Chapter 4 The Antiquity of Rhizomys and Independent Acquisition of Fossorial Traits in Subterranean Muroids

LAWRENCE J. FLYNN1

ABSTRACT

In parallel with the growing body of molecular data bearing on the relationships of muroids, particularly subterranean lineages, the relevant fossil record has improved to the point that its data constrain scenarios of evolution about both the timing and mode of evolution of burrowing muroids, especially bamboo rats, blind mole rats, and zokors. Morphologists have considered these groups phylogenetically distinct from each other, but the three lineages appear to be related as a monophyletic Family Spalacidae, sister taxon to all other living muroids, based on both nuclear and mitochondrial genes. Although living genera are fully subterranean, the fossil record shows that the three groups evolved burrowing characteristics independently. Bamboo rats (Rhizomyinae) have the longest fossil record, extending into the Late Oligocene, but do not show fossorial traits until the Late Miocene. Blind mole rats (Spalacinae) have a fossil record nearly that long, and its early members also lack burrowing traits. Zokors (Myospalacinae) show characteristics considered derived relative to other groups, and have a shorter fossil record. The fossil record of the Tribe Rhizomyini, living Asian bamboo rats, extends to about 10 million years ago, with early species distinct at the generic level from living Rhizomys. The oldest well- known species assignable to an extant genus is Rhizomys (Brachyrhizomys) shansius from the early Pliocene of Yushe Basin, China, north of the geographic range of modern Rhizomys.A hypothesis of close relationship of bamboo rats, blind mole rats, and zokors leads to a reevaluation of affinities of certain Asian fossil taxa and reevaluation of polarity of some features, but molecular data are not yet robust enough to clarify interrelationships of the groups. Morphological and fossil data suggest that myospalacines are more closely related to rhizomyines than to spalacines, and that known Early Miocene rhizomyines are close to the stem zokor morphotype.

INTRODUCTION like North American gophers and even ‘‘The panoply of adaptations necessary for superficially like lipotyphlan moles. Nevo living underground in tubular burrows has (1999) explored this very subject in his book dramatically defined the blind mole rat on subterranean mammals. Since phenotypic phenotype, which facilitates their diagnosis resemblance of rodent burrowers is not an but obscures phylogenetic connections to especially strong argument for close relation- other muroid rodents.’’ This wonderful state- ship, fossorial lineages have been classified ment from Musser and Carleton (2005) could separately, often at the family level. Ironical- be applied as readily to bamboo rats as to ly, recent molecular work suggests that Spalax, the blind mole rat. Truly subterra- several different burrowers constitute, in fact, nean small mammals are obliged to spend at a monophyletic group. least 95% of the time underground. Given My personal fascination with bamboo rats this condition, adaptive constraints are fo- and relatives began with research on the cused such that phylogenetically remote taxa Siwalik fossil record of the Indian subconti- converge in body form, limb proportions, nent, which contains the lion’s share of what and skull structure. Old World burrowers is known of the history of the group. The like bamboo rats and blind mole rats look Late Tertiary basins of China, in which I

1 Peabody Museum of Archaeology and Ethnology, Harvard University, Cambridge, MA 02138 ([email protected]).

128 2009 FLYNN: ANTIQUITY OF RHIZOMYS 129 have had the pleasure to work with col- the family-level category to recognize these leagues in the field, also produce bamboo subdivisions, others, including Carleton and rats, as well as the endemic fossorial muroids Musser (1984) and McKenna and Bell (1997), known as zokors. It is the fossil record of have elected to give them lower rank China that provides knowledge delimiting the (subfamily), pending more data to clarify origin of the bamboo rat Rhizomys, and interrelationships. Such data are emerging includes remains of other enigmatic muroid from molecular as well as anatomical studies. precursors. These tend to endorse the higher categories This study begins with description of the that are evidenced by morphology (15 oldest fossil skull and postcrania of the genus subfamilies according to Carleton and Mus- Rhizomys, which derives from Pliocene Ep- ser, 1984), and to define clusters of them, och rocks of Yushe Basin, Shanxi Province, which correspond to family-level taxa, fewer People’s Republic of China, and is housed in in number. Current molecular studies (Step- the American Museum of Natural History, pan et al., 2004; Jansa and Weksler, 2004) are New York. It demonstrates the generic beginning to survey Muroidea broadly, but identity of other less complete remains of individual efforts still sometimes miss key bamboo rats from the fossil record of China, taxa. The integrative analysis of Musser and and sheds light on the systematics of older Carleton (2005) draws on molecular and bamboo rats. This analysis leads to a morphological data and recognizes six fam- discussion of the phylogenetic relationships ilies within Muroidea: Platacanthomyidae of bamboo rats and the relevant fossil record, (usually not treated in molecular work), given that recent molecular work argues for Calomyscidae, Nesomyidae, Cricetidae, Mu- strong affiliation with Spalax and with the ridae, and Spalacidae. Steppan et al. (2004) diverse but local zokors of eastern Asia. and Jansa and Weksler (2004) show Spalaci- Previously, most systematists had con- dae to be the sister taxon to all other extant sidered the common origin of bamboo rats, Muroidea. Spalax, and zokors to be remote in time and The family-level taxon Spalacidae has likely not monophyletic. Finally, enigmatic come to be seen as a basal survivor of the elements in the fossil record of China are fabulous radiation leading to crown-group placed in the context of the evolutionary muroids. Spalacidae would be an ancient history of higher muroids, given a hypothesis branch of living Muroidea. Other more of close relationship of these burrowing primitive muroids, logically treated as inde- groups. pendent families, are all extinct. Hence the Eocene and early Oligocene radiations of muroids should be recognized at the family MUROID SYSTEMATICS rank (e.g., Eumyidae, Pseudocricetodontidae, Paracricetodontidae, Eucricetodontidae, Ta- Musser and Carleton (2005) give a thor- chyoryctoididae; see U¨ nay-Bayraktar, 1989), ough accounting of the history of muroid and not lumped in Cricetidae. Coming from classification. For the purposes of this paper, the molecular work, however, is the some- it is sufficient to note that the monophyly of what surprising linkage of Spalax and Muroidea is not a question, nor is (for the Rhizomys + Cannomys + Tachyoryctes with moment) the close relationship of South the fossorial zokor Myospalax. Musser and Asian bamboo rats (Rhizomys and Can- Carleton (2005) illustrate the colorful history nomys) and African mole rats (Tachyoryctes). of classification of these genera and show Relevant to the issues explored herein, that they have been associated in the past, however, is the family-level division of but not usually together to the exclusion of Muroidea and implied close relationships all other muroids. S¸en (1977) did suspect that within those families. these genera belonged together in Family Subdivisions of Muroidea have been evi- Spalacidae. This is an exciting and testable dent for many years, and these are treated hypothesis of relationship that makes review extensively by Musser and Carleton (2005). of the fossil record of these groups relevant. I While many authors have preferred to utilize begin with the bamboo rats. 130 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 331

ABBREVIATIONS olingual reentrant on m3, reduced incisive foramina, and a host of fossorial features (chisel incisor, strong coronoid process on AMNH American Museum of Natural deepened mandible, modified humerus with History, New York broad epicondyles and strong deltoid crest, F:AM Frick mammal fossil collection broad skull with flaring zygoma and lamb- of the American Museum of doid crest). The genera Rhizomys and Can- Natural History nomys are distinguished from earlier taxa by Fm. formation the apomorphic high, round infraorbital GSP Geological Survey of Pakistan foramen lacking a ventral slit, and the latter MCZ Museum of Comparative Zoolo- is derived in its short M3 and m2, reduced gy, Harvard University, Cam- posterolophid on m2, and three crests on M2 bridge, MA and m2. Rhizomys has a deep dentary and its THP Tianjin Huang Pei, fossil collec- masseteric crest terminates below the front of tions of the Tianjin Museum of m1 or beyond, without an anterior extension, Natural History, P.R.C. four-crested M1–2 (strong mesoloph), meso- lophid indistinct on m3. The subgenus COMPARISONS: Among a large suite of Brachyrhizomys is characterized by features specimens in the collections of the Vertebrate that are likely primitive: lower crowned than Zoology Division of the AMNH and in the living Rhizomys, m2 longer than wide until MammalDepartmentofMCZ,usefulcompar- late wear, posterolophid on m3 incompletely isons were made among Rhizomys pruinosus isolated, masseteric crest not extending past pruinosus (AMNH 113477), Rhizomys pruino- anterior root of m1, lower diastema not sus senex (MCZ 35935, 35936), Rhizomys shortened. Brachyrhizomys appears to be sinensis vestitutus (MCZ 7559, 7560), Rhiz- derived in proodonty, but shares this trait omys sinensis wardi (AMNH 115527, 115528), with Cannomys. Rhizomys sumatrensis (MCZ 61049), and Cannomys badius (AMNH 163644, 163645). Rhizomys (Brachyrhizomys) shansius The AMNH Division of Paleontology also Teilhard de Chardin, 1942 houses several series of elements attributed to HOLOTYPE: IVPP 31.096, right dentary fossil Rhizomys troglodytes, for example, with incisor, m1–3; coronoid process, condyle AMNH 18407 and 18409. Also examined and angle of the mandible missing. were Eospalax fontanierii (MCZ 19896), REFERRED MATERIAL: THP 14.183, right ‘‘Myospalax cf. cansus’’ (AMNH 135358, dentary fragment with incisor, m2–3; F:AM probably Eospalax fontanierii), Spalax mi- 117337, cranium and mandible with complete crophthalmus (AMNH 228), Spalax ehrenbergi dentition, atlas + axis, partial rib cage and (MCZ 61376) and Tachyoryctes sp. (AMNH clavicle fragment, right scapula, left and right 54305, 187461). humerus and ulna, right distal femur frag- ment, and tibia diaphysis. SYSTEMATICS LOCALITIES AND AGE: All specimens from Pliocene deposits of Yushe Basin, Shanxi Tribe Rhizomyini Winge, 1887 Province, China. North of the Yellow River, Genus Rhizomys Gray, 1831 in central Shanxi, this basin is north of the Subgenus Brachyrhizomys Teilhard de range of Rhizomys at present. The holotype is Chardin, 1942 of imprecise provenance; THP 14.183 is from Baihaicun, upper Gaozhuang Fm. or lower DIAGNOSTIC FEATURES: The Tribe Rhi- Mazegou Fm.; F:AM 117337 is from Zhaoz- zomyini is equivalent to Flynn’s (1982, 1990) huang, Mazegou Fm., age about 3.5 million subfamily Rhizomyinae, diagnosed by de- years ago (megaannum, Ma). The age range rived characters including reduction of the for known fossils, at most 4.5 – 3 Ma, is quite ventral slit of the infraorbital foramen, ridge possibly shorter, ,4–3.3 Ma (time scale of absent on lower incisor enamel, small poster- Cande and Kent, 1995). 2009 FLYNN: ANTIQUITY OF RHIZOMYS 131

Fig. 1. Cranium of Rhizomys (Brachyrhizomys) shansius (F:AM 117337), in dorsal (A), ventral (B), right lateral (C), and anterior (D) views. Scale bars equal 1 cm.

SUPPLEMENTED DIAGNOSIS: Brachyrhiz- skull of living species; the occiput is clearly omys about the size of smaller individuals wider than high, and less sloping. F:AM of Rhizomys sinensis or R. pruinosus but with 117337 represents a young individual, with lower crowned molars; skull with proodont last molar still erupting and not yet showing incisors, weak lambdoid crest, gently convex attrition (fig. 1). occiput nearly vertical and not expanded The tips of the incisors, nasals, and bits of dorsoventrally; lingual reentrant on M1 short the premaxillas are missing due to breakage. and not convergent with middle buccal The incisors were proodont unlike living reentrant, m2 not wider than long and with Rhizomys, in which they recurve and are persistent anterolingual enamel lake, mandi- directed ventrally. The incisors appear to be ble with diastema not shortened. rooted above M1. The snout (fig. 2) is broad DESCRIPTION OF CRANIUM: The skull of posteriorly, and narrows steadily anteriorly; R. shansius is very much like that of living conversely the nasals broaden anteriorly. The Rhizomys, showing the overprint of adapta- narrow posterior limits of the nasals are tions for a fossorial lifestyle. It is compact, adjacent to the posterior borders of the stoutly built, with heavy rostrum and broad premaxillae. Together, these make a jagged braincase. There is a narrow interorbital intersection with the frontals opposite the constriction and the occiput supports heavy posterior rim of the high infraorbital fora- musculature. The upper diastema rises ante- men. The dorsal aspect of the infraorbital riorly toward the heavy incisors. Tooth rows foramen is ringed by the maxilla. The are set close, with small incisive foramina foramen transmits the deep portion of the several mm anterior to them. Significantly masseter, whose scar extends slightly beyond different, however, is that the skull, especially the foramen onto the premaxilla. Anterior to the occiput, is lower than that of the high the infraorbital foramen, the premaxilla- 132 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 331

Fig. 2. Line drawing of the skull of Rhizomys (Brachyrhizomys) shansius (F:AM 117337), in ventral (above) and dorsal (below) views; 1 cm scale. View is slightly angled to show features of the orbit (deep fissure from ventral view; some of the orbitosphenoid from dorsal view, on the side of the orbit, above the maxilla floor). Abbreviations: al (alisphenoid), ab (auditory bulla), bo (basioccipital), bs (basisphenoid), f (frontal), i (interparietal), j (jugal), ma (mastoid), mx (maxilla), n (nasal), or (orbitosphenoid), p (parietal), pl (palatine), pm (premaxilla), pt (pterygoid), sq (squamosal). 2009 FLYNN: ANTIQUITY OF RHIZOMYS 133 maxilla suture drops vertically to intersect the convex, not flattened or surmounted by a middle of the incisive foramen. The muscle distinct flange of bone. From the occipital scar of the lateral masseter clearly extends condyles, barely visible in dorsal view, the anteriorly on the side of the rostrum beyond occiput rises vertically and curves gently this suture. Anterior to the scar is a shallow rostrally. The dorsal view of the strongly depression in the premaxillary bone, posteri- inclined occiput of living Rhizomys reveals or to the incisors. Because the origin of the the condyles prominently. lateral masseter is highly developed, it In posterior view the braincase is much suppresses completely the zygomatic plate, wider than high. Mastoids are prominent. fusing it onto the side of the snout. The Dorsolateral to them, a wing of the squamo- infraorbital foramen is high and rounded, sal extends above and medial to the bony fully modern, except that it is proportionately auditory tube (broken on the left side). The larger than in living species. The anterior root foramen magnum is large relative to modern of the zygoma of F:AM 117337 is broken on Rhizomys. Bullae are large and heavy, with a both sides and the lacrimal bone is not short bony auditory tube present. The bony preserved. The lacrimal may contribute to auditory tube terminates at the level of the the dorsal aspect of the anterior root of the posterior root of the zygoma, and does not zygoma in living species but, in any case, does extend dorsolaterally above the zygoma as in not touch the premaxilla. The heavy left living Rhizomys. Paroccipital processes, al- zygomatic arch flares broadly laterally. It was though broken, probably were not as big as broken during preparation, but the small in living species. In ventral view, the basioc- jugal (fig. 2) was located and replaced. The cipital extends to the mouth of the eustachian arch is not as deep dorsoventrally as in living canal of the bulla. Wings of the pterygoids species, and the contribution to it by the jugal are broken, and most of the basisphenoid is bone is smaller. missing (fig. 2). The basicranium is tele- In dorsal aspect, the interorbital region of scoped, shortened as in living Rhizomys. the skull is constricted, narrower than the The telescoping places the bullae adjacent rostrum, rounded around the braincase. to the pterygoid vacuities. The deep ptery- Scars for the temporalis muscle begin at the goid vacuities and mesopterygoid fossa ac- anterior margin of each orbit and converge commodate the pterygoid musculature that posteriorly but do not meet. Temporal ridges counters the strong temporalis. The vacuities can be traced caudally to the interparietal retain some matrix, obscuring details of the and nearly to the occiput before they turn foramen ovale, the middle lacerate foramen, laterally. Both the degree of interorbital and the alisphenoid canal. constriction and convergence of scars as The rostrum and palate are strongly built. parasagittal crests are a function of size and Penetrating the bone near the midline and individual age. The frontal bones contribute posterior to each incisor are small premaxil- to the anterior portion of the braincase and lary foramina. The rostrum is constricted on terminate somewhat posterior to the interor- each side by the heavy masseteric muscula- bital constriction. Bones of the cranial vault ture, which builds a flange of bone along the are thick with sutures reinforced by overlap- posterior half of the diastema. These ridges ping laminae of bone, particularly between flank short, closely paired incisive foramina, parietal, interparietal, supraoccipital, and whose posterior ends are 4.4 mm anterior to squamosal. Hence, tracing sutures is compli- the alveolus rim for M1. Molar rows are set cated by minor spalling of laminar bone from close and diverge somewhat posteriorly. A the fossil. The parietals roof most of the ridge down the center of the palate is braincase; the occipital plate does not en- bounded on each side by a narrow trough. croach upon them from behind. The large The palatine bone tightly rounds the maxil- interparietal is roughly triangular, not sub- lary alveolus for M3 and contributes to the rectangular as in many muroids, and it is posterior margin of the hard palate. It large for Rhizomys, not overlain posteriorly terminates with a V-shape at the maxilla, by the supraoccipital as it is nearly complete- with rostral apex and opposite the boundary ly so in living species. The occipital is gently between M2 and M3. Posterior palatine 134 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 331 foramina are weakly defined, apparently the dorsal aspect (fig. 2). These bones are located at the border with the maxilla. relatively larger than in living Rhizomys, Because the hard palate extends caudally which has a relatively more vertically ex- only to the middle of M3 in this young panded maxilla. Correlated to the more individual, the palatine contribution to the vertically expanded maxilla, the orbit is less hard palate is minor. The palate ends about deeply excavated above the molar roots in 1.7 mm anterior to the posterior rim of the living species. alveolus for M3. The otherwise concave MANDIBLE: The unfused mandible has a posterior margin of the palate has a midline large, rugose contact surface between den- knob. Laterally, the maxilla extends as a taries that brings the chisel-shaped incisors process behind each M3 and is continuous close together. Molar occlusal surfaces are with the pterygoid plate. tilted inward. The diastema (10.3 mm for the One’s first impression of the lateral view of left dentary) is neither as shortened nor as F:AM 117337 is that the skull is deep concave dorsally as in living species. Jaw dorsoventrally, a characteristic of Rhizomys. depth is considerable: measured externally, The skull roof is arched, highest above the 14 mm below m1, and 13.5 mm below m2. In back of the orbit. Deepening is accomplished the young individual represented by F:AM mainly by dorsoventral expansion of the 117337 (fig. 3), m3 is erupting, which makes maxilla, which accommodates the high- the measurement of 14.2 mm an underesti- crowned molars. The orbit and its structures mate for the length of the tooth row. (The are extended dorsoventrally, with simulta- distal wall of m3 slopes ventrally and neously increased basicranial flexion. The posteriorly, which makes measurements of expanded maxilla is consistent with the deep the occlusal surface dependent on wear rostrum, which rises steeply toward the stage). The ventral border of the mature, incisors. The anterior root of the zygoma is and larger, THP 14.183 is damaged so that anterior to M1. The posterior root begins the depth of 15.3 mm below the back of m1 above the back of M3, is expansive antero- in this specimen is an underestimate for the posteriorly, and rises gently rearward. It ends original; its tooth row length at the alveoli is at the posterolateral extent of the temporal 14.75 mm. The dentary has a high condyle ridge, which passes above the auditory tube (10.5 mm above the m2 alveolus in F:AM of the bulla. This tube is more elongate and 117337) and strong coronoid process, per- extends dorsally above the level of the haps stronger and more vertical than in living zygoma in living Rhizomys. The glenoid fossa species. A prominent coronoid correlates is wide and elongate. The heavy posterior with strong temporalis muscles which stabi- root of the zygoma is supported by a large lize the jaw as the head is lifted during chisel- squamosal bone. Relative to the squamosal, tooth burrow excavation. The incisor is the small alisphenoid is squeezed with the rooted in a capsule lateral to the condyle. maxilla and contributes to the back of the The capsule of this young bamboo rat is orbit. Posteriorly, the alisphenoid shows a smaller and does not extend as far posteriorly notch, which may represent the anterior as in observed living Rhizomys.(Cannomys opening of the alisphenoid canal. Anterior exhibits a condition in which the extended to this are two foramina, apparently equiv- capsule rises above and dominates the alent to the buccinator and masticatory adjacent condyle.) The strong upper and foramina of geomyids (Wahlert, 1985), which lower masseteric crests meet below the front also have a deep skull. The floor of the orbit, of m1 as a rounded scar without anterior maxillary bone, is excavated deeply into the extension. Anterior to it is the mental skull and includes pits for the sphenoidal foramen, above the midline of the dentary, (anterior alar) fissure and optic foramen. A at the back of the diastema. The ascending small foramen above M1 in the maxilla may ramus is broad anteroposteriorly, and rises be the sphenopalatine. The small orbitosphe- opposite m2, close to the tooth row and noid is located anteriorly at the midlevel of without a defined pocket next to the teeth. the orbit, and deep within it. The orbitosphe- The mandibular foramen is above the distal noid and alisphenoid are barely in view from part of m3, not well posterior to it as in living 2009 FLYNN: ANTIQUITY OF RHIZOMYS 135

Fig. 3. Internal (above) and external views of the left dentary, and dorsal view of the right dentary (below) of Rhizomys (Brachyrhizomys) shansius (F:AM 117337); scale bars equal 1 cm. 136 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 331 species. The angular process is more cau- TABLE 1 dally extended than in living Rhizomys, Cranial and Dental Measurements (mm)a of F:AM beyond the limit of the condyle and incisor 117337, Rhizomys (Brachyrhizomys) shansius capsule. Occipitonasal lengthb 55 DENTITION: The stout upper incisor is Zygomatic breadth 46 broad mediolaterally (3.45 mm) in relation to Interorbital breadth 9.7 its parasagittal length (4.25 mm). Its enamel Length of rostrumb 21 cap is gently rounded and wraps about three- Breadth of rostrum 15.6 tenths of the way onto the lateral side of the Breadth of braincase 24.9 incisor. There is no major ornamentation— Height of braincase 20.0 only gentle rugosities on the enamel. The Length of diastema 17.9 lower incisor (3.91 3 4.35 mm) is heavier, Length of incisive foramina 4.8 with the enamel cap somewhat less rounded Breadth of incisive foramina 2.5 Length of bony palate 17.6 than that of the upper incisor. Enamel wraps Breadth across bony palate at M1 3.1 just over a quarter of the way onto the lateral Postpalatal length 20.2 side of the incisor. Only minor rugosities Breadth of mesopterygoid fossa 6.5 ornament the enamel. Although the enamel is Length of bulla 10.7 considerably darkened, there are mottled Crown length of M1-M3c 12.2 traces of original orange pigmentation. The Breadth of M1, M2, M3c 3.9, 3.8, 3.2 more mature individual THP 14.183 has a Crown length of m1-m3c 14.2 c larger lower incisor, 4.55 mm wide. Breadth of m1, m2, m3 3.5, 4.4, 4.4 Molar dimensions (length 3 width) are a After Musser et al. (2005: fig. 3). cited here as maxima, including expanded b Underestimated due to breakage. bases of crowns (not occlusal wear surface; c Molar dimensions are at the occlusal surface, and are hence, measurements are greater than in therefore minimum measurements for these little-worn table 1). Measurements draw on information teeth (the occlusal wear surface is larger in later wear from the left and right sides of the skull stages). (differentially preserved and exposed). M1 (6.0 3 4.4 mm) is noticeably longer than wide and moderately high crowned (4 mm crown reentrant is confluent in early wear with both height visible lingually, less than its length or the anterior and middle buccal enamel lakes. width). The tooth is higher crowned lingually Lower molars are lower crowned, with less than buccally and appears to have four roots, noticeable unilateral height (higher crowned one anterolingual root well developed and buccally). The first molar (5.2 3 3.95 mm) is directed anteriorly for support. The rounded well worn although m3 is not fully erupted, occlusal outline has one lingual reentrant, and shows primarily five transverse lophs which extends nearly to the base of the (anterolophid, metalophid, mesolophid, hy- crown. In occlusal view, the reentrant is short polophid, and posterolophid). The mesolo- and directed anteriorly toward the first of phid is a long lingual continuation of the three buccal transverse enamel lakes. Conse- protoconid. A large enamel lake separates the quently, the tooth exhibits four lophs: ante- mesolophid from the metalophid and is just roloph, protoloph, mesoloph, and combined closing off from the buccal wall of the tooth metaloph-posteroloph. M2 is more squared at this stage of wear. This lake parallels the in its major dimensions (4.3 3 4.25 mm), major buccal reentrant between protoconid although it is narrower posteriorly than and hypoconid. The posterior enamel lake is anteriorly. It has the same four transverse small. The m2 is square (4.9 3 4.95 mm; THP lophs, but the lingual reentrant is directed 14.183 is somewhat bigger [Flynn, 1993]) and toward the middle enamel lake, and possibly has four major lophs. The short mesolophid joined it in early wear. M3 is just erupting, joins the anterior loph through the protoco- making its measurement (3.8 3 3.6 mm) nid. Posterolingual to the mesolophid is the difficult. It has the same four lophs, although hypolophid, which joins the posterolophid the posteroloph is short due to the rounded lingually. In early wear a small enamel lake in nature of the back of the tooth. The lingual the anterolingual corner of the tooth demar- 2009 FLYNN: ANTIQUITY OF RHIZOMYS 137

Fig. 4. Forelimb of Rhizomys (Brachyrhizomys) shansius (F:AM 117337). A, right humerus in medial view; B, left humerus in posterior view; C, lateral view of proximal right ulna; D, medial view of left proximal ulna. Scale bar equals 1 cm. cates a remnant of the lingual anterolophid. Rhizomys, rather than the more slender In later wear the mesolophid joins the clavicle of Tachyoryctes or Cannomys. The hypolophid and all lingual reentrants are five ribs preserved in near-anatomical posi- closed off. The m3 is much like m2 but is tion, plus association with other parts of the longer, especially due to the posterior wall skeleton and the nearly complete skull, being expanded backward toward its base suggest that the specimen was not transport- (7.2 3 5.3 mm). Its small anterolingual lake is ed or torn apart by a predator, but perhaps persistent and the mesolophid is less prom- was preserved in its burrow. inent, more rapidly merged by wear with the The appendicular skeleton includes several hypoconid. The posterolophid is a simple, parts of each forelimb. The right scapula, prominent, arcuate structure, completely with prominent spine but broken acromion isolated from the rest of the crown by a long and coracoid process, is elongate (24.4 3 11 reentrant that closes off lingually during mm); the supraspinous fossa is narrow, much moderate wear. The right m3 of F:AM less expanded than in living species. The ulna 117337 is somewhat more erupted than the (fig. 4) is heavy and anteroposteriorly deep. left, shows a bit of wear on its cusp tips Broken distally, it exceeds 31.5 mm in length. although they are below the level of m2, and The olecranon process is prominent, the is partially prepared out of its crypt (allowing length from the proximal surface of the crude measurement; crown height about 5.5 trochlear notch to the tip being 9.3 mm. mm). It has two lingual reentrants in early There is a heavy coronoid process and wear. prominent supinator crest running distally POSTCRANIA: Portions of the axial skele- from it. ton of F:AM 117337 are preserved. The The heavy humerus (fig. 4) has fused, or broad atlas-axis complex is stoutly built as in nearly fused epiphyses. With a length of 34.1 living bamboo rats. The proximal two-thirds mm, its head is large (maximum dimension of a clavicle, 21 mm long, is nearly straight and its perpendicular are 10.5 3 9.1 mm). and missing its curved distal end. It is The distal end of the humerus is wide, the comparable to the stout clavicle of living epicondylar width being 11.6 mm. The 138 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 331

TABLE 2 Comparative Forelimb Measurementsa (mm) and Ratios of F:AM 117337, Rhizomys (Brachyrhizomys) shansius, and Other Burrowing Muroids

HL HW HW/L HD HD/L UL OL OL/UL

Brachyrhizomys shansius (F:AM 117337) 34.1 11.6 0.34 19.0 0.56 .31.5 9.3 ,0.30 Rhizomys pruinosus senex (MCZ 35935) 43.0 14.5 0.34 26.1 0.61 48.4 13.7 0.28 R. pruinosus pruinosus (AMNH 113477) 47.2 14.1 0.30 26.1 0.55 55.5 13.5 0.24 Cannomys badius (MCZ 3617) 25.4 7.1 0.28 15.4 0.61 28.2 7.2 0.25 Tachyoryctes ibeanus (MCZ 8134) 25.6 8.0 0.31 14.1 0.55 30.3 6.5 0.21 Tachyoryctes sp. (AMNH 54305) 24.0 6.9 0.29 13.3 0.55 30.3 6.2 0.20 Tachyoryctes sp. (AMNH 187461) 27.0 7.8 0.29 16.0 0.59 35.0 6.7 0.19 Spalax microphthalmus (AMNH 228) 27.0 9.5 0.35 15.8 0.59 33.0 11.3 0.34 Spalax ehrenbergi (MCZ 61376) 18.0 5.8 0.32 12.2 0.68 24.8 8.8 0.35 Eospalax fontanierii (MCZ 19896) 32.0 14.9 0.47 21.2 0.66 43.2 18.0 0.42 Eospalax fontanierii (AMNH 135358) 31.0 15.1 0.49 19.3 0.62 — — —

a Measurement abbreviations: HL, humerus length; HW humerus width at the epicondyles; HD, distal extent of the deltoid crest of the humerus (measured from the proximal surface of the head of the humerus); UL, ulna length, OL, olecranon length (measured from the proximal surface of the trochlear notch). prominent deltoid crest is long, ending about absolute size, as judged from total humerus 19 mm from the proximal end of the length, and the features are consistent with humerus—over half the length of the bone burrowing, particularly for rodents that do (56%), similar to other Rhizomys (table 2). not rely solely on the forelimbs, but are For example, the larger humerus of AMNH primarily chisel-tooth diggers. However, the 113477 is 47.2 mm long with deltoid crest humerus and ulna of zokors, which depend length of 26.1 mm, 55% its length; its mainly on the forelimb for digging, are epicondylar width is 14.1 mm, proportionally strikingly modified. Zokor humeri have somewhat less than F:AM 117337. The prominent deltoid crests that project distinct- humerus of R. troglodytes AMNH 18407, ly from the shaft, and exceedingly broad nearly twice as long as that of R. shansius,has epicondyles; a tranverse axis through them is a deltoid crest ratio of 63%. angled with respect to the head of the shaft. The hind limb is represented by a right Humerus width through the epicondyles is distal femur fragment with total condyle nearly half the total length. The olecranon width of 11.1 mm. An associated right tibia process is long, 42% of ulna length in MCZ is missing its epiphyses. The tibia diaphysis is 19896, and with a massive proximal tubercle 34 mm long and relatively slender. The scar located medially. The forelimbs of zokors for the distal portion of the fibula begins 9 show strong adaptations for burrowing, since mm from the distal end of the diaphysis. at least the Late Miocene. Other elements include a proximal phalanx DISCUSSION: Teilhard de Chardin (1942) 8.5 mm long. based his taxon Brachyrhizomys shansius on The forelimb of F:AM 117337 is clearly several mandibular fragments from Yushe modified for burrowing, the ulna having a Basin. Scholars at the Institute of Vertebrate large olecranon process and strong supinator Paleontology and Paleoanthropology deter- crest, and the slightly twisted humerus having mined that the Tianjin collection dentary a heavy deltoid crest and wide epicondyles. 31.096 is the type specimen. Its provenance is Epicondyle width, deltoid crest development, ‘‘Pre-Villafranchian Pliocene beds of Yushe’’ and prominence of the olceranon process are (Teilhard de Chardin, 1942). Of all known developed as in the living rhizomyines specimens, only F:AM 117337 and a few Rhizomys, Cannomys, and Tachyoryctes (ta- collected in recent years have good strati- ble 2). Spalax is similar, although its olecra- graphic provenance. All are mid-Pliocene in non process is relatively longer. All of these age. Teilhard de Chardin (1942) noted the observations appear to be independent of long diastema and corresponding proodonty 2009 FLYNN: ANTIQUITY OF RHIZOMYS 139 of the incisors, although he determined this There are also named forms of Rhizomys without the benefit of a skull to observe! He from the Pleistocene fossil record of China. was also impressed by the lower crown All of these are clearly modern in grade, some height, hence the name Brachyrhizomys. In assignable to living species, and all can be reality, Brachyrhizomys specimens including classified as Rhizomys (Rhizomys). Rhizomys F:AM 117337 are high crowned, although troglodytes is like living R. sinensis in size, not hypsodont and not as high crowned as and was compared with the latter by living Rhizomys. This feature plus those Matthew and Granger (1923). R. ultimus noted above distinguish Brachyrhizomys Young and Liu (1950) was synonymized with from living species, but they are largely (if R. sinensis by Zheng (1993). Zheng (1993, not completely) primitive conditions for 2004) conserved R. troglodytes and added the Rhizomys. F:AM 117337 shows, on the other small form Rhizomys brachyrhizomysoides as hand, that Brachyrhizomys shansius shares extinct bamboo rats from eastern Sichuan many important derived features with living Province and from Hubei Province. Wei et al. Rhizomys. Fundamental is the infraorbital (2004) named Rhizomys fanchangensis from foramen confined to a high position and Anhui Province, assigning it to the living without ventral slit—it is sealed ventrally by genus based on similar dentition, which is anterior expansion of the deep masseter. consistent with incisor orthodonty of the Brachyrhizomys shares a high skull profile fossil material. This oldest described Rhi- with Rhizomys and other features of the zomys (Rhizomys) dates to the early Ma- basicranium and palate, plus traits related to tuyama magnetic chron, greater than 2 Ma. burrowing. Brachyrhizomys shansius is known for Because it is closely related to living remains from Yushe Basin. The species name Rhizomys species, but distinct, Brachyrhi- is applied by Zheng (1993) to fossils said to zomys shansius is treated appropriately as a be from the Zhoukoudian Cap Travertine, subgenus within Rhizomys. Living forms are but I have not seen these specimens. The all relatively close, distinguished by shared Yushe fossils are all Pliocene in age. Flynn advanced features, and I classify them as (1993) added a new, small species to the Rhizomys (Rhizomys). Features distinguish- Yushe fossil record, Rhizomys (Brachyrhi- ing Rhizomys (Rhizomys) are correlated with zomys) shajius, based on a single dentary the more elevated cranium, which yields from Late Miocene deposits. Like B. shan- dorsoventrally extended orbital structures sius, this species is primitive in its unshort- and a high occiput. The occiput is further ened diastema, moderate crown height, m2 exaggerated as a large flat surface area for longer than wide, and termination of the attachment of musculature used in burrowing. masseteric crest under m1 (not more anteri- Among the dozen or so named forms of orly). A recalibrated paleomagnetic age for extant Rhizomys (mostly subspecies), system- Brachyrhizomys shajius is 5.9 Ma. This atists consistently recognize three species, R. species is significantly older than B. shansius pruinosus, R. sinensis, and the larger R. and much smaller (estimated m1–m3 length, sumatrensis. Wang (2003) adds R. wardi at 9.3 mm). the species level. Dental morphology perhaps Flynn (1982) applied the name Brachy- distinguishes some of the subspecies, for rhizomys to large, fossorial bamboo rats from example, the large M3 of Rhizomys sinensis the Siwalik deposits of the Indian subconti- vestitutus. R. shansius is about the size of nent. The several Siwalik species share a small individuals of R. pruinosus or R. number of features with B. shansius (and sinensis, but its features do not argue for living Rhizomys), and are distinct from other special affinity with any living species (its M3 large fossil bamboo rats, such as Rhizomyides size is moderate, judgment made difficult by sivalensis. However, the Late Miocene Rhi- incomplete eruption). Some authors have zomyini, as exemplified by GSP 8362 (Flynn, placed the large species R. sumatrensis in its 1982, pl. 2), do not show the derived closure own subgenus Nyctocleptes, but Musser and of the ventral slit of the infraorbital foramen Carleton (2005) do not follow this; distinc- of B. shansius. Also, GSP 8362 is a low skull tions of this species are largely allometric. and not as broad and wedge shaped as 140 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 331

AMNH 117337. These Siwalik bamboo rats Indian subcontinent. Oldest records date to and Miocene age representatives from south- about 9.6 Ma, with age of actual appearance ern China cannot be maintained in the taxon likely to be no greater than 9.9 Ma (Barry et Brachyrhizomys, and require a new generic al., 2002). By about 9 Ma, four species were level name. present. The genus continued to be diverse through at least 8 Ma, probably later, and is Miorhizomys, new genus well represented in the Late Miocene of northern India. Flynn (1985) pointed out YPE PECIES Miorhizomys nagrii (Hin- T S : that high diversity and cladogenesis are not ton, 1933); illustrated in Flynn (1982). unexpected for a fossorial rodent. Flynn and INCLUDED SPECIES: Miorhizomys blacki Qi (1982) noted broad distribution of rhizo- (Flynn, 1982), Miorhizomys choristos (Flynn, myines across southern Asia into Southeast 1982), Miorhizomys harii (Prasad, 1968), Asia, and cited some of the same species at Miorhizomys micrus (Flynn, 1982), Miorhi- Lufeng, Yunnan Province, as in the Siwaliks. zomys nagrii (Hinton, 1933), Miorhizomys Recent reexamination of original, large sam- pilgrimi (Hinton, 1933), Miorhizomys tetra- ples of the Lufeng material (personal obser- charax (Flynn, 1982). vation) confirms that the Lufeng species are DIAGNOSIS: (after Flynn, 1982, character- very close to, but not all identical with the istics for ‘‘Brachyrhizomys,’’ which were Siwalik species. They are certainly congener- drawn mainly from these fossil species) ic. Rhizomyini with moderately deep dentaries Tribe Rhizomyini, all members of which and a masseteric crest lacking a strong show fossorial adaptations, appeared during anterior extension, lower crest not inflated; Late Miocene time around 10 Ma. The early lower incisor usually with flattened enamel genus Miorhizomys ranged throughout and no median ridge; molars relatively small southern Asia, from Pakistan to Yunnan, and of moderate crown height, with unilat- and was successful as measured by high eral hypsodonty; M1 rounded anteriorly, species richness. The species of Miorhizomys without anterolingual flexus; m3 with small are the ecological precursors to modern posterolingual enamel lake and often a deep Rhizomys, which appeared about 6 Ma in buccal reentrant; incomplete closure of ven- China and became established during the tral slit of the infraorbital foramen; pro- Pliocene. Given late fossil occurrences for a nounced fossorial adaptations. In addition, group considered to be of remote common Miorhizomys shows low elevation of the ancestry with other muroids (Steppan et al., skull, m2 with three roots and strong 2004; Jansa and Weksler, 2004), it is impor- mesolophid, and a longitudinal connection tant to review the record of older rats (mure) on m2 and m3. potentially related to them. AGE AND GEOGRAPHIC RANGE: Late Miocene of southern Asia, from Pakistan to KEY CONSTRAINTS FROM THE Yunnan. FOSSIL RECORD ETYMOLOGY: Mio- (prefix derived from the epoch during which this bamboo rat As so eloquently stated by Musser and genus thrived) plus rhizomys (‘‘root rat’’). Carleton (2005, above) the fossorial habits of DISCUSSION: Miorhizomys is a rhizomyine Rhizomys, Spalax, and Myospalax have led genus showing fossorial adaptations, but is to skeletal modifications that overprint the not as derived as either Rhizomys or Can- features primitively characterizing each clade. nomys. The Siwalik fossil genus Rhizomyides The similar adaptations across the genera, Bohlin, 1946, is a large tachyoryctine, and but lack of these in early members of each more distantly related. Anepsirhizomys clade, makes diagnosis of rhizomyines, spa- Flynn, 1982, is another large genus of unclear lacines, and myospalacines difficult. Review affinity to which Flynn (1990) transferred the of the fossils attributed to each group places species A. pinjoricus (Hinton, 1933). Miorhi- in context not only the antiquity of each, but zomys shows remarkable diversity shortly allows a judgment of the likelihood of their after its appearance in the fossil record of the interrelationships and focuses on any direct 2009 FLYNN: ANTIQUITY OF RHIZOMYS 141 evidence for burrowing adaptations in these origination) or showing low inclination of the lineages. The analysis below suggests that zygomatic plate. each group acquired full fossoriality indepen- Flynn et al. (1985) found rhizomyines to dently. cluster among derived muroids in their myomorphy—they were not a part of the early radiation of muroids that were typically SUBFAMILY RHIZOMYINAE hystricomorphous. The myomorphous con- dition of Rhizomyinae is modified further Flynn (1982) revised knowledge of the with respect to many living muroids by a evolution of rhizomyines, benefiting from a degree of closure of the ventral part of the greatly expanded fossil record, primarily keyhole-shaped infraorbital foramen. In later from the Siwaliks of Pakistan. His diagno- rhizomyini, the ventral notch of the keyhole sis for the higher taxon (his Family opening is totally eliminated, leaving only a Rhizomyidae) including extant Rhizomys, dorsal hole in Rhizomys and Cannomys Cannomys, and Tachyoryctes noted the (autapomorphy for this pair of genera). Less muroid condition of three cheek teeth, the modified rhizomyines are represented by advanced muroid condition of myomorphy, Tachyoryctes and 9 Ma Miorhizomys and proodonty (some forms orthodont), lopho- Kanisamys, which show abbreviated ventral donty (transverse crests as high as cusps), notches. These taxa can be reinterpreted as and differential hypsodonty, with upper showing an early stage in the transformation molars markedly higher crowned internally of myomorphy from hystricomorphy, while than externally. There is also a strong molar Rhizomys and Cannomys show secondarily wear gradient in all rhizomyines: first molars reduced myomorphy. They are a product of are always much more worn than posterior muroid evolution at a time when the myo- molars. Musser and Carleton (2005) cau- morphous condition had not reached full tiously regard Tachyoryctes as potentially development of an expanded, high-angle independent of Rhizomys and Cannomys. zygomatic plate with deep ventral slit. However, some South Asian fossils, such as Rhizomyinae are derived with respect to Eicooryctes argue that tachyoryctines were many Paleogene muroids in loss of the widespread in the Late Miocene (Flynn, archaic posterior connection of the protoco- 1982), and not an independent African nid and metaconid on m1. The posterior group. connection runs directly from the protoco- MORPHOLOGY: An important feature in nid, or from its posterior arm, transversely to tracing the evolution of muroids is the the metaconid. Early rhizomyines may dis- character state of the anterior part of the play an ephemeral posterior connection, but zygomatic arch, and the muscle scar of the also lack an anterior connection. A firm masseter. In hystricomorphs, the deep part of connection of the anterior arm of the the masseter penetrates the enlarged infraor- protoconid to the metaconid developed in bital foramen and originates on the snout; later Kanisamys (Flynn, 1982) in parallel to the lateral part of the masseter is confined to that trend in other Miocene muroids. This the ventral border of the arch. Muroids show observation agrees with the early stage of progressive myomorphy in which the lateral myomorphy in plotting rhizomyines among masseter extends anteriorly on an expanded derived, but not modern muroids. zygomatic plate. This plate rises anteriorly, Rhizomyines possess persistent longitudi- constricting the infraorbital foramen ventral- nal ornamentation on the enamel of the ly, but leaving a slit behind. Advanced lower incisor. As in many late Oligocene to myomorphs extend the origin of the lateral Early Miocene muroids, a double ridge masseter anteriorly by adding a projecting occurs on the incisor in early species. By the flange onto the plate. Lindsay (1977) hypoth- Late Miocene this is a single ridge, which esized gradational transformation of hystri- persists throughout Tribe Tachyoryctini. comorphy to myomorphy in muroids, early This condition is transformed again in Tribe taxa (Oligocene) being hystricomorphous Rhizomyini in which ridges are totally (barely any expansion of the lateral masseter absent. 142 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 331

These features imply origin of rhizomyines barely advanced over that of typical muroids. (and spalacids in general) from Late Oligo- The studied sample is from the Zinda Pir cene muroids that already displayed a degree Dome, western Pakistan, locality Z113. of myomorphy, suppression of the archaic Isolated incisors from Z113 show a double posterior connection between metaconid and ridge on the enamel. The age of this site is protoconid of m1, and the then widespread demonstrably younger than the classic Bugti occurrence of persistent ridges along the Oligocene sites of Paali Nala (Welcomme et midline of the enamel of the lower incisor. al., 1999) and Z108 and Y417 (Flynn and Muridae and Cricetidae are Miocene deriva- Cheema, 1994), and older than Early Mio- tives of a more advanced stock of muroids. cene sites in the area. Analysis of the FOSSIL RECORD: The Siwalik Group magnetic polarity stratigraphy of the Zinda deposits, primarily in Pakistan, contain a Pir Dome (Lindsay et al., 2005) allows two relatively complete series of bamboo rat reasonable correlations to the geomagnetic fossils, filling the gap between standard polarity time scale, which place the site at muroids and living rhizomyines. Small Mio- about 24 or 27 Ma (near the Oligocene- cene species of the genera Prokanisamys and Miocene boundary or during the Late Kanisamys are abundantly preserved in suc- Oligocene). Thus, the oldest rhizomyines cessive horizons spanning . 20 to 8 Ma (de appear to antedate other spalacid muroids. Bruijn et al., 1981; Flynn, 1986; Wessels and de Bruijn, 2001). Postcrania show that these SUBFAMILY SPALACINAE species were not obligate fossorial rodents. There are some digging modifications in the In summarizing the spalacine fossil record, humerus of Kanisamys sivalensis (about 9 U¨ nay (1999) noted that rhizomyines, spala- Ma), but they are not striking. In contrast, cines and other lineages of burrowers appear beginning after 10 Ma, other large-body to be distinctive throughout most of their species do show fossorial adaptations in the histories, at least in dental characteristics skull and forelimb. These have been gathered (fig. 5), although all converge in adaptations under the names Rhizomyides and ‘‘Brachy- to fossorial lifestyle. Potapova and Voront- rhizomys’’ (now Miorhizomys). Flynn (1985) sov (2004) agreed and noted convergence in saw the Late Miocene species-rich pattern of auditory structures correlated with subterra- occurrence of large-bodied Miorhizomys as nean life. Paleontologists, therefore, have consistent with fossorial lifestyle (under- considered the spalacines to be distinct at ground habitus promoting isolation and the family level from other muroid groups. splitting). Apparently, bamboo rats were Three morphological features have tended to initially terrestrial scratch diggers; commit- support the apparently remote ancestry of ment to subterranean life came later and Spalax: (1) The expanded infraorbital fora- independently in different groups. men is round (but not reduced and dorsal in The fossil record of rhizomyines is sub- position) and lacks an inclined myomor- stantially older than the first evidence for phous zygomatic plate; the lateral masseter fossoriality, but just how old is this lineage? origin is clearly defined, but restricted to the The emerging pre-Siwalik fossil record shows ventral border of the zygoma, close to the greater antiquity for the group than previ- condition seen in hystricomorphs. (2) The ously in evidence. Lindsay (1996) described lower first molar lacks the anteriorly placed the oldest rodent attributable at present to connection between the protoconid and the bamboo rat lineage. Under the name metaconid (the posterior metalophid is gen- Eumyarion kowalskii (in earlier works he erally lost as well). (3) The lower incisor mentioned this as an unnamed species of shows very weak if any longitudinal orna- Eucricetodon), he showed the beginnings of mentation, unlike most Late Oligocene– the lophodont rhizomyine dentition, modi- Miocene muroids. Given the data then at fied from the basic cricetid plan, and hand, Flynn et al. (1985) concluded that preserving, variably, the archaic posterior spalacines were of ancient origin from connection between protoconid and metaco- hystricomorphous muroids that lacked ad- nid on the m1. Its elevated crown height is vanced molar structure and incisor ridges. It 2009 FLYNN: ANTIQUITY OF RHIZOMYS 143

longitudinal ridges. Sinapospalax showed that ridges were indeed present on early spalacine incisors. Further, the robust den- taries suggest that these muroids were proodont burrowers. Generally considered separate but closely related are the extinct anomalomyines, char- acterized primarily by Anomalomys and Prospalax (Fejfar, 1972). These have a similar age range as spalacines, but are Fig. 5. Comparisons of upper and lower confined to Europe. Bolliger (1999) reviewed molars of a generalized rhizomyine (A, B), the group and noted the presumed fossorial spalacine (C, D), and myospalacine (E, F). Molar habitus of later species. Following de Bruijn rows are composites of fossils, except E and F, which are associated. Seen left to right, A (right and Sarac¸ (1991), he further drew compari- M3 to M1) and B (left m1 to m3) are based on son to the Early Miocene muroid Eumyarion, Kanisamys sivalensis from Flynn (1982). C (worn based on molar similarities and the double right M3–M2 and separate M1) and D (left m1 ridge of enamel along the length of the lower and m2 and worn m3) are based on Pliospalax incisor. Eumyarion is known from the Early macoveii from S¸en (1977). E (right M3 to M1) and Miocene of Turkey (U¨ nay-Bayraktar, 1989), F (left m1 to m3) are based on Mesosiphneus as well as Pakistan (above), older than praetingi from Teilhard de Chardin (1942). Scale is Anomalomys or Debruijnia. If anomalo- 1 mm. myines and spalacines are closely related, Late Oligocene origin of this group would be probable. appeared that rhizomyines were more derived Despite the projected Oligocene origin for in these features. spalacines and their apparently primitive The spalacines have a modern distribution morphology, the oldest clear spalacine from the Mediterranean area throughout fossils are younger than the earliest rhizo- eastern Europe to western Asia, with extinct myines. Still, given that spalacines and species of Spalax and Pliospalax recorded Anomalomys may share common ancestry widely in Eurasia, but not the Far East. The with a morphotype resembling Eumyarion, fossil record is Early Miocene and younger. one can hypothesize close relationship to the The Oligocene Rhizospalax is a beaver earliest rhizomyines, a hypothesis consistent (Hugueney and Mein, 1993). The late Early with molecular data. The features of early Miocene Heramys Klein Hofmeijer and de spalacines are now better known, and may Bruijn (1985) shows several spalacine features be reinterpreted as similar to those of early (lophodonty, high molar crowns, loss of rhizomyines, not more primitive. Early metalophid on m1), but retains a large m3. members of both groups show enamel U¨ nay (1996) added to the fossil record the ornamentation and similar m1 morphology. more primitive, lower-crowned genus Deb- The undeveloped zygomatic plate of Spalax ruijnia, which shows a variable metalophid and Pliospalax, possibly a reversal, does connection on m1. The age of the earliest differ from the stronger plate of early species of this genus is ca. 20 Ma. Neither rhizomyines. Heramys nor Debruijnia indicate fossoriality, Skull and postcranial remains, rare before but postcrania are unknown. the Late Miocene, show that Pliospalax was Sarica and S¸en (2003) added key fossil an active burrower (S¸en, 1977). The robust, remains from the early Late Miocene Sinap proodont dentary of Sinapospalax also sug- Formation of Turkey. These revealed im- gests burrowing. However, there is no portant anatomical details of spalacines in evidence for fossoriality in early spalacines, the 11 to 9 Ma age range. Sarica and S¸en and the conservative dentitions of Heramys (2003) named Sinapospalax for an array of and Debruijnia suggest that these were not species that share robust dentaries, lopho- obligate burrowers. Present evidence indi- dont molars, and incisors bearing multiple cates independent acquisition by the early 144 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 331

Late Miocene of the subterranean lifestyle in is clear that their prismatic (omega form is a rhizomyines and spalacines. better term used by Carleton and Musser, 1984) condition is derived convergently with SUBFAMILY MYOSPALACINAE respect to the much more complex molar shapes of arvicolines. The lower first molar Whereas spalacines and anomalomyines presents an anterior connection between the are European-Mediterranean and western anterior arm of the protoconid and the Asian rodents, and rhizomyines are primarily metaconid, which appears to be the derived South Asian–East African, myospalacines muroid condition. It could have been con- (zokors) are common eastern Asian burrow- structed independently, much the same way ers, typical of much of northern China and as it developed in rhizomyines. Also like Mongolia and ranging into southern Siberia, earlier rhizomyines and various muroids, the the Altai Mountains, and eastern Kazakh- lower incisor is always characterized by a stan. Living species, up to about 250 grams in single longitudinal ridge. These morphologi- body mass, are highly derived subterranean cal features suggest that myospalacines small mammals that dig with their forelimbs evolved from later muroids more derived (Nevo, 1999). They have adopted, therefore, than the ancestral stock for spalacines. a different strategy than rhizomyines and The fossil record demonstrates rapid spalacines, which utilize the incisors exten- diversification of zokors, with most of the sively in burrowing. splitting during the past 10 million years. Living myospalacines comprise at least The oldest species attributed to early seven species (Musser and Carleton, 2005), Prosiphneus date to over 11 Ma according which can be assigned to two genera, to Zheng et al. (2004). As research has Myospalax and Eospalax. The fossil record developed in China, scientists have traced the of myospalacines in China is extraordinarily origin of the myospalacines to the Middle rich, and appears to show geographic differ- Miocene muroids Plesiodipus Young, 1927, ences, with different species as contemporar- or Gobicricetodon Qiu, 1996, both in excess ies in eastern and central China. Most of 12 Ma (see Wang et al., 2003). It seems researchers acknowledge broad morphologi- logical to transfer these genera to the cal diversity among the array of fossil zokor Myospalacinae. However, Plesiodipus is species. Zheng (1997) recently rearticulated comparable to generalized cricetids, the his rationale for diagnoses of generic and family in which Qiu (1996) classifies it, and higher level groupings (here, tribes). In Gobicricetodon is low crowned, although Prosiphneini Leroy, 1940, he groups Prosiph- robust. Probably these earliest myospalacines neus Teilhard, 1926, Myotalpavus Miller, were not truly fossorial. More likely, myos- 1927, Pliosiphneus Zheng, 1994, Eospalax palacines acquired fossoriality independently Allen, 1938, Allosiphneus Kretzoi, 1961; in about 11 Ma, an independence consistent Mesosiphneini Zheng, 1994, he places Char- with characteristic forelimb-digging in con- dina Zheng, 1994, Mesosiphneus Kretzoi, trast to chisel-tooth digging in rhizomyines 1961, Yangia Zheng, 1997 (5 Youngia Zheng, and spalacines. 1994); in Myospalacini Lilljeborg, 1866, are Conclusions based on morphology and the Episiphneus Kretzoi, 1961, and Myospalax, fossil record are: (1) Zokors show derived Laxmann, 1769. This diversity reflects a high features (myomorphy, anterior metaconid- degree of cladogenesis, likely promoted by protoconid connection, incisor ridge) sug- isolating mechanisms correlated with solitary gesting derivation from Miocene muroids. (2) lifestyle and small home ranges. Zokors evolved forelimb burrowing indepen- Structural characteristics of myospalacines dently about the beginning of the Late include a zygoma interpreted as myomor- Miocene. (3) Zokor origin appears to be phous, although the infraorbital foramen is Middle Miocene; the common ancestor of not strongly keyhole shaped and the zygo- Myospalax and Spalax must be much older matic plate is not expansive or greatly than the known range of myospalacines, and inclined, a forelimb highly derived for bur- more basal among murids. The Middle rowing, and superficially prismatic molars. It Miocene zokor origin conflicts with a Plio- 2009 FLYNN: ANTIQUITY OF RHIZOMYS 145 cene estimate by Michaux and Catzeflis fossorial rodent. It is rather low and elongate (2000). Norris et al. (2004) discounted a (Kowalski, 1968), comparable perhaps to Pliocene split based on molecular grounds, that of a small Marmota. and found that mitochondrial genes place Pararhizomys is known from several Late Myospalax basal to Spalax and Rhizomys. Miocene sites of North China (Zhang et al., Jansa and Weksler (2004) reanalyzed the von 2005). The genus may also be recorded from Willebrand Factor (vWF) gene results of the Late Miocene of Tibet. Brachyrhizomys Michaux and Catzeflis (2000) and also placed hehoensis Zheng, 1980, from Biru, Bulong Myospalax outside the other spalacids. How- Basin, is unusual for rhizomyines (small m3, ever, a maximum-likelihood analysis of the even molar wear) and may be a member of first exon of the IRBP gene placed Myospa- the genus Pararhizomys (Jacobs et al., 1985). lax somewhat closer to rhizomyines (Jansa Zhang et al. (2005) evaluated proposed and Weksler, 2004) and showed that the Pararhizomys relationships to rhizomyines, geometry of the three taxa is not yet spalacines, myospalacines, and tachyoryctoi- understood robustly. A monophyletic rela- dines, and found two most probable scenar- tionship with Spalax and Rhizomys, in which ios: Pararhizomys is a primitive rhizomyine Myospalax is the most basal form, is or a derivative of independent early muroid unexpected from morphology and the fossil origin. I cannot improve on this suggestion at record. This relationship disrupts character present other than to emphasize that if distributions and calls for a zokor lineage Pararhizomys is a rhizomyine, it is not closely back to the end of the Oligocene, which is related to living bamboo rats. The odd m2 nowhere in evidence. As noted below, nesting reentrant and retention of the incisor ridge of zokors within Miocene rhizomyines seems would be primitive structures modified in more conceivable. Rhizomyini. The elevated molar crown height reminiscent of spalacines and Tribe OTHER CHINESE FOSSIL MUROIDS Tachyoryctini likely evolved independently. In building a hypothesis of sister-taxon status The enigmatic Pararhizomys and Tachyor- for spalacines and rhizomyines, it is conceiv- yctoides have been implicated in the evolu- able to include Pararhizomys as a member of tion of bamboo rats. Pararhizomys hippar- that clade, outside Tribe Rhizomyini. Fur- ionum is a Late Miocene species from ther, if myospalacines are indeed part of the Mongolia and North China that presents spalacine-rhizomyine radiation, then Para- rhizomyine-like features with some excep- rhizomys could root close to the origin of tions. Dentition is lophodont and rather high Myospalacinae, possibly among Middle Mio- crowned, and skull material shows modified cene rhizomyines. myomorphy (reduced ventral slit of the Oligocene to Early Miocene Tachyoryc- infraorbital foramen), as in late Neogene toides shows no special features to align it rhizomyines. Unlike rhizomyines the molar with either rhizomyines or spalacines. Its wear gradient is even (not markedly stronger characteristics illustrate primitive muroid anteriorly). Unlike most rhizomyines but like features. The cheek teeth are rather low spalacines, third molars are reduced, and m1 crowned. The crests are not transformed as lacks a metalophid. The lower second molar sweeping lophs, but rather, in early wear, as has two buccal reentrants, the anterior one thin interconnections with deep reentrants. bounded by a strong buccal extension of the The primitive muroid buccal arm of the anterolophid. This buccal anterolophid is anterolophid is well developed on m2 and normally suppressed in rhizomyines, but m3. Reminiscent of the spalacine condition present in early species, and conserved in of m1, there is no direct connection of the myospalacines. Unlike Rhizomys but like protoconid and metaconid. Tachyoryctoides Tachyoryctes, the near-flat incisor of Para- has almost no mesolophid on lower molars; rhizomys shows a single longitudinal ridge. upper molars have a very short mesoloph, Myospalacines conserve the ridge as well. but are four-crested due to a long poster- The skull of Pararhizomys is strongly built, oloph emanating from the hypocone. In the but not wedge shaped like a committed absence of cranial material, the masseteric 146 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 331 condition may be inferred from the mandible. Rhizomys is from the Late Miocene of The masseteric crest is quite heavy, flaring northern China, 5.9 Ma, but the oldest well- laterally and terminating opposite m2, al- represented species is Rhizomys (Brachyrhi- though a weak portion of the scar extends zomys) shansius from the early Pliocene of below m1. Probably a heavy lateral masseter Yushe Basin, Shanxi Province. Osteological originated from the bottom of the zygoma. material consisting of the skull and anterior The lower incisor is heavy and distinctly parts of the axial skeleton, parts of both flattened, lacking a ridge, but with low forelimbs, and a partial hindlimb shows that thickenings of enamel along both the medial by the Pliocene, Chinese bamboo rats were and lateral borders. The chisel-like incisor is fully modern with advanced burrowing ad- consistent with digging, but until postcrania aptations and derived skull structure, espe- or skull material are described, fossorial cially the modified, high and round, infraor- adaptation for Tachyoryctoides is poorly bital foramen of living species. supported. The pattern of evolution of burrowing Tachyoryctoides is distributed throughout rodents appears to be linked to opportunities northern China and Mongolia, and westward to exploit available subterranean ecological to the Aral Sea region, as a common space (Nevo, 1999; Cook et al., 2000). component of Oligocene and Early Miocene Controlling factors would include availability assemblages. It is thus a contemporary of the of underground food sources and quality of basal rhizomyine Eumyarion kowalskii. The the soil (including rockiness and moisture). differences in the structures of these taxa One hypothesis sees radiation of subterra- show that Tachyoryctoides is not a rhizo- nean rodents and emergence of hominids as myine. Flynn (1982) agreed with Fejfar phenomena connected to the evolution by (1972) that the genus could be a primitive plants of underground nutrient-storing or- spalacine, but that idea comes from the gans, which would be highly desirable food similar m1 structure. Otherwise, there are sources (Laden and Wrangham, 2005). The no special features shared with Heramys or data to test these ideas come from the fossil with Anomalomys. Consequently, Tachyoryc- record, and for rodents include constraining toides is not readily associated with either the timing of acquisition of fossorial traits rhizomyines or spalacines, and at present its across lineages. How muroids, in particular, sister status with rhizomyines + spalacines acquired subterranean lifestyles is relevant to cannot be defended. testing whether burrowing evolved in this group due to availability of a new food DISCUSSION source. If bamboo rats, the blind mole rat (Spalax), and zokors are confamilial (Step- The Tribe Rhizomyini arose late in bam- pan et al., 2004; Jansa and Weksler, 2004), boo rat evolution, after 10 Ma, and all fossil then early acquisition of burrowing traits members show burrowing adaptations. Like- might be proposed as a synapomorphy ly their success reflects radiation into the uniting them (Norris et al., 2004). It turns subterranean niche, and exploitation of out that the pattern of evolution of fossori- available underground resources. Early spe- ality in these muroids shows late and cies lack some of the advanced fossorial traits independent commitment to subterranean of extant taxa, and cranial fossils demon- lifestyles. Fossorial features cannot be used strate that they are not assignable to Rhiz- to define Spalacidae cladistically, and prob- omys or Cannomys. The new generic name, ably evolved in the early Late Miocene to Miorhizomys, is created for the Late Miocene exploit new food sources. radiation of subterranean bamboo rats. It is Rhizomyines have the longest known fossil characterized among other features by re- record of the three subfamilies. Early forms taining an abbreviated ventral portion of the (Late Oligocene and younger) clearly were myomorphous infraorbital foramen. Mior- not obligate subterranean dwellers. Full hizomys was distributed across southern Asia fossoriality appears only after 10 Ma. The from Pakistan in the west to Yunnan known time depth of spalacines rivals that of Province in the east. The oldest known rhizomyines, but osteological evidence for 2009 FLYNN: ANTIQUITY OF RHIZOMYS 147 burrowing is rather young. Dental remains lophs were considered underived (which they indicate lack of features, like high-crowned, are, as are early rhizomyines, relative to most lophodont teeth, that are correlated with Miocene taxa). burrowing. In contrast, the known time range The constructive challenge of new molec- of zokors is hardly half of the other groups, ular data leads to reappraisal of character and morphological features suggest that states, which properly entails a survey of the zokors represent a later radiation. Middle extant muroids (since spalacids are basal), Miocene zokor species show no evidence of plus well-known key fossil taxa (appendix 1). fossoriality, but fossil representatives of the This survey makes use of the mammal radiation after 11 Ma include humeri mod- collections of the AMNH and MCZ and of ified for burrowing. Independent acquisition available casts of fossil material. There are of full burrowing adaptations in each muroid far too few character states for the 400 living group appears to be the pattern. Indepen- and fossil muroid genera to allow informative dence is underscored by the forelimb bur- phylogenetic analysis using parsimony, but a rowing in zokors that contrasts with the tree can be generated as a testable hypothesis chisel-tooth digging of bamboo rats and (fig. 6) using the relationships of living blind mole rats. families established on molecular grounds. Despite independent and divergent means Injecting fossil taxa into a tree generated by of solving the problems of subterranean life, molecular data is provocative in stimulating a the three groups may have evolved from hard look at both character definition and generalized muroid scratch-diggers that polarity on the one hand, and assumptions lived in burrows, but were not fully subter- about molecular evolution on the other. It is ranean. The fossil record should be consis- hoped that this hypothesis will stimulate a tent with such a hypothesis, and is relevant deep and extended search for characters and for constraining the timescale of evolution. a new survey of fossil taxa to generate a more The oldest known spalacids are Late Oligo- rigorous test. cene, 27 Ma at most. Under this scenario, A tree (fig. 6) showing both living and the spalacid group would have originated in fossil groups allows reanalysis of character the Late Oligocene and the three lineages distributions. Salient morphological features would have adopted fully subterranean that can be observed on the tree include: lifestyles later, probably not before the Late early muroids are hystricomorphous and Miocene. later genera show modifications of the Given that molecular and morphological zygoma, including progressive inclination data place Spalacidae as a basal group and expansion of the plate, the posterior among all other (studied) living muroids, connection of m1 is lost, followed by origination dates for all other muroid groups construction of an anterior connection, and would be younger than 27 Ma, which is a incisor ornamentation is a conservative maximum estimate for the earliest known feature that appears among muroids and spalacid fossil. A younger, Miocene, age for then is lost. Other features of the dentition the appearance of other extant muroid are overlain on the tree (fig. 6) with polarity families would be expected. proposed by outgroup comparison. Herein, morphological features used to Character states for stem spalacids may be evaluate relationships have been reinterpret- reinterpreted as follows: (1) The common ed given the molecular hypothesis of spalacid ancestor had a zygomatic structure advanced monophyly. Impressed by differences be- over hystricomorphy but incompletely myo- tween spalacines and rhizomyines, Flynn morphous, with moderately inclined zygo- (1982) had considered the former as primitive matic plate not defining a narrow ventral slit and to have originated significantly earlier. and without an anteriorly expanded flange. The zygoma of Spalax was interpreted as (2) The first lower molar had lost its posterior hystricomorphous, instead of myomorphous connection but variably showed a remnant of with a weakly inclined plate, the smooth a protoconid-metaconid bridge, and lacked a incisor enamel was taken as primitive (not as new anterior connection. (3) The lower secondary loss of ornamentation), and molar incisor was ornamented with a pair of 4 ULTNAEIA UEMO AUA ITR O 331 NO. HISTORY NATURAL OF MUSEUM AMERICAN BULLETIN 148

Fig. 6. Phylogenetic tree depicting arrangement of major living groups of muroids plus key fossil taxa, with Dipodoidea as outgroup. This hypothesis depends on molecular data to arrange the living groups, and use s a set of characters (appendix 1) with proposed polarities to plot the foss il taxa. Homoplasy is clear and there are far too few characters to run an analy sis based on parsimony. Note that under this analysis, certain higher taxa in occasional usage such as Cricetodontidae need significant revision or abandonment. 2009 FLYNN: ANTIQUITY OF RHIZOMYS 149

Fig. 7. Known age ranges of the three spalacid subfamilies and key taxa discussed in the text. These are arranged, west to east, according to geographic range, although rhizomyines extend as far eastward (but to the south) as do myospalacines. Rhizomyinae have the longest known age range. Known myospalacines, beginning with Gobicricetodon at about 13 Ma, have a much shorter age range. This implies two hypotheses (dashed): either the group has a long ghost lineage, or it is in fact a derivative (infrafamily) of Subfamily Rhizomyinae. Pararhizomys may also be a member of Rhizomyinae. Tachyoryctoides, on the other hand, does not show spalacid synapomorphies and is excluded from Spalacidae. parallel, raised ridges. Later derived trans- cene, and that Spalacinae and Rhizomyinae formations included: (1) elimination of the are sister taxa that split shortly thereafter ventral portion of the infraorbital foramen in (fig. 7). Myospalacinae are more derived and later rhizomyines, but restriction of the a younger group that would be closer to lateral masseter to the zygomatic arch in Rhizomyinae than to Spalacinae. Character spalacines, (2) neomorphic establishment of a distributions are most consistent when myos- connection to the metaconid from the ante- palacines are considered as derived from rior arm of the protoconid in rhizomyines Early Miocene rhizomyines. This also reduc- and myospalacines, (3) loss of incisor ridges es the apparently long ghost lineage leading in later spalacines and successive reduction to the first zokor fossils (fig. 7). If this from two to one to none in rhizomyines, a hypothesis is maintained, myospalacines single ridge conserved in myospalacines. may be reconsidered as a taxon of lower Myospalacines present features derived rank within Subfamily Rhizomyinae. with respect to stem spalacids, but consistent Among enigmatic Asian fossil taxa, some with earlier rhizomyines. Muroids more may have been part of the spalacid radiation. derived than spalacids adopt the advanced Possibly Pararhizomys from the late Neogene anterior connection for m1, lose incisor of Mongolia and northern China is an early ridges, and progressively build the modern rhizomyine-zokor derivative. The archaic myomorphous zygoma, including a strongly Asian Tachyoryctoides, however, is distin- inclined plate, usually with ventral slit, and guished as a primitive muroid excluded from often expanded anteriorly with a flange to Spalacidae. offer greater origination area for the lateral masseter muscle. CONCLUSION The scenario favored here, given molecular constraints that unite these muroids, is that There is irony in the maturing analysis of Spalacidae originated during the late Oligo- basal muroid groups. It is those scientists 150 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 331 studying morphology (and paleontologists with any other living muroids, but do not yet necessarily emphasize morphology) who con- demonstrate which two of the three subfam- clude independence of bamboo rats, blind ilies are most closely related. Morphology mole rats, and zokors, and suspect that and the fossil record appear to argue against similar degrees of fossoriality and the atten- myospalacines being more basal than the dant ‘‘panoply of adaptations . . . obscure[s] others. Zokors share more features with phylogenetic connections’’ (after Musser and Rhizomyinae than with Spalacinae, but are Carleton, 2005). It is molecular biologists not part of the modern bamboo rat tribe. It is (e.g., Jansa and Weksler, 2004; Norris et al., reasonable to consider zokors and fossil 2004) who indicate that these lineages are Pararhizomys as derivatives of early rhizo- actually closely related, and imply shared myine evolution. predisposition to subterranean life. While not It is important to continue testing the falsified by morphology or the fossil record, hypothesis of monophyly of Spalax + Rhi- this scenario is constrained by the distribu- zomys + Myospalax. More genetic evidence is tion of features and by paleontology. There is likely to clarify the interrelationships of the no evidence for subterranean lifestyle among three taxa from a molecular point of view. early members of any of the groups. Each This will help to focus reanalysis of polarity group developed fossoriality in Middle to of morphological features and the pattern of Late Miocene and younger members, and evolution of these structures. Clearly the burrowing features cannot be used to diag- demands of a fossorial lifestyle encourage nose Spalacidae cladistically. Of the three morphological convergence. Could the evo- groups, zokors present more derived condi- lution of some molecules be channeled by tions in dentition and skull morphology, and selection? Studying more genes is welcome, they have the shortest fossil record. but also expanding the data set to capture The origin and radiation of Rhizomys is a other distantly related burrowing muroids late event in muroid evolution. Fossil repre- might help. Ellobius is perhaps the most sentatives close to extant species occur fossorial of several burrowing arvicolines, throughout much of China and Southeast and based on morphology, would not group Asia during the Pleistocene. Early Pliocene with spalacids. Inclusion of this and other fossils show the derived conditions of living taxa in the muroid genetic sampling serves as Rhizomys, especially the ventrally constricted a test of observed relationships. It is the infraorbital foramen and features of the skull integration of molecular, morphological, and and humerus associated with a subterranean natural history data, and continual testing of mode of life. Skull features are, however, hypotheses of relationship and pattern of generally less derived than those of living evolution that promise to advance under- species. The best represented Pliocene bam- standing of the course of muroid evolution. boo rat is from strata of Yushe Basin, Shanxi This is what Guy Musser knows so well. Province, dating to around 3.5 Ma, and is distinguished at the subgenus level as Rhi- ACKNOWLEDGMENTS zomys (Brachyrhizomys) shansius. The oldest known Rhizomys at 5.9 Ma is also from This work began as part of my Carter Yushe. Earlier during the Late Miocene, Fellowship at the AMNH in the mid-1980s. I more primitive bamboo rats were distributed was studying the Frick Collection of Chinese across southern Asia, from Pakistan to fossil mammals with Dick Tedford, which Yunnan. Diverse species, many gathered naturally led me to the fabulous comparative under the new name Miorhizomys, show collections of the Mammal Division. Guy fossorial characteristics but are less derived Mussser also became a mentor at that time, than Rhizomys. Still older nonfossorial Rhi- and I am grateful to the editors to be able to zomyinae can be traced to the late Oligocene show my appreciation by dedicating this cricetid-like species Eumyarion kowalskii. work to him. I also thank Judy Chupasko Molecular data strongly argue for closer at the MCZ, and Chris Norris and Eileen relationships of Spalacinae, Rhizomyinae, Westwig at the AMNH for access to and Myospalacinae with each other than collections, and Randal Collura for discus- 2009 FLYNN: ANTIQUITY OF RHIZOMYS 151 sion of molecular data. Miche`le Morgan, de Bruijn, H., S. Taseer Hussain, and J.J.M. Alexa Weingarden, John Barry, and David Leinders. 1981. Fossil rodents from the Murree Pilbeam of the Peabody Museum at Harvard, formation near Banda Daud Shah, Kohat, provided vigorous discussions of various Pakistan. Proceedings of the Koninklijke Ne- derlandse Akademie van Wetenschappen Series aspects of this work. Reviewers Louis Jacobs B Physical Sciences, 84: 71–99. and S¸evket S¸en offered constructive com- Fejfar, O. 1972. Ein Neuer Vertreter der Gattung ments that improved the manuscript. Many Anomalomys Gaillard, 1900 (Rodentia, Mam- thanks to Rob Voss and Mary Knight for malia) aus dem europa¨ischen Mioza¨n (Karpat). their help and patience. Neues Jahrbuch fu¨r Geologie und Pala¨ontologie Abhandlungen 141 (2): 168–193. Flynn, L.J. 1982. Systematic revision of Siwalik REFERENCES Rhizomyidae (Rodentia). Ge´obios 15: 327– 389. Allen, G.M. 1938. The mammals of China and Flynn, L.J. 1985. Evolutionary patterns and rates Mongolia. Natural History of Central Asia, vol. in Siwalik Rhizomyidae. Acta Zoologica Fen- 11 (1): 1–620. New York: American Museum of nica 170: 141–144. Natural History. Flynn, L.J. 1986. Species longevity, stasis, and Barry, J.C., M.E. Morgan, L.J. Flynn, D. Pilbeam, stairsteps in rhizomyid rodents. In K.M. Flanagan A.K. Behrensmeyer, S.M. Raza, I.A. Khan, C. and J.A. Lillegraven (editors), Vertebrates, phy- Badgley, J. Hicks, and J. Kelley. 2002. Faunal logeny, and philosophy (Contributions to Geol- and environmental change in the Late Miocene ogy Special Paper 3): 273–285. Laramie: Univer- Siwaliks of northern Pakistan. Paleobiology sity of Wyoming. Memoirs 3, Suppl. 28:1–71. Flynn, L.J. 1990. The natural history of rhizomyid Black, C.C. 1972. Review of fossil rodents from rodents. In E. Nevo and O.A. Reig (editors), the Neogene Siwalik Beds of India and Paki- Evolution of subterranean mammals at the stan. Palaeontology 15 (2): 238–266. organismal and molecular levels: 155–183. New Bohlin, B. 1946. The fossil mammals from the York: Wiley-Liss. Tertiary deposit of Taben-buluk, western Kansu. Flynn, L.J. 1993. A new bamboo rat from the Late Part II: Simplicidentata, Carnivora, Artiodacty- Miocene of Yushe Basin. Vertebrata Palasiatica la, Perissodactyla, and Primates. Sino-Swedish 31: 95–101. Expedition Publication 28, 6 (4): 1–259, 9 pls. Flynn, L.J., and I.U. Cheema. 1994. Baluchimyine Bolliger, T. 1999. Family Anomalomyidae. In G.E. rodents from the Zinda Pir Dome, western Ro¨ssner and K. Heissig (editors), The Miocene Pakistan: systematic and biochronologic impli- Land Mammals of Europe: 389–394. Mu¨nchen: cations. In Y. Tomida, C. Li, and T. Setoguchi Verlag Dr. Friederich Pfeil. (editors), Rodent and lagomorph families of Cande, S.C., and D.V. Kent. 1995. Revised Asian origins and diversification. Tokyo Na- calibration of the geomagnetic polarity time- tional Science Museum Monograph 8: 115– scale for the Late Cretaceous and Cenozoic. 129. Journal of Geophysical Research 100B: 6093– Flynn, L.J., L.L. Jacobs, and E.H. Lindsay. 1985. 6095. Problems in muroid phylogeny. In W.P. Luckett Carleton, M.D., and G.G. Musser. 1984. Muroid and J.L. Hartenberger (editors), Evolutionary rodents. In S. Anderson and J.K. Jones, Jr relationships among rodents: a multidisciplinary (editors), Orders and families of Recent analysis: 589–616. New York: Plenum. mammals of the world: 289–379. New York: Flynn, L.J., and G. Qi. 1982. Age of the Lufeng, Wiley. China, hominoid locality. Nature 298: 746–747. Cook, J.A., E.P. Lessa, and E.H. Hadly. 2000. Gray, J.R. 1831. Characters of three new genera, Paleontology, phylogenetic patterns, and mac- including two new species of Mammalia from roevolutionary processes in uubterranean ro- China. Proceedings of the Zoological Society of dents. In E.A. Lacey, J.L. Patton, and G.N. London 1: 94–96. Cameron (editors), Life underground: the biol- Hartenberger, J.-L. 1967. Contribution a`l’e´tude de ogy of subterranean rodents: 332–369. Chicago: l’anatomie cranienne des rongeurs. I. Principaux University of Chicago Press. types de Cricetodontine´s. Palaeovertebrata 1: de Bruijn, H., and G. Sarac¸. 1991. Early Miocene 47–64. rodent faunas from the eastern Mediterranean Hinton, M.A.C. 1933. Diagnoses of new genera area. Part 1. The genus Eumyarion. Proceedings and species of rodents from Indian Tertiary of the Koninklijke Nederlandse Akademie van deposits. Annals and Magazine of Natural Wetenschappen 94 (1): 1–36. History Series 10, 12: 620–622. 152 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 331

Hugueney, M., and P. Mein. 1993. A comment on Journal of Vertebrate Paleontology 27 (4): 1062– the earliest Spalacinae (Rodentia: Muroidea). 1065. Journal of Mammalian Evolution 1 (3): 215– Lazzari, V., J. Michaux, and J.P. Aguilar. 2008. 223. Les Megacricetodon du gisement karstique Jacobs, L.L., and L.J. Flynn. 2005. Of mice… Mioce`ne moyen de Blanquete`re1 (Pyre´rene´es- again: the Siwalik rodent record, murine distri- Orientales, Sud de la France): nouvelles espe`ces, bution, and molecular clocks. In D.E. Lieber- implications biochronologiques et phyloge´n- man, R.J. Smith, and J. Kelley (editors), iques. Ge´obios 40: 91–111. Interpreting the past, essays on human, primate, Leroy, P. 1940. Observations on living Chinese and mammal evolution in honor of David mole-rats. Bulletin of the Fan Memorial Insti- Pilbeam: 63–80. Boston: Brill. tute of Biology. Zoology 10: 167–193. Jacobs, L.L., L.J. Flynn, and C. Li. 1985. Lilljeborg, W. 1866. Systematisk O¨ fversigt af de Comments on rodents from the Chinese Neo- Gnagande Da¨ggdjuren, Glires. 59pp. Kongliga gene. Bulletin of the Geological Institutions of Akademiska Boktryckeriet, Uppsala. the University of Uppsala New Series 11: 59–78. Lindsay, E.H. 1977. Simimys and origin of the Jaeger, J.J. 1977. Les rongeurs du Mioce`ne moyen Cricetidae (Rodentia: Muroidea). Ge´obios 10 (4): et supe´rieur du Maghreb. Palaeovertebrata 8 597–623. (1): 1–166, 7 pls. Lindsay, E.H. 1988. Cricetid rodents from Siwalik Jansa, S.A., F.K. Barker, and L.R. Heaney. 2006. deposits near Chinji Village. Part I: Megacrice- The pattern and timing of diversification of todontinae, Myocricetodontinae and Dendro- Philippine endemic rodents: evidence from murinae. Palaeovertebrata 18 (2): 95–154. mitochondrial and nuclear gene sequences. Lindsay, E.H. 1996. A new eumyarionine cricetid Systematic Biology 55 (1): 73–88. from Pakistan. Acta Zoologica Cracoviensia 39: Jansa, S.A., and M. Weksler. 2004. Phylogeny of 279–288. muroid rodents: relationships within and among Lindsay, E.H., L.J. Flynn, I.U. Cheema, J.C. major lineages as determined by IRBP gene Barry, K.F. Downing, A.R. Rajpar, and S.M. sequences. Molecular Phylogenetics and Evolu- Raza. 2005. Will Downs and the Zinda Pir tion 31: 256–276. Dome. Palaeontologia Electronica 8 (1) 19A: Kalthoff, D. 2000. Die Schmelzmikrostruktur in 1–19 (available at http://palaeo-electronica.org/ den Incisiven der hamsterartigen Nagetiere und toc.htm). anderer Myomorpha (Rodentia, Mammalia). Matthew, W.D., and W. Granger. 1923. New fossil Palaeontographica Abteilung A Palaeozoologie- mammals from the Pliocene of Sze-Chuan, Stratigraphie 259: 1–193. China. Bulletin of the American Museum of Klein Hofmeijer, G., and H. de Bruijn. 1985. The Natural History 48 (17): 563–598. mammals from the lower Miocene of Aliveri McKenna, M.C., and S.K. Bell. 1997. Classifica- (Island of Evia, Greece) – the Spalacidae and tion of mammals above the species level. New Anomalomyidae. Proceedings of the Konink- York: Columbia University Press. lijke Nederlandse Akademie van Wetenschap- Michaux, J., and F. Catzeflis. 2000. The bushlike pen Series B Physical Sciences 88: 185–198. radiation of muroid rodents is exemplified by Kowalski, K. 1968. Pararhizomys hipparionum the molecular phylogeny of the LCAT nuclear Teilhard and Young, 1931 (Rodentia) from the gene. Molecular Phylogenetics and Evolution Pliocene of Altan Teli, western Mongolia. 17: 280–293. Palaeontologia Polonica 19: 163–168. Miller, G.S. 1927. Revised determinations of some Kretzoi, M. 1961. Zwei Myospalaxiden aus dem Tertiary mammals from Mongolia. Palaeonto- Nordchina. Vertebrata Hungarica 3 (1–2): 123– logia Sinica C Series 5 (2): 1–20. 136. Miller, G.S., and J.W. Gidley. 1918. Synopsis of Laden, G., and R. Wrangham. 2005. The rise of the supergeneric groups of rodents. Journal of the hominids as an adaptive shift in fallback the Washington Academy of Science 8: 431–448. foods: plant underground storage organs Musser, G.G., and M.D. Carleton. 2005. Super- (USOs) and australopith origins. Journal of family Muroidea. In D.E. Wilson and D.M. Human Evolution 49: 482–498. Reeder (editors), Mammal species of the Laxmann, M.E. 1769. Sibirische Briefe: 74–77. world: a taxonomic and geographic reference, Go¨ttingen: Gotha. 3rd ed.: 894–1531. Baltimore, MD: Johns Lazzari, V., and J.P. Aguilar. 2007. First Hopkins University Press. occurrence in Europe of Myocricetodontinae Musser, G.G., A.L. Smith, M.F. Robinson, and (Rodentia, Gerbillidae) during the lower D.P. Lunde. 2005. Description of a new genus Middle Miocene in the karstic locality of and species of rodent (Murinae, Muridae, Blanquete`re1 (southern France): implications. Rodentia) from the Khammouan Limestone 2009 FLYNN: ANTIQUITY OF RHIZOMYS 153

National Biodiversity Conservation Area in Lao U¨ nay, E. 1999. Family Spalacidae. In G.E. PDR. American Museum Novitates 3497: 1–31. Ro¨ssner and K. Heissig (editors), The Miocene Nevo, E. 1999. Mosaic evolution of subterranean land mammals of Europe: 389–394. Munich: mammals: regression, progression, and global Pfeil. convergence. Oxford: Oxford University Press. U¨ nay-Bayraktar, E. 1989. Rodents from the Norris, R.W., K. Zhou, C. Zhou, G. Yang, C.W. Middle Oligocene of Turkish Thrace. Utrecht Kilpatrick, and R.L. Honeycutt. 2004. The Micropaleontological Bulletins Special Publica- phylogenetic position of the zokors (Myospala- tion 5: 389–394. Netherlands. cinae) and comments on the families of muroids Vianey-Liaud, M. 1985. Possible evolutionary (Rodentia). Molecular Phylogenetics and Evo- relationships among Eocene to lower Oligocene lution 31: 972–978. rodents of Asia, Europe, and North America. In Potapova, E.G., and N.N. Vorontsov. 2004. W.P. Luckett and J.L. Hartenberger (editors), Taxonomic position of the genus Tachyoryctes Evolutionary relationships among rodents: a and mutual relations between Rhizomyidae and multidisciplinary analysis: 277–309. New York: Spalacidae families (Rodentia). Zoologicheski˘l Plenum. Zhurnal 83 (8): 1044–1058. [in Russian] Wahlert, J.H. 1985. Skull morphology and rela- Prasad, K.N. 1968. The vertebrate fauna from the tionships of geomyoid rodents. American Mu- Siwalik beds of Haritalyangar, Himachal Pra- seum Novitates 2812: 1–20. desh, India. Palaeontologia Indica New Series Wang, X., Z.-d. Qiu, and N.D. Opdyke. 2003. 39: 1–56. Litho-, bio-, and magnetostratigraphy and Qiu, Z.-d. 1996. Middle Miocene micromamma- paleoenvironment of Tunggur Formation (Mid- lian fauna from Tunggur, Nei Mongol. Beijing: dle Miocene) in central Inner Mongolia, China. Academic Press. [in Chinese, English summary] American Museum Novitates 3411: 1–31. Rummel, M. 1999. Tribe Cricetodontini. In G.E. Wang, Y.-x. 2003. A complete checklist of Ro¨ssner and K. Heissig (editors), The Miocene mammal species and subspecies in China: a land mammals of Europe: 359–364. Munich: taxonomic and geographic reference. Beijing: Pfeil. China Forestry Publishing House. [in Chinese Sarich, V.M. 1985. Rodent macromolecular sys- and English]. tematics. In W.P. Luckett and J.-L. Hartenber- Wei, G., Y. Kowamura, and C. Jin. 2004. A new ger (editors), Evolutionary relationships among bamboo rat from the Early Pleistocene of rodents: a multidisciplinary analysis: 423–452. Renzidong Cave in Fanchang, Anhui, central New York: Plenum. China. Quaternary Research 43 (1): 49–62. Sarica, N., and S¸. S¸en. 2003. Spalacidae (Roden- Welcomme, J.-L., L. Marivaux, P.-O. Antoine, tia). In M. Fortelius, J. Kappelman, S¸. S¸en, and and M. Bennami. 1999. Mammife`res fossiles des R.L. Bernor (editors), Geology and Paleontolo- collines Bugti (Balochistan, Pakistan): nouvelles gy of the Miocene Sinap Formation, Turkey: donne´es. Bulletin de la Societe´ d’Histoire 141–162. New York: Columbia University Press. Naturelle de Toulouse 135: 135–139. S¸en, S¸. 1977. La faune de Rongeurs plioce`nes de Wessels, W. 1996. Myocricetodontinae from the C¸ alta (Ankara, Turquie). Bulletin du Muse´um Miocene of Pakistan. Proceedings of the Ko- National d’Histoire Naturelle Se´rie 3, 465: ninklijke Nederlandse Akademie van We- 89–171. tenschappen 99: 253–312. Steppan, S.J., R.M. Adkins, and J. Anderson. Wessels, W. 1999. Family Gerbillidae. In G.E. 2004. Phylogeny and divergence-date estimates Ro¨ssner and K. Heissig (editors), The Miocene of rapid radiations in muroid rodents based on land mammals of Europe: 395–400. Munich: multiple nuclear genes. Systematic Biology 53 Pfeil. (4): 533–553. Wessels, W., and H. de Bruijn. 2001. Rhizomyidae Teilhard de Chardin, P. 1926. Mammife`res ter- from the Lower Manchar Formation (Miocene, tiaires de Chine et de Mongolie. Annales de Pakistan). Annals of Carnegie Museum 70: Pale´ontologie 15: 1–51. 143–168. Teilhard de Chardin, P. 1942. New rodents of the Winge, H. 1887. Jordfundne og nulevende Gna- Pliocene and lower Pleistocene of north China. vere (Rodentia) fra Lagoa Santa, Minas Geraes, Publications of the Institut Ge´o-biologie 9: Brasilien. E Museo Lundii, University of 1–101. Copenhagen 1: 1–178. U¨ nay, E. 1996. On fossil Spalacidae (Rodentia). In Young, C.C. 1927. Fossile Naetiere aus Nord- R.L. Bernor, V. Fahlbusch, and H.W. Mitt- China. Palaeontologia Sinica Series C 5 (3): mann (editors), The evolution of Western 1–82, + 3 pls. Eurasian neogene mammal faunas: 246–252. Young, C.C., and Liu, P.T. 1950. On the New York: Columbia University Press. mammalian fauna at Koloshan near Chung- 154 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 331

king, Szechuan. Bulletin of the Geological Zheng, S.-h. 1997. Evolution of the Mesosiphnei- Society of China 30: 43–90. nae (Siphneidae, Rodentia) and environmental Zhang, Z.-q., L.J. Flynn, and Z.-d. Qiu. 2005. New change. In Y. Tong, Y. Zhang, W. Wu, J. Li, materials of Pararhizomys from northern China. and L. Shi (editors), Evidence for evolution— Palaeontologia Electronica 8 (1) 5A: 1–9 (avail- essays in honor of Prof. Chungchien Young on able at http://palaeo-electronica.org/toc.htm). the hundredth anniversary of his birth: 137–150. Zheng, S.-h. 1980. The Hipparion fauna from the Beijing: China Ocean Press. [in Chinese with Bulong Basin of Biru County, Tibet. Paleontol- English summary] ogy of Tibet 1: 33–47. Beijing: Science Press. [in Zheng, S.-h. 2004. Mammalia: Rodentia. In S.-h. Chinese, English summary] Zheng (editor), Jianshi hominid site—state key Zheng, S.-h. 1993. Quaternary rodents of Sichuan- project of the 9th five year plan—origin of early Guizhou area, China. Beijing: Kexue Chu- humans and environmental background. Series banshe. [in Chinese, English summary] monograph 1: 121–181. Beijing: Science Press. Zheng, S.-h. 1994. Classification and evolution of [in Chinese, English summary] the Siphneidae. In Y. Tomida, C.K. Li, and T. Zheng, S.-h., Z.-q. Zhang, and N. Cui. 2004. On Setoguchi (editors), Rodent and lagomorph some species of Prosiphneus (Siphneidae, Ro- families of Asian origins and diversification. dentia) and the origin of Siphneidae. Vertebrata Tokyo National Science Museum Monographs Palasiatica 42 (4): 297–315. [in Chinese, English 8: 57–76. summary]

APPENDIX 1 D. Eucricetodon presents moderate zygomatic plate inclination, and is thus modernized; this is also the first taxon to show the widespread double NOTES FOR CLADISTIC RELATIONSHIPS OF ridges on lower incisors. 2 SELECTED MUROIDEA E. Eumyarion plots as the immediate sister taxon to all living spalacids; here I extend the strong case NODES: of de Bruijin and Sarac¸ (1991) for affinity of Eumyarion with anomalomyines, to affinity with A. As outgroup to Muroidea, Dipodoidea illustrate all spalacids. Within Spalacidae, anomalomyines character polarity (P4 present, hystricomorphy, are close to spalacines, and myospalacines share poorly developed anterior cusps on first molars). features with rhizomyines. These two superfamilies constitute a well-defend- F. Cricetodon and the closely allied Deperetomys ed infraorder Myodonta Schaub (see, e.g., show a more modern zygoma than do Eucriceto- Vianey-Liaud, 1985). don or Eumyarion: more inclined plate with a ventral trough or slit. This condition is observed B. This polytomy does not differentiate between in nesomyids. The double ridge on the incisor is Cricetops, Pseudocricetodon, or the American retained. taxon Eumys. All three lack derived character G. Lartetomys is placed here to signify stable states in m1 morphology, zygomasseteric struc- presence of the new anterior connection of m1; ture, and incisor ornamentation, which are that is, a crest running from the anterior arm of tracked in more derived taxa. Cricetops appears the protoconid to the metaconid. to conserve the primitive zygomasseteric condi- H. Nesomyids present the modern m1, but variably tion of hystricomorphy; Pseudocricetodon has a developed myomorphy. Here the primitive char- weakly inclined masseteric plate (Lindsay, 1977), acter states for the group are interpreted to be and Eumys-Leidymys show a plate inclined m1 anterior connection present, one incisor ridge toward a very high infraorbital foramen (the retained, moderate myomorpy. The Malagasy height seems to be an autapomorphy, Harten- nesomyines retain the incisor ridge, but develop berger, 1967). myomorphy variably, some genera showing a C. Tachyoryctoides shows low zygomatic plate strongly inclined plate; continental cricetomyines inclination, hence weak myomorphy. Its flat lost the incisor ridge (a very weak ridge is incisor has no median ridges. These features are sometimes observed), but do not develop myo- taken to indicate a more basal position than morphy (some are nearly ‘‘hystricomorphous’’). Eucricetodon, but it is also conceivable that these I. Based on AMNH specimens of Typhlomys character states are overprinted by modifications cinereus, the inclined zygomatic plate is not and the position could be reversed with that of developed in keyhole fashion; the ventral slit is Eucricetodon. not compressed and there is no flange of the plate expanded anteriorly. Neither fossil nor living 2 Letters refer to comments for nodes and chosen taxa, and genera are known to show incisor enamel numerals represent key characters plotted on the cladogram. ornamentation (Kalthoff, 2000; personal com- This tree presents an explicit hypothesis for testing; it is not an mun.). Tentatively placed in a position more attempt at full phylogenetic analysis of the superfamily (see derived than nesomyids, but lower than calomys- fig. 6). cids, platacanthomyids have not yet been evalu- 2009 FLYNN: ANTIQUITY OF RHIZOMYS 155

ated with other muroids on the basis of molecular Myocricetodontines and gerbils present a char- data; they could lie outside nesomyids. The group acteristic pattern of alternating cusps in m1. displays extreme lophodonty. Special features of murines (and some gerbils and J. Calomyscidae is represented by Calomyscus, deomyines) include reduction of longitudinal which is of modern cricetid-murid grade. It is crests and realignment of the power stroke in modern in zygomasseteric structure (strongly chewing (Lazzari et al., 2008). Distinctive murine inclined plate, with ventral slit), in m1 morphol- molar morphology with stable accessory cusp ogy (anterior protoconid connection to metaco- pattern and propalinal mastication either ap- nid; metaconid relatively anterior in position), peared independently among certain deomyines, and in suppression of the incisor enamel orna- or this complex suite of features was simplified mentation (there remains a faint striation on C. secondarily in gerbils (and some deomyines). baluchi). Wessels (1999) lists Late Miocene records for the genus, based primarily on small CHARACTERS: third molars and lack of mesoloph/-id, and argues for gerbil relationship. However, the simple 1. Double anterocone on M1. Some muroids have anterocone/-id and strong longitudinal crest of transversely twinned anterocone cusps. This these fossils are not consistent with gerbil derived condition is found in isolated genera, relationship. Placement of Calomyscidae here plus a radiation of advanced muroids, including follows Jansa et al. (2006) and Steppan et al Muridae. (2004). 2. Double anteroconid on m1. This feature does K. Fossil Democricetodon represents stem cricetids. not occur without presence of character 1. Its myomorphy is advanced, with the ventral slit However, not all muroids with a double ante- of the infraorbital foramen pinched in keyhole rocone have a double anteroconid. fashion, and the M1 has an expanded antero- 3. Pinnate incisor ornamentation. Autapomorphy cone. Relationships of groups of cricetids are for Pseudocricetodon among genera surveyed informed by both the fossil record and molecular here. data. NWC (New World cricetids) have a late 4. Low inclination of zygomatic plate. Lindsay Early Miocene record, younger than the oldest (1977) cites a measure of 10 degrees, barely Democricetodon (Jacobs and Flynn, 2005). advanced over hystricomorphy. L. Megacricetodon is a derived muroid showing a 5. Zygomatic plate inclined to high infraorbital double anterocone on M1, an elongated ante- foramen. Hartenberger (1967) shows that the roconid on m1, and reduced mesoloph/-ids. Third precociously inclined zygomatic plate of Oligo- molars are generally reduced. These features cene North American genera reflects the high place Megacricetodon close to murid ancestry position of the infraorbital foramen. (Flynn et al., 1985), and as outgroup to murines 6. Central anterocone on M1. All muroids share a plus deomyines, could reasonably be seen as a distinct anterocone that is near the midline of the basal murid, not assigned to a subfamily. This tooth (not in extreme buccal position). hypothesis seemed weak in 1985, but Lazzari and Aguilar (2007) have documented high variability 7. Expanded anteroconid on m1. Dipodoids have a in large samples of Middle Miocene (ca. 16 Ma) small anteroconid (if any). Megacricetodon, including occasional presence of 8. P4 absent. Correlated with characters 6 and 7, accessory cusps in the positions of the classic muroids lack the small, peglike P4 of dipodoids. murid morphology (anterostyle on M1 lingual to 9. Moderate zygomatic plate inclination. Lindsay double anterocone, enterostyle adjacent to pro- (1977) measures 29u and higher for Eucricetodon tocone), which makes this genus a reasonable and other Miocene genera. murid outgroup. Potwarmus Lindsay (1988) 10. Double longitudinal ridge on the enamel of the shows M1 morphology near that of murids with lower incisor (present through node F). stable presence of the enterostyle, but m1 is 11. Anterocone on M1 expanded in transverse conservative; the genus would fall near node M anteroloph. on the cladogram. 12. Posterior connection on m1 weak or absent. The M. Jaeger (1977) laid the morphological groundwork widespread primitive condition of a crest joining for realization that gerbils had something to do the metaconid directly to the protoconid or its with murid origination. Inclusion of extinct posterior arm is reduced and lost in Early myocricetodontines among gerbils provided a Miocene lineages. Later changes include anterior deep fossil record to envision close relationship. shift of the metaconid cusp and construction of a Wessels (1996) documented the richness of new connection from the anterior arm of the myocricetodontine diversity, including taxa with protoconid to the metaconid. This new connec- accessory cusps; this Early and Middle Miocene tion is not the same as linkage to the metaconid radiation is consistent with closeness of true from the lingual side of the anteroconid. gerbils and the small Deomyinae group, includ- 13. Lophodonty. This condition unites spalacids, ing very murinelike genera. Sister status of but arose independently in other groups, includ- Deomyinae and Gerbillinae was found by ing arvicolines, gerbils, platacanthomyids, some Steppan et al. (2004) and Jansa et al. (2006). murines, and some New World cricetids; it is Sarich (1985) had discovered 20 years earlier that widespread but not universal in nesomyines, and the deomyine Acomys plotted outside murines. occurs in cricetomyines. 156 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 331

14. Reduced anterior cingulum on M2 and M3. The sion of the lateral masseter onto an expanded primitive cingulum is suppressed in many zygomatic plate; this plate is often built anteri- genera. Its suppression unites spalacids and is orly as a flange of bone lateral to the foramen. usual in later muroids. This condition is plotted for muroids based on 15. Infraorbital foramen expanded and origin of widespread distribution. It is important to note lateral masseter confined to ventral border of the high frequency of variant zygomatic struc- zygomatic arch. This condition appears to be a tures, some of which represent character rever- reversal from the trend of developing full sals. The feature is not distributed consistently myomorphy, and characterizes Spalacinae (un- even within small, well-defined clades. Within known in Anomalomyini) and several other cricetines (hamsters) for example, Cricetus shows living genera, e.g., Deomys. rather typical myomorphy, but Cricetulus lacks 16. Anterior connection on m1 from the anterior the zygomatic flange. arm of the protoconid to the metaconid. This 24. Prominent, elongated anteroconid on m1. condition characterizes derived muroids after Whereas the anterocone of M1 is expanded Node F, and appears to have been developed and prominent along much of this tree, the independently in Rhizomyinae + Myospalacinae. anteroconid is often small, and its longitudinal 17. Single ridge on lower incisor enamel. Reduction extent is not great. All murids have a prominent from two ridges to one is observed within the anteroconid and most have twinned cusps genus Kanisamys (Flynn, 1982) and is evident in (character 2). Nesomyinae. 25. Frequent accessory lingual cusps on upper 18. Absence of ridges on lower incisor. This trait molars, and buccal shelf with cusps on lower evolved in parallel in Spalacinae, Rhizomyini, molars. This appears occasionally in Megacrice- and Cricetomyinae. It would characterize all todon and Sindemys as a variable trait. higher muroids, except that some south Asian 26. Anterocone on M1 and accessory cusps simpli- Democricetodon show the double ridge. fied. In gerbils the M1 is simplified by variable 19. Infraorbital foramen transformed as a high, rounded opening in the anterior root of the reduction of the doubled anterocone and loss of zygoma. lingual cusps. 20. Zygomatic plate large and inclined. Cricetodon 27. Murine plan for first molars: M1 with lingual and later muroids are clearly more derived in cusps stable in position (anterostyle linked with myomorphy (relative size, anterior expansion, double anterocone, enterostyle always adjacent inclination) than any spalacid. Although zygo- to protocone) and m1 with X-shaped double matic plate size is difficult to evaluate given anteroconid-protoconid-metaconid). These fea- allometric scaling (Cricetodon is larger than tures are present in some deomyines (e.g., Eucricetodon) its prominent zygomatic plate Acomys). resembles that of nesomyines. 28. Anteroposterior power stroke in chewing. 21. Thick, wrinkled enamel on molars links the Most rodents, including many muroids, have clade of cricetodontine genera (Rummel, 1999). an oblique component to mastication, and 22. Reduced inclination of the zygomatic plate. chew preferentially with alternate jaws. A few Molecular data link nesomyines and criceto- lineages of rodents develop fully propalinal myines. Whereas nesomyines generally retain the mastication, nearly perfectly anteroposterior in zygomatic plate, it is reduced in most criceto- direction. This feature characterizes true mu- myines, greatly so in some, such as Steatomys. rines, and deomyines for the most part, but 23. Full myomorphy. Full myomorphy means pres- gerbils preserve an oblique chewing compo- ence of a keyhole infraobital foramen, the upper nent (Lazarri et al., 2008). Although propal- part of which transmits the medial masseter inal mastication arises in parallel among muscle onto the side of the snout, and the lower distantly related rodents, the character serves part of which is compressed by anterior expan- as a derived condition for murids.