species of Metatitan in which the incisors form a straighter row and are reduced to a very small and essentially vestigial condition. The lower incisors of M. khaitshinus more closely resemble those of Brachydiastematherium transylvanicum. In general the lower incisors have a semispatulate morphology. The apex of the i1 is worn flat. The i2 and i3 have arched apices. The i2 is the largest incisor. Each lower incisor has a distinct lingual cingulid. Labial cingulids are absent. There are no diastemata between the incisors or canines. The postcanine gap seen in PIN 3745-31 is the space for the missing p1, for which signs of the alveolus are still present. A genuine postcanine diastema (gap between canine and p1) is not present.

The remaining premolars (p2–p4) show a degree of molarization similar to other species of Metatitan and Brachydiastematherium transylvanicum, although some distinctions can be found and are noted below. The trigonids of p2–p4 are similar in length and width to their talonids. The paralophid of p2 arches strongly lingually generating a small lingual trigonid notch. The protolophid is straight, but it is angled in a posterolingual direction. There is no p2 metaconid although it might have been variably present as in M. relictus. The labial notch formed by the p2 protolophid and cristid obliqua is broad, shallow, and points lingually. This trait resembles B. transylvanicum but differs from both M. relictus and M. primus in which the labial p2 notch forms a deep narrow posterolingually angled groove. The right p3 that is shown in close up (fig. 111c) is damaged. However, the left p3 (fig. 111b) has a fully lingually arched paralophid and protolophid, creating a nearly molariform trigonid basin. The trigonid of the p4 is equally molariform. Both p3 and p4 have large lingually positioned metaconids. The talonids of p2–p4 are well developed with long hypolophids, long cristids obliqua, and broad talonid basins. The p4 of M. khaitshinus lacks the unusual crest of enamel extending posteriorly from the middle of the cristid oblique that is seen in B. transylvanicum. Labial premolar cingulids are weak and lingual premolar cingulids are absent.

The molars of Metatitan khaitshinus are typical with shallow trigonid and talonid basins and thin lingual enamel. The m3 is very elongate. Labial molar cingulids are distinct. Lingual molar cingulids are absent. The m3 cingulid does not wrap completely around the hypoconulid.

Remarks

Three diagnosable brontothere species can be assigned to Metatitan: M. primus Granger and Gregory (1943) (the type species), M. relictus Granger and Gregory (1943), and M. khaitshinus (Yanovskaya, 1980). Granger and Gregory (1943) erected Metatitan for a group of brontotheres characterized by extremely widened skulls, a short basioccipital, small frontonasal horns, small incisors, no postcanine diastema, and relatively advanced premolars. Granger and Gregory (1943) considered these species to be derived from Rhinotitan and to have bridged the morphological gap between the bizarre battering ram of Embolotherium and the more typical paired frontonasal horns of most other horned brontotheres. The type species, Metatitan primus, “points the way to Embolotherium, since its horns are displaced forward in front of the orbit and joined by a transverse connecting crest” (Granger and Gregory, 1943: 367). Granger and Gregory (1943) seem to have been referring to the fact that the frontal and nasal bone forms a single large process that rises above the orbits at a steep angle. This was thought to be the precursor of the ram of Embolotherium, a hypothesis that is further supported by the phylogenetic analysis of Asian horned brontotheres in Mihlbachler et al. (2004a). However, unlike Embolotherium, the nasal processes are retained in Metatitan and they are elevated to the peak of the frontonasal pillar, whereas in Embolotherium they are lost. Aktautitan hippopotamopus, a taxon not known to Granger and Gregory (1943), shares the Metatitan-like frontonasal morphology, but differs from Metatitan primarily in having large incisors, a postcanine diastema, less molarized premolars, and a more saddle-shaped cranium.

In addition to the type species, Metatitan primus, Granger and Gregory (1943) assigned two other species to Metatitan, M. relictus and M. progressus. Both of these species are valid, although the latter species is now placed in a new genus, Nasamplus. Most of the brontothere material collected in the Camp Margetts area during the 1930 Central Asiatic Expedition of the American Museum of Natural History represents M. relictus. M. relictus can be distinguished from M. primus by its less swollen cranium, more prominent parasagittal ridges, and deeper nasal incision. It is possible that that premolar hypocones occur less frequently in M. relictus, but this cannot be confirmed without additional specimens of M. primus.

Additional Metatitan fossils from Mongolia were described by Yanovskaya (1980) although she misidentified much of the material. Yanovskaya (1980) erroneously referred a skull, mandible, and numerous postcranial elements from the Ergilin Dzo of Inner Mongolia to M. relictus. Mihlbachler et al. (2004a) recognized that this material was not Metatitan and found it to be more similar to Parabrontops gobiensis. In this revision, that material is assigned to Eubrontotherium clarnoensis.

In the same monograph, Yanovskaya (1980) described numerous Metatitan fossils but misidentified them as Protitan. Yanovskaya (1980) erected two species based on a large collection of brontothere material from the Khaichin Formation of southwestern Mongolia; P. khaitshinus (Yanovskaya, 1980) was based on a complete skull (PIN 3745-1) (fig. 106) while P. reshetovi was based on a subadult skull (PIN 3745-11) (fig. 109). Mihlbachler et al. (2004a) first demonstrated the close similarity of these species with Metatitan, and they considered Yanovskya's Protitan synonymous with Metatitan. Working from descriptions and figures published by Yanovskaya (1980), Mihlbachler et al. (2004a) preliminarily suggested that (1) P. khaitshinus is a junior synonym of M. relictus, (2) that P. reshetovi is possibly a valid species, and that (3) Brachydiastematherium transylvanicum (the only brontothere species named from Europe) might actually be synonymous with a species of Metatitan.

After directly examining the material in the PIN collection these conclusions must be modified. Based on more substantial observations taken directly from the specimens, M. khaitshinus can be differentiated from both M. relictus and M. primus and thus constitutes a valid species. Although similar to M. relictus and M. primus, M. khaitshinus differs in substantial ways. M. khaitshinus has more anteriorly positioned posterior nares, a more arched lower incisor row, larger lower incisors, and it lacks the long narrow groove-like labial notch in the p2 that characterizes M. primus and M. relictus. M. khaitshinus appears to have a more vertical occiput than M. relictus. It has a less constricted face than M. primus and its lacks the autapomorphic dorsal dome of that species.

Secondly, Metatitan reshetovi cannot be differentiated from M. khaitshinus and is therefore a junior synonym of the later. Yanovskaya (1980) diagnosed M. reshetovi as having a narrower skull, narrower nasal, and more molarized premolars. However, the differences in relative skull width and nasal width between M. khaitshinus and the type of M. reshetovi (PIN 3745-11) are barely noticeable. For instance, Yanovskaya (1980) found only a 5% difference in skull width (calculated as a ratio with skull length). Considering the possible effects of subtle distortion and the subadult age of PIN 3745-11, these differences are of dubious taxonomic significance. Among the specimens attributed to M. khaitshinus and M. reshetovi there is conspicuous variation in the relative degree of premolar molarization. Specimens without premolar hypocones or poorly developed premolar hypocones were assigned to M. khaitshinus by Yanovskaya (1980), while others with more developed premolar hypocones were assigned to M. reshetovi. However, the variation in premolar molarization in these specimens is arguably continuous. Moreover, conspicuous intraspecific variation in lingual premolar morphology is a common pattern of intraspecific variation that is seen within many other brontothere species.

Thirdly, to the extent that they can be compared, M. khaitshinus is very similar to Brachydiastematherium transylvanicum from Europe. Currently B. transylvanicum is known only from its type specimen, a partial mandible that is almost identical to the mandible of M. khaitshinus. They are of similar size, both lack postcanine diastemata, and both differ from M. relictus and M. primus in the same ways: namely, each has larger, less globular incisors with lingual cingulids and each has a broader and more lingually pointed labial notch on the p2. B. transylvanicum and M. khaitshinus differ in only one detectable way: B. transylvanicum has an unusual crest extending posteriorly from the cristid oblique that is absent in M. khaitshinus. For this reason I consider these species to be distinct, but further discoveries may indicate that these species are synonymous.

Nasamplus progressus (Granger and Gregory 1943), new genus

Holotype

AMNH 26014, a skull fragment including portions of the right orbit, frontal, nasal, maxilla, and P4–M1.

Type Locality

Ulan Gochu, Jhama Obo, East Mesa, Shara Murun Region, Inner Mongolia, China. (Ulan Gochu [used in quotes] denotes uncertain correlation with the type Ulan Gochu beds at Baron Sog and refers to a faunal zone rather than a formation; see Radinsky, 1964.)

Age

Late Eocene (Ulangochuian land mammal “age”).

Etymology

The genus name, Nasamplus, is a combination of the Latin terms nasus (“nose”) and amplus (“large”). This name refers to the elevated nasal process of this species, which results in a large nasal cavity.

Diagnosis

Nasamplus progressus is a large brontothere in which the frontonasal horns are fused into a single transverse crest. The frontonasal crest and nasal process are elevated high above the orbits on a frontonasal process (ram). The nasal incision is very deep and extends posteriorly about to the anterior margin of M1. The orbit is positioned above the posterior portion of M1. The elevated nasal process is horizontal and with lateral walls that are deep proximally. The P4 of Metatitan primus has a distinct hypocone that is well separated from the paracone. A central fossa is present on M1.

The skull fragment representing Nasamplus progressus is morphologically intermediate between Aktautitan and Metatitan on the one hand and Protembolotherium and Embolotherium on the other. Nasamplus progresses resembles Aktautitan and Metatitan in having an elevated horizontal nasal process, but like Protembolotherium and Embolotherium the horns are fused into a transverse frontonasal crest at the peak of the frontonasal ram.

Description

Skull and upper dentition

The single fragmentary specimen of Nasamplus progressus (AMNH 26014) (fig. 112) indicates a large brontothere with a frontonasal morphology that is intermediate between the condition seen in Aktautitan and Metatitan, and the more derived condition of Protembolotherium and Embolotherium. As in Metatitan and Aktautitan the frontonasal protuberance is elevated high above the orbit, although it is more highly elevated in N. progressus. The angle of the superorbital frontonasal process upon which the frontonasal protuberance rests is steeper than 45°. The frontonasal contact is visible from the lateral view on the upper half of the frontonasal process. The frontal bone forms the posterodorsal surface of the frontonasal process while the nasal bone forms the anteroventral surface. The frontonasal contact is even more distinct from the dorsal view of the cranial fragment. At the peak of the frontonasal process the posterolateral border of the nasal bone is bordered by a thick section of frontal bone. The line of frontonasal contact recedes posteromedially and at the midline the frontal bone is thinnest. The frontal bone does not form a pair of distinct protuberances. Instead, there is a single transverse crest formed at the peak of the frontonasal process that is formed by both the frontal and nasal bones. The surface of this crest is roughened and from the anterior view it is dorsally arched. The configuration of the frontal and nasal bones and the transverse frontonasal crest at the peak of the frontonasal process of Nasamplus progressus closely resembles the rams of Protembolotherium and Embolotherium.

Figure 112

The holotype cranial fragment of Nasamplus progressus (AMNH 26014). (A) Right view, (B) anterior view, (C) posterior view, slightly rotated laterally, (D) dorsal view, (E) right P4–M1.

However, unlike Protembolotherium and Embolotherium, Nasamplus progressus retains a large horizontal nasal process that, like Metatitan and Aktautitan, is elevated to the peak of the frontonasal process. Only the proximal end of the nasal process is preserved. The lateral walls of the proximal end of the nasal process are very deep.

The orbit is positioned above the posterior portion of M1 with the anterolateral root of M1 below the anterior orbital rim. Due to the elevated nasal process the nasal incision is very deep and it extends posteriorly about to the anterior margin of M1.

Only two teeth, P4 and M1, are preserved in the fragmentary holotype and these are heavily worn. P4 is essentially rectangular with a labially oriented parastyle and a straight metastyle. A labial paracone rib is not discernable. On the lingual side of the tooth, there is both a protocone and a smaller but well-developed hypocone. The hypocone is positioned on the very distal side of the crown and connected to the protocone by a minor lingual crest. The labial cingulum of P4 is very weak, and the lingual P4 cingulum is thick but slightly discontinuous. M1 is extremely worn and the ectoloph is not preserved. A small remnant of the central fossa is visible in the center of M1.

Remarks

Granger and Gregory (1943) originally assigned Nasamplus progressus to Metatitan. Mihlbachler et al. (2004a) found that Metatitan progressus did not group into a monophyletic clade with other species of Metatitan, but that it grouped closer to Embolotherium. Unfortunately, no specimen other than the fragmentary holotype has ever been reported and although the holotype is only a very small cranial fragment, the right combination of features are preserved to determine that this specimen represents a distinct taxon that is essentially intermediate between Metatitan and Embolotherium. Therefore, this species is assigned a new genus name, Nasamplus.

This specimen led Granger and Gregory (1943) to conclude that the distinctive “battering ram” of Embolotherium was derived from a Metatitan-like morphology, involving the increasing height of the frontonasal process and the eventual loss of the nasal process. Nasamplus progressus seems to represent an intermediate stage in which the frontonasal process is increased in height and forms a battering ram–like crest at the peak but the horizontal nasal process has not yet been reduced or lost.

Protembolotherium efremovi Yanovskaya, 1954

Holotype

PIN 473-311, a skull lacking the nasal elements with right P2–M3 and left P3–M3.

Type Locality

Ergilin Dzo (lower part), Dornogobi Province, Outer Mongolia

Age

Late Eocene (Ulangochuian [Ergilian] land mammal “age”).

Referred Specimens

(From Ergilin Dzo [lower part, Outer Mongolia) PIN 473-310, a partial skull missing its left and ventral surfaces; PIN 3109-40, a partial skull missing its left and ventral surfaces.

Diagnosis

Protembolotherium efremovi is a large brontothere that is very similar to Embolotherium. Its most distinctive features are a pair of small secondary horns positioned above the orbits, and a tall, nearly vertical process of bone (a “ram”), probably composed of the frontal and nasal bones. The ram rises from above the orbits and arches backward. The ram tends to be shorter than those of Embolotherium andrewsi and E. grangeri. The peak of the ram is a transverse crest that forms the distal margin of the anteroventral nasal channel, sometimes with rugose swellings at the distolateral corners. The nasal incision extends posteriorly to the anterior margin of P4 and the orbits are positioned above the M2 and the posterior portion of M1. A horizontal nasal process can be absent or it persists in a diminished form as a flat triangular process that is elevated to the peak of the ram. The dorsal margin of the rostrum is strongly sloped so that the premaxillomaxillary rostrum deepens posteriorly. The rostrum is not enclosed by bone dorsally. Other cranial characteristics include a saddle-shaped cranium, a robust occiput, and weakly curved zygomatic arches with prominent lateral swellings. The external auditory pseudomeatus enters the skull in a posteromedial direction and is ventrally constricted.

Dentally, Protembolotherium efremovi is characterized by a distinct P2 metacone, and hypocones on P2–P4 that are connected to the protocone (to varying degrees) by lingual crests. The molars of P. efremovi have tall, lingually angled ectolophs with weak labial ribs, and thinned lingual ectoloph enamel with wedge-shaped paracones and metacones. A cingular parastyle shelf is absent. Deep central molar fossae and large anterolingual cingular cusps are present. Paraconules and metalophs are absent.

Protembolotherium efremovi is one of three species with an enlarged frontonasal battering ram. It can be differentiated from Embolotherium in the following ways: there is a pair of small secondary horns positioned above the orbits, the dorsal margin of the rostrum is steeply sloped upward, the skull is less deeply saddle-shaped, the occiput is shorter and more erect, the ram is somewhat shorter, and occasionally a reduced remnant of the nasal process is seen in an elevated position at the peak of the ram. It can be further differentiated from E. andrewsi by its less robust occiput, less prominent parasagittal ridge, and more laterally bowed zygomatic arches. It can be further differentiated from E. grangeri by its shorter nasal incision, more anterior position of the ram, and the backward curvature of the ram.

Description

Skull

The holotype skull of Protembolotherium efremovi (PIN 473-311) is missing the frontonasal region and a small portion of the right basicranium and occiput (fig. 113). In other respects the skull is essentially complete and not greatly distorted. Additional skulls referred to P. efremovi are PIN 473-310 (fig. 114) and PIN 3109-40 (fig. 115). These three skulls indicate a species that is large, but perhaps smaller on average than Embolotherium andrewsi and Embolotherium grangeri. The two referred skulls (PIN 473-310 and PIN 3109-40) have relatively complete and undistorted right and dorsal sides but they are lacking complete left and ventral surfaces. Both of these specimens exhibit two conspicuous features in the frontonasal region that are not preserved on the holotype specimen but that are, nonetheless, critical to the identification and characterization of this species. These features are (1) a large battering ram–like structure similar to that of Embolotherium and (2) an autapomorphic pair of short and widely spaced secondary hornlike bony protuberances positioned directly above the orbits. The hornlike protuberances of both of these specimens are rather narrow and they project dorsally, posteriorly, and laterally. In Protembolotherium efremovi the ram is massive and well developed although it is shorter than those of Embolotherium andrewsi and Embolotherium grangeri. Several aspects of the ram specifically resemble Embolotherium andrewsi. For instance, the ram originates from a position above and slightly anterior to the orbits, it protrudes from the skull in an anterodorsal direction at an angle greater than 45°, and curves slightly backward, thus resulting in a concave posterodorsal surface and a convex anteroventral surface. However, because of its shorter length the backward curvature of the ram is somewhat less pronounced and the distal end does not achieve the vertical angle that is seen in some specimens of Embolotherium andrewsi (such as AMNH 26001, see fig. 117). The ram of Protembolotherium efremovi broadens distally and a deep nasal channel runs along the anteroventral surface and continues into the nasal cavity of the skull. Like E. andrewsi, the nasal cavity extended to the distal end of the ram and was extraordinarily tall, although nostril position (elevated to the peak of the ram, or positioned more normally in a low position) remains unknown (Mihlbachler and Solounias, 2004). This nasal channel is bordered laterally by thick but rather shallow walls. The lateral walls moderately constrict the nasal channel at the proximal end of the ram, but not to the degree seen in Embolotherium andrewsi, nor is the nasal channel extremely deepened at the proximal end of the ram.

Figure 113

The holotype skull of Protembolotherium efremovi (PIN 473-311). (A) Dorsal view, (B) left view.

Figure 114

A skull referred to Protembolotherium efremovi (PIN 473-310). (A) Dorsal view, (B) right view.

Figure 115

A skull referred to Protembolotherium efremovi (PIN 3109-40). (A) Right view, (B) dorsal view, (C) anterior view, (D) posterior view.

Figure 117

The holotype specimen of Embolotherium andrewsi (AMNH 26001). (A) Right view, (B) dorsal view, (C) anterior view, (D) posterior view.

Protembolotherium efremovi has a pair of small secondary horns positioned above the orbits that are completely separated from the ram. On first impression the horns of P. efremovi appear to be homologous to the paired frontonasal horns typical of many brontotheres while the ram has the superficial appearance of an enlarged and upturned nasal process, the portion of nasal bone that in less derived brontotheres typically forms the dorsal margin of the nasal incision. Evidence of the sutural configuration of the facial bones would be valuable in determining the best hypothesis of homology of these structures, but unfortunately none of the relevant sutures are discernable on any of these specimens. However, there seems little doubt that the ram itself is homologous to that of Embolotherium andrewsi, in which there is clear evidence that the distal surface of the ram is homologous to the tips of the frontonasal horns of other horned brontotheres. Therefore, the small secondary horns of P. efremovi do not appear to be homologous with the paired frontonasal horns of other horned brontotheres.

The distal end of the ram of PIN 3109-40, in particular, is unique and takes on an intermediate condition between Embolotherium andrewsi and less derived species, such as Nasamplus, Metatitan, and Aktautitan. From the anterior view, the distal surface of the ram is concave. At each distal corner is a large pachyostotic swelling. From the lateral view the pachyostotic swelling is continuous with the lateral emargination of the ram. A more peculiar feature present in PIN 3109-40 but not seen in PIN 473-310 is the relatively flat process that projects anteriorly and slightly ventrally from between the two distolateral swellings. This process appears to be the free end of a reduced nasal process that in less derived species (e.g., Protitan, Metatitan) extends anterior to the frontonasal horn and forms the dorsal margin of the nasal incision. In Embolotherium andrewsi this structure is entirely lost. In the other specimen, PIN 473-310, the lateral swellings at the distal end of the ram are less pronounced; however, the free end of the nasal bone is essentially absent (or absorbed by the ram) as in E. andrewsi. If Yanovskaya (1954, 1980) and I are correct in assigning these skulls to a single species, the reduced nasal process was variable within P. efremovi, sometimes appearing in diminished state as in PIN 3109-40, and sometimes completely diminished as in PIN 473-310. Unfortunately both of these specimens are edentulous, so it is difficult to gauge their ontogenetic ages, although variation at the distal end of the ram might have been related to the ontogenetic process.

In the holotype (PIN 473-311) the posterior border of the nasal incision extends as far back as the posterior margin of P3. This is similar to Embolotherium andrewsi. Likewise, the position of the orbit is similar; it is positioned directly above M2 with the anterior rim of the orbit positioned approximately over the anterolateral root of M1. The nasal incisions of PIN 3109-40 and PIN 473-310 are complete and their posterior margins are entirely anterior to the orbits; however, the position of the nasal incision and the orbits cannot be determined with respect to the dentition due to the edentulous condition of these specimens.

None of the three skulls of Protembolotherium efremovi has a complete premaxillomaxillary rostrum although most of it is preserved with the type skull (PIN 473-311). Unfortunately premaxillomaxillary sutures are not clearly discernable. From a lateral view the rostrum is rather short. The ventral surface of the rostrum is curved slightly upward. The dorsal margin of the rostrum is flat and steeply angled posterodorsally so that it rises to a level just slightly higher than the orbits. From anterior and dorsal views the rostrum is broad and the lateral margins diverge posterolaterally to create a broad cavity on the dorsal side of the rostrum that is continuous with the nasal cavity. In these respects the rostrum is unspecialized and different from the rather shallow and abbreviated rostrum of Embolotherium andrewsi or the elongate Tapirus-like rostrum of Embolotherium grangeri. PIN 3109-40 has a less completely preserved rostrum; its dorsal surface appears to have been flatter than that of PIN 473-311, although it is so incomplete that I am not confident that the shape of this specimen can be accurately interpreted.

The entire dorsal surface of the skull of Protembolotherium efremovi is concave (saddle-shaped) though not as deeply as that of Embolotherium andrewsi. The parasagittal ridges extend along the sides of the skull from the postorbital process of the frontal bone to the upper corner of the occiput. These ridges are not as prominent as those of E. andrewsi. From a dorsal view the parasagittal ridges are widely separated and they barely constrict the dorsal surface posteriorly.

In the holotype (PIN 473-311), the zygomatic arches are massive and laterally expanded, with a large swelling at the junction of the jugal and squamosal bones. From the dorsal view the zygomatic process of the jugal is relatively straight anterior to the swelling and it projects posterolaterally. The degree to which the zygomatic arches bow out laterally is intermediate between Embolotherium andrewsi on the one hand, in which the jugal processes project in a predominantly posterior direction, and Embolotherium grangeri on the other, whose zygomatic arches are more strongly bowed. The remaining skulls of Protembolotherium efremovi demonstrate conspicuous intraspecific variation in the size of the zygomatic swellings. In both PIN 473-310 and PIN 3109-40 the zygomatic arches are thinner with lesser amounts of swelling in the center of the arch. Similar intraspecific variation in the thickness of the zygomatic swellings is documented in other species that possess such swellings and probably represents sexual dimorphism (Mihlbachler et al., 2004b). From lateral views of the skulls the jugal portion of the zygomatic arch anterior to the swelling is shallower than the squamosal portion of the zygomatic arch. There is a small ventral flange on the jugal below the massive swelling. The jugal portion of the zygomatic is essentially horizontal while the squamosal portion is angled slightly upward, thus giving the zygomatic arch a weak curvature.

From a dorsal view the nuchal crest is thickened with posterolaterally angled winglike expansions. The posterior margin of the occiput is concave, although it lacks the deep median notch seen in Embolotherium andrewsi. The overall degree of robustness of the occiput is more or less typical of large horned brontotheres, although it is not nearly as massive as the occiput of E. andrewsi. From a lateral view the occiput is rather short and it is not as strongly tilted backward as the occiputs of E. andrewsi and E. grangeri. The best-preserved occiput is that of PIN 3109-40. The dorsal margin of the occiput is slightly concave, the width of the occiput is slightly constricted in the middle, and the dorsal and ventral portions of the occiput are of similar width. The posterior surface of the occiput has two distinct but relatively weak occipital pillars with a shallow central depression between them. It these respects the occiput most closely resembles E. grangeri.

The ventral surface of the skull of Protembolotherium efremovi can only be described from the holotype (PIN 473-311) (fig. 116). The anterior margin of the posterior nares is positioned near the posterior margin of M3; this position is slightly more anterior to the posterior nares than those in Embolotherium andrewsi and E. grangeri in which the anterior margin is typically a short distance behind M3. Other characteristics of the ventral surface of PIN 473-311 are essentially indistinguishable from E. andrewsi and E. grangeri. The anterior and lateral sides of the posterior narial opening are rimmed by a wide U-shaped emargination. The ridge demarcating this emargination is faint, but it can be seen upon examination of the specimen. This emargination is widest on the anterior margin and tapers along the lateral margins. The vomer, which would have bisected the posterior narial canal, is not preserved although a remnant in the form of a small ridge of bone can still be seen running along the upper (dorsal) surface of the posterior narial canal. The posterior narial canal seems to intrude slightly into the sphenoid bone (a typical condition of brontotheres), but large sphenoidal pits like those of Metatitan or Protitan, for instance, are not present in P. efremovi. The basicranium seems rather narrow, although this effect is exaggerated because part of the right side is broken away. The total width of the basicranium at the position of the mastoid processes does not exceed the distance across the right and left M3s and the basicranium is not nearly as widened as in Metatitan. The configuration of the basicranial foramina is typical as well with a widely separated foramen ovale and foramen lacerum. The mastoid process is shorter than the postglenoid process and it curves anteroventrally forming a tube-shaped external auditory pseudomeatus. (This character can also be seen in PIN 3109-40.) The external auditory pseudomeatus extends into the basicranium in a posteromedial direction, a condition shared with E. andrewsi and E. grangeri.

Figure 116

The holotype skull of Protembolotherium efremovi (PIN 473-311). (A) Ventral view, (B) left P3–P4, (C) right P2–P4, (D) right molars.

Upper Dentition

Of the three known skulls of Protembolotherium efremovi only the holotype (PIN 473-311) includes upper dentition, consisting of an incomplete set of moderate to heavily worn and rather poorly preserved upper teeth, which includes the right P2–M3 and the left P3–M3 (fig. 116). Neither the upper incisors nor their alveoli are preserved. The crowns of the canines are not preserved, although bits of the roots of both right and left canines remain inside their alveoli. The canines appear to have been small though well developed. Apparently the canines have not reached the tiny essentially vestigial state seen in Embolotherium andrewsi. The transverse distance across the roots of the canines is about the same as the distance across the P1s, further indicating that the rostrum is not as reduced as that of E. andrewsi. (In E. andrewsi the transverse distance across the canines is notably less than that of the P1s.) Like E. grangeri there is a short postcanine diastema, although unlike E. grangeri the width of the rostrum is not constricted at this point.

The P1s are missing although well-preserved alveoli indicate that single-rooted P1s were present in life. The remaining premolars are nearly rectangular in outline with nearly parallel anterior and posterior margins. The parastyle of P2 is missing. The parastyles of P3 and P4 are directed anterolabially. The metastyles of the P2, P3, and P4 are essentially straight. Minor labial paracone ribs can be found on P3 and P4 (P2 is too damaged to discern this character). The labial paracone rib of P3 is broader than that of P4. None of the premolars has a distinct mesostyle. Labial cingula are discernable on the premolars, but they are very thin and poorly developed.

The lingual features of P2 consist of a protocone and nearly equally sized hypocone. Each of these cusps is embedded in a strongly developed lingual connecting crest. The connecting crest arches around the anterolingual corner of the crown forming a small preprotocrista. The protocone of P3 is a well-developed tall cusp, while the hypocone is significantly smaller. The lingual morphology of P4 is similar to that of P3, although the hypocone is more completely disconnected from the protocone. The protocone and hypocone of P3 are barely connected by a poorly developed lingual crest. Preprotocristae are not discernable on either P3 or P4, indicating that they were either absent or so poorly developed that they have been worn off. The lingual premolar cingula tend to be discontinuous or barely continuous around the lingual margins of the protocones and they tend to be particularly thick next to the hypocones, most notably on P4.

The upper molars are heavily worn and fragmented although many typical brontotheriine traits can be discerned on them, including tall (though now very worn) and lingually angled ectolophs, very weak (and on some teeth absent) labial ribs, thin lingual ectoloph enamel, and wedge-shaped lingual sides of the paracone and metacone in the least worn molars. The anterior cingulum (though impossible to see on the photographs) is thin and passes proximally to the apex of the parastyle. Deep central molar fossae and tall anterolingual cingular cusps are present. Parts of the central molar fossae can still be seen on all three molars. The anterolingual cingular cusp, however, is completely worn away on the M1s, but it is still discernable on M2 and M3. There are no traces of paraconules or metalophs on any of the molars. Both right and left M3 appear to have had a rather small hypocone. Labial molar cingula are very weak and lingual molar cingula are mostly absent or have been worn off. A very thin cingulum appears to wrap around the anterolingual corner of M3.

Remarks

Yanovskaya (1954) initially described Protembolotherium efremovi from two skulls and a variety of other specimens from the late Eocene Ergilin Dzo Formation of Outer Mongolia. P. efremovi is a large Asian brontothere that is most easily recognized by its Embolotherium-like “battering ram” that protrudes upward from the anterior end of the skull, as well as an autapomorphic pair of short secondary horns that are positioned above the orbits. It is a curious fact that Yanovskaya (1954) assigned holotype status to a skull, PIN 473-311, in which the frontonasal portion, and consequently all evidence for these diagnostically useful characters, is not preserved. Another skull (PIN 473-310) that preserves the battering ram and secondary horn was referred to P. efremovi by Yanovskaya (1954). Decades later, Yanovskaya (1980) referred a third skull (PIN 3109-40) from Ergilin Dzo to P. efremovi. This specimen is unique in retaining an anteriorly projecting nasal process at the peak of the ram.

In addition to these three skulls several additional fragmentary specimens were initially referred to Protembolotherium efremovi (PIN 473-139, 473-141, 473-216, 473-217, 473-397, 473-654, 473-656, PIN 473-708) by Yanovskaya (1954). I was unable to find these additional specimens in the PIN collection, however, I am reasonably certain that it would be difficult or impossible to narrow down the specific identity of most of them due to their fragmentary condition. One of these, PIN 473-139, is a partial mandible with left p2–m2 that Yanovskaya (1954: fig. 7 and pl. 3) figured; however, based on these figures I am unable to differentiate that specimen from mandibles of Embolotherium andrewsi or E. grangeri. Presently, there are no mandibles or lower dentitions associated with diagnostic P. efremovi fossils.

To summarize, the specimens that are relevant to the diagnosis and characterization of Protembolotherium efremovi are three skulls: the holotype (PIN 473-311) is missing the entire frontonasal region; a second specimen (PIN 473-310) has an Embolotherium-like ram and an additional horn above the orbit. Finally, the third specimen (PIN 3109-40), in addition to having a ram and additional horns, retains a short nasal process projecting from the peak of the ram. It is necessary to assess the cranial morphology of the type (PIN 473-311) in more detail to determine what features (if any) distinguish it from other brontothere species, and if there is truly enough evidence to conclude that the remaining skulls (with the more diagnostic elements that are not preserved in the type) are indeed the same species as the type.

Despite the lack of preservation of the ram in the type (PIN 473-311), certain characteristics of this specimen show a combination of states very similar to Embolotherium andrewsi and E. grangeri; this includes the very widened dorsal cranial surface, the fully saddle-shaped skull, prominent swellings in the zygomatic arches, the high relief of the lingual premolar cusps, the relatively deep central molar fossa, and the well-developed anterolingual cingular cusps on the molars and the sharp posteromedial angle of the external auditory pseudomeatus. No other brontotheres share this last condition except two small North American brontotheres (Dolichorhinus, Sphenocoelus) that are otherwise clearly very different from PIN 473-311. Therefore, it is evident that the type of P. efremovi belongs to an Embolotherium-like species.

On the other hand, the type of P. efremovi exhibits a number of features that are structurally very different from both Embolotherium andrewsi and E. grangeri. For instance, the skull is not as deeply saddle-shaped. The occiput is shorter and not tilted backward to the degree seen in E. andrewsi or E. grangeri. In other respects the occiput is similar to E. grangeri, but it is not nearly as massive as that of E. andrewsi. The nuchal crest is not as thick and rugose, the occipital pillars are not nearly as prominent, and the dorsal portion of the occiput is not broader than the ventral portion. Further differences with E. andrewsi include the much less prominent parasagittal ridges and more laterally bowed zygomatic arches. The rostrum differs substantially from both species of Embolotherium. The dorsal margin of the rostrum is steeply sloped posterodorsally, so that the rostrum deepens proximally while those of E. andrewsi and E. grangeri do not. Furthermore, the rostrum of P. efremovi is much broader than that of E. andrewsi. At the same time, the rostrum of PIN 473-311 is not nearly as elongate as that of E. grangeri and there is no constriction at the postcanine diastema. Moreover, the nasal cavity of E. grangeri is much longer with the nasal incision extending above the orbits, whereas in PIN 473-311 the nasal incision appears to be entirely anterior to the orbit. Finally, though they are not completely preserved, the canines of PIN 473-311 seem larger and more developed than the globular incisiform or cuplike vestigial canines of E. andrewsi and E. grangeri.

The additional skulls referred to Protembolotherium efremovi (PIN 473-310 and 3109-40), to the extent that they are preserved, share the same combination of similarities and differences with Embolotherium andrewsi and E. grangeri that are seen in the type specimen of P. efremovi (PIN 473-311). Based on the morphological congruence of these three skulls it is probable that they represent the same species. Unfortunately, it is not feasible to compare the type of P. efremovi with Nasamplus progressus from the “Ulan Gochu” faunal zone (sensu Radinsky, 1964) of Inner Mongolia, a species known only from a single cranial fragment. However, the two skulls referred to P. efremovi, PIN 473-310 and PIN 3109-40, are quite distinct from Nasamplus, primarily (to the extent that they can be compared) because of their well-developed rams with diminished (or absent) nasal processes.

Protembolotherium efremovi is the third valid species found to possess the “battering ram” that was first discovered in the genus Embolotherium. The extremely derived and bizarre ram structure was initially interpreted by Osborn (1929b) as having a separate origin from the paired frontonasal horns of other brontotheres. Later Granger and Gregory (1943) interpreted the ram as homologous with the frontonasal horns of other brontotheres. More recently reported evidence suggests that Granger's and Gregory's (1943) idea is the better hypothesis (Mihlbachler et al., 2004a; Mihlbachler and Solounias, 2004). The evolution of the ram of Embolotherium and Protembolotherium is best explained by a morphological transformation series starting from a condition similar to that of Metatitan where the paired frontonasal horns and the free end of the nasal process are elevated on a frontonasal pillar. The ram itself is a grossly amplified frontonasal process that initially formed the base of the horns and the frontonasal pillar, the portion of the skull upon which the horns rested. The original free end of the nasal process is lost or, in the words of Granger and Gregory (1943), “absorbed” by the ram. The distal surface of the ram is homologous to the peaks of the frontonasal horns, fused into a transverse crest. Some evidence for this is seen in the apparent remnants of a sutural pattern of facial bones in Embolotherium andrewsi and Nasamplus progressus. Nasamplus represents an intermediate condition one step derived from Metatitan and Aktautitan, in which the frontonasal horns have been fused into a single transverse crest at the top of an enlarged frontonasal process, but the free end of the nasal process is still present and unreduced. Protembolotherium efremovi represents a further derived condition in which the ram is more fully developed, although it is still shorter than that of E. andrewsi and has shallower lateral margins that do not constrict the nasal cavity as strongly. Perhaps more significantly, the process that protrudes from the peak of the ram of PIN 3109-40 seems to be the reduced remnant of the free end of the nasal bone. If this interpretation is correct the small hornlike protuberances above the orbits of PIN 3109-40 and PIN 473-310 are secondary horns and they are not homologous with the paired frontonasal horns of other horned brontotheres.

If Yanovskaya (1954, 1980) and I are correct in referring PIN 3109-40 and PIN 473-310 to the same species, this implies that the nasal process, which is still present in Nasamplus but lost or absorbed in Embolotherium, was occasionally present in a diminished state (PIN 3109-40) but also was occasionally absent (PIN 473-310) in Protembolotherium efremovi. It seems obvious that, as with Embolotherium, the ram of this species would have undergone considerable ontogenetic change throughout life. After all, it seems very unlikely that embolotheres could have given birth to calves with fully developed rams. It is entirely possible that the loss of an already diminished nasal process was part of the ontogenetic process. Unfortunately, both PIN 473-310 and PIN 3109-40 are edentulous, rendering it impossible to determine the relative ontogenetic ages of these skulls even though the lack of unfused sutures in these skulls suggests that all of the skulls were well into their adult years. The North American species Megacerops coloradensis (sensu Mihlbachler et al., 2004b) is known from hundreds of specimens and exhibits considerable variation in the size and shape of the nasal process. In that species the variation is clearly related to the size of the frontonasal horns (Mihlbachler, 2004). In smaller specimens of M. coloradensis with more gracile horns the nasal process is well developed and takes on more typical proportions; however, in many of the larger specimens with huge swollen frontonasal horns the nasal process is reduced to a small triangular remnant and most of it seems to have been absorbed by the bony growth related to the massive horns. In Megacerops, this variation seems related to sexual dimorphism. Given these possible explanations (ontogeny, sexual dimorphism) for variation in the nasal process, it is difficult to justify the differences between PIN 473-310 and PIN 3109-40 as taxonomically significant (i.e., indicating two species). Therefore I presently accept Yanovskaya's (1954, 1980) conclusions that these specimens are conspecific and that both belong to Protembolotherium efremovi.

Embolotherium andrewsi Osborn, 1929b

Holotype

AMNH 26001, a skull missing the premaxillomaxillary rostrum and right zygomatic arch, with fragmentary right M2–M3 and left M3.

Type Locality

“Ulan Gochu”, Urtyn Obo, “middle white” or “gray” beds, 125 feet below Baron Sog unconformity, Shara Murun Region, Inner Mongolia, China. (Ulan Gochu [used in quotes] denotes uncertain correlation with the type Ulan Gochu beds at Baron Sog and refers to a faunal zone rather than a formation. [Radinsky, 1964])

Age

Late Eocene (Ulangochuian land mammal “age”).

Synonyms

Embolotherium ultimum Granger and Gregory, 1943; Embolotherium ergilense Dashzeveg, 1975.

Referred Specimens

(From the “Ulan Gochu” fanual zone [sensu Radinsky, 1964] of the Shara Murun Region, Urtyn Obo and East Mesa, Inner Mongolia) AMNH 20352, an occipital fragment; AMNH 26011, a left mandibular ramus and symphysis with right p3 and p4 (partial), left p3–p4 (partial), and m1–m3; AMNH 26000 (specimen lost, but photos exist in vertebrate paleontology archives at the AMNH), a skull with right P1–M3 and left P2–M3; AMNH 26003, a skull with right P2–M3 and left C–M3; AMNH 26006, a left ramus with p2–m3; AMNH 26007, a partial mandible with partial right ramus and complete left ramus with right p2–p4, left p2 (partial), and p3–m3; AMNH 26008, a right ramus with p2–m3; AMNH 26009, a skull with right I1–M3, left I2 (?), P1–M3, and a mandible with right i1, i2–i3 (roots), c–m3, left i2, i2–c (roots), and p2–m3; AMNH 26010, a crushed skull missing the frontonasal process with right I3–M3 and left C–M3; IVPP V11959, a skull with right P2–P4, M2–M3, left M2–M3, and a mandible with right i1?, p2–m3 and left i1–i3, and p2–m3; PIN 2200-1, an edentulous skull; PIN 2200-2, a mandible with right and left m3; (from the Baron Sog Formation at Baron Sog Mesa, Inner Mongolia) AMNH 21604 (holotype of Embolotherium ultimum), a crushed basicranium with right M2 (partial) and M3; AMNH 22114: left maxillary fragment with P3–P4, M1 (partial); (from the Ergilin Dzo Formation, Outer Mongolia) National Museum in Ulan-Bator, item 10 (holotype of Embolotherium ergilense), a skull missing the premaxillae and most of the maxillary dentition.

Diagnosis

Embolotherium andrewsi is a very large brontothere with a tall and nearly vertical process of bone (“ram”) that originates above the orbits and appears to be composed of the frontal and nasal bones. At the peak of the ram is a rugose transverse crest that forms the distal margin of nasal channel that runs along the anteroventral surface of the ram. The nasal incision extends posteriorly to the anterior margin of P4. The orbits are positioned above the M2 and the posterior portion of M1. A horizontal nasal process is absent. The premaxillomaxillary rostrum is small and short, it slightly thickens posteriorly, and is not enclosed by bone dorsally. A pair of tall, thin bony ridges is seen on the dorsal surface of the rostrum. Other cranial characteristics include a deeply saddle-shaped cranium, an extremely robust occiput, very prominent parasagittal ridges that overhang the sides of the cranium and do not constrict the dorsal surface of the skull posteriorly, and weakly curved zygomatic arches with massive swellings. The external auditory pseudomeatus enters the skull in a posterolateral direction and is ventrally constricted. The posterior nares are shifted behind M3. Large ventral sphenoidal fossae are absent.

Dentally, Embolotherium andrewsi has three very small incisors that form a semicircular row. The canines are irregular and they are not always fully erupted. There is a distinct P2 metacone. Hypocones are present on P2–P4 and strongly connected to the protocone by lingual crests. The molars of E. andrewsi have tall, lingually angled ectolophs with weak labial ribs, and thinned lingual ectoloph enamel with wedge-shaped paracones and metacones. A cingular parastyle shelf is absent. Deep central molar fossae and large anterolingual cingular cusps are present. Paraconules and metalophs are absent. The lower dentition of E. andrewsi includes three small incisors that form an arch anterior to the canines. Metaconids are present on p3, p4, and occasionally on p2. The p2 trigonid is only marginally longer than the talonid. The lower molars have deep valleylike basins and m3 is elongate.

Embolotherium andrewsi can be distinguished from E. grangeri in the following ways: ram more vertical, positioned above orbits; nasal channel of ram distally emarginated; rostrum and incisors much smaller; dorsal premaxillary ridge smaller; occiput more massive; zygomatic arches not strongly bowed laterally; parasagittal ridges more prominent; premolar hypocones less strongly separated from protocones; M3 hypocone is smaller; and cingulum of M3 continuous around the distolingual corner of the crown.

Description

Skull

The holotype of Embolotherium andrewsi (AMNH 26001) is a very large skull that is missing the left zygomatic arch, the rostrum, and most of the dentition (fig. 117). Except for the missing portions, the skull is well preserved and undistorted. A few areas are filled in with plaster where the skull surface is not preserved. In particular, the dorsal surface of the skull is made up of numerous fragments with small plaster-filled gaps between them, although more intact specimens show a similar morphology. More complete skulls of E. andrewsi include AMNH 26003 (fig. 118b, c, and fig. 119a), AMNH 26009 (fig. 118a, and see Mihlbachler et al., 2004a: fig. 19k, l), and IVPP V11959 (see Wang [2000] for figure). (Note that the published photo of AMNH 26003 in Osborn [1929b: fig. 5] is unreliable because the specimen was actually photographed while it was wrapped with plaster).

Figure 118

Skulls of Embolotherium andrewsi. (A) right face of AMNH 26009, (B) Right view of AMNH 26003, (C) dorsal view of AMNH 26003.

Figure 119

Ventral view of skull and upper dentition of Embolotherium andrewsi. (A) Ventral view of AMNH 26003, (B) right molars of AMNH 26003, (C) left premolars of AMNH 26003, (D) right premolars of AMNH 26010.

Embolotherium andrewsi is one of the largest brontotheres, rivaled in size only by E. grangeri, Megacerops coloradensis (sensu Mihlbachler et al., 2004b), and possibly Gnathotitan berkeyi. The most conspicuous feature of E. andrewsi is the large battering ram– like structure (henceforth referred to as the “ram”) that has the superficial appearance of an enlarged and upturned nasal process. From the lateral view the ram originates just above the orbits. It curves posteriorly as it rises from the skull so that at its peak it is nearly vertical. The rams of the referred skulls tend to be less vertical than the holotype. The anteroventral surface of the ram is convex while the posterodorsal surface is concave. The ram broadens distally. A deep nasal channel runs along the anteroventral surface of the ram. The nasal channel is bordered laterally by deep and thick walls that line the side of the ram. Toward the proximal end of the ram the lateral sides are deep and constrict the nasal channel. At the distal end of the ram a thick rim of bone forms a distinct distal border for the nasal channel. The nasal channel broadens distally and is bifurcated slightly at the distal end by a short wedge of bone below the distal margin. The inner surface of the nasal channel is bisected its entire length by a thin ridge of bone that runs along the midline; this appears to represent the osteological marker for the cartilaginous nasal septum that extended to the distal peak of the ram (Mihlbachler and Solounias, 2004).

Reconstructions of Embolotherium andrewsi in Osborn (1929a, 1929b) depict the ram as a hornlike process, with the nostrils positioned very low in a normal rhinolike position just above the premaxilla. However, the deep channel on the anteroventral surface of the ram and the ossified marker for the nasal septum indicate that the nasal cavity extended to the peak of the ram. Wang (2000) depicted the nostrils as elevated to the peak of this structure although the true position of the nostrils is uncertain.

The distal margin of the ram forms a broad transverse crest that is marked by a coarse rugosity on the anterior and dorsal surfaces of the distal margin. The upper surface of the distal margin of the ram is slightly convex from the anterior view. From the dorsal view the distal margin is thinner medially and thicker laterally. Closer inspection of the peak of the ram reveals that in most specimens a weak but distinct groove on the posterolateral side of the distal rim continues down the side of the ram for several centimeters. The groove cannot be traced farther down the ram beyond a few centimeters from its peak; however, this groove resembles the frontonasal contact seen on the frontonasal crest of Nasamplus progressus. If this groove does represent a sutural remnant, then it indicates that the frontal bone rises to the peak of the ram and that the ram is actually a frontonasal process, with the posterodorsal surface formed from the frontal bone and the anteroventral surface from the nasal bone. This frontonasal process is structurally similar to those of Aktautitan, Metatitan, and Nasamplus, although it is much taller and more massive, as are those of Protembolotherium efremovi and E. grangeri. The actual nasal process (which normally extends horizontally from the peak of the frontonasal process) appears to have been lost in Embolotherium andrewsi.

Due to the large frontonasal ram the nasal incision of Embolotherium andrewsi is dorsoventrally very deep. In specimens such as AMNH 26003 and AMNH 26009, which have the premaxillomaxillary rostrum complete, the nasal incision extends posteriorly to the anterior margin of P4. The posterior margin of the nasal incision is anterior to the orbit, and unlike E. grangeri the nasal incision does not extend over the orbit. The orbits of E. andrewsi are positioned directly above the M2 and the posterior portion of M1. The anterolateral root of M1 is positioned directly below the anterior rim of the orbit.

The premaxillomaxillary rostrum is complete in AMNH 26003 and AMNH 26009, although more details can be seen in the latter. From a lateral view the rostrum of AMNH 26003 is strongly curved upward, but in AMNH 26009 the rostrum is straighter. Premaxillomaxillary sutures are not discernable on any specimen. As a whole, the premaxillomaxillary process is short, shallow, and narrow. It is small in comparison to those of Protembolotherium efremovi and Embolotherium grangeri. The rostrum deepens slightly posteriorly. The premaxillary symphysis is not ossified and the premaxillae are separated by a small median gap. On the dorsal surface of the rostrum of AMNH 26009 are two pairs of thin, roughened ridges of bone. The smaller pair is at the distal end of the rostrum and the larger pair is positioned at about the midpoint of the rostrum. The medial surfaces of these ridges are flat. These ridges of bone probably served as osteological supports for the tall, cartilaginous nasal septum that would have extended to the peak of the ram, although they are not as thickened and rugose as the equivalent dorsal rostral ridges seen on E. grangeri.

The entire dorsal surface of the skull of Embolotherium andrewsi is deeply saddle-shaped. The parasagittal ridges are very prominent and greatly overhang the sides of the cranium. The extent to which the parasagittal ridges overhang the skull is most pronounced in the holotype (AMNH 26001) and not quite as pronounced in other specimens. From a dorsal view the parasagittal ridges are widely separated throughout their length, although they barely constrict the dorsal surface posteriorly.

The zygomatic arches are massive. From the dorsal view the zygomatic arches are essentially straight and parallel except for a large swelling at the junction of the jugal and squamosal bones. Among specimens of E. andrewsi the size of the swelling is variable. They are largest in AMNH 26001 and somewhat smaller in AMNH 26009 and AMNH 26010.

From the lateral view the jugal and squamosal portions of the zygomatic arch have similar depths. There is a small ventral flange below the massive swelling. The jugal portion of the zygomatic is essentially horizontal, although the squamosal portion is angled slightly upward, thus giving the zygomatics a weak curvature.

The occiput of Embolotherium andrewsi is far more massive than that of E. grangeri and Protembolotherium efremovi and rivals the occiputs seen in the largest specimens of Megacerops (sensu Mihlbachler et al., 2004b). From a dorsal view the nuchal crest is thick and rugose. It is also concave and very deeply notched medially. From a lateral view the occiput is strongly titled backward. From a posterior view the dorsal margin of the occiput is notched medially. The dorsal portion of the occiput is wider than the ventral portion. The lateral wings of the massive nuchal crest are supported by massive bony pillars and the center of the occiput is deeply recessed in the middle.

The ventral surface of AMNH 26003 is the most complete and best preserved among the available specimens (fig. 119a). The posterior nares of Embolotherium andrewsi are positioned completely behind the M3s in all specimens. Direct inspection of the specimens reveals a faint ridge of bone arching around the anterior and lateral margins of the posterior nares. This ridge indicates a horseshoe-shaped emargination similar to that seen, for instance, in Metatitan relictus, although in E. andrewsi this emargination is wider. The posterior narial canal itself is rather short in comparison to most other brontotheres. The vomer, which bisects the posterior narial canal, can be seen in AMNH 26003. The posterior narial canal seems to extend slightly into the sphenoid, but large ventral sphenoidal pits like those of Protitan or Metatitan are distinctly absent. The basicranium is not greatly widened or anteroposteriorly abbreviated like that of Metatitan. The configuration of the basicranial foramina is typical with a widely separated foramen ovale and foramen lacerum. The mastoid process is shorter than the postglenoid process and it curves anteroventrally, thus contacting the postglenoid process and forming a tube-shaped external auditory pseudomeatus. The external auditory pseudomeatus also extends into the basicranium in a posteromedial direction, a condition shared with E. grangeri and Protembolotherium, and a few other more primitive taxa such as Rhinotitan kaiseni, Dolichorhinus, and Sphenocoelus.

Upper Dentition

The teeth of the holotype (AMNH 26001) were not recovered except for some of the molars, but even these are fragmented. Therefore, the description of the dentition is based primarily on referred specimens, AMNH 26003 (fig. 119a–c), AMNH 26010 (fig. 119d), and AMNH 26009 (fig. 120).

Figure 120

Anterior upper dentition of Embolotherium andrewsi (AMNH 26009).

The number of incisors is unreduced (three pairs). There are six intact small incisor alveoli (not shown) in AMNH 26010. The complete left incisor row is preserved in AMNH 26009. The incisors of Embolotherium andrewsi are very small, are separated by large gaps, project straight down, and form a semicircular row anterior to the canines. Because of the narrow premaxillomaxillary rostrum the diameter of the anterior tooth row is less than the width across the P1s. In AMNH 26009 a fragment of enamel is impacted into the left side of the palate medial to the diastema between I3 and P1. This is presumably a vestige of the canine. In AMNH 26003 and 26010 the canines are present, but they are only slightly larger than the incisors, and they appear to be only partially erupted, even though these specimens are clearly adults. The canine crowns have a peculiar cup-shape. The tendency for the canines to be small, highly irregular, and incompletely unerupted suggests that these teeth are truly vestigial.

In AMNH 26010 and AMNH 26003 there are very short diastemata between the I3 and canine, and between the canine and P1. Most of the available specimens of Embolotherium andrewsi exhibit heavy cheek-tooth wear. AMNH 26003 and AMNH 26010 have the least worn cheek teeth. The P1 is much smaller than the other premolars. The P1 crown is nearly as wide as it is long, but further details of its morphology are obliterated by wear. The remaining premolars are nearly rectangular in outline, although the anterior margins of the premolars are concave. The parastyle of the P2 is directed anteriorly, although the parastyles of P3–P4 are oriented anterolabially. The P2–P4 metastyles are heavily worn, but they appear to have been directed slightly lingually or they were straight. Very small labial paracone ribs are present on P2–P4. These ribs become progressively narrower in consecutively posterior premolars. None of the premolars has a mesostyle. Labial premolar cingula are generally very faint or absent.

The lingual features of P2 always include a protocone and a slightly smaller hypocone. Preprotocristae are absent or so weak that they are not readily discernable. In AMNH 26010 the hypocone of P2 is slightly smaller than the protocone and completely disconnected from it and there is no trace of a lingual crest. However, on AMNH 26003 the protocone and hypocone are connected by a lingual crest. In AMNH 26010, the P3 protocone and hypocone of P2 are more closely positioned and are fully connected by a wide, tall lingual crest. The crest absorbs the cusps to such a great degree that they are barely discernable as distinct cusps. In other specimens, such as AMNH 26003, the paracone and hypocone of the P3 are more widely separated and more distinct, but a tall lingual crest always connects them. In the P4 of AMNH 26010 the protocone and hypocone are distinct cusps, they are positioned closely together, the hypocone is slightly smaller than the protocone, and it is not connected to it by a lingual crest. In other specimens, such as AMNH 26003, a distinct P4 connecting crest is always present. The texture of the surface of the enamel on the lingual side of the premolars crowns varies from highly crenulated to smooth. The lingual premolar cingulum varies from continuous around the lingual side of the protocone to slightly discontinuous.

One specimen, AMNH 22114, a maxillary fragment with an unworn P3–P4, has a small crest of enamel that stretches from the posterior slope of the hypocone to the posterior cingulum. This crest is anomalous and does not appear on any other Embolotherium andrewsi specimen.

The upper molars of Embolotherium andrewsi show typical brontotheriine traits, including tall, lingually angled ectolophs (the ectolophs of E. andrewsi are among the tallest among brontotheres), very weak labial ribs, thin lingual ectoloph enamel, and wedge-shaped lingual sides of the paracone and metacone in molars that are not heavily worn. The anterior cingulum is thin and passes proximally to the distal peak of the parastyle. Deep central molar fossae and large anterolingual cingular cusps are present. All evidence of paraconules and metalophs is lost. There is always a hypocone on the M3 of E. andrewsi although it varies in morphology from a functional cusp, though smaller and shorter than the paracone (e.g., AMNH 26003), to a series of two or three tiny cusps (e.g., AMNH 26010). Labial molar cingula are weak while lingual molar cingula are absent. The cingulum of the M3 continuously wraps around the distolingual corner of the crown in all specimens of E. andrewsi.

Mandible and Lower Dentition

Two Embolotherium andrewsi skulls are associated with mandibles, AMNH 26009 (fig. 121e, f) and IVPP V11959 (see Wang, 2000). Based on comparison with these mandibles several other mandibles are referable to E. andrewsi, including AMNH 26008, a specimen with lightly worn cheek teeth (fig. 121a–d). The horizontal ramus of AMNH 26008 is shallow but it deepens posteriorly. The inferior margin of the symphysis is angled slightly less than 45°. Overall, the ascending ramus is very short and the coronoid process is short and moderately curved backward. The symphysis, most nearly intact in AMNH 26009, is broad and extends to the m1 trigonid, but in other specimens it extends only to the p4 metaconid (e.g., AMNH 26008).

Figure 121

Mandibles and lower dentition of Embolotherium andrewsi. (A) Right view of AMNH 26008, (B) right p2–p4 of AMNH 26008, (C) medial (lingual) view of right ramus of AMNH 26008, (D) left molars of AMNH 26008, (E) dorsal view of AMNH 26009, (F) lingual view of incisors and canine of AMNH 26009.

The incisors of AMNH 26009 are small and form a semicircular arch anterior to the canines. Only the crowns of the i1s are preserved, but they are so worn that they cannot readily be described although they appear to have been somewhat semispatulate. The less worn incisors of IVPP V11959 also suggest a semispatulate shape. The lower canines are fully erupted (unlike the upper canines which sometimes do not fully erupt) and they are not much larger than the lower incisors. The canine crown is blunt. All of the incisors, the canine, and the p1 are separated by very short diastemata. The anterior tooth row forms a broad semicircular arch. In other words, the canines are anterior to p1 and they are not shifted lingually as are the upper canines.

AMNH 26009 includes a small p1 although it is too worn to describe. No other specimen of E. andrewsi has a p1 preserved with it. The description of the remaining cheek teeth (p2–m3) is based on the more nearly pristine lower dentitions of AMNH 26006, AMNH 26007, and AMNH 26008. The trigonid of the p2 is only slightly longer than the talonid, but the talonids of p3 and p4 are somewhat longer than the trigonids. The talonids of p2–p4 are all wider than the trigonids. The paralophid and protolophid of the p2 are angled somewhat lingually, thus creating a narrow but long lingual trigonid notch. The trigonids of p3 and p4 are essentially molariform with fully lingually arched paralophids and protolophids and a deep and broad lingual notch. A small lingual bulge in the p2 at the junction of the protolophid and cristid obliqua of AMNH 26008 can be interpreted as a metaconid, although other specimens lack a distinct p2 metaconid. Large, lingually positioned metaconids can be found on the p3 and p4. The p2–p4 talonids of E. andrewsi have well-developed cristids obliqua and hypolophids. The talonid basins of the p2–p4 progressively broaden posteriorly and have deep talonid valleys similar to those of the molars. Labial premolar cingulids are weak while lingual premolar cingulids are absent.

The thin lingual enamel of the lower molars and the elongate m3 are typical brontotheriine characters. However, in Embolotherium andrewsi the trigonids and talonids of the lower molars form deep valleys rather than the broad shallow basins that characterize most other brontotheres. Lingual molar cingulids are absent and the labial cingulids tend to be discontinuous around the labial bases of the trigonids, talonids, and hypoconulid heel, but the cingulids can be relatively thick between them.

Embolotherium grangeri Osborn, 1929b

Holotype

AMNH 26002, a skull with right and left P1–M3, and three isolated upper incisors.

Type Locality

“Ulan Gochu”, East Mesa, “middle red” beds, Shara Murun Region, Inner Mongolia, China. (Ulan Gochu [used in quotes] refers to a faunal zone rather than a formation. [Radinsky, 1964])

Age

Late Eocene (Ulangochuian land mammal “age”).

Synonyms

Embolotherium louksii Osborn, 1929b; Titanodectes ingens Granger and Gregory, 1943; Titanodectes minor Granger and Gregory, 1943; Embolotherium insigne Yanovskaya, 1980.

Referred Specimens

(From the “Ulan Gochu” faunal zone [sensu Radinsky, 1964] of the Shara Murun Region, East Mesa, Inner Mongolia) AMNH 26004, the complete ventral surface of a skull with left and right P2–M3 and the distal portion of the ram; AMNH 26040, the anterior portion of a juvenile skull with right DP2–DP4, M1, left P1, DP2–DP4, M1, an associated pair of premaxillae with unerupted incisors, numerous isolated deciduous incisors, and a mandible with canines and incisors (unerupted), right and left p1, dp2–dp4, and m1; (from the “Shara Murun” Formation [sensu Radinsky, 1964] of the Shara Murun Region, East Mesa, Inner Mongolia) AMNH 26005 (holotype of Titanodectes ingens), a mandible in three pieces, right ramus, left ramus, and a symphysis with right i1–m3, left i1–c, and p3–m3; AMNH 26132 (holotype of Titanodectes minor), a partial mandible with right p3–m1, left p2–m1, and a separate symphysis fragment with right i1–c and left i2–i3; (From the Ulan Gochu Formation of the Shara Murun Region, Baron Sog Mesa, Inner Mongolia) AMNH 21610 (holotype of Embolotherium louksii), a partial skull with erupting ?M3; (from the “Baron Sog” Formation [sensu Radinsky, 1964] of the Shara Murun Region, East Mesa, Inner Mongolia) AMNH 26018, a mandible fragment with right p4–m3 and isolated (?)left incisors and canine; (from the Shara Murun Formation of the Shara Murun Region, Baron Sog Mesa, Inner Mongolia) AMNH 21600, a mandible with right i1–i2, p2–m3, left i2–c, and p2–m3 (from the Ergilin-Dzo Svita at Khoer Dzan, Outer Mongolia) PIN 3110-52 (holotype of Embolotherium insigne), a partial skull with left molars. The PIN also holds a large but uncatalogued collection of Embolotherium grangeri from Khoer Dzan (to be described in a later publication), including skulls from individuals of numerous ontogenetic stages.

Diagnosis

Embolotherium grangeri is a very large brontothere with a tall process of bone (ram) that rises from behind the orbits at approximately a 45° angle. The rugose transverse crest at the peak of the ram does not form a distal emargination for the nasal channel. The nasal incision extends above the orbits to the anterior margin of the M3. The orbits are positioned above M2. The premaxillomaxillary rostrum is long, deepens slightly proximally, and is not fully enclosed by bone dorsally. A thick U-shaped ridge of bone is seen on the dorsal surface of the premaxilla. Other cranial characteristics include a saddle-shaped cranium, an unthickened occiput, prominent parasagittal ridges that do not constrict the dorsal surface of the skull posteriorly, and weakly curved and laterally bowed zygomatic arches with massive swellings. The external auditory pseudomeatus enters the skull in a posterolateral direction and is ventrally constricted. The posterior nares are shifted posteriorly behind M3. Large ventral sphenoidal fossae are absent.

Dentally, Embolotherium grangeri has three relatively large but globular upper incisors that form a semicircular row. The canines are very small. There is a distinct P2 metacone. Hypocones are present on P2–P4 and are either weakly connected to the protocone by a lingual crest or they are completely separated from it. The molars of E. grangeri have tall, lingually angled ectolophs with weak labial ribs, and thinned lingual ectoloph enamel with wedge-shaped paracones and metacones. A cingular parastyle shelf is absent. Deep central molar fossae and large anterolingual cingular cusps are present. Paraconules and metalophs are absent. The lower dentition of E. grangeri includes three large semispatulate incisors that form an arch anterior to the canines. The i2 is the largest incisor. A metaconid is present on p3, p4, and occasionally on p2. The p2 trigonid is only marginally longer than the talonid. The lower molars have deep valleylike basins and the m3 is elongate.

Embolotherium grangeri can be distinguished from E. andrewsi in the following ways: ram less vertical, positioned behind orbits; nasal channel of ram not distally emarginated; rostrum and incisors much larger; dorsal rostral ridge larger; occiput less massive; zygomatic arches strongly bowed laterally; parasagittal ridges less prominent; premolar hypocones more strongly separated from the protocones; M3 hypocone larger; and cingulum of M3 discontinuous around the distolingual corner of the crown.

Description

Skull

The holotype of Embolotherium grangeri (AMNH 26002) is a complete and nearly undistorted cranium with only minor fragments of missing bone that have been replaced with plaster (fig. 122). The most significant damage to the holotype specimen is a large crack on the dorsal and lateral surfaces of the skull just behind the base of the ram. The bone anterior to this crack is depressed about a centimeter, suggesting that the ram has been forced downward. However, comparison of this skull with other specimens seems to confirm that the angle of the ram is essentially intact. Embolotherium grangeri has never been described or figured from the fully prepared type specimen. Unfortunately, the only previously published photos and drawings of AMNH 26002 (Osborn, 1929a: fig. 797; 1929b: figs. 3b, 7) were made while the specimen was still in its plaster wrappings and some aspects of its shape in these older figures are misleading (relating to pieces of sediment that had been wrapped with the specimen). In addition to the holotype there is an additional adult specimen of a larger individual that consists of a ventral surface of the skull and the distal segment of the ram (AMNH 26004) (fig. 123), and three partial subadult skulls (AMNH 21610, AMNH 26040, and PIN 3110-52). E. grangeri is similar in size to E. andrewsi and shares with it a massive upturned ram. The ram of E. grangeri originates from behind the orbits, rather than from above as in E. andrewsi. The ram projects from the skull at approximately 45°. From a lateral view the ram is nearly straight, but it is slightly curved downward. The posterodorsal surface is convex and the anteroventral margin is slightly concave. Like E. andrewsi the lateral walls of the ram are deep and thick, forming a ventral nasal channel that runs to the distal end of the ram. The lateral walls are about the same depth throughout the length of the ram although they become slightly shallower distally. Proximally, the lateral walls are shallower than those of E. andrewsi and they do not constrict the nasal channel. A thin ossified ridge (osteological marker for the cartilaginous nasal septum) runs along the anteroventral surface of the ram and extends to the distal end. As in E. andrewsi, the nasal channel and ossified marker for the nasal septum indicate that the nasal cavity would have extended to the distal end of the ram, although the actual position of the nostrils is uncertain. However, because of the more posterior position of the ram and its shallower angle, the space between the ram and the rostrum is tremendous, suggesting an enormous nasal cavity (Mihlbachler and Solounias, 2004). The nasal channel is bifurcated distally by a wedge-shaped process that extends from the anteroventral surface at the distal end of the ram. The wedge-shaped process and distal bifurcation of the nasal channel is best seen in the ram of AMNH 26004 (fig. 123b). This structure appears as a median bulge from the dorsal view of AMNH 26002 (fig. 122b). The strong rim of bone that emarginates the distal end of the nasal channel in the ram of E. andrewsi is absent in adult specimens of E. grangeri. Instead, the nasal channel is open distally. However, like E. andrewsi, the distal surface of the ram is very rugose, particularly in AMNH 26004.

Figure 122

The holotype of Embolotherium grangeri (AMNH 26002). (A) Left view, (B) dorsal view, (C) dorsal view of rostrum, (D) anterior view, (E) posterior view.

Figure 123

Distal ram fragment of Embolotherium grangeri (AMNH 26004). (A) Posterodorsal surface, (B) distal surface, (C) anteroventral surface.

The ram of Embolotherium grangeri superficially resembles the nasal processes of other brontotheres. In particular, the lateral walls, deep ventral nasal channel, and distally widening shape resemble the nasal processes of brontotheres such as Epimanteoceras formosus, although the ram is obviously longer, more massive, and more steeply angled upward. There are no discernable sutures on any of the previously described specimens of Embolotherium grangeri, and this has led to some ambiguity in interpreting the structural makeup of the ram of E. grangeri (Mihlbachler et al., 2004a). However, the uncatalogued collection of E. grangeri from Khoer Dzan at PIN contains specimens that clearly reveal the ram of E. grangeri to be structurally homologous to the rams of E. andrewsi and Protembolotherium efremovi, and with the paired frontonasal horns of other brontotheres.

The nasal incision of Embolotherium grangeri is very deep. The posterior margin of the nasal incision extends above the orbit and is directly above the anterior margin of M3. The orbit is positioned above M2, with the anterolateral root of M2 and posterolateral root of M1 situated below the anterior rim of the orbit.

The dorsal margin of the rostrum is long and deepens slightly posteriorly. The dorsal margin is nearly horizontal and does not rise above the midpoint of the orbit. The premaxillae are completely fused at the symphysis and the premaxillomaxillary sutures are not discernable. The premaxillary symphysis is nearly vertical. Above the symphysis is a tall and thick U-shaped ridge of bone. This ridge is thicker and more rugose than the anterodorsal rostral ridge seen in Embolotherium andrewsi. The sidewalls of the rostrum are parallel and constrict the dorsal opening of the premaxillomaxillary chamber to form an elongate narrow channel along the dorsal rostral surface.

The dorsal surface of the cranium is deeply saddle-shaped. The parasagittal ridges are widely separated throughout their length and do not significantly constrict the dorsal surface posteriorly. The parasagittal ridges are less prominent than those of Embolotherium andrewsi and only slightly overhang the sides of the cranium. The zygomatic arches of E. grangeri resemble those of E. andrewsi in having massive swellings at the junction of the jugal and squamosal. From a lateral view, the jugal portion of the zygomatic is horizontal and it is not as deep as that of E. andrewsi. The squamosal portion of the zygomatic is deeper and slightly sloped upward, giving the zygomatic arch a weak curvature. There is a small ventral flange below the large zygomatic swelling. From the dorsal view the zygomatic arch is thinner than that of E. andrewsi except for the large central swelling. Additionally, the zygomatics are more strongly bowed laterally.

From a lateral view the posterior end of the cranium is similar to that of Embolotherium andrewsi. The dorsal surface is strongly curved upward and the occiput is strongly tilted backward. However, the occiput and nuchal crest of E. grangeri are neither as massive nor as rugose as those of E. andrewsi. From the dorsal view of the skull the nuchal crest is concave, but it is not as deeply notched as that of E. andrewsi, and the nuchal crest is comparatively thin. From the posterior view the occiput is nearly square. The dorsal margin of the occiput is essentially flat. The dorsal portion of the occiput is as wide as the posterior portion and the occiput is not constricted in the middle. The bony pillars on the surface of the occiput are not as massive as those of E. andrewsi and the central depression of the occiput is not as deep.

Like Embolotherium andrewsi the posterior nares of E. grangeri are positioned completely behind M3 (fig. 124a). Paired ventral sphenoidal fossae are absent. The proportions of the basicranium are essentially normal. The transverse width across the mastoid processes is less than the transverse distance across the M3s. The configuration of the basicranial foramina are typical, with a widely spaced foramen ovale and foramen lacerum. Like in E. andrewsi the external auditory pseudomeatus enters the skull at a posteromedial angle. The mastoid process is shorter than the postglenoid process; it curves anteroventrally, contacts the postglenoid process, and forms a tube-shaped external auditory pseudomeatus.

Figure 124

Ventral view of skull and upper dentition of Embolotherium grangeri (AMNH 26002). (A) Ventral surface of skull, (B) right molars, (C) left premolars, (D) labial view of isolated incisors, (E) lingual view of isolated incisors.

Upper Dentition

The description of the upper dentition of E. grangeri is based primarily on the holotype, AMNH 26002 (fig. 124), although important taxonomic information on the dental formula and morphology of the upper teeth of E. grangeri is provided by a juvenile specimen (AMNH 26040) that is discussed further below.

Although no incisors or canines are preserved within the holotype skull (AMNH 26002), the intact alveoli indicate an unreduced dental formula (3-1-4-3). The incisor and canine alveoli form a semicircular arch. The diameter of the anterior tooth row arch (measured transversely across the canines) is similar to the width across the P1s, so that the canines are positioned anteriorly to the P1s rather than anterolingually as in E. andrewsi. A small diastema is present between the I3 alveolus and canine. The sizes of the incisor alveoli indicate that the incisors were significantly larger than those of Embolotherium andrewsi.

Three complete isolated incisors that are consistent in size with the incisor alveoli of the holotype skull were assigned the same field number (770) and catalog number (AMNH 26002) (fig. 124d, e). The association of these incisors with the holotype is uncertain because the canine alveoli of the skull are filled with sediment, indicating that the incisors of the holotype must have fallen out of the skull before burial. The isolated incisors were probably found somewhere in close proximity to the skull, but they do not necessarily belong with the holotype. Nonetheless, similar incisors found in the juvenile skull (AMNH 26040; see below) confirm that these isolated incisors belong to the species E. grangeri. The ovoid-globular crowns are angled lingually. One of them has a small irregular lingual cingulum. Another incisor crown has a shallow pit on the lingual side of the crown apex and the third has a completely smooth surface. Otherwise, the crowns are essentially featureless and globular, although they are abnormally large in comparison to other brontotheres with similar globular incisors. Nonetheless, the absence of regular wear facets on these teeth suggests they may have been vestigial.

The canine alveoli of AMNH 26002 suggest that the canines were not much larger than the incisors, and the juvenile specimen described below (AMNH 26040) confirms this. The postcanine diastema is similar in length to the P2. The P1s are heavily worn, but the left P1 reveals a paracone, a metacone, and a small posteriorly shifted lingual heel with a small lophlike cusp. The remaining premolars (P2–P4) are nearly as rectangular (not oblique), although the anterior margins of these premolars are concave. The parastyle of P2 is angled slightly lingually. The parastyles of P3 and P4 are directed somewhat labially. The metastyles of these teeth are worn, but appear to have been straight or slightly curved labially. There are distinct but weak labial paracone ribs on each of these premolars, although they are relatively smaller on progressively posterior premolars. A mesostyle is not present on any premolar. Preprotocristae are not seen on the premolars of AMNH 26002. There are two distinct lingual cusps on P2–P4. The P2 hypocone is about the same size as the protocone and it is barely connected to it. The hypocones of P3 and P4 are slightly smaller then the protocones. These cusps are completely separated because there are no lingual connecting crests on the premolars. The labial premolar cingula are weak while the lingual cingula of P2–P4 are weak and discontinuous around the lingual base of the paracone.

The premolars of another adult specimen, AMNH 26004, differ from those of AMNH 26002 in the following ways: the hypocone and protocone of P2 are connected by a stronger, taller lingual crest; in P3 and P4 the hypocones are smaller and connected to the protocone by a thin connecting crest; the lingual cingulum is thicker and continuous.

The upper molars of Embolotherium grangeri show typical brontotheriine characteristics, including tall and lingually angled ectolophs, very weak labial ribs, thin lingual ectoloph enamel, and wedge-shaped lingual sides of the paracone and metacone in molars that are not heavily worn. The anterior cingulum is thin and passes proximally to the distal peak of the parastyle. Deep central molar fossae and large anterolingual cingular cusps are present. All evidence of paraconules and metalophs is lost. The M3 hypocones of E. grangeri are nearly the same in size and shape as those of the M1 and M2. Labial molar cingula are weak while lingual molar cingula are absent. The cingulum of M3 is discontinuous and does not wrap around the distolingual base of the crown of M3 as it does in Embolotherium andrewsi.

Mandible and Lower Dentition

None of the adult skulls of Embolotherium grangeri is associated with a mandible, although a juvenile skull, AMNH 26040, is associated with a mandible that includes a complete set of unerupted adult anterior dentition. That specimen, described further below, is an important link for identifying other mandibles of E. grangeri. Among these are AMNH 26005, AMNH 26132, and AMNH 21600 (fig. 125). The largest of these specimens, AMNH 26005, consists of three pieces, two separate mandibular rami (not shown), and a crushed fragment of the symphysis (fig. 125e). The most complete specimen is AMNH 21600, a nearly complete and uncrushed mandible that is missing some incisors and cheek teeth (fig. 125a–c). Finally, AMNH 26132 consists of the anterior portion of the mandible with a separately preserved anterior fragment of the symphysis with lightly worn incisors and canine (fig. 125f–h).

Figure 125

Mandibles and lower incisors of Embolotherium grangeri. (A) Right view of AMNH 21600, (B) dorsal view of AMNH 21600, (C) right p2–p4 of AMNH 21600, (D) reflection of left p1 of AMNH 26040, (E) lingual view of incisors and canines of AMNH 26005, (F) dorsal view of AMNH 26132, (G) lingual view of incisors and canines of AMNH 26132, (H) labial view of incisors and canines of AMNH 26132.

The horizontal ramus of AMNH 21600 is shallow and deepens slightly posteriorly. The ventral margin of the symphysis has a shallow angle (< 45°). The symphysis extends to the talonid of the p4.

The incisors of these specimens form semicircular arches in front of the canines. The i2 is the largest incisor. The crowns of i1 and i2 are semispatulate with convex labial sides and flat or slightly concave lingual sides. The i3 crown is more conular. Lingual cingulids are very weak or absent. Labial cingulids are not seen, although faint labial cingulids can be seen in the unworn incisors of AMNH 26040 (fig. 128). The canines of AMNH 26005 and AMNH 21600 are about the same size as the incisors, no greater in diameter or height than i3. However, the canines of AMNH 26132 are taller than the incisors and they are more lingually curved. Nonetheless, the canines of all specimens can generally be described as very small. There are no diastemata between the anterior dentitions of these specimens. However, a postcanine diastema is present.

Figure 128

Mandible associated with juvenile skull of “Embolotherium” grangeri (AMNH 26040). (A) Left view, (B) anterior (labial) view before fragments were removed, (C) anterior (labial) view with left i1 removed, revealing unexposed left i2, (D) anterior view after fragments containing the incisors were removed showing unexposed canines.

The p1 is not preserved in AMNH 21600, nor is there a p1 alveolus, although the p1 could have fallen out in life with the alveolus being lost due to bone remodeling. In AMNH 26005 the p1 is tiny in comparison to the posterior cheek teeth and it is heavily worn. The p1 of the juvenile specimen (AMNH 26040, fig. 125d) is unworn. Its crown consists of a single cusp and a talonid heel. In AMNH 21600, the p2 trigonid and talonid are similar in length, although the trigonid is slightly narrower. The trigonids of p3 and p4 are slightly shorter and narrower than their talonids. The p2 paralophid of AMNH 21600 is broken. The p2 paralophid of AMNH 26132 is strongly curved lingually. The p2 protolophid of AMNH 26100 is straight and not angled lingually, although the entire lophid is positioned lingually. The p2 protolophid of AMNH 26132 is angled lingually and the trigonid has a broad lingual notch. Additionally, AMNH 26132 has a small p2 metaconid while AMNH 26100 does not. Similar intraspecific variability in the presence and absence of a p2 metaconid is seen in Embolotherium andrewsi. The paralophids and protolophids of p3 and p4 arch lingually 90°, forming nearly molariform trigonid valleys. The talonids of p2–p4 have well-developed cristids obliqua and hypolophids with deep valleylike talonid basins similar to those of the molars. Labial premolar cingulids are weak while lingual premolar cingulids are absent.

The thin lingual enamel of the lower molars and the elongate m3 of Embolotherium grangeri are typical brontotheriine traits. However, like E. andrewsi, the trigonids and talonids form deep valleys rather than broad shallow basins. Lingual molar cingulids are absent and the labial cingulids tend to be discontinuous around the labial bases of the trigonids, talonids, and hypoconulid heel, but the cingulids can be somewhat thick between them.

Juvenile Skull and Mandible

AMNH 26040, a juvenile specimen, is the only skull of Embolotherium grangeri associated with a mandible (figs. 126–Figure 127128). It is a critical specimen for linking the adult skulls and mandibles described above because it includes numerous isolated deciduous anterior teeth and complete sets of unerupted adult incisors embedded within the premaxilla and mandible. Parts of the specimen are described in detail by Granger and Gregory (1943), although their interpretation of the anterior dentition is erroneous. Gregory (Granger and Gregory, 1943: figs. 10, 11) worked out an arrangement of the isolated deciduous dentition of AMNH 26040; however, there are clearly errors in that arrangement. For instance, the root of the specimen referred to as a left deciduous I3 (diameter  =  8.5 mm) is too wide to fit within the left dI3 alveolus (diameter  =  7.1 mm). Any attempt to correctly arrange the deciduous incisors of AMNH 26040 at this point would be guesswork.

Figure 126

Juvenile skull of Embolotherium grangeri (AMNH 26040) with detached premaxilla.

Figure 127

Premaxillary fragment associated with the juvenile skull of Embolotherium grangeri (AMNH 26040) seen in Fig. 126. (A) Interpretation of the dentition by Granger and Gregory (1943) from left view, and (B) ventral view. The revised interpretation is shown (C) from the left view and (D) the ventral view.

Other errors in Granger and Gregory's (1943) interpretation of the unerupted adult anterior dentition of AMNH 26040 led them to mistakenly assign adult mandibles (described above) to a completely different genus, Titanodectes. The original and corrected interpretations of the dentition of the premaxillary fragment of AMNH 26040 can be seen in fig. 127. Granger and Gregory (1943) identified a ridge running along the side of the premaxillary fragment as the premaxillomaxillary suture, leading them to conclude that there were only two pairs of adult incisors. The large unerupted crowns preserved in the premaxillary fragment were interpreted as I2, I3, and the canine. The first incisor (I1) was presumed to be lost. However, all six of the unerupted adult teeth in this fragment are incisors. The canine itself can be seen in an unerupted state on the left maxillary of the main skull piece and not on the detached premaxillary fragment. The portion that was labeled as a maxillary in the original figure of Granger and Gregory (1943; seen here in fig. 127a) is actually the contact surface on the premaxilla for the maxilla. No part of the maxilla is actually attached to this fragment. Therefore, the three unerupted teeth visible in the fragment are actually I1, I2, and I3. The unerupted incisors of AMNH 26040 resemble the isolated incisors that are associated with the holotype (AMNH 26002) with large globular crowns and short lingual cingula on I2 and I3, but not on I1.

Granger and Gregory (1943) also mistakenly interpreted the mandible of AMNH 26040 as having two pairs of incisors when there are actually three. Removal of bone fragments from this specimen reveals additional unexposed teeth that were not known to Granger and Gregory (1943) (fig. 128). The unaltered specimen contains two pairs of unerupted but exposed crowns of large adult incisors (fig. 128b). Removal of the left i1 reveals an additional unexposed incisor directly behind the i1 and i3 (fig. 128c). Although not shown, the right i2 is also buried in bone behind the right i1 and i3. Finally, the anterior portion of the jaw containing all of the incisors was removed. The piece that was removed is that portion anterior to the large crack just in front of the p1 that can be seen from the lateral view of the specimen in figure 128a. Removal of this piece revealed two unerupted canine crowns (fig. 128d). The lower adult incisors of AMNH 26040 are very large, tall, and semispatulate with rounded occlusal edges and faint traces of labial cingulids. The canines are small and are not nearly caniniform due to their very blunt rounded tips. These incisors are distinctly different from the very small lower incisors of Embolotherium andrewsi, but the incisors and canines closely resemble those of adult mandibles of E. grangeri.

Remarks

Osborn (1929b) described the genus, Embolotherium, and three species, E. andrewsi, E. grangeri, and E. louksii, from material collected in the Shara Murun Region of Inner Mongolia. He erected a new subfamily for this genus, Embolotheriinae. The most conspicuously diagnostic feature of Embolotherium is the massive, upturned, and elongated “battering-ram nose” for which this beast is named and which Osborn believed functioned in “battering, assaulting, attacking, and tossing” (Osborn, 1929b: 19). The three species recognized by Osborn (1929a, 1929b) were distinguished primarily on characteristics of the ram and the rostrum. Granger and Gregory (1943) recognized the same three species, and added an additional species, Embolotherium ultimum, based on a particularly large partial cranium (AMNH 21604). Granger and Gregory (1943) mistakenly erected a new genus, Titanodectes, and two species, T. ingens and T. minor, for mandibles that turn out to belong to E. grangeri. Additional species of Embolotherium were added by Dashzeveg (1975) (E. ergilense) and Yanovskaya (1980) (E. insigne) based on skulls from the Ergilin Dzo of Mongolia. Most recently, Wang (2000) described an additional skull (IVPP V11959) of E. andrewsi.

Reanalysis of the known specimens reveals, however, that among these six species, only the first two, Embolotherium andrewsi and E. grangeri are valid. E. andrewsi was designated as the type species by Osborn (1929b). E. grangeri differs from E. andrewsi in several respects. The ram of E. andrewsi is curved upward while that of E. grangeri is straight or slightly curved downward. The ram of E. grangeri originates from the skull at a point posterior to the orbits, while that of E. andrewsi rises from above the orbits. Consequently, the nasal incision of E. grangeri is posteriorly much deeper. The large rim of bone that emarginates the distal border of the ventral nasal channel of the ram in E. andrewsi is not seen in adult specimens of E. grangeri. The rostrum of E. grangeri is longer and more robust than that of E. andrewsi with a much thicker dorsal premaxillary ridge. However, the nuchal crest and occiput of E. grangeri are far less robust. The zygomatic arches are shallower and more strongly bowed laterally. The upper and lower incisors of E. grangeri are much larger than those of E. andrewsi, and the anterior tooth row has a greater diameter. The hypocone of the upper premolars is more separated from the protocone. The M3 hypocone is taller and larger, and finally, the cingulum of the M3 is not continuous around the distolingual corner of the crown as it is in E. andrewsi.

Embolotherium ultimum (AMNH 21604) is a junior synonym of E. andrewsi. It was diagnosed primarily by its larger size. In particular, Granger and Gregory (1943) noted, “M3 of great size and relative width”. However, the size of M3 is exaggerated by numerous wide cracks in the tooth. Moreover, AMNH 21604 shares numerous characteristics with E. andrewsi. These include massive occipital pillars, a small M3 hypocone, and a continuous cingulum wrapping around the distolingual corner of M3. The holotype of Embolotherium ergilense (National Museum in Ulan-Bator, item 10) is also morphologically congruent with E. andrewsi, although appears to be larger and more robust than other specimens.

The remaining four species are synonyms of Embolotherium grangeri. These include Embolotherium louksii, Titanodectes ingens, Titanodectes minor, and Embolotherium insigne. E. louksii (AMNH 21610) and E. insigne (PIN 3110-52) are based on partial crania. Although the teeth of AMNH 21610 are mostly unpreserved, part of an unerupted right molar (M3?) is present, indicating that this individual is a subadult. Likewise, PIN 3110-52 appears to be a young individual with a newly erupted M3. Both specimens are consistent with E. grangeri in the posteriorly positioned ram behind the orbit, its ∼45° angle of orientation, and the posteriorly deep nasal incision. Originally, E. louksii and E. insigne were distinguished from E. grangeri by the fact that the rams of their holotypes are shorter and appear straight rather than slightly curved downward. When one considers the fact that the proximal portion of the ram of AMNH 21610 consists of small bone fragments and significant amounts of plaster, this difference is not significant. Furthermore, it is hard to imagine that there were not radical ontogenetic changes in the size and the shape of the ram. Therefore, minor variation in the size and shape of the ram among subadults or young adults is undoubtedly ontogenetic and probably not taxonomically relevant. New data from the undescribed Embolotherium grangeri collection from Khoer Dzan currently housed at PIN will help resolve this problem.

Granger and Gregory (1943) erected a new genus, Titanodectes, for adult mandibles of E. grangeri because of their erroneous interpretation of the anterior dentition of Embolotherium. They described Embolotherium adults as retaining the deciduous incisors. It was thought that not only were the adult incisors reduced to two pairs, but also they remained unerupted throughout adulthood. The origin of this puzzling interpretation originates primarily from a juvenile skull and mandible (AMNH 26040) of E. grangeri (though referred to E. louksii by Granger and Gregory [1943]). This juvenile specimen (described more fully above) includes a set of deciduous incisors and a full set of six unerupted adult incisors. Granger and Gregory (1943) recognized the juvenile age of this specimen but for unknown reasons inferred that its condition characterizes adults, despite the fact that numerous adult specimens completely contradict their interpretation. All adult skulls of Embolotherium indicate three, not two, incisors. In adult skulls of E. grangeri the large incisors are fully erupted and the small deciduous incisors are not retained.

In recognizing Granger and Gregory's (1943) error it is apparent that the mandibles previously referred to Titanodectes actually belong to Embolotherium grangeri. Although Granger and Gregory (1943) named two species of Titanodectes, T. ingens and T. minor, the holotypes of these species were distinguished only by their size differences. Embolotherium grangeri skulls (particularly those from the undescribed Khoer Dzan collection at PIN) seem to demonstrate about as much size variability as the “Titanodectes” jaws. Therefore, both Titanodectes species can be regarded as synonyms of Embolotherium grangeri.

A final note is made on the stratigraphic occurrences of Embolotherium. Most of the Embolotherium material from the AMNH expeditions was collected from the “Ulan Gochu” and “Shara Murun” faunal zones (sensu Radinsky, 1964) of the East Mesa and Urtyn Obo of the Shara Murun region. In reviewing field records of these expeditions, Radinsky (1964) found that the stratigraphic data associated with specimens from those areas was questionable due to difficulties correlating the strata with the type Ulan Gochu and Shara Murun sections at Baron Sog Mesa. However, Embolotherium materials are more definitively known from the Shara Murun Beds at Baron Sog Mesa (AMNH 21600, E. grangeri), the Ulan Gochu of Baron Sog Mesa (AMNH 21601, E. grangeri), as well as from the Baron Sog Formation at Baron Sog Mesa (AMNH 21604, E. andrewsi). The occurrence of Embolotherium in all three of these formations at Baron Sog Mesa as well as their occurrence in the Ergilin Dzo of Outer Mongolia suggests a Sharamurunian-Ulangochuian age for Embolotherium.

Pachytitan ajax Granger and Gregory, 1943

Holotype

AMNH 21612, a poorly preserved anterior fragment of a skull with left I3, C, and P2–M3.

Type Locality

Shara Murun Formation, four miles north of Baron Sog Lamasery, Baron Sog Mesa, Shara Murun region, Inner Mongolia, China.

Age

Middle Eocene (Sharamurunian land mammal “age”).

Diagnosis

Pachytitan ajax is a very large brontothere with short robust horns. The nasal incision is dorsoventrally deep and extends to the anterior edge of M1. The nasal process is unelevated, relatively broad, strongly rounded, thick, downturned anteriorly, and with very thick lateral walls that deepen proximally and are angled ventromedially. The premaxillomaxillary rostrum deepens posteriorly.

Dentally, Pachytitan ajax is characterized by three large or intermediate-sized upper incisors. There is a distinct P2 metacone. Premolar hypocones are well developed and only weakly connected to the protocone. The molars of Pachytitan ajax have tall, lingually angled ectolophs with weak labial ribs, and thinned lingual ectoloph enamel with wedge-shaped paracones and metacones. A cingular parastyle shelf is absent. Central molar fossae are present, but anterolingual cingular cusps are absent. Paraconules and metalophs are absent.

Pachytitan ajax is most similar to Rhinotitan andrewsi and Diplacodon elatus. It differs from the former primarily in the more massive nasal process with a more strongly rounded and downturned anterior margin, and by its more molariform premolars. Pachytitan ajax differs from Diplacodon elatus most clearly in the deep proximal lateral walls of the nasal process and the flat rather than upturned distal lateral margins of the nasal process.

Description

Skull

Pachytitan ajax is a large horned brontothere that is known only from the holotype specimen (AMNH 21612), which consists of a poorly preserved anterior portion of a skull (fig. 129). The right view shows a complete horn and nasal process; the left side includes a partial premaxillomaxillary rostrum, a poorly preserved I3 and canine, and a complete but heavily damaged cheek-tooth row. The two primary pieces, (1) the right horn and nasal and (2) the left maxilla are plastered together, but these two pieces do not actually appear to contact each other. Therefore, some aspects of the shape of the skull, such as the orientation of the nasal process are not necessarily reliable.

Figure 129

The holotype of Pachytitan ajax (AMNH 21612). (A) Right view, (B) left view, (C) dorsal view, (D) anterior view, (E) ventral view, (F) left premolars.

No sutures are discernable on this specimen, although it is probable that the horn is composed of both frontal and nasal components. The surface of the horn is rugose. The horn is large, projects in a dorsal direction, and is elliptical in cross section with its longest axis in an anteroposterior direction. However, the shape of the horn and its orientation seem to be affected by lateral crushing. The lateral side of the horn appears to have been smashed and dislocated medially, so that shape and orientation of the horn has been altered. The position of the horn with respect to the orbit is not clear, but it appears to have been located above and behind the posterior margin of the nasal incision. Judging from the left side, the nasal incision extended nearly to the anterior edge of M1. The nasal incision appears to have been rather deep, like that of Rhinotitan. The nasal process is reconstructed as if it had been angled slightly upward; Granger and Gregory (1943) described the nasal process of this specimen as being “sharply upturned” (p. 366). However, considering the copious amount of plaster in the specimen, the original orientation of the nasal process is uncertain. It is clear, however, that the dorsal surface of the nasal process is not turned upward with respect to the dorsal surface of the skull that is posterior to the horn. This suggests, rather, that the nasal process was not sharply upturned (contra Granger and Gregory, 1943).

The external surface of the nasal process is nearly intact on the right side. The nasal process is massive. The dorsal surface of the nasal process is flat, and the distal end is curved downward. The lateral walls are very thick and are very deep at the proximal end of the nasal process. However, the lateral walls shallow distally so that the anterior end of the nasal process lacks lateral walls. From the anterior view (fig. 129d) it can be seen that the thick lateral walls curve ventromedially and nearly partition the nasal chamber of the skull into two openings. In this respect the nasal processes of Pachytitan ajax resemble Rhinotitan andrewsi, although the lateral walls of the nasal process of P. ajax are much deeper and do not curve medially as sharply as in R. andrewsi. The distal end of the nasal process is strongly rounded and the anterior margin is thick and quite rugose.

The premaxillomaxillary rostrum is similar in length to the nasal process. From a lateral view the rostrum deepens posteriorly and its lateral dorsal surface is sloped dorsoventrally. Little else of the cranial morphology of Pachytitan ajax can be described from the fragmentary holotype.

Upper Dentition

The crowns of I1 and I2 are not preserved and the crown of I3 is too poorly preserved to precisely judge its morphology. However, the incisors appear to have been either large or intermediate in size, but they do not appear to have reached a small vestigial state. Additionally, the incisors are positioned completely anterior to the canine rather than between the canines. The canine itself is of moderate size. There is a short precanine diastema. The postcanine diastema is longer than the P2.

P1 is not preserved. P2–P4 are complete but badly damaged. In outline, P2–P4 are nearly rectangular although the anterior margin of P2 is slightly angled more posterolingually. The parastyle and metastyle of P2 are straight. The parastyles of P3 and P4 are more strongly angled labially. The P3 metastyle is straight while the P4 metastyle is deflected labially. The labial side of the P2 paracone is rounded while P3 and P4 have slight labial paracone ribs. P4 has a well-developed mesostyle that is situated closer to the metacone than the paracone. The other premolars lack mesostyles. Although it is possible that a P4 mesostyle is a characteristic of this species, this seems unlikely considering that specimens of other brontothere species occasionally show a P4 mesostyle, but it never seems to occur in any brontothere species as a fixed state.

The lingual margins of the P2–P4 are nearly flat. The lingual morphology of P2 is uncertain, but there appears to have been two lingual cusps or a single lingual crest. P3 and P4 have two distinct lingual cusps that are weakly connected. The hypocones of P3 and P4 are positioned well behind the protocones and are similar in size to the protocones. Labial premolar cingula are weak. P4 has a continuous lingual cingulum, but the lingual cingulum of P3 is slightly discontinuous.

The molars are too damaged for precise description and measurement. However, the shape of the wear facet is consistent with those of other brontotheriines. The ectoloph is substantially taller than the lingual cusps, the inner band of enamel is thinner than the outer band, and the lingual sides of the paracone and metacone are wedge-shaped. Central molar fossae are present, but anterolingual cingula cusps appear to have been absent. There is no evidence of paraconules on the molars, nor does the M3 have a hypocone.

Remarks

Granger and Gregory (1943) based Pachytitan ajax on a single specimen (AMNH 21612) and differentiated it from Rhinotitan by its larger size, more massive upturned nasals, large horns, and more advanced premolars. Due to the poor condition of the specimen, the interpretation of the nasal bone as upturned is conjectural. Likewise, when one considers the apparent variability of horn size in other brontothere species the larger horns of AMNH 21612 are not strong evidence for a distinct species. However, the premolars of Pachytitan ajax are distinctly more molarized than those of Rhinotitan due to the well-developed hypocones that are only weakly connected to the protocones. In this respect the premolars of P. ajax more closely resemble those of Diplacodon elatus and more advanced brontotheres such as Embolotherium or Parabrontops.

Cranially, Pachytitan ajax seems most similar to Rhinotitan andrewsi and Diplacodon elatus. Pachytitan ajax shares with Rhinotitan andrewsi (and differs from Diplacodon elatus) in its deep lateral nasal walls that are strongly angled ventromedially. However, the nasal bone of P. ajax is more massive than R. andrewsi and it has a thick, strongly rounded, and downwardly curved distal end; in these respects the nasal process of P. ajax more closely resembles Diplacodon elatus.

Diplacodon elatus Marsh, 1875

Holotype

YPM 11180, a flattened skull with right P2 (partial), P3–M3, left C (partial), P2 (partial), P3–P4, M1 (partial), and M2–M3.

Type Locality

Myton Member (Uinta C) of the Uinta Formation, Uinta Basin, Utah.

Age

Middle Eocene (late Uintan land mammal “age”).

Synonyms

Eotitanotherium osborni (Peterson, 1914b); Diplacodon (Pseudodiplacodon) progressus Peterson, 1934.

Referred Specimens

(From the Myton Member of the Uinta Formation of Utah) AMNH 21887, a skull with right and left P1–M3; CMNH 2858, an anterior portion of a skull with right P2 (partial), P3, left I1-canine, and P2; CMNH 2859 (holotype of Eotitanotherium osborni), an anterior portion of a skull with right I1–C, left I1–M3, a partial mandible with right i3–m3, and numerous postcranial elements; CMNH 10200, a skull with right and left P1–M3; CMNH 11819, a fragment of a palate with right M1–M3; CMNH 11876, a mandible with right c–m3 and left i3–m3; CMNH 11877, a complete mandible with right c, p2–m3, and left p1–m3; CMNH 11879 (holotype of Diplacodon progressum), a skull with right I2–I3, P1, P4, M2–M3, left P2–M3, a partial mandible with left p2 (partial), p3–m3, and numerous postcranial elements; CMNH 11895, an anterior portion of a skull with right I3, P2, P4, M1–M3, and left P2–M3; CMNH 11828, an anterior portion of a skull with right P2–M3, and left C–M3; CMNH 11881 skull (partially prepared) with right C, P2–M3, and left C–M3; CMNH 11967, a partial mandible with right i2–i3, c, p1–m3, and left i2; FMNH P14632 skull with right C–M3 and left I3–M3; FMNH P14633, a skull with right C–M3, left C–M3 and isolated right incisors; FMNH P14634, a complete mandible with complete lower dentition; FMNH P14638, a left mandibular ramus with p2–m3 and a vertebral atlas; FMNH P14640 (in part), a partial mandible with right i3, c, p1, m2 (partial), m3, and left p2–m3; FMNH P14799 dorsoventrally crushed skull with right P1–M3 and left P4–M3; FMNH P15446, a laterally crushed skull with right I1–I2, C–M3, left I1–C, and P2–M3.

Diagnosis

Diplacodon elatus is somewhat larger than Protitan minor but smaller than Protitanotherium emarginatum with horns of variable size that are positioned high above the orbits. The nasal incision is dorsoventrally deep and it extends posteriorly to the P4. The nasal process is slightly angled downward; it is broad, with thickened and upturned sides, and with a strongly rounded distal edge with a downturned distal tip. The nasal process is not elevated to the peaks of the horns. The orbits are positioned above the M2 and the posterior part of M1. The premaxillomaxillary rostrum hangs down posteriorly and it is not enclosed by bone dorsally. Other cranial characteristics include a saddle-shaped cranium, separate parasagittal ridges that moderately constrict the dorsal width of the cranium posteriorly, lack of a bony emargination surrounding the posterior nares, large ventral sphenoidal fossae, nearly straight zygomatic arches, and a ventrally constricted and mediolaterally angled external auditory pseudomeatus.

Dentally, Diplacodon elatus has three intermediate-sized upper incisors, globular I1 and I2, a more subcaniniform I3, a complex P1, and a distinct P2 metacone. Premolar hypocones are present although the protocone and hypocone of P2 and P3 sometimes take the form of a single lingual crest. The molars of Diplacodon elatus have tall, lingually angled ectolophs with weak labial ribs and thinned lingual ectoloph enamel with wedge-shaped paracones and metacones. A cingular parastyle shelf is absent. Distinct central molar fossae are present. Anterolingual cingular cusps tend to be weakly developed. Paraconules and metalophs are absent. The lower dentition includes three intermediate-sized incisors. The i1 and i2 are semispatulate while the i3 is more subcaniniform. There is a distinct postcanine diastema. There is a distinct metaconid on p3 and p4 but not on p2. The p2 trigonid and talonid are of similar length. The lower molars have shallow basins and the m3 is slender.

Diplacodon elatus can be distinguished from Protitanotherium, Rhinotitan, and more plesiomorphic brontotheres such as Protitan by its elevated horns, deep nasal incision, significantly more molarized premolars, and by its autapomorphic nasal morphology with upturned sides. Other brontotheres with premolars that are undifferentiated from those of Diplacodon elatus have smaller incisors, taller lower premolars, and/or a radically different skull shape.

Description

Skull

The holotype of Diplacodon elatus (YPM 11180) consists of a skull that is preserved in two pieces, an anterior portion (fig. 130), and a posterior portion (not shown). The specimen is extremely flattened and the only parts of the skull that are readily describable are the cheek teeth. However, D. elatus is also known from many complete and undistorted skulls and jaws. Two of the best skulls are FMNH P14632 and FMNH P14633 (figs. 131 and 133) from which the following description is primarily based. However, other specimens offer numerous important details on variation in this species.

Figure 130

The holotype of Diplacodon elatus. (Division of Vertebrate Paleontology, YPM 11180. © 2005 Peabody Museum of Natural History, Yale University, New Haven, Connecticut, USA. All rights reserved.) (A) Ventral view of palate, (B) left P2–P4, (C) right P2–P4.

Figure 131

Skulls referred to Diplacodon elatus. (A) Left view of FMNH P14632, (B) dorsal view of FMNH P14632, (C) anterior view of FMNH P14633, (D) posterior view of FMNH P144633.

Figure 133

Ventral view of a skull (FMNH P14633) of Diplacodon elatus.

Diplacodon elatus is rather small (table 10) for a horned brontothere. It is larger than Protitan minor but is smaller than Protitanotherium emarginatum or Protitan grangeri. There is no discernable frontonasal suture on any specimen. However, it is probable that like other horned brontotheres, the horns are composed of frontal and nasal elements, but these elements are completely fused together. Juvenile specimens may provide a more definitive answer, but none are currently known.

Table 10

Summary statistics for selected morphometric variables of Diplacodon elatus See Methods for measurement definitions

In cross section the horns vary from being round to somewhat elliptical. The surfaces of the horns are rugose, although some specimens have smooth horn surfaces. The horns range in size from being very small to quite massive, although they are never extremely tall as in some specimens of Megacerops (sensu Mihlbachler et al., 2004b). Many of the specimens have horns that are about the same size as either those of CMNH 2859 (fig. 132a), a gracile specimen, or CMNH 11879A (fig. 132b), a more robust specimen. However, numerous specimens (e.g., AMNH 21887, CMNH 2858, CMNH 11881, FMNH P14632, FMNH P14632, FMNH P14633) have horns that are intermediate in size. A single specimen, CMNH 11828, has horns that are very small and exist only as small rugose ridges.

Figure 132

Skulls referred to Diplacodon elatus illustrating opposite extremes in horn and nasal development and robustness. (A) Reflection of left view of CMNH 2859, (B) right view of CMNH 11879A.

Despite conspicuous variation in horn size and horn surface texture, the position and orientation on the horns is much less variable. The horns tend to project anterodorsally and laterally to varying degrees. They are positioned high above the orbits and are elevated upon tall superorbital pillars that project from above the orbits in an anterodorsal direction at about a 45° angle. This differs from Protitanotherium and Protitan whose horns are lower on the skull, a placement that more closely resembles more derived taxa such as Aktautitan and Metatitan. However, unlike these taxa, the nasal process of Diplacodon elatus is not elevated to the very peak of the frontonasal protuberances.

The nasal incision extends posteriorly to a point that fluctuates between the anterior and posterior margins of the P4. Although the nasal process is not elevated to the apex of the horns as in Metatitan and Aktautitan, it is nonetheless higher than those of Protitanotherium and Protitan, resulting in a dorsoventrally deep nasal incision where the posterior notch of the nasal incision rises higher than the orbit. The orbits of Diplacodon elatus are shifted anteriorly; all the roots of M2 and the posterolateral root of M1 are positioned directly below the orbit. The anterolateral root of M1 is positioned directly below the anterior orbital rim.

The nasal bones of Diplacodon elatus are completely fused. The length of the nasal process varies; it is typically a little shorter than the premaxillomaxillary rostrum, but occasionally, such as in CMNH 2859 (fig. 132a), it is nearly as long as the rostrum. The nasal process is as broad as the rostrum and it has thickened sides. From a dorsal view the nasal process of CMNH 14632 is distally tapered and the anterior end is strongly rounded, although in some specimens (e.g., AMNH 21887) the sides are more nearly parallel and somewhat constricted proximally. The distal tip of the nasal process is roughened and strongly curved downward. The sides of the nasal process are autapomorphically upturned rather than turned downward. The upturned sides of the nasal process are most easily seen from the anterior view where the dorsal surface of the nasal process is concave in a transverse direction (fig. 131c). In more gracile specimens, such as CMNH 2859 (fig. 132a), the nasal process is thinner and more nearly flat in cross section, although the dorsal surface is still convex in a mediolateral direction. (Note that previously published figures of another specimen, CMNH 2856 [Peterson, 1914b: pl. 8, and reprinted in Osborn, 1929a: fig. 365] give the misleading impression that the lateral margins are not upturned.) In the most robust specimens, such as CMNH 11879A (fig. 132b), a distinct rugosity traces around the edges of the nasal process, the dorsal surface is more strongly concave in cross section, and the distal end is more strongly curved downward. The variation in shape and thickness of the nasal process is generally related to the size and robusticity of the horns although the correlation is not perfect. For instance, CMNH 11828 (not shown) has a relatively thick nasal bone, though the horns are small.

From a lateral view (fig. 131a), the dorsal surface of the premaxilla is flat and steeply angled posterodorsally. Behind the symphysis the dorsal surface of the premaxilla is nearly horizontal. From the anterior view it can be seen that the premaxillary symphysis is rather long. Behind the symphysis the dorsolateral margins of the rostrum diverge laterally and the rostral cavity is not enclosed by bone dorsally. A distinct premaxillomaxillary suture has not been observed in any available specimen of Diplacodon elatus.

The dorsal surfaces of the skulls of Diplacodon elatus are strongly convex or saddle-shaped. The parasagittal ridges remain separate throughout their length and moderately constrict the dorsal surface of the cranium posteriorly but to a lesser degree than seen in Protitanotherium emarginatum. From the lateral view of the skull the zygomatic arch is not very deep. The jugal portion of the zygomatic arch is horizontal, while the squamosal portion is slightly sloped upward posteriorly, but as a whole, the zygomatic arch is nearly straight. From a dorsal view the zygomatic arch is relatively thin, straight, and strongly angled posterolaterally.

The occiput is moderately tilted backward. From a dorsal view the nuchal crest is slightly convex. From the posterior view the dorsal margin of the occiput is barely arched dorsally. The widths of the dorsal and ventral portions of the occiput are nearly the same and the occiput is very slightly constricted in the middle. There are strong occipital pillars of the surface of the occiput and there is a shallow central depression.

The position of the anterior margin of the posterior nares is somewhat variable. It is most often positioned slightly anterior to the M3 protocones as in CMNH 14633 (fig. 133). In a few specimens, such as FMNH P14632, it is slightly anterior to the M3. The bony horseshoe-shaped rim that typically emarginates the posterior nares of many brontotheres is not seen in Diplacodon elatus.

The posterior narial canal is elongate and continues far into the sphenoid via a pair of ventral sphenoidal fossae. In CMNH 14633 the elongate posterior narial canal has been fully cleared of sediment on the left side, thus exposing the deep ventral sphenoidal fossa and the elongate vomer that joins the septumlike basisphenoid to form a continuous partitioning septum. Every specimen of Diplacodon elatus in which the basicranium is preserved has distinct ventral sphenoidal fossae.

The short mastoid process arches anteroventrally and contacts the longer postglenoid process, thus creating a tubelike external auditory pseudomeatus that enters the skull in a mediolateral direction. The configuration of the basicranial foramina is consistent with those of other brontotheres with widely separate foramina of the alar canal, foramen ovale, and foramen lacerum.

Upper Dentition

In all of the known skulls the upper incisors are either not preserved or they are heavily worn and/or damaged. One specimen (CMNH 2859) has a set of worn but nearly complete incisors (fig. 134a). Additionally, isolated incisors are associated with FMNH P14633 (fig. 134b, c). Two nearly intact sets of cheek teeth are present in FMNH P14632 and FMNH P14633 (fig. 135). These specimens, along with others including the holotype (YPM 11180, fig. 130), indicate conspicuous intraspecific variation in cheek-tooth morphology that is further described below.

Figure 134

Incisors and canines of Diplacodon elatus. (A) Lingual view of CMNH 2859, (B) lingual, and (C) labial views of isolated incisors associated with FMNH P14633.

Figure 135

Cheek teeth of Diplacodon elatus. (A) Left premolars of FMNH P14632, (B) reflection of right premolars of FMNH P14633, (C) left molars of FMNH P14632, (D) reflection of right molars of FMNH P14633.

The dental formula of Diplacodon elatus is unreduced (3-1-4-3). The incisors are small in comparison to Protitan and other more plesiomorphic brontotheres, but they have not reached the vestigial state as seen, for example, in Duchesneodus. The incisors are not clearly differentiated from those of Protitanotherium emarginatum although they arch slightly more anterior to the canines. The I1 and I2 appear to have been subglobular, particularly those of FMNH P14633. The I1 and I2 of CMNH 2859 retain small lingual cingula. The substantial amount of wear on this specimen obscures other details. The I3 is larger than I1 or I2 and though worn in both specimens, it is clearly taller and more caniniform. The left incisors of CMNH 2859 are separated by very short diastemata. There is always a distinct diastema between the I3 and canine. The canines tend to be small with a weak distal cingulum. The postcanine diastema varies in length from slightly longer than the P2 to slightly shorter. The canines tend to be smaller than those of Protitanotherium, the postcanine diastema tends to be longer, and the distal end of the rostrum is somewhat narrower.

The P1 of FMNH P14632 is well worn, but two labial bulges in the crown indicate that both a paracone and metacone were present. The P1 crown is rounded. A lingual heel was clearly present, but further details are obliterated by wear. The P1 of FMNH P16633 is less worn. In that specimen there is single large protocone on the lingual heel. This protocone is connected to the paracone by short preprotocrista. Additionally, there is a short lingual crest that arches around the posterolingual border of the crown.

The P2–P4 of FMNH P14632 are subrectangular with nearly flat lingual margins. The anterior and posterior sides are nearly rectangular, but the lingual sides of P2 and P3 are slightly narrower than the labial sides. In other specimens, such as FMNH P14633, the P2 can be slightly more oblique in shape due to a more strongly distolingually angled anterior margin. The parastyle and metastyle of the P2 are straight. The parastyle of P3 is slightly angled labially while the metaconid is straight. The parastyle and metastyle of P4 are strongly labially directed. The labial sides of the P2 paracone and metacone form small swellings. In P3 and P4 there are more distinct labial paracone ribs. Although a P4 mesostyle is occasionally seen among specimens of brontotheres such as in Pachytitan or Protitanops, none of the P4s of Diplacodon elatus has a mesostyle.

The lingual features of the P2–P4 tend to be higher in relief in comparison to Protitan, Protitanotherium, or Rhinotitan. A distinct preprotocrista can be seen on P2–P4, although this structure tends to be larger on more anterior premolars. The P2–P4 of Diplacodon elatus tend to have two lingual cusps, a protocone and a hypocone. The exact morphology of the lingual side of the crown is variable. In AMNH 14632 the P2–P4 have distinct protocones and smaller hypocones. The P2 hypocone is not connected to the protocone by a lingual crest. However, the P3 and P4 hypocones are connected to the protocones by a thin lingual crest. In each of these premolars the hypocone is positioned distantly from the protocone. FMNH P14633 exhibits a different morphology. In that specimen, there are no distinct lingual cusps on the P2, although a rounded crest arches around the lingual margin of the crown. In P3, the lingual crest is much thicker than that of P2 and it completely absorbs the protocone and hypocone. There is a distinct protocone and hypocone on P4, although, in this instance, the cusps are positioned very closely together and are of approximately the same size.

The upper premolars of the holotype specimen (fig. 130) demonstrate bilateral asymmetry. In that specimen there is a lingual crest on the right P2, but there are no distinct lingual cusps. On the left P2 there are two widely spaced cusps connected by a lingual crest. On the right P4 the small hypocone is connected to the protocone by a crest, while on the left P4 there is no connecting crest.

Variation in the size and distinctness of the lingual cusps, and their degree of separation is seemingly random in each individual tooth. However, when one examines all the specimens the lingual side of P2 ranges from having a single tall lingual crest with no distinct cusps to having two distinct lingual cusps that are weakly connected by a thin crest. P3 most often has a distinct protocone and a smaller hypocone. The connecting crest itself is sometimes nearly as tall as the lingual cusps, but it can also be much shorter. P4 is more variable in this respect. There are always two distinct cusps that are usually connected to varying degrees by a connecting crest of variable height, but occasionally, as in FMNH P14632, the hypocone is much smaller and it is positioned on the base of the posterior slope of the protocone. Despite the overall variability, a general pattern does emerge; P2 shows a tendency to have large lingual crests that often absorb the lingual cusps, while P4 tends to have more distinct labial cusps with the connecting crest reduced or absent. P3 is usually intermediate.

Labial premolar cingula are always weak or absent in Diplacodon elatus. The lingual cingula of the P2–P4 are always thicker. It is consistently thickest on P2 and it is never discontinuous. Occasionally the lingual cingulum of P3 and P4 is slightly discontinuous around the lingual margin of the crown.

The upper molars of Diplacodon elatus show numerous advanced features, including tall, lingually angled ectolophs, very weak labial ribs, thin lingual ectoloph enamel, and wedge-shaped lingual sides of the paracone and metacone in molars that are not heavily worn. The anterior cingulum is thin and passes proximally to the distal peak of the parastyle. Distinct central molar fossae are present. In the holotype specimen (fig. 130) the anterolingual cingulum of the molars thickens and rises slightly but does not form a strong anterolingual cingular cusp. Other specimens, such as FMNH P14632 (fig. 135c), reflect the condition seen in the holotype. However, other specimens, such as FMNH P14633 (fig. 135d) show strong anterolingual cingular peaks on the molars.

Mader (1989, 1998) noted that paraconules or metaconules are occasionally present on the molars of Diplacodon ( =  Pseudodiplacodon). However, I found no evidence of paraconules or metaconules in any of the specimens. Nor is there any evidence of a vestigial metaloph. Mader (1989, 1998) also noted the occasional presence of a hypocone (or “pseudhypocone”) on M3, but none of the specimens has a distinct hypocone. Although no hypocone is present, the lingual side of the M3 is broad and squarish with a large space posterior to the protocone and a thick cingulum that wraps around the distolingual corner of the crown. Labial molar cingula are generally weak and lingual molar cingula are very weak between the protocone and hypocone, or they are absent altogether.

Mandible and Lower Dentition

The only skulls of Diplacodon elatus that are associated with mandibles are CMNH 2859 and CMNH 11879. However, each of these mandibles is incomplete. The mandible of CMNH 2859 is missing its incisors, but that specimen has the most unworn and complete set of lower premolars available for Diplacodon elatus, and is therefore figured in close up (fig. 136c). There are also several mandibles and partial mandibles from the Myton Member of the Uinta Formation that are not associated with skulls but that are consistent with CMNH 2859 and CMNH 11879 and are referable to D. elatus. Among these, FMNH P14634 is the most complete D. elatus mandible (fig. 136a, b, d, e).

Figure 136

Mandible and lower dentition of Diplacodon elatus. (A) Right view of FMNH P14634, (B) dorsal view of FMNH P14634, (C) right lower premolars of CMNH 2859, (D) labial view of right incisors and canine of FMNH P14634, (E) lingual view of incisors and canines of FMNH P14634.

The mandible of FMNH P14634 has a thin coronoid process that rises above the mandibular condyle. The inferior margin of the symphysis is angled slightly less than 45°. From the dorsal view of the mandible the mandibular symphysis is longer and more slender than that of Protitanotherium emarginatum and with a longer postcanine diastema. The posterior portion of the symphysis extends to the talonid of p4 in FMNH P 14634, but the symphysis more typically ends at the posterior edge of the p3 (CMNH 11876, 11877, 11967).

Like the upper incisors, the lowers are smaller than those of Protitan grangeri, but they have not reached a vestigial stage. The lower incisors of Diplacodon elatus are not clearly differentiated from those of Protitanotherium, although the incisor row is more arched. The i2 is the largest incisor; this is most obvious from the labial view (fig 136d). The crowns of the i1 and i2 are semispatulate with rounded apices and narrow lingual cingulids. The i3 is shorter and more mesiodistally elongate in comparison to i1 or i2. A single left i3 is also preserved with CMNH 2859 (not shown). The i3 of that specimen seems taller and somewhat more subcaniniform than the i3 of FMNH P14634. There are no diastemata between any of the lower incisors or canines. The lower canines are of small to moderate size and the postcanine diastemata of FMNH P14634 and other specimens of D. elatus are consistently longer than those of Protitanotherium emarginatum.

The lower premolars of Diplacodon elatus are relatively slender and low-crowned. In general, their proportions do not differ notably from Protitanotherium emarginatum. The p1 is a simple, single-cusped tooth with a talonid heel and a weak lingual cingulid. The trigonid of p2 is slightly longer than the talonid. This differs from Protitanotherium, whose p2 trigonid is nearly twice the length of the talonid. The p3 trigonid of D. elatus is slightly longer than the talonid, while the trigonid and talonid of p4 are of similar length. The trigonids of p2–p4 are all barely narrower than their talonids. The paralophid of p2 arches lingually but at an angle less than 45°, thus creating a small lingual trigonid notch. The protolophid of p2 is nearly straight, but it is lingually positioned. The paralophid of p3 is relatively long and arches lingually about 45°, while the p4 paralophid is molariform and arches 90° lingually. The protolophids of p3 and p4 are molariform and arch almost completely lingually. The trigonid of p3 has a very broad lingual notch, while the trigonid of p4 has an essentially molariform basin. The p2 lacks a metaconid, but p3 and p4 possess a large lingually positioned metaconid. The talonids of p2–p4 have well developed cristids obliqua and hypolophids. These elements are more elongate in more posterior premolars, creating increasingly molariform talonid basins. The labial premolar cingulids are weak to absent and lingual premolar cingulids are absent.

The lower molars of Diplacodon elatus are typical. They have relatively thin lingual enamel, shallow trigonid and talonid basins, and the m3 is elongate. There are no lingual cingulids, but labial molar cingulids are distinct although they tend to be discontinuous around the paraconids and metaconids.

Remarks

Marsh (1875) based Diplacodon elatus on a dorsoventrally crushed skull (YPM 11180) with comparatively undistorted cheek teeth (fig. 130). Marsh (1875) asserted that D. elatus was hornless. The dorsal surface of the holotype does appear to lack horns as Marsh (1875) had observed, but the specimen is severely flattened and it was apparent to others that the specimen was far too damaged to realistically determine the presence or absence of horns in D. elatus from its holotype (Hatcher, 1895; Osborn, 1929a). The only readily describable parts of the holotype are the cheek teeth, which indicate a brontothere that is somewhat smaller than Protitanotherium emarginatum, also from the Myton Member of the Uinta Formation, and with significantly more molarized premolars.

Several complete and less distorted specimens of Diplacodon elatus from the Myton member of the Uinta Formation have since been discovered, although many of them have been assigned to a variety of other names. Peterson (1914b) named a new species, Diploceras osborni, from the anterior portion of a skull and mandible (CMNH 2859) (figs. 132a, 134a, 136b). Peterson (1914b) relegated another partial skull (CMNH 2858) as the cotype. The later specimen differs distinctly from the former with its larger and more rugose horns, which Peterson (1914b) attributed to sexual dimorphism. Peterson (1914c) later renamed the genus to Eotitanotherium after discovering that Diploceras was preoccupied by a mollusk. Peterson (1914b) did not clearly differentiate Eotitanotherium osborni from Diplacodon elatus, although he mentioned that Eotitanotherium osborni had shorter, blunter canines, a less ridgelike deuterocone (protocone) on the P2, and more distinct lingual cusps on the P3. None of these observations warrant a clear taxonomic distinction considering (1) the poor state of preservation of the canine in the holotype of D. elatus, (2) and the large degree of essentially continuous morphological variation seen in the upper premolars of brontotheres.

Peterson (1934) named yet another species, Diplacodon progressum, based on a skull (CMNH 11879A) (fig. 132b) and nearly complete skeleton (CMNH 11879). The horns of that specimen are robust and similar to those of Peterson's paratype of Eotitanotherium osborni. Peterson (1934) differentiated D. progressum from D. elatus with the following observations, (1) “Facial region is shorter in D. progressum”, (2) “the premolar series in the new species is farther advanced in molarization than in D. elatus, the individual teeth being more perfectly quadrate in D. progressum, P2 being especially advanced” (Peterson, 1934: 353). The first observation was based only on the slightly shorter premolar length and might simply relate to the fact that the specimens are of slightly different sizes. Moreover, the extremely crushed and incomplete condition of the holotype of D. elatus (YPM 11180) complicates any comparison. The second observation is questionable as well; the slight difference in the squareness of the premolars is consistent with a pattern of intraspecific premolar variation generally found in brontotheres and does not represent a taxonomically significant difference.

Peterson (1934) distinguished Diplacodon progressum from Eotitanotherium osborni on the observations that the (1) nasals “are shorter and heavier”, and (2) “the alveolar border of the premaxillary border is noticeably shorter” (Peterson, 1934: 353). Upon reexamining the specimens, the second observation cannot be confirmed. The first observation relates to the fact that the holotype of D. progressum is more robust in comparison to the holotype of E. osborni. This distinction contradicts Peterson's (1914a) earlier concept of E. osborni whereby differing degrees of robustness were attributed to sexual dimorphism. Specimens that have since been referred to either E. osborni or D. progressum vary only in the general robustness of the skull (including horn size and nasal thickness) and in the degree to which the lingual cusps of the premolars are separated or connected by a lingual crest.

Based on these aspects of variation Mader (1989, 1998) continued to accept Eotitanotherium osborni and Diplacodon progressum. Mader (1989) suggested that E. osborni was probably a junior synonym of D. elatus but he preferred to view the later as a nomen dubium because of the crushed nature of the holotype of that species. Because Diplacodon was not considered valid, Mader (2000) erected a new genus, Pseudodiplacodon, for the progressum species. Mader (2000) continued to distinguish E. osborni and P. progressum based on horn size, nasal thickness, and the degree of separation of the lingual cusps on the P4.

In considering all of the specimens, Mader's (2000) distinctions of Eotitanotherium and Pseudodiplacodon seem to represent continuous intraspecific variation. For instance, many of the specimens assigned to Pseudodiplacodon progressum by Mader (2000) have horns that are intermediate in size between the holotypes of E. osborni and P. progressum (e.g., AMNH 21887, CMNH 2858, CMNH 11881, CMNH 14632, FMNH P14632, and FMNH P14633) (for example, compare specimens in fig. 131 to fig. 132). In addition, specimens assigned to P. progressum by Mader (2000) distinctly vary in the degree to which the lingual premolar cusps are separated. For example, in FMNH P14632 the hypocones and premolars are well separated, although in FMNH P14633 the lingual cusps are very poorly separated (fig. 135). Finally, the nasal bones of one of the three specimens that was earlier referred to E. osborni by Mader (1989) has a nasal bone that is as thick as those specimens referred to P. progressum by Mader (2000).

The variation in horn size, nasal thickness, and lingual premolar morphology among these specimens is seemingly continuous and is generally consistent with a pattern of intraspecific cranial variation found in other brontotheres. Additionally, the lingual premolar morphology is found to be variable in most brontothere species. Given the continuous nature of this variation, the specimens in question cannot be easily partitioned into discretely diagnosable units. Therefore, if specimens previously referred to Eotitanotherium osborni or Pseudodiplacodon progressum represent a single taxon, as is concluded here, Diplacodon elatus Marsh is a valid species. E. osborni (Peterson) and P. progressum (Peterson) are junior synonyms of D. elatus.

Specimens referred to Diplacodon elatus are confined to the Myton Member of the Uinta Formation. However, a large undescribed Diplacodon-like skull (AMNH 117163) with very heavily worn dentition was collected by Malcolm McKenna from Mudstone Peak in the Wiggins Formation of Wyoming. Previously, Mader (1998) suggested this skull belongs to Protitanotherium, but this is almost certainly not correct. The skull very closely resembles Diplacodon, particularly in its dual lingual premolar cusps, deep nasal incision, and distinctively upturned lateral margins of the nasal process. The skull, AMNH 117163, differs from typical Diplacodon elatus from the Uinta Formation in only a few ways. It is considerably larger, has more elliptical horns (with the longest axis anteroposterior), and the parasagittal ridges constrict the dorsal surface of the skull more severely than typical D. elatus skulls. This specimen could represent a new species of Diplacodon, but it has not yet been thoroughly studied. Some or all of these differences may relate to sexual dimorphism or intraspecific variation. For the time being its taxonomic identity is uncertain.

Parabrontops gobiensis (Osborn, 1925)

Holotype

AMNH 20354, a severely distorted skull with right P1–P4 (broken but complete), M1–M3 (severely damaged), left P3–M1 (partial), and M2–M3.

Type Locality

Urtyn Obo Formation, Urtyn Obo, Shara Murun Region, Inner Mongolia, China.

Age

Late Eocene (Ulangochuian land mammal “age”).

Referred Specimens

(From the “Ulan Gochu” faunal zone [sensu Radinsky, 1964], Urtyn Obo, Shara Murun Region, Inner Mongolia); AMNH 26020, a partial skull lacking the occiput, basicranium and left zygomatic arch, with right C–M3. (Ulan Gochu [used in quotes] refers to a faunal zone rather than a formation; see Radinsky, 1964.)

The following mandibles could belong to Parabrontops gobiensis: (from the Urtyn Obo Formation, Urtyn Obo, Shara Murun Region, Inner Mongolia) AMNH 26019, a mandible with right and left incisors and canines (roots only), and p1–m3; AMNH 26021, a partial mandible with left p1–m3; (from the “Shara Murun” Formation [sensu Radinsky, 1964], East Mesa, Shara Murun Region, Inner Mongolia) AMNH 26131, a mandible with right and left i1–m3.

Diagnosis

Parabrontops gobiensis is a large horned brontothere with short massive elliptical frontonasal horns. The horns are positioned low on the skull and directly above the orbits. The nasal incision extends to the anterior margin of the P4. The anterior rim of the orbit is above the anterolateral root of M1 and the posterolateral root of P4. The nasal process is very thick, unelevated, horizontal, strongly rounded anteriorly, with thickened lateral walls, and with a downturned distal tip. The premaxillomaxillary rostrum thickens posteriorly and it is not enclosed by bone dorsally. Other cranial characteristics include a saddle-shaped cranium, a dorsal cranial surface that is moderately constricted posteriorly by parasagittal ridges, a narrow emargination surrounding the posterior nares, nearly straight zygomatic arches, and a ventrally constricted and mediolaterally angled external auditory pseudomeatus. Ventral sphenoidal fossae and postzygomatic processes are absent. Large zygomatic swellings are absent.

Dentally, Parabrontops gobiensis is characterized by three small upper incisors that form a straight row, a postcanine diastema, a complex P1, a distinct P2 metacone, and distinct premolar hypocones on P3 and P4. The molars of Parabrontops gobiensis have tall, lingually angled ectolophs with weak labial ribs, and thinned lingual ectoloph enamel with wedge-shaped paracones and metacones. A cingular parastyle shelf is absent. Distinct central molar fossae and anterolingual cingular cusps are present. Paraconules and metalophs are absent.

Parabrontops gobiensis shares with Eubrontotherium clarnoensis the combination of very small upper incisors, paired horns that are widely spaced and unelevated, and relatively straight zygomatic arches that lack conspicuous swellings. Parabrontops gobiensis differs from Eubrontotherium most clearly in the retention of three upper incisors.

Description

Skull