Paleo Bios Museum of Paleontology University of California, Berkeley NUMBER 26 JUNE 20, 1977 A NEW SPECIES OF GREGORYMYS (RODENTIA, GEOMYIDAE) FROM THE MIOCENE OF COLORADO by JENS MUNTHE1 ABSTRACT Gregovymya larsoni n. sp. is a very late survivor of the Gveg~ orymys lineage. It is known only from the Derby Peak fauna of late Barstovian or early Clarendonian age. Radiometric dates ob- tained from volcanic rocks above and below the sediments from which the Derby Peak fauna was collected range from approximately 13 to 10 million years before present. G. larsoni is temporally dis- junct from all previously described species of the genus and may represent a population surviving only within a persistent refugium. INTRODUCTION A small collection of mammals was obtained from Derby Peak in Garfield County, Colorado by a University of Colorado party in 1967. The geology of the area was studied in subsequent years and radiometric ages were determined for rocks above and below the fos- siliferous sediments. Because the stage of evolution of one of the rodent taxa represented in the collection appears anomalous with respect to the remainder of the fauna and the radiometric dates, this rodent was studied in detail to determine its affinities. It is described here as a new species and the geologic and faunal ev- idence bearing on the age of the deposit is discussed. Measurements of the holotype were made with an EPOI Measuring Microscope and are shown in Table 1. Acronyms used are AMNH (¦ American Museum of Natural History), CM (= Carnegie Museum), FMNH (= Field Museum of Natural History), UCM <» University of Colorado Museum), and UCMP (= University of California Museum of Paleontology) Department of Paleontology, University of California, Berkeley, California 9^720. PROTOCONID METACONID ENTOCONID METALOPHID HYPOLOPHID Figure I. Cheek tooth cusp terminology for Gregovymys lavsoni. The major metalophid and hypolophid cusps occupy the same relative positions on the molars as on P^. The cheek tooth cusp terminology employed here (Fig. 1) is that of Rensberger (1971:110). The terminology for P^, however, is used only in a topographic sense. Cusp homologies are not implied. If ijovipnys is ultimately descended from Heliscom-is, which had only three cusps on Pi,, then one of Gregorymys' principal cusps is a neomorph not homologous with the four principal cusps of general- ized rodent teeth. Several attempts have been made to homologize geomyoid P^ cusps assuming that Hrliseomyo had lost one principal cusp {c. ;,-., Shotwell, 1967; Lindsay, 1972), but none of these is entirely satisfactory. I have therefore used names topographically for simplicity and clarity. SYSTEMATICS Superfamily Geomyoidea Weber, 1 go't Family Geomyidae Gill, 1872 Subfamily Entoptychinae Miller and Gidley, 1918 Gvegorymys Wood, 1936 Gvegoripnyo lat'soni new species (Figure 2) PALE0B10S NO. 26 ANTEROSTYLIDS PROTOSTYLID HYPOSTYLID HYPOCONID CINGULAR CUSP 1977 NEW GEOMYID RODENT FROM COLORADO 3 Figure 2. Gvegorymys larsoni n. sp., UCM 297^5- A. Lateral view. B. Lingual view. C. Occlusal view. D. Cross-section of incisor. Scale bars equal 1 mm. ll PALEOBIOS NO. 26 Holotype: UCM as?2*?, right mandibular ramus with I -M2. Hypod igm: Type only. Etymology: Patronym for Dr. Edwin E. Larson, who discovered the type locality and has been instrumental in deciphering the Ceno- zoic history of the area. Local ity: UCMP locality V76UI, Derby Peak, Garfield County, Co- lorado. Detailed locality information may be obtained from the Museum of Paleontology, University of California, or the University of Colorado Museum. Age: Late Barstovian or early Clarendonian. Diagnos i s: Very small species of Gvegovymys (Table 1); mandible shallow (Table 1); diastema short (Table 1); protoconids and hypo- conids weakly developed. Descript ion: The mandible is shallow for the genus. Its depth is 10% of P4-M3 length as opposed to 833; in AMNH 12895, referred by Wood (1936) to G. aurtus, and 107% in CM 8999, the type specimen of G. kayi (Wood, 1950). The diastema is shallow and much shorter than in any previously described specimen of Gregorymys. Its length is less than half the P^-M-j length, whereas the diastema is approximately two-thirds as long as the cheek tooth row in FMNH 12221, the type specimen of G. viggsi (Wood, 1936) and CM 8999- The masseteric crest is low but sharply defined. It extends in a smooth arc from below the posterior portion of the diastema, im- mediately posterior to the small mental foramen, to a point below the hypolophid of M2. A weak, rounded crest begins at the poster- ior end of the masseteric crest and follows the ventral margin of the incisor alveolus into the ascending ramus. This is distinctly different from the condition In AMNH 12895, where the masseteric crest is snarply angled and there is no crest posterior to it a- long the incisor alveolus. The angle is missing, but it extended anteriorly as far as Mj. The symphysis is rugose. A shallow geni- oglossal fossa lies immediately posterior to its dorsoventral mid- point. A narrow ventrolingual crest, presumably the area of in- sertion of M. digastricus, extends from the symphysis to a small ventral process below Pj,. A rugose swelling near the ventral mar- gin of the mandible below the hypolophid of Pi, may have been the area of origin of M. transversus mandibulae. A similar swelling occurs in some specimens of G. aureus. The prominent chin process seen in some specimens of G. aurtus is lacking. P/, has three anterostyl ids, of which the most lingual is the largest, arranged in an arc anterolinguad of the metalophid. The lingual anterostylid differs from that in other described specimens of Gregorymys in its greater size and more labial position. In this respect it most closely approaches the type specimen of G. riggsi. The metalophid has only a very weak protoconid that has not yet united with the protostylid. The hypolophid has equally 1977 NEW GEOMYID RODENT FROM COLORADO b developed entoconid and hypostylid but a weak hypoconid. The pos- terior margin of the hypolophid is straight, in contrast to the more arcuate hypolophids of other specimens, particularly AMNH 12909, referred to G. formosus by Wood (1936) and to G. eurtua by Macdonald (1970). With increased wear a small enamel lake would develop between the anterosty1 ids and the metalophid as in AMNH 1290S. At a still later wear stage, a lake would form between the metalophid and hypolophid as they would unite first labially and then lingually. This is a result of the weak development of the central cusp in each of these lophids. M| and M2 also show weak development of the protoconid and hy- poconid in contrast to other species of the genus. The cingular cusp of the metalophid is slightly lingual to the protostylid, causing the labial margin of the metalophid to lie oblique to the anteroposterior axis of the tooth row on both M] and M2. This is typical of M2 in other species of Gregorymys, but in most previ- ously described specimens the anterolabial cusp and protostylid are equally labial on M|. With increased wear the metalophids and hypolophids would unite medially and 1abially at almost exactly the same time. A very short-lived enamel lake would thus be form- ed on the labial half of the tooth. The lophids would then unite progressively more lingually. The lophid unions in FMNH 12221 (6". riggsi) would be similar, although in that specimen the first connection is definitely at the labial margin and the medial con- nection is slightly linguad of that in G. lavsoni. In all other species, the valley between the lophids is very deep in the center of the tooth so that a persistent central enamel lake is formed. Cement is present around the roots and slightly above on P^ and it extends well up on the crowns of Mj _2» particularly labially. This development of cement exceeds that seen in specimens of 6'. auPtus, is very similar to G. riggsi, and is less than in G. doug- lassi (Wood, 1936) and G. kayi (Wood, 1950). The lower incisor is broad, flat anteriorly, and triangular in cross-section. The enamel is thin and is almost entirely con- fined to the anterior surface. The triradiate pulp cavity is i- dentical to that in AMNH 12895 (G. auvtus). The root is not pre- served, but the incisor extended posterodorsally at least above the occlusal surfaces of the cheek teeth. The incisor is larger relative to the cheek teeth than in any other species of Gregory- mys (Table 1). Pi scussion: Certain aspects of geomyoid superspecific taxonomy are fundamental to an understanding of G. larsoni's relationships to other members of this superfamily. Shotwell (1967:10) included the subfamilies traditionally grouped as Geomyidae within Hetero- myidae. Rensberger (1971:151) advanced additional reasons for this arrangement and acknowledged the probable artificiality of the geomyid-heteromyid dichotomy, but retained the dichotomy in order to emphasize "the fossorial attributes of the Entoptychinae and Geomyinae" (1971:66). However, a recent study of the skeleton 6 PALEOBIOS NO. 26 Table 1. Measurements (in millimeters) of Gregorymys lavsoni and selected comparative measurements of other species. UCM 297*15 AMNH 12895 FMNH 12221 CM 8999 G. larsoni G. ourtus G. riggsi G. kayi P/j-Mo alveolar distance Pj, anteroposterior (occlusal) transverse metalophid transverse hypolophid Mj anteroposterior (occlusal) transverse metalophid transverse hypolophid M2 anteroposterior (occlusal) transverse metalophid transverse hypolophid Mj anteroposterior (alveolus) transverse (alveolus) I anteroposterior transverse anteroposter ior/ transverse enamel thickness Length of diastema Depth of mandible below Pi, of Schizodontomys (Munthe, 1975). an entoptychine senr.u Wood (1936), indicates that this animal was not fossorial, but semi- sal tatorial. Gregorymys is a pivotal genus with regard to familial taxonomy of these rodents. Galbreath (1967:278-280) has summarized many of the similarities of Gregorymys to geomyines on one hand and heteromyines on the other. He concluded that placing Gregorymys in either Geomyidae or Heteromyidae might obscure its true rela- tionships. This seems certainly to be the case even with the limited suite of morphologic characters available for comparison in this study. While its overall size, length of diastema, and size and shape of incisor argue for G. larsoni being a geomyid, its Pj, morphology, depth of mandible, and certain characters of the molars appear to place it in the Heteromyidae. Despite the fact that all the above arguments support the in- clusion of heteromyids and geomyids within a single family, such 6.60 1.52 1.47 1.60 1.33 1.64 1.62 1.41 1.60 1.66 1.24 1.29 1.42 1.53 0.928 0.03 3.17 4.62 2.31 1.75 7-30 6.70 7.45 2.39 1.96 5.05 6.50 7.80 2.30 1.99 5.10 8.00 1977 NEW GEOMYID RODENT FROM COLORADO 7 an arrangement is never likely to win general acceptance. Both family names are very solidly entrenched in the scientific liter- ature. The most compelling argument against fusing the two fami- lies is the ease with which their modern representatives may be morphologically differentiated, although, as noted by Wood (1937), this morphologic disparity is largely a reflection of different habits. Neozoologists would be understandably skeptical of plac- ing Thomomys and Dipodomys in the same family based solely on evi- dence provided by extinct taxa whose exact relationships to modern subfamilies are uncertain. With some misgivings, I have attempted to place 6". larsoni within the traditional geomyoid taxonomic framework. The subfamily assignment of Gregorynvjs to the Entoptychi nae is based on its close similarity to Entoptychus. However, the very prominent anterior cingulum on Gregorymys' Pi) clearly separates it from EntopU/ohus and has led some authors (Wood, 1936:4; Shotwel 1 , 1967:^8-49) to place it closer to the heteromyines than the ento- ptychines in dental anatomy. Reeder (1956:411) considered Gregory- mys to be a heteromyid and closely related to Heteromys on the ba- sis of similarities in their DPi,s. The same evidence, however, was used by Hibbard (1954:357) to suggest that Gregorymys might be an- cestral to the geomyines. Rensberger (1971), who considered more morphologic characters than any previous worker, retained Gregory- mys within the Entoptychinae and that conclusion is followed here. No genus other than Gregorymys is closely comparable to the Derby Peak specimen. The distinctive cusp arrangement of P^, the relative sizes of the cheek teeth, the development of cement on the cheek teeth, and the broad, triangular incisor are found in combination only in Gregorymys. Certain heteromyines, notably some specimens of Pevidiomys and Diprionomysi approach G. larsoni's peculiar method of uniting the lophids of Pi, more closely than pre- viously described species of Gregorymys, but characters of their mandibles and especially their incisors clearly set them apart from 0. "Larsoni. Similarly, the molar lophid connections of some heteromyines and some geomyines proceed as in G. larsoni, but char- acters either of the mandible-and incisor or of Pj, exclude these taxa from close relationship. In most respects, G. Larsoni is more similar to the type of G. riggsi than to the other previously described specimens of the genus. None of these specimens approach G. larsoni in overall size, length of diastema, or depth of mandible, but G. riggsi is similar in P^ morphology, sequence of lophid unification in M,,, height of crown, and development of cement. I consider all G. larsoni's diagnostic characters to be derived (autapomorphies). This certainly seems to be true with respect to commonality of characters, since all other described specimens of Gregorymys are alike in being much larger than G. larsoni and hav- ing deeper mandibles, longer diastemas, and more prominent proto- conids and hypoconids. G. larsoni's small dimensions would appear 8 PALEOBIOS NO. 26 to place it in a primitive (plesiomorphic) position with respect to the other species of the genus when size trends in other groups of mammals are considered. The trend toward larger size through phylogeny is observed frequently enough to be considered an evolu- tionary "law", and G. larsoni appears to violate this law. How- ever, some geomyoid rodents are notoriously variable in size. This is true of modern gophers and the sample of Greg* '. . ;: j'or- mosUB studied by Macdonald (1970) is adequate to establish that it is also true of Gregovymys. The lengths of four P^s from the same locality are recorded by'Macdonald (1970:37) as I. U, 1.52, 1.95, and 2.^6 mm. Given this magnitude of potential variability and the unusual biogeographic circumstances under which (',. larsoni may have evolved, its size seems less of an evolutionary anomaly, although it is presently impossible to determine whether the Derby Peak specimen is a small individual or a typical member of a dwarf taxon. It should be noted that the three dimensional diagnostic characters are probably at least partially correlated and are pos- sibly united in a single size-dependent functional-morphologic complex. They may therefore be considered aspects of a single feature, but each nevertheless serves to distinguish .'. larsoni. I do not propose to construct a formal phylogeny of Gr&gorymys or attempt to place G. larsoni within such a phylogeny at this time. The lack of detailed population studies, the poor strati- graphic control for many of the published specimens, and the abun- dance of unpublished material prevent an understanding of the in- terspecific relationships within the genus. In fact, the genus needs a revision. It has not been extensively studied since it was described in 1936 and several species evidently referable to the genus remain to be described (c. g., Robinson, 1968:195)- The only population of Gregorymys which has ever been studied consists of 15 specimens assigned to G. fovmosue by Macdonald (1970). This sample shows very great dimensional variation in tooth measurements and no statistical treatment was attempted, presumably because the sample was so small and variable, so that it does not provide an understanding of potential ranges in size or morphologic charac- ters within Gregorymys' populations. Large samples of Greg are evidently available for study {e. g., Macdonald, 1972:30), but such studies have not been undertaken. Storer (1975:105) suggests that Gregorymys may be ancestral to Ligninrus, a dominantly Clarendonian genus. Transitional forms which might support this relationship have not been reported, but it is at least clear that G. larsoni has nothing to do with a Gregorymys-Lignvnus morphocline. G. larsoni may be characterized as a persistent survivor of the Arikareean Gregorymys morphotype. AGE AND CORRELATION The Derby Peak fauna containing Gregorymys larsoni has not been studied in detail. It consists of Equidae, Rhinocerotidae, Cameli- dae, Antilocapridae, and Mylagaulus laevis (Larson, 1968; P. Robin- son, personal communication). These taxa are not particularly 1977 NEW GEOMYID RODENT FROM COLORADO 9 useful for correlation purposes, but the presence of antilocaprids and Mylagaulus precludes an age earlier than Hemingfordian. Morris Skinner of the American Museum of Natural History has examined a horse mandible from Derby Peak (UCM 297^7) and considers it most similar to specimens from the Lower Snake Creek fauna of Nebraska. Based upon the stage of evolution of this horse, he believes the Derby Peak fauna to be approximately 13-5 m.y. old (personal com- municat ion). K-Ar ages have been determined for many of the volcanic rocks interbedded with vertebrate-bearing sediments in the area surround- ing Derby Peak (Larson, 1968; Larson et al., 1975)- An age of 10.3 * .5 m.y. was obtained from the flow immediately above the unit con- taining the Derby Peak fauna. No rocks below the vertebrate local- ity on Derby Peak itself have been dated, but flows immediately un- derlying and interbedded with similar sediments containing poorly preserved vertebrate remains on "W" Mountain and Trapper's Peak nearby have produced dates. Their ages range from 12.k + .5 m.y. to 13-4 i -5 m.y. These ages indicate that the Derby Peak fauna lived between 13 and 10 million years ago. This faunal and radiometric age evidence strongly suggests that the Derby Peak fauna is equivalent to late Barstovian or early Cla- rendonian faunas in age (Evernden et al., 1964; Lindsay, 1972). Gregorymus has previously been reported only from late Arikareean and early Hemingfordian deposits. A single "entoptychine" tooth has been reported from sediments as young as Clarendonian or Hemp- hill ian in Wyoming (Love, 1961:35), but the data are insufficient to evaluate this occurrence. The species with the closest apparent relationship to G. larsoni, G. riggsi, is known only from "Arikaree Beds" of Wyoming. Therefore, G. larsoni is anomalous in stage of evolution and the possibility that the specimen was redeposited from older sediments must be considered. There are older fossili- ferous sediments in the area of Derby Peak. These are the Group 1 rocks of Larson et al. (1975), which range in age from 2k to 20 m. y. This age would be- consistent with the stage of evolution of G. larsoni and a number of unstudied specimens of Gpegovymys have been collected from deposits of this age east of Derby Peak (Robinson, 1968). However, there is no evidence that the specimen of G. lar- soni was redeposited. It shows the same type of preservation and does not seem to be more abraded than the other specimens from Derby Peak, which all were apparently transported prior to depo- si tion. BI0GE0GRAPHY The Gregorymys lineage, based upon its previously recorded oc- currences, appeared to have become extinct by middle Hemingfordian time, approximately 18 million years ago. Its presence in the Der- by Peak fauna is unexpected. Persistence of a primitive morpholo- gical type such as G. larsoni could occur through isolating mechan- isms which protect it from competition. Topographic isolation in a high-altitude mountain basin may be suggested as such a mechanism 10 PALEOBIOS NO. 26 in this case, but the geologic evidence for isolation of this kind is not strong- The existence of isolated early Miocene deposition- al basins in northwestern Colorado is well documented, but the per- iod of volcanism during which the Derby Peak sediments were deposi- ted seems to have been characterized by widespread steppes which connected previously isolated basins and were inhabited by a uni- form biota (Izett, 1975; Larson ct at., 1975; earlier work cited therein). The widespread tectonism which resulted in the present intermontane basins of northwestern Colorado did not begin until nine or ten million years ago. However, the White River Plateau, of which Derby Peak is presently an outlier, has been a positive structural and topographic feature since its origin in the late Cretaceous. Kucera's (1968) studies show that it was shedding sed- iments into the basins to the north during deposition of the Derby Peak sediments. Furthermore, his observations on the depositional characteristics of the basalt flows in the Derby Peak area indicate that many of the flows are very localized. It is therefore possi- ble that Gregorymys laraont could have inhabited one or several very small intermontane basins where it was effectively isolated from competitors. It is interesting to note that Robinson (1968) has observed spe- ciation of early Miocene entoptychines in the area east of Derby Peak which was apparently at least partially controlled by the same isolated intermontane basin type of topography suggested above. If a population of the type mentioned by Robinson became isolated in a high valley on the slopes of the White River Plateau, and if that valley provided a refugium while the lower basins to the east grad- ually fused, bringing their previously isolated entoptychines and other later-evolved rodents into competition with one another, the population remaining in the high valley might well develop into the sort of taxonomic compromise represented by G. larsoni. ACKNOWLEDGEMENTS I thank Dr. Peter Robinson for allowing me to study the Derby Peak material in the Colorado University Museum collection. Dr. John Wahlert made type specimens of Gregorymys in the collections of the American Museum of Natural History available for study. Drs. James Patton and Albert E. Wood read preliminary versions of the manuscript and suggested improvements incorporated here. Mrs. Pat Robinson provided the specimen illustrations. LITERATURE CITED Evernden, J. F., D. E. Savage, G. H. Curtis, and G. T. James. 196^4. Potassium-argon dates and the Cenozoic mammalian chronology of North America. Amer. J. Sci., 262:1^5-198. Galbreath, E. C. 1967- A skeleton of the geomyoid rodent, Grego- rymys eurtus (Matthew) from the early Miocene of South Dakota. Trans. 111inois Acad. Sci., 60:272-281. Hibbard, C. W. 1951*- A new Pliocene vertebrate fauna from Oklahoma. Papers Michigan Acad. Sci., Arts and Ltrs., 39:339-359- 1977 NEW GEOMYID RODENT FROM COLORADO II Izett, G. A. 1975- Late Cenozoic sedimentation and deformation in northern Colorado and adjoining areas. Mem. Geol . Soc. Amer., 144:179-209. Kucera, R. E. 1968. Geomorphic relationship of Miocene deposits in the Yampa district, northwesl Colorado. Pp. I 16-134, in Field conference guidebook for the high altitude and mountain basin deposits of Miocene age in Wyoming and Colorado. Univ. Colorado Museum, Boulder, 209 PP- Larson, E. E. 1968. Miocene and Pliocene rocks in the Flat Tops Primitive Area. Pp. 135-144, in Field conference guidebook for the high altitude and mountain basin deposits of Miocene age in Wyoming and Colorado, Univ. Colorado Museum, Boulder, 209 PP- Larson, E. E., M. Ozima, and W. C. Bradley. 1975. Late Cenozoic basic volcanism in northwestern Colorado and its implications concerning tectonism and the origin of the Colorado River sys- tem. Mem. Geol. Soc. Amer., 144:155-178. Lindsay, E. H. 1972. Small mammal fossils from the Barstow For- mation, California. Univ. California Publ. Geol. Sci. , 93:1-104. Love, J. D. 1961. Split Rock Formation (Miocene) and Moonstone Formation (Pliocene) in central Wyoming. U. S. Geol. Sur. Bull., 1121-1:I 1 - 139- Munthe, L. K. 1975- The osteology of the Miocene rodent Sohizo- dontomys. Unpublished M.A. thesis, Univ. California, Berkeley, 110 pp. Macdonald, J. R. 1970. Review of the Miocene Wounded Knee faunas of southwestern South Dakota. Bull. Los Angeles County Museum Nat. Hist., No. 8:1-82. Macdonald, L. J. 1972. Monroe Creek (early Miocene) microfossils from the Wounded Knee area. South Dakota. South Dakota Geol. Sur. Report of Investigations, No. 105:1-43. Reeder, W. G. 1956. A review of Tertiary rodents of the family Heteromyidae. Unpublished Ph.D. dissertation, Univ. Michigan, Ann Arbor, 618 pp. Rensberger, J. M. 1971- Entoptychine pocket gophers (Mammalia, Geomyoidea) of the Early Miocene John Day Formation, Oregon. Univ. California Publ. Geol. Sci., 90:1-209. Robinson, P. 1963. Comments on the smaller mammals of Miocene age from Middle Park, Colorado. Pp. 194-203, in Field conference guidebook for the high altitude and mountain basin deposits of Miocene age in Wyoming and Colorado. Univ. Colorado Museum, Boulder, 209 PP- Shotwell, J. A. 1967- Late Tertiary geomyoid rodents of Oregon. Bull- Univ. Oregon Mus. Nat. Hist., 9:1-51- Storer, J. E. 1975- Tertiary mammals of Saskatchewan Part III: The Miocene fauna. Life Sci. Contrib. Royal Ontario Mus., 103:1-134. Wood, A. E. 1936. Geomyoid rodents from the middle Tertiary. Amer. Mus. Novitates, 866:1-31. ___________- 1937- Parallel radiation among the geomyoid rodents. J. Mamm., 18:171-176. 12 PALEOBIOS NO. 26 Wood, A. E. 1950. A new geomyid rodent from the Miocene of Mon- tana. Ann. Carnegie Mus., 31:335-338. Manuscript received April 25, 1977", accepted June k, 1977- Manuscript reviewed by W. A. Clemens, J. M. Rensberger and P. Robinson.