Description

Cranium: Of the cranium (IVPP V 22640) only the basicranium and a partial left portion (fig. 2), which was broken off anterior to the orbit, are preserved.

The orbit is rounded and large with the anterior border at the level of the M1 metacone. A small lacrimal tubercle is present on the anterior border of the orbit. Along the anterior border of the orbit there is an incomplete lateral trough presumed for the meatal diverticulum as in Tapirus, bordered by a sharp ridge laterally (fig. 2A, B). The lateral trough extends posteriorly over the supraorbital process of the frontal. The infraorbital foramen opens at the level of the anterior border of P4 (fig. 2A, B). The jugal forms the smooth ventral border of the orbit without the postorbital process, and prominently rises backward. A small tubercle is present at the anteroventral border of the jugal above M2. The skull roof is incomplete, although it is clear that a pair of medial troughs are separated by the anterior sagittal crest and extend posteriorly to the frontal lines (fig. 2B); the frontal lines converge posteriorly and likely form a long sagittal crest as in other early perissodactyls. Several nutrient foramina are discernable on the lateral side of the parietal and squamosal. The occipital bone is also fragmentary. The occipital condyles protrude more posteriorly than ventrally, and are widely separated dorsally and closely situated ventrally. The lateroventral border of the condyle (linea divisa condyli) is a somewhat blunt ridge, separating the dorsal and ventral surfaces of the condyles. The preserved ventral part of the nuchal crest on the left side is rather sharp and diverges into two ridges ventrally (fig. 2A): one extending anteriorly above the external acoustic meatus, the other running ventrally along the lateral side of the posttympanic process. Ventrally, the glenoid fossa is flat, and the postglenoid process faces anteriorly and slightly laterally with a postglenoid foramen (fig. 2C). The posttympanic process is short and widely separated from the postglenoid process, while the paracondylar process, preserved on the right side, is stouter and longer than the posttympanic process and curves posteriorly (fig. 2A, C). Medial to the paracondylar process there is a medium-sized hypoglossal foramen (fig. 3). The basioccipital bears a sharp median ridge, which terminates anteriorly at the two parallel basal tubercles (fig. 2C). The basisphenoid extends anterodorsally from the basioccipital.

FIG. 3.

Right petrosal of Paracolodon fissus (IVPP V 22640). Parallel lines indicate damaged surfaces. The anteroposterior length of the promontorium is about 14.66 mm. Abbreviations: bg, basicapsular groove; bo, basiocciptial; bs, basisphenoid; ci, crista interfenestralis; ctp, caudal tympanic process; eam, external acoustic meatus; fc, fenestra cochleae; fs, facial sulcus; fv, fenestra vestibuli; gpns, greater petrosal nerve sulcus; hf, hypoglossal foramen; mlf, middle lacerate foramen; oc, occipital condyle; p, promontorium; pgp, postglenoid process; pcop, paracondylar process; s, squamosal; sf, stapedial muscle fossa; sff, secondary facial foramen; ts, transpromontorial sulcus; ttf, fossa for the tensor tympani muscle.

The right petrosal is preserved in a relatively good condition with its ventrolateral (tympanic) surface exposed on the ventral side (figs. 2C, 3). The promontorium is almond shaped and pointed anteriorly; however, the bone is damaged at the posterolateral margin (indicated by crosshatching in fig. 3). The surface of the promontorium is uneven with the lateral portion situated more toward the cerebellar surface and the medial part relatively flat. A depression in the shape of a groove runs along the posterolateral side. We interpret this depression as the fossa for the tensor tympani muscle. Lateral to this fossa is a narrow and short groove, which is separated from the former by a distinct ridge. This groove may have been for the greater petrosal nerve, the anterior branch of the facial nerve. An unambiguous hiatus Fallopii is not discernable on the specimen either because the hiatus was absent or not preserved. Thus, whether the greater petrosal nerve branched within the petrosal bone and then passed through a hiatus Fallopii or the hiatus Fallopii was absent and the greater petrosal nerve branched within the middle ear and extended forward cannot be determined. On the medial side of the promontorium, there is another groove that extends from the anteromedial side of the fenestra cochleae (fenestra rotundum), and parallels the lateral border of the promontory. This groove has a pronounced lateral ridge. We tentatively interpret this groove as the transpromontorial sulcus, and it may have been made by the internal carotid artery, internal carotid plexus, tympanic plexus, or all of these structures. We note, however, that it is also possible that a pronounced lateral ridge of the transpromontorial sulcus was the attachment site for a tympanic bulla that is not preserved. Such an arrangement of the bulla might have meant that the internal carotid artery would have been external to the tympanic bulla (extrabullar). A short and narrow groove present medial to the transpromontorial sulcus may be the basicapsular groove for the inferior petrosal sinus. Between these two grooves the surface is swollen and has a pronounced bump.

The fenestra cochleae is situated on the posterolateral corner of the promontorium, rounded, and facing posteriorly. On the anterolateral side of the fenestra cochleae and considerably smaller than the fenestra cochleae is an oval-shaped fenestra vestibuli (fenestra ovalis) that faces ventrolaterally, indicating a relatively small base of the stapes. The crista interfenestralis is relatively protruded, somewhat swollen between the two fenestrae, becoming laterally compressed posteriorly, and terminating at the anterior margin of the stapedial muscle fossa. The caudal tympanic process is mediolaterally broad. A jugular foramen, which would have been located on the posteromedial side of the caudal tympanic process, is filled by matrix. The fossa for the stapedial muscle is relatively large and roughly triangular, situated on the posterolateral side of the fenestra cochleae and posterior side of the fenestra vestibuli. Anterior to the stapedial muscle fossa is a facial sulcus, which is partially broken off on the right side but more complete on the left side. The facial sulcus emerges anteriorly from a relatively large secondary facial foramen, which is situated anterolaterally relative to the fenestra vestibuli. Lateral to the facial sulcus is an incomplete external acoustic meatus, widening laterally. The epitympanic recess, which would have housed the articulation of the malleus and incus in life, is not preserved. The tegmen tympani is broken and not complete, however, its dorsolateral surface appears to have been convex as deduced from the preserved part.

Mandible: The horizontal ramus of mandible anterior to p2 was broken off, and the ventral border below the premolars was broken (fig. 2A). The preserved horizontal ramus is relatively slender with a slightly convex ventral border. The vessel notch is wide and shallow. The angle is rounded with thin margins, extending posteriorly beyond the condyle. The condyle is relatively high, and the distance between the condyle and the alveolar border is roughly equal to the length of m1–3. The condyle has a tear-shaped and nearly flat articular surface for the glenoid fossa with an apex toward the medial and slightly ventral side. The facet curves onto the posterior surface of the condyle on the medial end. The lateral side of the posterior surface of the condyle is rugose. The coronoid process is higher than the condyle with a truncated dorsal rim, and nearly vertically placed. The mandibular notch is generally concave with a slight convexity in the middle. The anterior border of ascending ramus is straight and slightly posteriorly slanted from the lateral view. The ventral half of the anterior border of the ascending ramus is a deeply concave fossa. On the lateral surface of the ascending ramus there is a relatively deep masseteric fossa.

Upper Teeth: P1: A broken alveolar and an appressed surface anterior to the parastyle of P2 indicate the presence of P1 (figs. 2C, 4A).

P2: The tooth is roughly quadrilateral in outline with the width considerably greater than the length (fig. 4A). The paracone and metacone are nearly equal in size and closely situated with convex buccal surfaces. The parastyle is smaller and slightly lower than the paracone, situated mesial to the latter. The metastyle is relatively distinct, but much smaller than the paras tyle. The protoloph extends from the protocone toward the joint of the parastyle and the paracone, and stops before its contact with the ectoloph. The hypocone is separated from the protocone by a deep furrow on the lingual side, and is slightly more lingually situated than the protocone. The metaloph is indistinct. The weak mesial and distal cingula are present.

FIG. 4.

Occlusal view of upper cheek teeth of Paracolodon fissus, Paracolodon inceptus, and Desmatotherium mongoliense. A–B, P. fissus; A, left P2–M3 (IVPP V 22640); B, left P2–P4 (AMNH FM 20161); C–D, P. inceptus; C, left P1–P3 (AMNH FM 20355); D, left P3–M3 (AMNH FM 20357); E–G, D. mongoliense; E, left P1–M2 (IVPP V 14692); F, left P3–P4 (AMNH FM 20156); G, left P2–M3 (AMNH FM 19161, reversed).

P3–4: The ectoloph of P3 was broken off. The morphology of P3 is similar to that of P2 (fig. 4A). However, the hypocone of P3 is as lingually situated as the protocone, and the metaloph is distinct extending from the mesiolingual slope of the metacone to the protocone, bypassing the hypocone. The P4 is similar to P3, but the former is much wider and relatively shorter than P2 and P3. The paracone and metacone are conical and closely placed, with the metacone slightly smaller and more lingually situated than the paracone. The parastyle is relatively small and slightly buccal to the paracone. The hypocone is separated from the protocone by a shallow furrow on the lingual side.

M1–2: These teeth are similar in morphology, except M2 is larger (fig. 4A). The parastyle of M1–2 is large, slightly buccally situated to the paracone, and compressed to the protoloph. The paracone is conical, relatively high and sharp, and vertically placed. A broadly convex crest is present on the lingual side of the paracone. The metacone is strongly lingually depressed with a concave buccal surface and a rather reduced postmetacrista. The position of the metacone is nearly midway between the paracone and the hypocone. The protocone and the hypocone are equal in size. The protoloph extends from the protocone to the parastyle, joining the preparacrista buccally, which is mesiodistally extended. The metaloph is much shorter than the protoloph, joining the ectoloph at the apex of the metacone. The buccal cingulum adjacent to the metacone is well developed and almost forms a small platform. The mesial and distal cingula are also distinct, while the lingual cingulum is absent. The buccal base of the paracone also lacks the cingulum.

M3: The tooth is roughly similar to M1–2 (fig. 4A), however, M3 differs from M1–2 in having an apex of paracone distally curved, a metacone less lingually depressed and more reduced, and a centrocrista relatively shorter than the metaloph and aligned with the metaloph. Furthermore, the platformlike buccal cingulum adjacent to the metacone on M3 extends to the distal border, and a prominent cingulum is present at the lingual base of the central valley.

Lower Teeth: p2: The tooth bears a conical protoconid in the middle, from which a short paracristid extends anteriorly to a high paraconid (fig. 5A). The paraconid is separated from the protoconid by a deep groove on the lingual side. The metaconid is absent. The hypoconid is large, nearly as high as the paraconid, and situated in the middle of the posterior border. The cristid obliqua is slightly lingually concave. A small, rudimentary entoconid is appressed to the lingual side of the hypoconid. The cingulum is absent.

p3: The crown of the tooth is partially broken. The tooth is rectangular in outline (fig. 5A). It can be deduced from the preserved portion that the paraconid is relatively high, the metaconid is smaller than the protoconid, the hypoconid is prominent, and the entoconid is small but distinct.

p4: The crown of the tooth is rectangular in outline (fig. 5A). The trigonid is wider than long, and slightly shorter and narrower than the talonid. The protoconid and the metaconid are conical, equal in size with the latter more distally situated than the former. The protolophid is moderately notched in the middle. The paracristid extends from the protoconid mesially and slightly lingually, and then curves lingually along the mesial border. The hypoconid is conical, large, and lower than the protoconid, extending the cristid obliqua to the distolingual side of the protoconid. The entoconid is small but as distinct as in p3. The hypolophid and the cingulum are absent.

FIG. 5.

Occlusal view of lower cheek teeth of Paracolodon fissus and Desmatotherium mongoliense. A–B, P. fissus; A, left p2–m3 (IVPP V 22640); B, left p2–m3 (AMNH FM 81802); C–D, D. mongoliense; C, left p3–p4 (AMNH FM 81718, reversed); D, left p3–m3 (AMNH FM 20155, reversed).

m1–2: The m1 is similar to m2 in morphology, but m2 is slightly larger than m1 (fig. 5A). The trigonid of m1–2 is similar to that of p4, except that the metaconid is more separated from the protoconid and the protolophid is nearly transversely extended and slightly notched. The talonid is slightly narrower and considerably longer than the trigonid, and the hypolophid is more oblique than the protolophid. The cristid obliqua is rather reduced. A weak ridge anterobuccally extends from the entoconid. A small hypoconulid is present. An anterobuccal cingulum is present.

m3: The tooth is similar to m1 and m2 except for a more distinct hypoconulid (fig. 5A).

Comparisons and Discussion

Skull and Teeth: Osborn (1923) erected Desmatotherium mongoliense based on materials from Irdin Manha Formation, Irdin Manha escarpment. Some of the specimens included in the species were later recognized as belonging to Lophialetes (Matthew and Granger, 1925a). The main diagnostic feature for D. mongoliense, according to Osborn (1923), is that P2–3 has a “duplicate internal cusp” similar to Desmatotherium intermedius (= D. guyotii). Similarly, Matthew and Granger (1925a) figured and discussed Desmatotherium mongoliense in more detail, and named another species Desmatotherium fissum.

Matthew and Granger (1925b) erected Colodon inceptus and Paracolodon curtus based on specimens from Ergilin Dzo (Ardyn Obo), Mongolia. They distinguished Colodon inceptus from C. occidentalis on the grounds that the former has a nearly straight centrocrista on molars connecting the paracone and the metacone, in contrast to the centrocrista that curves buccally just behind the paracone in C. occidentalis. Matthew and Granger (1925b) distinguished Paracolodon from Colodon mainly based on the absence of P1, short postcanine diastema, and molar relatively larger as inferred from the partial alveolus.

Radinsky (1963) regarded Desmatotherium as a junior synonym of Helaletes, and later he (Radinsky, 1965a) assigned both D. mongoliense and D. fissum to Helaletes based on the degree of premolar molarization and relatively long and narrow proportions of M1, P2–4, and p2–3 (for “H.” mongoliensis), or P2–4 metalophs less prominent than the protolophs and extending toward the protocones (for “H.” fissus). However, some authors still considered Desmatotherium a valid genus (Schoch, 1989; Colbert and Schoch, 1998). Radinsky (1965a) also considered that specimens on which Colodon inceptus and Paracolodon curtus were respectively based actually represent a single species; he combined them into Colodon inceptus.

Since the revision of Asian Paleogene tapiroids by Radinsky (1965a), Dashzeveg and Hooker (1997) have continued to work on Helaletes and Colodon and assigned the specimens of “Helaletes” fissus to Colodon and suggested its close relationship with C. inceptus; they erected Irdinolophus for “Helaletes” mongoliensis, suggesting its close relationship with Deperetellidae (discussed below).

The new material reported here is nearly identical to the holotype of “Helaletes” fissus (AMNH FM 20161), represented by a left maxilla with P2–4 (fig. 4B; table 1). However, AMNH FM 20161 has a more separated and larger hypocone compared with the protocone on P2 and relatively weaker metalophs on P3–4. In our view, these subtle differences are due to individual variation of the species. The mandible of “H.” fissus? (AMNH FM 81802) is very similar to that of IVPP V 22640 (fig. 5A, B), except that the former is slightly smaller (table 2) and has a smaller and lower paraconid on p2–3, an incipient metaconid and a more basined talonid on p2, and a slightly more reduced paralophid on m2–3. We attribute these differences to individual variation, and assign the specimen (AMNH FM 81802) to “H.” fissus.

Dashzeveg and Hooker (1997) recognized some similarities between “Helaletes” fissus and “Colodon” inceptus in that both shared premolars with incipient molarization and were lacking lingual cingula, with lower molars relatively narrow, and m3 hypoconulid reduced to a cingular bulge. We agree with Dashzeveg and Hooker (1997) that “Helaletes” fissus and “Colodon” inceptus are closely related, and consider that they share several other features, including M1–2 centrocrista less aligned with the metaloph, the upper molar centrocrista straight, the metacone considerably lingually depressed and short with a prominent buccal cingulum, and P2–4 paracone and metacone convex buccally, the protoloph extending toward the parastyle and fading out before reaching to the ectoloph (fig. 4A–D; table 1). We note that P3–4 of “Helaletes” fissus is more molariform than that of “Colodon” inceptus in having the protocone and hypocone more separated by a distinct lingual groove, whereas P2 of “Colodon” inceptus is more molariform than that of “Helaletes” fissus in having a quadrate outline in occlusal view and a stronger metaloph.

TABLE 1.

Measurements of upper cheek teeth of Paracolodon fissus, Paracolodon inceptus, and Desmatotherium mongoliense (mm).

It is uncertain whether “Helaletes” fissus and “Colodon” inceptus should be included in the genus Colodon. Radinsky (1965a: 232) thought that “Colodon” inceptus is different from Colodon occidentalis in having a small upper canine, a relatively longer P1 with separated paracone and metacone, more convex P2–4 paracone not so merged into the ectoloph, P3–4 hypocone not so well differentiated from the protocone, and M2–3 paracone narrower and less extended distally. These differences also distinguished “Helaletes” fissus from Colodon occidentalis, except that C1 and P1 were unknown in “Helaletes” fissus. The lower cheek teeth of “Helaletes” fissus further differ from those of Colodon occidentalis in being relatively longer and narrower, and in having a small c1, premolars much less molariform with smaller entoconids and talonids slightly wider than the trigonids, and m3 with much smaller hypoconulid.

TABLE 2.

Measurements of lower cheek teeth of Paracolodon fissus and Desmatotherium mongoliense (mm).

The skull of “Helaletes” fissus (IVPP V 22640) further differs from that of Colodon described by Colbert (2005) in having the following characters (fig. 2A–C): the lateral trough on the ascending process of the maxilla extending posteriorly over the supraorbital process of the frontal rather than terminating in a fossa, the frontal with a pair of anteroposteriorly extended medial troughs separated by a median anterior sagittal crest rather than dorsally inflated, the anterior border of the orbit above the boundary of M1 and M2 rather than above M1, and a relatively small lacrimal tubercle. Although the rostral part of the skull in IVPP V 22640 is not preserved, another mandibular specimen (AMNH FM 81802) with the symphysis has a relatively short symphysis and postcanine diastema, suggesting a relatively shorter preorbital skull length of “Helaletes” fissus to that of Colodon. “Colodon” inceptus also has a rather short postcanine diastema, indicating a similar short preorbital skull length as in “Helaletes” fissus.

In conclusion, both dental and especially cranial characters indicate that “Helaletes” fissus and “Colodon” inceptus belong to the same genus that differs from Colodon. Thus, we resumed the genus Paracolodon for these two species: P. fissus and P. inceptus. The possibility that the middle Eocene Irdinmanhan P. fissus is generically distinct from the later Eocene Ergilian P. inceptus cannot be ruled out, pending discovery of more complete material of P. inceptus.

Although the genus Desmatotherium was synonymized with Helaletes by Radinsky (1963), some authors still regarded Desmatotherium as a valid genus based on its larger size, the metalophs bypassing the hypocones on P3–4, and highly reduced hypoconulid on m3 (Schoch, 1989; Colbert and Schoch, 1998). Paracolodon fissus is similar to Desmatotherium intermedius in having the P3–4 metaloph extending toward the protocone, bypassing the hypocone. However, P. fissus is mainly distinguished from D. intermedius by teeth relatively shorter and wider (figs. 6–8), P2 more molariform with distinct hypocone, P2–4 paracone and metacone more convex and more closely situated, P3–4 protoloph extending toward the parastyle and fading out before its contact with the ectoloph, P4 metacone slightly more lingually depressed, upper molar metacone more lingually depressed, more concave, and shorter, protoloph and metaloph straight and less mesially convex, and p3–4 talonid nearly as wide as the trigonid.

In terms of size and morphology, the upper molars of P. fissus are rather similar to those of “Colodon” kayi and “Colodon” woodi, both of which were originally assigned to Desmatotherium by Hough (1955) and Gazin (1956). Radinsky (1963) referred “Desmatotherium” kayi and “Desmatotherium” woodi to Colodon mainly based on their molariform P3–4, and this assignment was followed by following authors (Eaton, 1985; Tabrum, 2012). On the other hand, upper molars of “Colodon” kayi and “Colodon” woodi are similar to those of Paracolodon fissus, but different from those of Colodon occidentalis, in having approximately equal size, straight centrocrista, the paracone relatively higher and sharper, and M2 metacone slightly less reduced (Matthew and Granger, 1925b; Hough, 1955; Radinsky, 1963). The upper premolars of “Colodon” kayi and “Colodon” woodi are further similar to those of P. fissus in having more buccally convex paracone and metacone and less separated protoloph and metaloph compared with those of Colodon occidentalis. The lower premolars of “Colodon” kayi are somewhat intermediate between P. fissus and C. occidentalis in proportion and morphology. However, the presence of the lower canine, the deep buccal fold on p2–4 (Radinsky, 1963), and the m3 hypoconulid strongly reduced are features that are more similar to those of P. fissus. However, P. fissus is mainly different from “C.” kayi and “C.” woodi in having less molariform upper premolars with metaloph extending toward the protocone, upper molars with more lingually depressed metacone, and lower premolars relatively longer and narrower with smaller entoconid. As a result, we are skeptical of the referral of kayi and woodi to Colodon, and we suggest assigning the species in Colodon? kayi and Colodon? woodi.

Petrosal: Because the new specimen (IVPP V 22640) preserves the relatively complete petrosal, it gives us an opportunity to investigate the morphology of petrosals of early perissodactyls. There are not many previous studies on early perissocdactyl petrosals. Paleogene perissodactyl petrosals that were mentioned or described with or without illustrations include Hyracotherium, Orohippus (Cifelli, 1982), Pachynolophus (Savage et al., 1965), Heptodon (Radinsky, 1965b; Cifelli, 1982), Schlosseria (Li and Wang, 2010), Hesperaletes (Colbert, 2006), Litolophus (Bai et al., 2010), and cf. Protitanotherium (Mader, 2009). Furthermore, petrosals of extant Equus and Tapirus have also been described (Sisson et al., 1975; O'Leary, 2010). Detailed comparisons and studies of perissodactyl petrosals are beyond the scope of present paper, so we focus our discussion on certain characters related to our new specimens.

The petrosal of Paracolodon fissus is characterized by the relatively medially situated transpromontorial sulcus, while in other known perissodactyls the transpromontorial sulcus is either absent or more laterally situated. The promontorium of perissodactyls were usually considered to be smooth, without traces of grooves for the internal carotid artery/plexus or stapedial artery (Wible, 1986). The internal carotid artery is medial to the auditory bulla and middle ear cavity (extrabullar) (Wible, 1986). However, Li and Wang (2010) described the petrosal of Schlosseria with three grooves on the promontorium, the sulcus for medial ramus of internal carotid artery, the stapedial artery sulcus, and promontory artery sulcus. We suspect that “the sulcus for medial ramus of internal carotid artery” in Schlosseria needs to be further investigated, because eutherians usually have a single internal carotid artery in the lateral groove, and the medial groove was usually for the inferior petrosal sinus (Wible, 1983; 1986). Colbert (2006) mentioned that the petrosal of Hesperaletes has a distinct transpromontory sulcus, running anteriorly from the fenestra cochleae. In Paracolodon fissus, we suggested that a medial groove probably represents the transpromontorial sulcus for the following reasons: First, the groove is too ventrally displaced to be the basicapsular groove, which is for the inferior petrosal sinus. In Tapirus and Equus, as well as many artiodactyls, the basicapsular groove is on the dorsal surface (O'Leary, 2010). On the other hand, the basicapsular grooves in a few groups (such as hippopotamids) with ventrally placed positions run along the medial border (O'Leary, 2010). Furthermore, we considered that a more medially situated, narrower groove probably represents the basicapsular groove. Second, the groove is relatively wide, which was likely not made by a single soft tissue, but may have been made by internal carotid artery, internal carotid plexus, and/or tympanic plexus corresponding with the transpromontorial sulcus.

The promontorium of Paracolodon is almond shaped, which is similar to that of Schlosseria and Hesperaletes, but different from hemiellipsoid promontorium in Heptodon and Tapirus. The petrosal of Paracolodon is similar to those of Heptodon and Tapirus in having the fenestra cochleae considerably larger than the fenestra vestibuli. The hiatus Fallopii is ventrally situated in Hyracotherium, Orohippus, and Heptodon, whereas that of Tapirus is situated anterior to the tegmen tympani. Unfortunately, the hiatus Fallopii in Paracolodon was uncertain, either absent or not preserved. The dorsolateral tegmen tympani of Paracolodon fissus is prominently convex, whereas that of Tapirus is flat.

Description

DP1 (P1): The tooth is anteriorly slanted as compressed by P2 (fig. 4E). The color of the enamel is light brown, lighter than black-brown enamels of other teeth, suggesting a deciduous tooth. DP1 is roughly oval in occlusal outline with two roots. A main cusp is present in the middle of the crown, extending a short ridge mesially to the indistinct parastyle, and then slightly curving lingually. A ridge extends distally from the main cusp terminating in a more distinct and slightly higher cusp compared with the parastyle. This distal cusp probably represents the rudimentary metacone. The buccal side of the ectoloph is slightly convex. A ridge extends from the lingual base of the main cusp distolingually, and another ridge extends from the apex of the metacone lingually along the distal border. Two ridges are separated by a shallow notch on the lingual side. A cingulum is absent.

P2: The tooth is roughly rectangular in outline and relatively long and narrow (fig. 4E). The paracone and metacone are equal in height and closely situated. Both of them are slightly convex on the buccal surface and merged with the ectoloph. The parastyle is distinct and mesial to the paracone, whereas the metastyle is indistinct and lower. The strong protoloph extends toward the parastyle, fading out before its contact with the ectoloph. A ridge extends from the protocone distally and even beyond the hypocone. The protocone and the hypocone are separated by a faint groove on the lingual side. A tiny cuspule is present in the valley between the hypocone and the metacone, representing the metaconule. A weak ridge connects the hypocone and the metaconule. The mesial cingulum is weak, whereas the distal cingulum is considerably stronger. The buccal cingulum is weak and somewhat interrupted at the paracone, and the lingual cingulum is absent.

P3: The P3 is similar to P2, but is relatively shorter and wider (fig. 4E). It further differs from P2 in having the protoloph in contact with the parastyle, the hypocone slightly more lingually situated compared with the protocone, and the metaloph completely absent.

P4: The P4 is relatively shorter and wider than P2–3 (fig. 4E). The paracone and metacone are equal in height, but more separated from each other than those of P2–3. The metacone is less convex and slightly more lingually situated than the paracone. The parastyle is more readily discriminated from the ectoloph than that of P2–3. The protoloph extends toward the junction of the parastyle and the preparacrista. The protocone and the hypocone are separated by a relatively distinct groove on the lingual side. The metaloph is as strong as the protoloph, extending from the hypocone to the mesiolingual base of the metacone, but separated from the latter by a shallow valley.

M1: The tooth is moderately worn and roughly square in outline (fig. 4E). The paracone is broadly convex, while the metacone is strongly lingually depressed and rather short with a slightly convex buccal surface. The postmetacrista is distobuccally extended. The parastyle is large, situated mesially to the paracone and closely appressed to the latter. The protoloph extends toward the preparacrista. The metaloph is about half of the protoloph length, joining the ectoloph at the apex of the metacone. A faint crest is present on the lingual surface of the paracone and nearly vertically placed. A weak cingulum is present on the mesial side, distal side, and the lingual opening of the middle valley. A prominent buccal cingulum is present adjacent to the metacone, although this portion is partially broken.

M2: The M2 is similar to M1 in morphology, but M2 is considerably larger than M1 (fig. 4E). M2 is further different from M1 in that the buccal side of the tooth is longer than the lingual side, the apex of the paracone curves distally as deduced from the direction of the lingual crest, and the parastyle is relatively and absolutely larger.

Comparisons and Discussion

The new material is very similar to the holotype of “Helaletes” mongoliensis (figs. 4G, 6; table 1), especially in their degree of premolar molarization, in which P2–3 protocone is distally extended, the hypocone is rudimentarily separated from the protocone, and the metaloph is much reduced as compared with the protoloph (fig. 4E, G). They are further similar in having premolars relatively long and narrow, paracone and metacone merged with the ectoloph, M1–2 paracone broadly convex with the apex of the paracone nearly vertically situated on M1, and distally inclined on M2, metacone strongly lingually depressed and short, and infraorbital foramen opening above the posterior border of P4. However, the new material is different from the holotype in having P2 protocone and hypocone less separated on the lingual side, and P3 lacking the metaloph—all due to individual variation.

The type of “Helaletes” mongoliensis (AMNH FM 19161), the right maxilla with P2–M3, was collected from Irdin Manha Formation at Irdin Manha escarpment, 23 miles south of Iren Dabasu in 1922. All other referred specimens housed at the AMNH were collected from Irdin Manha Formation at Irdin Manha escarpment in 1923. The new materials expanded the distribution of “Helaletes” mongoliensis to the Huheboerhe area.

Radinsky (1965a) determined that the lower cheek teeth of “Helaletes” mongoliensis are more similar to those of Colodon in having relatively wider p4–m2 except for the relatively long and narrow p2–3, while the upper cheek teeth are more similar to Helaletes (assumed for “Helaletes” intermedius) in the degree of premolar molarization and proportion of M1, P2–4, except for the more reduced and depressed metacones on M1–3. That Radinsky (1965a) assigned these materials to Helaletes rather than to Colodon was “an unusually subjective decision” (Radinsky, 1965a: 230).

Dashzeveg and Hooker (1997) erected the genus Irdinolophus for “Helaletes” mongoliensis, and reported an isolated upper preultimate molar of the species from eastern Gobi, Mongolia. They considered Irdinolophus as the most primitive member of the family Deperetellidae, because its p3–4 talonid is slightly wider and p4 trigonid shorter than other species of the family; in addition, its p3 paraconid is distinct and premolars show initial elongation (fig. 5C, D). These characters typify advanced members of the Deperetellidae, but they differ from those of Helaletes nanus and Desmatotherium intermedius. Nevertheless, the wider p3–4 talonid is also encountered in Desmatotherium intermedius (AMNH FM 12672) (Radinsky, 1963; Schoch, 1984) and Colodon occidentalis, and the distinct or weak p3 paraconid is variable as indicated by Paracolodon fissus (see above). The p4 trigonid is not as shortened as in Colodon and deperetellids, but somewhat intermediate between Helaletes and those two groups. Furthermore, the upper cheek teeth of Teleolophus are obviously distinguished from those of “Helaletes” mongoliensis in having P2–4 with a protoloph joining the equally developed metaloph a short distance before its lingual end (hypocone), and M1–3 with protoloph, paracone, and metaloph forming an inverted “U” and a metacone greatly reduced. Last, the earliest known deperetellid is from the Arshantan, earlier than Irdinmanhan “Helaletes” mongoliensis (Radinsky, 1965a; Meng et al., 2007; Wang et al., 2010). As a result, we are skeptical of considering “Helaletes” mongoliensis the most primitive member of the family Depetellidae, and the main reason for erecting the new genus “Irdinolophus” is not tenable. However, another species Irdinolophus? tuiensis could hold the validity of Irdinolophus by its S-shaped ectoloph.

“Helaletes” mongoliensis is similar to Paracolodon fissus in size (table 1, 2; figs. 6–8) and general morphology, and both are from Irdin Manha Formation, Erlian Basin. Radinsky (1965a) even suggests that Paracolodon fissus was probably an advanced variant of “Helaletes” mongoliensis. On the other hand, Radinsky (1965a) shows that P2–4 of P. fissus are more advanced than those of “Helaletes” mongoliensis in being relatively shorter and wider (figs. 6, 7) and in having the hypocone better separated from the protocone. Furthermore, the new material shows that “Helaletes” mongoliensis is different from P. fissus in having the M2 paracone distally inclined, M3 metacone slightly more lingually depressed (fig. 4A, B, E–G), p2 metaconid more separated from the protoconid, p2–4 buccodistal cingulum more prominent, p3–4 entoconids stronger, cristid obliqua slightly more buccally extended, p4 talonid relatively wider than trigonid, and m3 with slightly larger hypoconulid (fig. 5A–D).

Unfortunately, the skull of “Helaletes” mongoliensis (AMNH FM 19161) preserves only the preorbital portion (fig. 9), and this part is broken on the skull of Paracolodon fissus (IVPP V 22640). However, the long postcanine diastema of “Helaletes” mongoliensis suggests a relatively long preorbital portion, which contrasts with the relatively short preorbital portion of Paracolodon fissus as deduced from a mandible with a complete symphysis (AMNH FM 81802). The infraorbital foramen of “Helaletes” mongoliensis opens above the posterior end of P4, whereas that of Paracolodon fissus opens above the posterior end of P3. Furthermore, in “Helaletes” mongoliensis, based on the specimens AMNH FM 19161 and AMNH FM 20156E, the lateral border of the lateral trough of the ascending process of the maxilla seems completely composed of the maxilla and close to the anterior border of the orbital, and the facial part of the lacrimal is probably narrow (fig. 9), a condition similar to that of Helaletes nanus (AMNH FM 11635). By contrast, in Paracolodon fissus (IVPP V 22640), the lateral border of the lateral trough seems composed of the maxilla and lacrimal and prominently separated from the anterior border of the orbital, and the facial part of the lacrimal is probably larger (fig. 2A). Thus, these cranial differences between “Helaletes” mongoliensis and Paracolodon fissus are probably at the genus level.

Compared with Helaletes nanus, besides above mentioned characters, “Helaletes” mongoliensis is further similar to H. nanus (AMNH FM 11635) in having a relatively wide and shallow lateral trough, an anterior portion of the maxilla rising upward, the infraorbital foramen opening above the posterior end of P4, P2–4 metaloph weaker compared with protoloph, P4 protoloph extending toward the preparacrista, and the canine relatively large (Radinsky, 1965a). On the other hand, the facial part of “Helaletes” mongoliensis bears a shallow maxillary fossa and a relatively large depression for the attachment of the buccinator muscle above the postcanine diastema, and the canine is closely appressed to the I3 (fig. 9), whereas in H. nanus (AMNH FM 11635) the maxillary fossa and depression above the diastema are absent, and the canine is separated from I3 by a diastema. In terms of dentition, “Helaletes” mongoliensis mainly differs from Helaletes nanus in having P2–4 hypocone slightly more separated from the protocone, M1–3 parastyle more closely compressed to paracone, metacone shorter and more lingually depressed, p2–3 entoconid more prominent, talonid relatively wider than trigonid, m1–3 cristid obliqua slightly more reduced, and m3 hypoconulid smaller.

FIG. 6.

Scatter plot for P3 size of Paracolodon and Desmatotherium (in mm).

“ Helaletes” mongoliensis is similar to Desmatotherium intermedius in having a relatively long postcanine diastema, an infraorbital foramen opening above the posterior border of P4, and a distinct depression for the attachment of the buccinator above the postcanine diastema. They are further similar in having upper cheek teeth relatively long and narrow (figs. 6–8), P2–4 metaloph weaker compared with protoloph, P4 protoloph extending toward preparacrista, M1–3 parastyle compressed to the paracone, metacone less reduced, protoloph and metaloph slightly convex mesially, M2 paracone distally curved, p3–4 talonid wider than trigonid, and entoconid distinct. On the other hand, Desmatotherium intermedius is different from “Helaletes” mongoliensis in having P3–4 somewhat more molariform, M1–3 metacone slightly less lingually depressed, M2–3 centrocrista less aligned with the metaloph, p3–4 talonid relatively wider compared with trigonid, and m3 hypoconulid highly reduced.

In conclusion, considering the differences of cranial characters between Paracolodon fissus and “Helaletes” mongoliensis, and similarities of dental characters and degree of premolar molarization between “Helaletes” mongoliensis and Desmatotherium intermedius, we assigned “H.” mongoliense to Desmatotherium as originally named by Osborn (1923).

FIG. 7.

Scatter plot for P4 size of Paracolodon and Desmatotherium (in mm).

Radinsky (1965a) also assigned a right lower jaw with p2–m2 (AMNH FM 81803, field no. 927) to Helaletes sp. from “Houldjin gravels,” 10 miles southwest of Camp Margetts. In fact, this specimen was unearthed from Irdin Manha Formation in Chaganboerhe, where the new material was found (see discussion below). As noted by Radinsky (1965a), the specimen of AMNH FM 81803 is different from P. fissus in having “the termination of the symphysis well anterior to p2 and in having relatively long p3–4 trigonids” (Radinsky, 1965a: 231). Furthermore, p3–4 of AMNH FM 81803 has a cristid obliqua buccally extended and a distinct buccodistal cingulum, which suggests its affinity with Desmatotherium mongoliense.

Qi (1987) named Helaletes medius from the Arshanto Formation, Wulanboerhe, based on fragmental lower jaws (IVPP V 5729, V 5730). However, its specific character of the m1 and m3 having a well-developed cristid obliqua excludes the specimen from Helaletidae, which has a relatively reduced cristid obliqua. On the other hand, the small size, long trigonids of p4–m3, and small hypoconulid on m3 of IVPP V 5729 are very similar to those of AMNH FM 81788 (field no. 931), which was unearthed from Arshanto Formation, 10 miles southwest of Camp Margetts and assigned to Schlosseria cf. S. magister by Radinsky (1965a). However, the p4–m3 of IVPP V 5729 are relatively wider than those of AMNH FM 81788, and p4 of IVPP V 5729 has a more lingually situated paralophid. The affinity of IVPP V 5729 and the specimens of Schlosseria from the Camp Margetts area need further investigation.

Cladistic Analyses

We performed phylogenetic analyses using PAUP 4.0 and TNT 1.1 with a parsimony criterion (Swofford, 2002; Goloboff et al., 2008). The heuristic search algorithm was used with 1000 replications of random stepwise addition and tree-bisection-reconnection (TBR) branch swapping. Of the 72 characters, three are parsimony uninformative.

The analyses result in a single most parsimonious tree (fig. 11). The tree length is 268; the consistency index (CI) is 0.4216; the homoplasy index (HI) is 0.5784; the retention index (RI) is 0.5032; the rescaled consistency index (RC) is 0.2122.

FIG. 10.

Longitudinal section of the ectoloph of upper right P4 of Paracolodon fissus and Desmetotherium mongoliense showing the curved HSB configuration. A, P. fissus with continuous and relatively narrower HSB pattern; B, D. mongoliense with waved and relatively wider HSB pattern. Abbreviations: if, interface; me, metacone; pa, paracone; pas, parastyle.

The most parsimonious tree shows some results partially supporting our analyses above. The analysis indicates that both Paracolodon and Colodon are valid genera. The Paracolodon clade (node 20) includes only Asian P. fissus and P. inceptus and is supported by three synapomorphic characters (appendix 3). The Colodon clade (node 25) is supported by four synapomorphic characters (appendix 3), and Colodon stovalli is at the base of the clade. The species C.? woodi and C.? kayi are also included in Colodon clade instead of Paracolodon, which is consistent with statements proposed by Radinsky (1963). Nevertheless, the species C.? woodi and C.? kayi are more reasonably assigned in either Colodon or Paracolodon rather than in Desmatotherium as originally referred to. Furthermore, the result shows Paracolodon and Colodon form a sister group, which indicates that Paracolodon is more closely related to Colodon than to Hesperaletes, Protapirus, or Plesiocolopirus, in contrast to the above analysis based on cranial characters. The Paracolodon-Colodon clade (node 26) is supported by four synapomorphic characters (appendix 3). Plesiocolopirus and Protapirus form successive sister-taxa lineages to Paracolodon-Colodon clade. The high-latitude late Wasatchian Thuliadanta mayri is more primitive than Helaletes (Eberle, 2005), and they form successive sister-taxa lineages to Plesiocolopirus, Protapirus, Colodon, and Paracolodon clade. Furthermore, the result also supports that the assignment of the species Desmatotherium mongoliense and Paracolodon fissus from Irdin Manha Formation, Erlian Basin, into two different genera, since Desmatotherium mongoliense and Desmatotherium intermedius form a sister group that is more closely related to Heteraletes than to any other helaletids. The Desmatotherium clade (node 31) is supported by six synapomorphic characters (appendix 3). Hesperaletes is placed in a relatively basal position, more derived than Heptodon but less derived than any other helaletids. To sum up, the result of the cladistic analysis is partially consistent with the results based on the morphologic comparisons, although some discrepancies are present. It is premature to resolve these discrepancies considering the lack of cranial materials for many helaletids.

FIG. 11.

The single most parsimony tree and the distribution of selected taxa. The left column showing Polarity Chrons, NALMA, and ALMA was modified from Vandenberghe et al. (2012). Numbers on the cladogram are node numbers, and numbers in the parentheses are Bremer Support which is greater than 1. Abbreviations: ALMA, Asian Land Mammal Ages; Ck., Clarkforkian; Du., Duchesnean; Er., Ergilian; Ga., Gashatan; Hs., Hsandagolian; NALMA, North American Land Mammal Ages; Or., Orellan; Pa., Paleocene; Ta., Tabenbulakian; Th., Thanetian; Ul., Ulangochuian; Wt., Whitneyan.

Irdinmanhan ALMA

The phylogenetic analysis combined with morphological comparisons is also informative for correlation between Irdinmanhan ALMA (Asian Land Mammal Ages) and related NALMA (North American Land Mammal Ages). The Irdinmanhan ALMA is commonly correlated to the Uintan NALMA, but precise correlation is not clear (Romer, 1966; Vandenberghe et al., 2012). Tong et al. (1995) suggested that Irdinmanhan should be correlated with early Uintan, whereas the late Uintan is equal to Sharamurunian ALMA. Ni et al. (2010) suggested that the Irdinmanhan is probably slightly older than Ui1 subage of the Uintan, based on the fact that Tarkops from the Irdin Manha Formation is more primitive than Tarka from Ui1. The rodent assemblage (Li, 2016) from the lower part of the Irdin Manha Formation also suggested that the Irdinmanhan may be earlier than late Eocene and should be correlated with early Uintan (Li and Meng, 2015).

The Ulan Shireh fauna in the Shara Murun region has been roughly correlated to the Irdin Manha fauna (Radinsky, 1965a, 1967; Ye, 1983; Wang et al., 2012), but Li et al. (2016) recongnized a most probably Sharamurunian of the upper horizon of the Ulan Shireh Formation. To better define the Irdin Manha fauna, we list the following perissodactyls that are exclusively from the Irdin Manha Formation at the Irdin Manha escarpment and Huheboerhe area where the Irdin Manha Formation is well identified. These perissodactyls include: Brontotheriidae: Microtitan mongoliensis, Protitan grangeri (= P. robustus, P. obliquidens), Protitan minor, Protitan? cingulatus, Gnathotitan berkeyi, Metatitan relictus, Hyotitan thomsoni, Metatelmatherium cristatum, Metatelmatherium parvum; Deperetellidae: Teleolophus medius, Deperetella sp.; Helaletidae: Paracolodon fissus, Desmatotherium monogoliense; Lophialetidae: Lophialetes expeditus; Hyracodontidae: Triplopus? proficiens, Paraceratheriidae Fostercooperia totadentata, Amynodontidae Rostriamynodon grangeri (Granger and Gregory, 1943; Radinsky, 1965a, 1967; Russell and Zhai, 1987; Wall and Manning, 1986; Mihlbachler, 2008). Lophialetidae and Deperetellidae are endemic families in Asia, and many brontotheres are indigenous (Granger and Gregory, 1943; Radinsky, 1965a). Among brontotheres, Metatelmatherium cristatum was considered as a junior synonym of the Uintan Metatelmatherium ultimum from North America (Mihlbachler, 2008). Among hyracodontids, the North American Triplopus had a temporal distribution ranging from the Uintan to Duchesnean, and Uintaceras, which is similar to Fostercooperia to some extent, is from the Uintan (Radinsky, 1967; Holbrook and Lucas, 1997; Prothero, 1998). On the other hand, Rostriamynodon was considered more primitive than the Uintan Amynodon from North America (Wall and Manning, 1986).

Among helaletids from the Erlian Basin, Paracolodon fissus is more primitive than C.? kayi and C.? woodi from late Uintan Sage Creek, Montana, and Wind River Basin, Wyoming (Radinsky, 1963), in degree of premolar molarization, and thus supports a probable date for the Irdinmanhan earlier than late Uintan (fig. 11). Furthermore, upper molars of Desmatotherium mongoliense from Irdin Manha Formation are slightly more derived than those of D. intermedius from the late Bridgerian of the Bridger Basin and probably Washakie Basin, Wyoming (Radinsky, 1963; Schoch, 1984); thus, the Asian Irdinmanhan is probably younger than the late Bridgerian. Based on all available evidence, we suggest that the Irdinmanhan can be restricted in the interval roughly correlative to Ui1 and Ui2 subages of the North Amerian Uintan (fig. 11).