A New Species of Megacricetodon (Cricetidae, Rodentia, Mammalia) from the Middle Miocene of Northern Junggar Basin, China

Abstract Dental, mandibular, and postcranial specimens of Megacricetodon yei n. sp., are described. The new specimens, including the complete dentition, mandible, and partial forelimb and hindlimb, represent the most complete materials known for the genus, provide valuable information concerning the interspecific variation of the genus, and lead to the reassessment of the suprageneric position of Megacricetodon. Megacricetodon yei is characterized by having medium-size, clearly split anterocone of M1, presence of the labial spur of the anterolophule and the posterior spur of the paracone in some M1s, medium to long mesoloph in M1-2, frequent occurrences of double protolophules, transverse or posteriorly directed metalophule of M2, and single anteroconid of the m1. Megacricetodon yei is more closely related to Megacricetodon (= Aktaumys) dzhungaricus than to any other species of Megacricetodon, but is more derived than the latter. Based on the new information, the validity of the genus Aktaumys is discussed. The postcranial features of Megacricetodon yei show clear adaptations for terrestrial habits, but as in many ground-dwelling rodents living in burrows, it could also climb or dig. The associated fauna has been correlated to Tongxin fauna from the adjacent part of China and the Belometchetskya fauna of north Caucasus, equivalent to early Middle Miocene age, or MN 6 correlative. The stage of evolution of Megacricetodon yei is consistent with the faunal correlation.


INTRODUCTION
Cricetid rodents of the genus Megacricetodon are among the most abundant mam¬ malian species in the Old World Cenozoic fauna. Fahlbusch (1964) first recognized Megacricetodon as a subgenus of the genus Democricetodon, which he established at the time, but later researchers have elevated the subgenus to the genus level (Mein and Freudenthal, 1971a). The genus Megacri¬ cetodon is species-rich. To date, more than 25 fossil species have been recognized from the Miocene of Europe and Asia. Despite their great diversity and widespread distribu¬ tions, all the species of Megacricetodon are represented primarily by isolated teeth, often collected by screen washing. Very few cranial and postcranial bones are known. In addi¬ tion to numerous specimens of dentitions, here we describe the best preserved postcra¬ nial skeletal materials known for this genus from the northern Junggar basin, Xinjiang, China. The Junggar materials shed new light on the interspecific variation and anatomy of Megacricetodon and help to clarify several taxonomic issues of the genus.
Our field investigations, run from 1995 through the present, have greatly improved the documentation of Tertiary strata in this area. Based on these new discoveries, Ye et al. (2001aYe et al. ( , 2001bYe et al. ( , 2003 redefined the stratigraphical units of TCC, in ascending order, as: (1) the Eocene-Oligocene Ulunguhe Formation, (2) the Late Oligocene Tieersihabahe Forma¬ tion, (3) the Early Miocene Suosuoquan Formation, (4) the Middle Miocene Halamagai Formation, and (5) the late Middle Miocene Kekemaideng Formation. The mate¬ rials described here were collected from the Halamagai Formation between the Tieersihabahe section and Chibaerwoyi sec¬ tion in 1998 and 2000. The bones were excavated mainly from a mudstone lens (about 2 m2) imbedded in grayish medium¬ grained sandstone. Dozens of maxillae and mandibles with complete dentitions, 25 isolat¬ ed teeth, and about 27 postcranial elements were recovered. Although no articulated elements were found, the ratio of the speci¬ mens, morphology, and size suggest that they belong to the same species. Numerous speci¬ mens of a distinctively larger taxon, Cricetodon n. sp. were also recovered from the same spot and have been described by Bi (2005). Comparisons are specifically made to Cricetodon from Tieersihabahe since it represents the most complete skele¬ tal materials of extinct cricetids known to date.

MATERIALS AND METHODS
Terminology for dental morphology is after Mein and Freudenthal (1971a). Because their terminology has been well known, we prefer not to include a diagram to illustrate the tooth structure. Positional abbreviations of teeth follow the common alphanumeric convention of using uppercase versus lowercase letters to identify maxillary or mandibular teeth, respectively, and num¬ bers to indicate their placement in the tooth row (for example, Ml and ml). The cranial and postcranial skeletal terminologies, wher¬ ever appropriate, follow those by Howell (1926), Rinker (1954), Cooper and Schiller (1975), Carleton and Musser (1989), Carleton and Olson (1999), and Gebo and Rose (1993).
Teeth and mandibles were imaged and measured using a Nikon SMZ 8 microscope set at 20 X magnifications, and measure¬ ments were recorded to the nearest 0.01 mm. The SEM photographs of some teeth were taken from uncoated specimens using a Hitachi scanning electron microscope. All postcranial measurements were taken to 0.05 mm using digital calipers. IVPP is the abbreviation of the Institute of Verte¬ brate Paleontology and Paleoanthropology, Beijing. Etymology: The species name, yei, is in honor of our colleague, professor Jie Ye, for his contribution to the study of Cenozoic stratigraphy and mammals in northern Xinjiang and discovery of the locality.
Repository: The specimens are housed in the collections of the Institute of Vertebrate Paleontology and Paleoanthropology, Chinese Academy of Sciences, Beijing.
Diagnosis: A Megacricetodon species of medium size, Ml with clearly split anterocone, presence of the labial spur of the anterolophule and the posterior spur of the paracone in some Mis, medium to long mesoloph in Ml-2, frequent occurrences of the double protolophules in M2 of the collection, M2 metalophule transverse or posteriorly directed, ml anteroconid simple, mesolophid of medi¬ um length in ml-2, and presence of the entepicondylar foramen in the humerus.
Differential diagnosis: Differs from Megacricetodon (= Aktaumys) dzhungaricus in having slightly smaller size; better bifurcated Ml anterocone; shorter mesoloph in Ml and M2; fewer occurrences of the labial spur of the anterolophule in Ml; fewer occurrences of double protolophule and metalophule in M2. Differs from any other known species of Megacricetodon in having the following combi¬ nation of features: well-bifurcated Ml antero¬ cone; presence of a labial spur of the ante¬ rolophule in many Mis, double protolophules in most M2s, the presence of the posterior spur of paracone, and long mesoloph in Ml and M2, and single ml anteroconid. Of the cranial elements, only the left fragmentary maxilla is preserved in IVPP V15349.9 ( fig. 1). The zygomatic plate is broad and has a well-defined fossa on the ventral surface for the origin of the anterior part of the muscle masseter lateralis profundus. The ante¬ rior margin of the plate is partially broken, but the oval patch for the superficial masseter is preserved; it is prominent and located immedi¬ ately posteroventral to ventral end of the anterior margin. The posterior margin of the plate extends anterior to Ml. The posterior portion of the incisive foramen is preserved, of which the posterior edge levels with the protocone of Ml, as in the Cricetodon, but differing from those in most living cricetines (Cricetus and Mesocricetus) in which the incisive foramen terminate anterior to Ml.
Upper teeth: Seven upper incisors are preserved. Its anterior surface is gently convex and smooth. The enamel wraps slightly around onto both lateral and medial surface, but extends farther on the lateral surface than on the medial one ( fig. 2). The cross section of the incisor is oval.
Eight complete Mis are in the collection and five of them are preserved in the maxilla ( fig. 1, 3, 4). Ml is the largest upper cheek tooth with three roots, a major lingual one and two minor buccal ones; the root support¬ ing the anterocone extends slightly anteriorly. The occlusal outline is longer than wide and slightly widens posteriorly. The anterocone is divided by a deep, longitudinal groove into labial and lingual conules, of which the former is larger. There is a low ledge on its anterior face in five of eight specimens. The anterolophule is connected to the lingual conule in six of eight specimens or extends to a point between the two conules in two Mis. The low labial spur of the anterolophule is long and transverse in two specimens, of medium length and joins the base of the paracone in two specimens, and absent in the remaining specimens. The posterior spur of the paracone is present, connecting to the mesoloph in three specimens. The mesoloph is of medium length in one specimen but is long and reaches the labial border of the occlusal surface in seven specimens. The entomesoloph is strong, reach¬ ing the lingual border of the occlusal surface only in one specimen ( fig. 4A). The metalophule is directed posteriorly, joining the posteroloph immediately behind the hypocone. The posteroloph is long, extending to the posterolabial part of the metacone. A cingulum is usually developed and inflated to form a small cuspate between major cusps. The sinus is transverse.
Seven M2s are in the collection. The M2 is also triple-rooted and its occlusal outline is subquadrate ( fig. 3, 4). The lingual and labial branches of the anteroloph are well developed and are of about equal lengths. In three M2s, the protolophule is doubled; in the remaining four, only the anterior protolophule is present 7 and projects anterolingually. The single metalophule extends anterolingually in one M2, transversely in two M2s, and posterolingually joining the posterior arm of the hypocone in four M2s ( fig. 4). The mesoloph is long in five specimens and is of medium length in the remaining two. In two M2s, the posterior spur of the paracone is present and connects to the mesoloph. The sinus is transverse.
Eight M3s are in the collection. The M3 is much smaller than Ml-2 and is also triplerooted. The labial anteroloph is well devel¬ oped in all M3s, but the lingual one is present only in two specimens. The protocone and paracone are prominent, but the posterior cusps are reduced. The metalophule extends anterolingually to connect the longitudinal crest. An axioloph originating from the longitudinal crest is directed anteriorly and connects to the protolophule in two M3s ( fig. 3). The hypocone and metacone are nearly merged into the posteroloph, which encircles the posterior sinus.
Mandible: IVPP VI5349.23 is a nearly complete mandible except for the broken angular process ( fig. 5A, B, C). The lower incisor is displaced from the alveolus and protrudes more anterodorsally than its original position. The original position of the incisor is preserved in V15349.24 ( fig. 5D, E, F). The dentary is thin; its ventral rim is convex under ml and is concave under the last two molars. The anterior slope of the diastema is gently curved, whereas its posterior slope, anterior to ml, is steep. The length of diastema is 3.1 mm, less than that of the lower tooth row. On the lateral surface of the mandible, the masseteric crests are weaker than those of Cricetodon; the superior and inferior crests converge anteriorly as a V shape, with its apex terminating at the level of the anterior root of ml. An oval mental foramen is at the position anteroventral to the ml; it opens anterolaterally and is best seen in lateral view. On the medial surface of the mandible, the large mandibular foramen lies anterovental to the base of the condyloid process. The ascending ramus is lateral and oblique to the tooth row, similar to that of M. similis (the only other Megacricetodon species preserving the medial side of the mandible) and extant cricetines (Cricetus and Mesocricetus), but differing from that of Cricetodon. In M. yei. M. similis, and extant cricetines (Cricetus and Mesocricetus), the coronoid process is falciform and very pronounced, with its tip higher than the condyloid process. In contrast, the coronoid process is reduced and its tip is slightly below or about the same level with the condyloid process in Cricetodon. The sigmoid notch is deep, distinctive from the shallow notch in Cricetodon. The capsular process of the lower incisor alveolus is conspicuous, lying below the base of the coronoid process. In dorsal view, the condyloid process is longitudinally long in a teardrop shape with its anterior tip bending slightly medially. The angular process is missing from all the specimens. But judged from the breakage, the angular notch is broad and somewhat oval. The pterygoid fossa for the insertion of the medial pterygoid muscle is deep.
Lower teeth: Nine lower incisors are pre¬ served in mandibles. The lower incisor is delicate compared to the upper incisor. It extends posteriorly under the cheek teeth and terminates posteriorly at the level below the base of the coronoid process. The tip of the lower incisor is slightly lower than the occlusal surface of the cheek teeth ( fig. 5F). The enamel extends only to the labial side, covering about one third of the labial surface. The incisor is oval in cross section, but the anterior and medial sides are flat. The anterior surface is smooth and has no longitudinal ornaments. The wear facet of the incisor tip is long ( fig. 1C).
Fourteen mis are present ( fig. 6). The lower molars are all double-rooted. The ml anteroconid is a high, conical cusp and situated on the longitudinal axis of the tooth. Both the labial and lingual anterolophids are well devel¬ oped and reach the base of the protoconid and metaconid, respectively. The anterolophulid extends from the labial side of the anteroconid and is connected to the anterior arm of the protoconid. The metalophulid and hypolophulid extend slightly anterolabially. The mesolophid is absent in seven, of medium length in four, and long and reaches the lingual edge of the tooth in three specimens. The ectomesolophid is absent except for a weak one present in one specimen. The sinusid is transverse. The long posterolophid closes the posterosinusid.
Sixteen m2s are in the collection. The labial anterolophid is long and connected to the base of the protoconid. The lingual anterolophid is short or absent in 14, except a strong one present in one specimen. The metalophulid and the hypolophulid are directed slightly anterolabially. The mesolophid is of medium length in nine, and short or absent in six specimens. The ectomesolophid is absent. The posterolophid is strong and descends to the base of the entoconid. The wear of the lower m2 from a relatively old individual (IVPP VI5349.24) is illustrated in figure 7A and B. The SEM image shows that the occlusal surface, particularly the enamel ridges, bears numerous pits and some striations. There is no shearing facet on the lingual and posterior sides of the tooth, suggesting that grinding was its primary function during mastication. The orientation of the striations indicates a primarily anteroposterior movement of the lower jaw during mastication, with a minor component of lateral shift.
Fourteen m3s are in the collection. The labial anterolophid is long and descends to join the 10 AMERICAN MUSEUM NOVITATES NO. 3602 base of the protoconid, but the lingual one is short or absent. The mesolophid is of medium length in three, but absent in 10 specimens. The transverse hypolophulid reaches the longitudi¬ nal crest anterior to the hypoconid. The posterolophid is connected to the entoconid and encloses the posterosinusid. The entoconid is reduced. A lingual cingulid almost connects the entoconid and the metaconid.

Baculum:
The baculum has a relatively long shaft with an anchorlike base (IVPP VI 5350.1; fig. 8). On each side of the base is a distally directed spike to which the corpora cavernosum attaches. The shaft is wider than deep and expands slightly distally. In lateral view, the two ends of the baculum bend dorsally. The tip, to which the cartilaginous tissue adheres, is broken.

Humerus:
The humerus is represented by three immature specimens and two fragmen¬ tary distal ends. The proximal end was detached from all the immature specimens at the proximal epiphyseal line, so that the shape and extent of the head, greater tubercle, and lesser tubercle cannot be determined. The remaining length of the humerus measures 10.9 mm from the proximal epiphyseal line to the distal articular surface, and 2.7 mm wide between the lateral and medial edges at the distal end (IVPP V 15350.2).
The humeral shaft is straight, with a wide distal end, large medial epicondyle, and prom- inent brachial crest ( fig. 9). The deltoid crest, which marks the insertion of the deltoid and several related muscles (panniculus carnosus, pectoralis profundus anterior, pectoralis superflcialis, and brachialis), is set on the anterior margin of the shaft and bends slightly laterally, its point of maximum projection approximately 40% of the shaft's length from the preserved proximal end. The crest is triangular and as elevated as that of Cricetodon. The medial ridge is not observed in M. yei and extant cricetines such as Cricetus and Mesocricetus, whereas it is distinct in many humeri of Cricetodon.
The distal epiphysis closely resembles that of Cricetodon and is transversely expanded and anteroposteriorly flattened, with a large distomedially projected medial epicondyle. There is no distinct ridge separating the capitulum from the trochlea. The spindlelike capitulum is broad anteriorly and tapers posteromedially.
The trochlea is shallowly grooved; its articular surface diminishes anteriorly but expands pos¬ teriorly. The medial trochlear rim is sharply angled and projected distally. The radial fossa is wide and well developed. The olecranon fossa is shallow. A well-delineated pit is present be¬ tween the medial epicondyle and trochlea, probably for the attachment of the ulnar collateral ligament.
On the medial aspect of the medial epicon¬ dyle there is a well-developed muscle scar for the flexor muscles of the carpus and digits. The entepicondylar foramen is large and located above the medial epicondyle, as in most living genera of the Cricetinae, but absent in Mesocricetus and Cricetodon. The brachial crest for the insertion of the brachialis, brachioradialis and extensor muscles of the wrist extends proximally 35% of the humeral length. The crest resembles that of Mesocricetus and Cricetus, but is more prominent and flares posterolaterally more than in Cricetodon. The lateral epicondyle is weakly developed and is the distal end of the brachial crest.

Ulna:
The ulna is represented by four proximal and two distal fragments. The olecranon process is moderately developed and deflected slightly anteriorly ( fig. 10A, B), as in Cricetodon. On its proximal aspect there is a distinct groove for the tendon of triceps longus. The trochlear notch is well developed with an anteriorly oriented anconeal process and an anterodistally directed coronoid pro¬ cess. The notch is somewhat less open than in Cricetodon. The triangular radial notch is flat and faces anterolaterally, being transversely narrower than in Cricetodon. Just distal to the coronoid process along the medial aspect of the shaft is a pronounced brachial ridge, upon which the brachialis is inserted.
The diaphysis is slender and mediolaterally compressed; its medial surface is slightly convex and the lateral surface is more deeply grooved than that of Cricetodon. The distal epiphysis is rounded, bearing a faint styloid process ( fig. 10C, D). the distal epiphyseal line and 2.65 mm wide at the distal epiphyseal line. The femur is similar to that of Cricetodon except for having a slightly weaker third trochanter. The head is semispherical and the articular surface does not extend onto the femoral neck; on its medial central articular surface there is a welldefined fovea for the attachment of the ligamentum capitis femoris. The neck is elongate, set at about 130° to the shaft. The greater trochanter rises nearly vertically from the shaft, and is higher than the head; its proximal end exhibits a large muscle scar for the insertion of deep gluteal muscles. The intertrochanteric fossa, the attachment site for obturator muscles, is very deep. The lesser trochanter, the attachment site for the quadratus femoris and iliopsoas, projects medially to the lateral edge of the femoral head. The third trochanter is moderately developed in contrast to the prominent one in Cricetodon. It is triangular and lies opposite and just distal to the lesser trochanter. The midshaft is some¬ what anteroposteriorly flattened and slightly expanded distally. There is little or no anteroposterior curvature along the shaft.
The distal epiphysis is detached from the distal epiphyseal line, indicating that the specimen belongs to an immature individual. The preserved portion of the distal end Seven fragmentary tibiae are in the collection. The proximal epiphysis is not preserved in any specimen. The tibia of IVPP VI5350.22 measures 16.6 mm from the prox¬ imal epiphyseal line to the distal; slightly longer than that of the femur. In lateral view, it exhibits a sigmoid profile: its proximal end is deflected posteriorly whereas its distal end is bowed anteriorly (fig. 12).
The tibial crest appears sharp and extends distally about 30% of the preserved tibial length. It is situated less laterally than in Cricetodon, so that the lateral tibial fossa for the origin of the tibialis anterior is not as deeply excavated as in Cricetodon. The medial aspect of the proximal tibia is relatively flat owing to the absence of any muscle attach¬ ment. Posteriorly on the proximal part of the shaft, both lateral and medial edges are prominent, enfolding a deep posterior tibial fossa for the origin of the tibialis posterior.
The distal tibia is mediolaterally com¬ pressed. The articular surface of the distal end is divided into a lateral and a medial articular facet by a distinct median ridge. The lateral articular facet is broad and faces slightly laterally, as in Cricetodon whereas the medial one faces distally, as in Cricetodon, but is shallower than the latter. The medial malleolus is distinct. The posterior tibial process is large and lies behind the medial articular facet. Between the two processes lie the badly defined sulci for the tendons of muscles tibialis posterior and flexor digitorum longus. On the posterior aspect of the distal epiphysis is a broad and well-defined sulcus, which is probably left by the tendon of muscle flexor tibialis.

Fibula:
Only portion of fused fibula is preserved. Judged from the breakage, starting from a point about two thirds distal along the tibia, the fibula fuses with the tibia over a distance of several millimeters, and then the two diverge again distally to form the medial and lateral malleolus ( fig. 12).

Calcaneus:
A right calcaneus (IVPP VI5350.27) is perfectly preserved (fig. 13). The calcaneus is 4.0 mm long and 2.6 mm wide from the peroneal process to the medial tip of the sustentaculum. The calcaneal tubercle is long and extends distally half the calcaneal length. In dorsal view, it is trans¬ versely compressed with a slightly expanded proximal end. A broad groove for the attachment of the Achilles tendon separates the lateral and more proximally projecting medial margin of the proximal aspect of the tubercle. The calcaneoastragalar facet is con¬ vex and smoothly rounded. Proximolateral to the protuberance that bears the calcaneoas¬ tragalar facet is a distinct tubercle for the calcaneofibular ligament. The sustentacular facet is flat and faces dorsally. The sulcus calcaneus separating the sustentacular facet from the calcaneoastragalar facet is narrow and poorly defined. The sustentaculum talus is prominent and projects medially. The perone¬ al process of M. yei is more proximodistally expanded and displaced more proximally than that of Cricetodon. It is shelflike and almost at 14 AMERICAN MUSEUM NOVITATES NO. 3602 the same level as the sustentaculum talus. The groove for the passage of the muscle peroneus longus is poorly delineated in contrast to the deep one in Cricetodon.
In distal view, the calcaneocuboid facet is gently concave and oriented almost 90° to the long axis of the calcaneus, whereas it is obliquely oriented in Cricetodon. The anterior plantar tubercle is faint, and positioned medial to the calcaneocuboid facet. Between the anterior plantar tubercle and the calca¬ neocuboid facet lies a broad groove for the tendon of the muscle flexor hallucis longus, which extends proximally to the sustentacu¬ lum talus.

COMPARISON
A direct comparison of Megacricetodon yei with Aktaumys dzhungaricus (Kordikova and de Bruijn, 2001) from adjacent Aktau Mountain clearly indicates these two species are closely related and belong to the same genus (see discussion below for generic allocation). They are similar in sharing many features, including the split anterocone of Ml, presence of labial spur of the anterolophule in some Mis, presence of double protolophules in some M2s, long mesoloph in Ml and M2, single anteroconid of ml, and well-developed mesolophid in ml and m2. However, M. yei differs from A. dzhungar- icus by being slightly smaller and having a more clearly bifurcated anterocone, a shorter mesoloph in Ml and M2, fewer occurrences of the labial spur of the anterolophule in Ml, fewer occurrences of double protolophules and single metalophule in M2 , and fewer occurrences of the ectomesolophid in ml.
Other Megacricetodon species from China include M. sinensis from Chetougou Formation of Qinghai Province (Qiu et al., 1981) and Tunggur Formation of Inner Mongolia (Qiu, 1996), M. pusillus from Tunggur Formation of Inner Mongolia (Qiu, 1996), and M. aff. collongensis from Xiacaowan Formation of Jiangsu (Li et al.,1983). M. yei is larger than all of these species. In addition, M. yei is distinct from them in having a longer mesoloph in Ml and M2, a labial spur of anterolophule in Ml, a ga, groove for the Achilles tendon; gfhl, groove for the flexor hallucis longus; gpl, groove for the peroneus longus; pp, peroneal process; sc, sulcus calcaneus; su, sustentacular facet; sust, sustentaculum talus; tub, calcaneal tuber.  well-developed posterior spur of paracone in Ml, double protalophules in M2, a longer mesolophid in ml and m2, and an elongate m3. Megacricetodon yei is comparable in size to several European species, such as M. similis Fahlbusch, 1964, M. rafaeli Daams and Freudenthal, 1988, and M. lopezae Garcia Moreno, 1986. M. yei differs from M. similis by the deeply split anterocone of Ml, the higher frequency of posteriorly directed metalophule in M2, the simple ml anteroconid, and longer mesolophids on lower cheek teeth. M. yei differs from M. rafaeli by possessing the deeply split anterocone, the presence of labial spur in some Mis, and the simple anteroconid of ml. M. yei differs from M. lopezae in having a longer mesoloph of the upper molars, the double protolophules of M2, a simple anteroconid of ml, and a longer mesolophid in ml-2.
Megacricetodon andrewsi, from Pa §alar, Turkey (Pelaez-Campomanes and Daams, 2002), differs from M. yei in having an incompletely bilobed anterocone on Ml and bilobed anteroconid on ml as well as in lacking the labial spur of the anterolophule and the anteriorly or transversely directed metalophule of M2.

DISCUSSION
Validity of the genus Aktaumys: The ge¬ nus Aktaumys was named by Kordikova and de Bruijn (2001) based on isolated teeth recovered from the Aktau Mountains, Southeastern Kazakhstan. The genus contains only the type species, Aktaumys dzhungaricus. According to Kordikova and de Bruijn (2001), Aktaumys was differentiated from Megacricetodon in having the following characters: (1) long, transverse mesolophs(ids) in the first and second upper and lower molars; (2) a long labial spur of the anterolophule in most Ml; (3) double proto¬ lophules in all M2 (in the material that we were able to study, courtesy of Dr. Hans de Bruijn, most, but not all, M2s have this feature) and double metalophules in most M2s; and (4) a shorter, more triangular, occlusal surface in the ml. The authors suggested that the main differences precluding Kazakhstan materials from being allocated to Megacricetodon are double protolophules and metalophules in most M2s (Kordikova and de Bruijn, 2001). We observed subtle differences between M. yei and Aktaumys in the overall dental morphology and their geographical occurrence is relatively close, about 150 miles apart. The morphological similarities and geographical vicinity indicate that these two species should be placed into the same genus. In M. yei, double protolophules of M2 are present in 43% (3 out of 7) of the specimens, and the anterolingually directed single protolophule occurs in 57% (4 out of 7) of the specimens. The presence of double protolophules of M2 also varies greatly within Megacricetodon. In some species, such as M. andrewsi, M. ibericus, and M. crusafonti, double protolophules are uncommon (0%-10%), but can be as high as up to 63% in others, such as M. minor (Daams and Freudenthal, 1988). Because the variation of double protolophules is common in Megacricetodon, this character does not distinguish Aktaumys from Megacrice todon.
The same pattern is also true for the double metalophules of M2. In Aktaumys dzhungar¬ icus, double metalophules are present in many M2s but the exact percentage has not been available. In M. yei only a simple metalophule is present in the collection. In other species of Megacricetodon, however, there is a consider¬ able variation in this character. Most species have a transverse or posteriorly projecting, simple metalophule, whereas in M. collongensis, for instance, 30% of the M2s have the double metalophules. The available evidence suggests that the double metalophules are variable within the species of Megacricetodon. Because none of the morphological differences listed above is unique for Aktaumys, we consider that the generic name Aktaumys a junior synonym of Megacrice todon.

dae:
The suprageneric classification of the Megacricetodon has long been controversial. Mein and Freudenthal (1971a) placed the genus Megacricetodon in the subfamily Cricetodontinae with the tribe Megacricetodontini. Rieg (1972) considered that Megacricetodon has a closer affinity with the subfamily Cricetinae rather than with Cricetodontinae on the basis of dental similarities. Although this idea has been shared by various authors (Fahlbusch, 1996;Kalin, 1999), it has never been universally accepted. McKenna and Bell (1997), for instance, placed Megacricetodon in its own subfamily, Megacricetodontinae. Mein and Freudenthafs (1971a) classification relies on only a few cranial and postcranial characters, frequently surveyed from just one or a few species. The cranial character of the genus Megacricetodon, however, is based only on undescribed materials in the collection of the Museum of Natural Science in Lyon; no interspecific variation has been docu¬ mented in publication. Uncertainties about the relationships of the subfamily Cricetinae and Cricetodontinae primarily resulted from the use of some characters, such as the incisive foramina and the entepicondylar foramen of the humerus, that have not been fully known within relevant species.
It is known that the subfamily Cricetodon¬ tinae has the incisive foramina penetrating the palate beyond the anterior borders of the first molars and that the subfamily Cricetinae is characterized by short incisive foramina, which end before the anterior border of the first molar. The incisive foramina of M. yei extend further back, to the protocone of Ml, as reported by Mein and Freudenthal (1971a), corroborating their conclusion that Megacricetodon would be placed in the subfamily Cricetodontinae. However, Reig (1972) suggested that the relative development of the incisive foramina is variable at the generic level and is not useful above the generic level. Mein and Freudenthal (1971a) used the entepicondylar foramen of the humerus as a diagnostic character to distinguish the subfam¬ ily Cricetodontinae from the Cricetinae. They pointed out that in the Cricetodontinae the entepicondylar foramen is absent, whereas all genera of Cricetinae have this foramen in the humerus. The entepicondylar foramen, howev¬ er, is variable within the genus Megacricetodon. Unlike the undescribed specimens in the collec¬ tion of Lyon examined by Mein and Freudenthal (1971a), the large entepicondylar foramen occurs in all humeri of M. yei. Therefore this character does not appear to be particularly useful for distinguishing species of Megacricetodon.
Based on the cranial and postcranial char¬ acters, Mein and Freudenthal (1971a) inter¬ preted the dental similarities between the Cricetodontinae and the Cricetinae as conver¬ gence. However, one can argue equally well that the available cranial and postcranial features result from parallel evolution. While considering the craniomandibular, dental, and postcranial features as a whole, we think Megacricetodon is better placed within the subfamily Cricetinae, as suggested by Reig (1972). Megacricetodon lacks the characteristic features of Cricetodon, such as the longitudi¬ nal ridges on the anterior surface of the lower incisor. In contrast, the enamel surface of Megacricetodon is smooth and resembles those of Democricetodon and living genera of the Cricetinae. The mandible of Megacricetodon is also similar to those of Democricetodon and living genera of the Cricetinae, and differs from that of Cricetodon. The coronoid process is very pronounced and its tip is obviously higher than the condyloid process as in recent cricetines, whereas in Cricetodon the coronoid process is minute and its tip is about the same level or slightly lower than the condyloid process. The sigmoid notch is deep in contrast to the shallow notch in Cricetodon.

Species of Megacricetodon:
The species assigned to the genus Megacricetodon have been unstable historically, making their taxonomy and phylogenetic reconstruction complicated. Future research may reveal many synonyms of named taxa, as pointed out by Kalin (1996). In Europe, 20 species have been recognized (Kalin, 1996), of which the oldest representative is M. primitivus from the European Neogene land mammal zone MN4 (Freudenthal, 1963) (Daams and Freudenthal, 1988). Major evolu¬ tionary trends seen in European species include size increase, bifurcation of the ml anterconid, and reduction of the mesoloph(id). Although most European species could be fit into these lines as various evolutionary stages, some, such as M. rafaeli, may have evolved independently. In Turkey, M. andrewsi is from Pa §alar (MN6) (Pelaez-Campomanes and Daams, 2002 M. sinensis was originally described from Chetougou Formation of Qinghai, China (Li and Qiu, 1980;Qiu, Li, and Wang, 1981) and also occurs in Tunggur (MN8), Inner Mongolia (Qiu, 1996). The second species from Tunggur is M. pusillus, which is very similar to M. sinensis in size and morphology and may prove to be a synonym of M. sinensis. Based on their sizes and ages, one may entertain the idea that M. sinensis and M. pusillus could have derived from the M. dzhungaricus-M. yei lineage. However, the bifurcated anterocone, the presence of the posterior spur of the paracone, the protolophule of M2, and the shape of ml suggest that M. dzhungaricus and M. yei represent a distinct lineage from other Chinese species of the genus.
The evolution of the Megacricetodon is complex and displays a mosaic pattern. Although evolutionary trends of European species have been widely used in regional biostratigraphical correlation, Wessels et al. (2001) cautioned that the correlations might apply only to local sequences. The new materials we present here indicate that Asian lineages evolved independent of the European species, echoing the caution of Wessels et al. (2001).

BIOCHRONOLOGY
The Halamagai Formation contains more than five fossil beds, but the fossil mammals were recovered mainly from the two lower beds. Ye et al. (2001a, b) published the most updated faunal list for fossils collected from these beds and concluded that the faunas from different beds are all similar in composition, and cannot be differentiated in age; it was therefore assumed that fossils collected from these beds within the Halamagai Formation represent one fauna.
Preliminary study has identified the following taxa associated with M. yei from the Halamagai Formation of the  (Qiu et al., 1999;Pickford et al., 2000). Representatives of Megacricetodon have been considered as a good biostratigraphic markers in local sequence (Wessels et al., 2001). Geographical occurrence of M. yei and Megacricetodon (= Aktaumys) dzhungaricus is so close so that these two species bear significantly on the biostratigraphy of the region. M. yei is more advanced than M. dzhungaricus from the Chul'adyr Formation of Aktau Mountains (MN4), and is therefore indicative of a later appearance, probably MN5/6. This concurs with the estimation that the Halamagai fauna is early Middle Miocene age, roughly MN 6 equivalent.

PALEOBIOLOGICAL RECONSTRUCTION
Modern species of the family Cricetidae are primarily terrestrial in habits; most forms either scamper, jump, or burrow, but a number of species are semiarboreal or semiaquatic.
Similar to Cricetodon from Tieersihabahe, M. yei exhibits a number of postcranial features observed in recent terres¬ trial cricetines. As noted in the description, the deltoid crest of the humerus, which serves for the insertion of the pectoralis and deltoid muscle, is pronounced in M. yei and reflects powerful shoulder flexion in parasagittal movements. The spindle-shaped capitulum, the posteriorly concaved trochlea with prom¬ inent lateral and medial edges, and the continuation of the capitulum and the trochlea restrict movements more to the parasagittal plane during terrestrial quadrupedal walking and running in M. yei (Jenkins, 1973;Harrison, 1989); these are characters, common to extant cricetids, that promote stabilities at the elbow joint by restricting radial rotation while allowing flexion and extension in the parasagittal plane.
The medial epicondyle is the site of origin for the wrist and digital flexors (e.g., flexor digitorum profundus muscle) (Sargis, 2002a). Hence, a strong medial epicondyle may represent a powerful flexion of the digits during grasping of branches, or digging (Taylor, 1974;Hildebrand, 1985;Sargis, 2002a). As in the living cricetines, such as Mesocricetus and Microtus (good runners, climbers, and diggers), the medial epicondyle is well developed in M. yei, which probably indicates a mixture of climbing and digging abilities. The brachial crest, which serves as the insertion sites for the brachialis, brachioradialis, and extensor muscles, is distinct in M. yei and stronger than that in Cricetodon, indicating more sophisticated climbing and digging abilities.
The flat and anterolaterally oriented radial notch is indicative of a limited range of the pronation and supination at the elbow as in ground-dwelling mammals. The radial notch in M. yei is transversely narrower than in Cricetodon, suggesting more capacity for supination.
On the femora, a broad muscle scar of the greater trochanter and a laterally expanded third trochanter in M. yei reflect a welldeveloped gluteus muscle mass, which pro¬ vides the powerful extension and abduction of the thigh for propulsion during terrestrial running. The more proximal projection of the greater trochanter of M. yei is another indicator of terrestrial habits of the animal. It restricts the mobility of the hip joint and particularly limits the range of abduction. This, in turn, makes the parasagittal hind-limb movements of terrestrial locomotion more efficient (Sargis, 2002b).
The prominent tibial crest for the patellar ligament of the quadriceps femoris indicates that muscles acting to extend knee joint are well developed. The emphasis on extension, reconstructed for M. yei, is typical of runners (Muizon, 1988;Szalay and Sargis, 2001). On the distal end of the tibia, the posterior tibial process of M. yei is relatively long. This is clearly related to efficiency in terrestrial locomotion because a relatively long posterior tibial process restricts mobility at the upper ankle joint, so that only parasagittal move¬ ments are possible at this joint (Sargis, 2002b).
In conclusion, M. yei was basically a terrestrial rodent with a propensity for climb¬ ing or digging. Compared to Cricetodon from Tieersihabahe, M. yei exhibits more speciali¬ zations toward climbing, such as stronger brachial crest and transversely narrower radial notch.