Palaeoecology and Palaeoenvironments of Late Cenozoic Mammals: Tributes to the Career of C.S. (Rufus) Churcher 9781487574154

Citation preview

Palaeoecology and Palaeoenvironments of Late Cenozoic Mammals: Tributes to the Career of C.S. (Rufus) Churcher

C.S. (Rufus) Churcher

Palaeoecology and Palaeoenvironments of Late Cenozoic Mammals Tributes to the Career of C.S. (Rufus) Churcher

Edited by Kathlyn M. Stewart and Kevin L. Seymour

UNIVERSITY OF TORONTO PRESS Toronto Buffalo London

© University of Toronto Press Incorporated 1996 Toronto Buffalo London Printed in Canada Reprinted in 2018 ISBN 0-8020-0728-7 ISBN 978-1-4875-8559-4 (paper) Publication of this volume was made possible by a grant from the Canadian Geological Foundation.

Printed on acid-free paper

Canadian Cataloguing in Publication Data Main entry under title: Palaeoecology and palaeoenvironments of late Cenozoic mammals : tributes to the career of C.S. (Rufus) Churcher ISBN 0-8020-0728-7 1. Churcher, C.S., 1928- . 2. Paleontology - Pleistocene. 3. Paleoecology- Pleistocene. I. Churcher, C.S., 1928- . II. Stewart, Kathlyn Moore, 1952- . III. Seymour, Kevin L. (Kevin Lloyd). QE741.2.P3 1995

560'.178

C95-931846-1

University of Toronto Press acknowledges the financial assistance to its publishing program of the Canada Council and the Ontario Arts Council.

Contents

Contributors / ix Preface/ xv C.S. 'Rufus' Churcher - The man from Aldershot / xvii K.L. Seymour Churcher bibliography / xxii K.L. Seymour

North American Quaternary Mammals:

Palaeoecology and Palaeoenvironments

Diversity bottlenecks, oddball survivors, and negative keys / 3 WA. Akersten Comparison of mammalian response to glacial-interglacial transitions in the middle and late Pleistocene/ 16 A.D. Barnosky, T.l. Rouse, E.A. Hadly, D.L. Wood, F.L. Keesing, and V.A. Schmidt Review of Pleistocene zoogeography of prairie dogs (genus Cynomys) in western Canada with notes on their burrow architecture / 34 J.A. Burns

vi Contents Pleistocene and Holocene vertebrates as models for the study of evolutionary patterns / 54

R.L. Carroll Selective mortality of mastodons (Mammut americanum) from the Port Kennedy Cave (Pleistocene; lrvingtonian), Montgomery County, Pennsylvania / 83 E.B. Daeschler Middle Pleistocene (early Rancholabrean) vertebrates and associated marine and non-marine invertebrates from Oldsmar, Pinellas County, Florida / 97 P.F. Karrow, G.S. Morgan, R. W. Portell, E. Simons, and K. Auffenberg Population structure of the late Pliocene (Blancan) zebra Equus simplicidens (Perissodactyla: Equidae) from the Hagerman Horse Quarry, Idaho / 134

H.G. McDonald Current status of North American Sangamonian local faunas and vertebrate taxa / 156

J.D. Pinsof Distribution and size variation in North American Short-faced bears,

Arctodus simus I 191 R.L. Richards, C.S. Churcher, and W.D. Turnbull

Origin of the vertebrate fossil sites near Medicine Hat, Alberta / 247 A. Macs. Stalker

North American Quaternary Mammals: Faunas and Morphological Analyses

A preliminary report on the Camivora of Porcupine Cave, Park County, Colorado / 259

E. Anderson Force generation by the jaw adductor musculature at different gapes in the Pleistocene sabretoothed felid Smilodon I 283

H.N. Bryant

A review of Pleistocene giant armadillos (Mammalia, Xenarthra, Pampatheriidae) / 300

A.G. Edmund

Contents vii A spectacular specimen of the elk-moose Cervalces scotti from Noble County, Indiana, U.S.A. / 322 J.O. Farlow and J. McClain Tracking Ice Age felids: Identification of tracks of Panthera atrox from a cave in southern Missouri, U.S.A. / 331 R. W. Graham, J.O. Farlow, and J.E . Vandike Pleistocene mammals of Dublin Gulch and the Mayo District, Yukon Territory / 346 C.R. Harington The masticatory apparatus of the American mastodon (Mammut americanum) I 375 R.S. Laub Pleistocene caribou (Rangifer tarandus) in the eastern United States: New records and range extensions/ 406 J.N. McDonald, C.E. Ray, and F. Grady Dental evolution and size change in the North American muskrat: Classification and tempo of a presumed phyletic sequence/ 431 R.A. Martin Early Rancholabrean mammals from Salamander Cave, Black Hills, South Dakota / 458 J.I. Mead, C. Manganaro, C.A. Repenning, and L.D. Agenbroad Late and middle Pleistocene vertebrate fossils from Old Crow Basin, Locality CRH 15, northern Yukon Territory / 483 R.E. Morlan

Late Cenozoic Mammals of Africa Palaeoecology and Palaeoenvironments

Antlers in America: Ossicones in Africa / 522 E.H. Colbert Sexual dimorphism in Antidorcas recki from Bolt's Farm, South Africa, in the University of California collections / 537 H.B .S. Cooke A review of Dietrich's hipparions from South Serengeti (Tanzania) and a comparison with similar materials / 554 A. Forsten

viii Contents A fossil Budorcas (Mammalia, Bovidae) from Africa / 571 A.W. Gentry Basicranial anatomy of the giant viverrid from 'E' Quarry, Langebaanweg, South Africa /588 R.M.Hunt,Jr The identification of Equus skulls to species, with particular reference to the craniometric and systematic affinities of the extinct South African quagga / 598 R.G. Klein and K. Cruz-Uribe Temporal variability in horn-core dimensions of Damaliscus niro from Olduvai, Sterkfontein, Cornelia, and Florisbad / 631 J.F. Thackeray, J.S. Brink, and I. Plug The fossil and living Hyaenidae of Africa: Present status / 637 L. Werdelin and A. Turner Is the rodent Acomys a murine? An evaluation using morphometric techniques / 660 X. Xu, A.J. Winkler, and L.L. Jacobs

Contributors

Larry D. Agenbroad Quaternary Studies Program Department of Geology P.O. Box 4099 Northern Arizona University Flagstaff, Arizona 86011 USA

Anthony D. Bamosky Mountain Research Center Montana State University Bozeman, Montana 59717-0348 USA

William A. Akersten Idaho Museum of Natural History Idaho State University Campus Box 8096 Pocatello, Idaho 83209 USA

J.S. Brink National Museum P.O. Box266 Bloemfontein 9300 South Africa

Elaine Anderson Department of Earth Sciences Denver Museum of Natural History 2001 Colorado Boulevard Denver, Colorado 80205-5798 USA Kurt Auffenberg Florida Museum of Natural History University of Florida Gainesville, Florida 32611-2035 USA

Harold N. Bryant Mammalogy Program Provincial Museum of Alberta 12845-102 Avenue Edmonton, Alberta Canada T5N0M6 James A. Bums Department of Quaternary Paleontology Provincial Museum of Alberta 12845-102 Avenue Edmonton, Alberta Canada T5N0M6

x Contributors Robert L. Carroll Department of Biology McGill University 1205 Docteur Penfield Avenue Montreal, Quebec Canada H3A lBl C.S. Churcher R.R. 1, Site 42, Box 12

Gabriola Island, British Columbia Canada V0R lX0 Edwin H. Colbert Museum of Northern Arizona Route 4, Box 720 Flagstaff, Arizona 86001 USA H.B.S. Cooke 2133 154th Street White Rock, British Columbia Canada V4A 455 Kathryn Cruz-Uribe Department of Anthropology Northern Arizona University Box 15200 Flagstaff, Arizona 86011 USA Edward B. Daeschler Department of Vertebrate Biology Academy of Natural Sciences 1900 Benjamin Franklin Parkway Philadelphia, Pennsylvania 19103-1195 USA A. Gordon Edmund Department of Geological Sciences University of Toronto Toronto, Ontario Canada M5S 3B3 James 0 . Farlow Department of Geosciences Indiana-Purdue University 2101 Coliseum Boulevard East Fort Wayne, Indiana 46805-1499 USA

Ann Forsten Zoological Museum P.B. 17 (P. Rautatiekatu 13) FIN-00014 University of Helsinki Helsinki, Finland Alan W. Gentry

cl o Department of Palaeontology

Natural History Museum Cromwell Road London, SW7 5BD, England

Frederick Grady Department of Paleobiology National Museum of Natural History Washington, DC 20560 USA Russell W. Graham Research and Collections Center Illinois State Museum 1011 East Ash Springfield, Illinois 62703 USA Elizabeth A. Hadly Department of Biology Montana State University Bozeman, Montana 59717 USA C.R. Harington Department of Paleobiology Canadian Museum of Nature P.O. Box 3443, Station D Ottawa, Ontario Canada KlP 6P4 Robert M. Hunt, Jr Division of Vertebrate Paleontology University of Nebraska State Museum Lincoln, Nebraska 68588-0549 USA

Contributors xi Louis L. Jacobs Shuler Museum of Paleontology Department of Geological Sciences Southern Methodist University P.O. Box 750395 Dallas, Texas 75275 USA Paul F. Karrow Department of Earth Sciences University of Waterloo Waterloo, Ontario Canada N2L3Gl Felicia L. Keesing Department of Integrative Biology University of California Berkeley, California 94720 USA

H. Gregory McDonald Hagerman Fossil Beds National Monument P.O. Box570 Hagerman, Idaho 83332 USA Jerry N. McDonald McDonald & Woodward Publishing Co. 6414 Riverland Drive Fort Pierce, Florida 34982 USA Robert A. Martin Department of Biological Sciences Murray State University Murray, Kentucky 42701 USA

Richard G. Klein Department of Anthropology Stanford University Stanford, California 94305-2145 USA

Jim I. Mead Quaternary Studies Program Department of Geology P.O. Box 4099 Northern Arizona University Flagstaff, Arizona 86011 USA

Richard S. Laub Division of Geology Buffalo Museum of Science 1020 Humboldt Parkway Buffalo, New York 14211-1293 USA

Gary S. Morgan New Mexico Museum of Natural History 1801 Mountain Road, NW Albuquerque, New Mexico 87104 USA

Carol Manganaro Quaternary Studies Program Department of Geology P.O. Box 4099 Northern Arizona University Flagstaff, Arizona 86011 USA

Richard E. Morlan Canadian Museum of Civilization P.O. Box 3100, Station B Hull, Quebec Canada J8X 4H2

Jana McClain Department of Geology Northern Illinois University Dekalb, Illinois 60115 USA

John D. Pinsof Department of Natural Sciences Daemen College 4380 Main Street Amherst, New York 14226-3592 USA

xii Contributors Ina Plug Transvaal Museum P.O. Box 413 Pretoria 0001, South Africa Roger W. Portell Florida Museum of Natural History University of Florida Gainesville, Florida 32611-2035 USA Clayton E. Ray Department of Paleobiology National Museum of Natural History Washington, DC 20560 USA Charles A. Repenning United States Geological Survey Stratigraphy and Paleontology, MS919 Denver Federal Center Denver, Colorado 80225 USA Ronald L. Richards Department of Natural Resources Indiana State Museum 202 North Alabama Street Indianapolis, Indiana 46204 USA Tina I. Rouse Department of Integrative Biology University of California Berkeley, California 94720 USA Victor A. Schmidt Department of Geology University of Pittsburgh Pittsburgh, Pennsylvania 15260 USA

Kevin L. Seymour Department of Vertebrate Palaeontology Royal Ontario Museum 100 Queen's Park Toronto, Ontario Canada M5S 2C6 Erika Simons Florida Museum of Natural History University of Florida Gainesville, Florida 32611-2035 USA A. Macs. Stalker 2126 Strathmore Boulevard Ottawa, Ontario Canada K2A 1M7 J. Francis Thackeray Department of Paleontology Transvaal Museum P.O. Box413 Pretoria 0001, South Africa William D. Turnbull Field Museum of Natural History Roosevelt Road at Lakeshore Drive Chicago, Illinois 60605 USA Alan Turner Department of Human Anatomy University of Liverpool P.O. Box 147 Liverpool, L69 3BX, England James E. Vandike Department of Natural Resources Division of Geology and Land Survey P.O. Box250 Rolla, Missouri 65401 USA

Contributors xiii Lars Werdelin Section of Palaeozoology Swedish Museum of Natural History Box 50007 S-104 05 Stockholm, Sweden

XiaofengXu Shuler Museum of Paleontology Department of Geological Sciences Southern Methodist University Dallas, Texas 75275 USA

Editors Alisa J. Winkler Shuler Museum of Paleontology Department of Geological Sciences Southern Methodist University P.O. Box 750395 Dallas, Texas 75275-0395 USA

David L. Wood VA Medical Center (151) 3350 La Jolla Village Drive San Diego, California 92161 USA

Kathlyn M. Stewart Deparhnent of Paleobiology Canadian Museum of Nature P.O. Box 3443, Station D Ottawa, Ontario Canada KlP 6P4 Kevin L. Seymour Department of Vertebrate Palaeontology Royal Ontario Museum 100 Queen's Park Toronto, Ontario Canada M5S 2C6

Preface

This volume was initiated as a tribute to our mentor, colleague, and friend, Rufus Churcher. It was originally intended to coincide with his retirement, but as these projects inevitably take more time than expected, it is appearing somewhat later. This, it turns out, is ironically appropriate, as Rufus's 'retirement' so far has not appeared to us to have affected his activities in the least. He has been teaching classes, publishing papers, doing fieldwork, and supervising students. His life work goes on. Rufus's career so far has spanned 40 years and three continents, as well as a diversity of research interests. When we first conceived the idea of a festschrift volume for Rufus, we sent out letters to his colleagues asking for titles. We were overwhelmed by both the number of titles we received (65) and the range of interests represented, which of course reflect Rufus's diverse interests. Sheer practicality of publishing the volume forced us to restrict both the scope and the number of papers. We chose to focus on Rufus's principal research interest, Quaternary mammals, in the two continents where he has conducted the vast majority of his research, North America and Africa. This focus is reflected in the three sections of the volume. After a brief biography and bibliography, the first two sections contain papers on Quaternary mammals in North America, divided into (1) palaeoecology and palaeoenvironments and (2) faunas and morphological analyses. The final section contains nine papers on Late Cenozoic African mammals. In the African section, Alan Gentry pays tribute to Rufus by naming a species after him: Budorcas churcheri. Altogether we have brought together 30 contributions from 53 colleagues. It takes many people to produce a volume of this magnitude. In particular, we would like to thank the Canadian Geological Foundation for a timely grant which enabled this project to get off the ground. Additional financial support from the Department of Vertebrate Palaeontology

xvi Preface at the Royal Ontario Museum was also much appreciated. Joan Burke, Sandra Shaul, and Glen Ellis, all at the ROM, and Bonnie Livingstone of the Canadian Museum of Nature provided advice and encouragement. We are very grateful to Donna Naughton (CMN) for her help in assembling and editing the volume. We thank Harold Bryant, Jim Bums, Gerry De Iuliis, Dick Harington, Chris McGowan, and Hans-Dieter Sues for helpful discussions. We are also very grateful to the numerous reviewers who gave their time to critique papers for the volume. The success of this project was due in large part to the support and encouragement of a great number of people along the way, including D.M. Avery, A. Azzaroli, 0. Bar-Yosef, C.K. Brain, P.M. Butler, K.E. Campbell Jr, J. Clutton-Brock, W.W. Dalquest, G. De luliis, T. Downs, V. Eisenmann, L. Flynn, A. Gautier, D.D. Gillette, F. Grady, J.A. Holman, G. Hurlburt, H. Johnson, A. Kemp, A.C. Kitchener, G.E. Lammers, M.G. Leakey, J.A. Lillegraven, D.G. Matthiesen, J.E. Mawby, W.E. Miller, V.L. Naples, S.J. Olsen, E.C. Olson, S. Payne, C.A. Reed, D.A. Russell, S. Sampson, J. Shoshani, F.B. Stangl Jr, M.S. Stevens, J.E. Storer, M.R. Voorhies, D.B. Weishampel, M.C. Wilson, and D.G. Wyckoff. And, finally, both editors wish to thank the contributors to the volume for their time and patience throughout the long process from initial conception through to publication. Kathlyn Stewart Kevin Seymour

C.S. 'Rufus' Churcher - The man from Aldershot

Kevin L. Seymour

Charles Stephen Churcher was born in Aldershot, England, on 21 March 1928. Early on, the headmaster at Stowe School, J.F. Roxburgh, encouraged his interest in natural sciences. The Churcher family moved to Kenya when he was 18, and his early love of natural history flourished there. An interest in the living African flora and fauna has continued to this day, in spite of or, strangely, maybe because of, his Canadian connection. His knowledge of natural-history (as well as human history) is prodigious and it is well known that he seems to have an answer for any natural-history question; in truth, if he doesn't actually know the answer, he knows at least something about the topic and he will share it with you. Not all people viewed this as a good thing, however, as one of his early teachers said to him, 'You'll never be anything - you have too many interests.' This broad interest will become a recurring theme, and ultimately will serve him well in his chosen career of vertebrate palaeontology. As we will refer to him by his preferred name, Rufus, throughout this introduction, the story of how this name came to be attached to him should be told here. His father, in a somewhat old-school yet endearing sense, felt all pets and children should have a name. So he labelled each of his children with a 'pet' name, regardless of the fact that they had perfectly good names already. The name Rufus came from the American comedy team Rufus and Rastus Johnson-Brown, also known as 'The Two Black Crows.' As Rufus's parents were hoping for twins, it seemed appropriate to dub the unborn pair Rufus and Rastus; when only Rufus was born, the Rastus disappeared and the Rufus stuck. Perhaps it was expected that once the children grew into adults they could or would assume their 'proper' names, but such was not the case for Rufus. At age seven, when he was given the choice to change to his real name 'Charles,' he would have nothing to do with it; his name was Rufus! This frequently

xviii

K.L. Seymour

has caused confusion for those who have met him as Rufus Churcher, but later find publications by a C.S. Churcher (another Churcher palaeontologist?). It is for this reason that he often writes his name, as we have in the title of this book, as C.S. 'Rufus' Churcher. Between 1942 and 1946 he served first as a lance-corporal in the Officers Training Corps, then as a private in the Home Guard, next as a flight-sergeant in the Air Training Corps, and finally as an aircraftsman, 2nd class, with the Royal Air Force. Between 1950 and 1954 he served as an assistant inspector with the Kenya Police Reserve. This military experience has left him with a keen interest in military history. His undergraduate career was completed in biology at the University of Natal in Pietermaritzburg, South Africa, in 1950, with the encouragement of Professor S.F. Bush and Doctors D.W. and R.F. Ewer. He spent the next year as a student at the Imperial Forestry Institute and Exeter College, Oxford University, England. This foray into 'practical botany' certainly left its mark. Rufus's eyes are always open and questioning, resulting in several publications during his career which would seem out of place only on someone else's resume [7, 15, 83, 88). A return to Pietermaritzburg in 1951 saw the completion of an honours degree in zoology in 1952. The course was now set, and a master's degree on 'The fossil Hyracoidea of the Transvaal Cave deposits' was completed in 1954 under the direction ofJ.T. Robinson (and resulted in Rufus's first refereed publication [1]). With this study came the realization that individual specimens were generally inadequate for the description of variation in a normal population of mammals. After a stint as a preparator and research assistant in the Department of Palaeontology at the Transvaal Museum in Pretoria, South Africa, Rufus made the big step and emigrated to Canada in 1954. His original intention was to work under Loris Russell at the Royal Ontario Museum (ROM), but, as fate would have it, Loris had just left Toronto to take the position of chief of the Zoology Section of the National Museum of Canada in Ottawa. Consequently, Rufus embarked on a dissertation that would extend the concept elucidated in his master's, by giving him training that involved consideration of geographic ranges and subspecies in a living mammal. Rufus completed his Ph.D. on the 'Variation in the red fox, Vulpes vulpes Linnaeus' under R.L. Peterson in 1957. This resulted in two publications [3, 4) and set a trend of interest in carnivores. Although he did not pursue much neomammalogical work in his career, his dissertation gave him an appreciation for variation in populations of living mammals, which enabled him to better appreciate what kinds of information could, and more importantly could not, be gleaned from palaeontological material. From 1954-7 Rufus was a demonstrator, and from 1957-60 a lecturer, in

'Rufus' Churcher - The man from Aldershot xix the Department of Zoology at the University of Toronto. Assistant professorship came in 1960. He became an associate professor in 1965, a full professor in 1970. During his tenure at the University of Toronto, he has been involved with teaching a variety of courses, including vertebrate palaeontology, vertebrate zoology, evolutionary theory, mammalogy, comparative anatomy, and comparative dental anatomy, to name a few. Rufus married W.B.M. Lindsay, known as 'Bee,' in 1959. They had met in 1948 in Kenya, where their parents' farms were within a few miles. A return visit from Canada in 1958 moved the relationship to another level. Bee has been a frequent companion on his many field trips, and in addition three children (Jaclyn, Nigel, and Stephanie) have enriched their lives. With A.G. Edmund's 1958 excavations at the Pleistocene tar pits of Talara, Peru, tens of thousands of bones came to the ROM. As Rufus had just completed his dissertation on canids, it was only natural for him to be interested in this new fossil material. In it he recognized the first record of Canis dirus from South America [2], as well as a diverse fauna that would spawn a number of papers over the years [6, 8, 11, 14, 15]. This also led him to address other issues in South American vertebrate palaeontology [16, 18] and to supervise or serve on the committees of several students who studied parts of the Talara fauna, including L.S. Kisko (canids), B.F. Beebe (canids), H.G. McDonald (scelidothere sloths), and K.L. Seymour (felids). Nevertheless, his main interest became the large herbivorous mammals of Africa and North America. Rufus often heard it said that Canada had no Pleistocene fossils and so he accepted the challenge, in 1966, to prove this incorrect. The Geological Survey of Canada (GSC) invited him to accompany one of their field geologists, A. MacS. Stalker, on a 17-day field excursion between Calgary, Alberta, and Regina, Saskatchewan, to evaluate a series of supposed Pleistocene fossiliferous deposits. They found 21 sites; three were recognized as being major (Cochrane, Alberta; Medicine Hat and Swift Current, Saskatchewan). A number of publications resulted [20, 23, 26, 34, 37] and opened the door for other, often collaborative, Quaternary vertebrate work on the Canadian prairies [25, 28, 31, 32, 33, 40, 41, 42, 45, 53, 58, 59, 72, 74, 85]. As well, over 1000 pages of report were penned for the GSC. Rufus was essentially the pioneering Quaternary vertebrate palaeontologist of the Canadian prairies. At the invitation of P.E.L. Smith, he tackled the fauna from the late Palaeolithic sites at Korn Ombo, Egypt [29, 30, 35]. This work ultimately led him to become involved in the Dakhleh Oasis Project in Egypt starting in 1979, and has resulted in other publications on the fauna from this area [65, 86, 92].

xx K.L. Seymour A sabbatical leave in 1969 allowed Rufus the time to study in Nairobi under L.S.B. Leakey and in Pretoria with his old friend C.K. Brain, resulting in publications on giraffids (24, 36) and equids (22). These works, among others, gained him expertise in these groups, so that he was the natural choice to write the chapters on Giraffidae and Equidae in the Evolution of African Mammals (42, 43). Further works on African equids (49, 50, 51, 64, 89, 90) and giraffids [44, 82) followed. He began to come to the belief that an understanding of the Equidae of the world might present a key to the correlation, at least on an equivalent faunal basis, of the Quaternary and possibly the later Cenozoic of the Old and New Worlds. Rufus elucidated the striking contrasts and similarities between the Canadian prairie sites and those from other 'prairie' environments in eastern and southern Africa (20, 26, 31, 35, 37, 58, 60, 67, 68, 76, 77). For this reason, he continues to work on two continents, using the recent African ecological interactions with their Pleistocene origins for the interpretations of the Quaternary faunas of Canada and North America [31, 32, 37, 41, 45, 58, 59, 68). Besides equids and giraffids, he has had a long-standing interest in other large mammals. He has investigated the taxonomy and relationships of sabretoothed animals, primarily felids [10, 17, 18, 55, 61, 70), as well as various cervids [8, 16, 52, 56, 71, 78, 84), proboscideans [19, 28, 48, 66, 73), and ursids [38, 87, this volume]. The aforementioned Dakhleh Oasis Project in Egypt consists of a large, interdisciplinary team, including participants from England, Poland, Australia, and Canada. Although Rufus was originally their fauna} analyst, there was not really enough Neolithic material to keep him busy. As a result, he couldn't resist investigating nearby petroglyphs or the primarily marine Cretaceous sediments and their fossils, which were everywhere! A large volume reporting on the survey of the area is currently in press, edited by Rufus and A.J. Mills. In addition, he always found time to give some attention to local specimens and faunas [5, 9, 12, 13, 19, 21, 38, 40, 79). Although some people didn't think much of his predilection for 'bits 'n pieces' (as Rufus himself likes to call them), the time was well spent. Two of the more significant local finds included the discovery of pika (Ochotona) in the Pleistocene of Ontario [46) and the discovery of a new genus of Pleistocene deer (Torontoceros), excavated from Toronto [52). Not only was Rufus's career full of teaching and writing papers, but he managed to fit in numerous administrative and professional duties. Although his appointment was in zoology at the University of Toronto (U of T), he also taught in the Department of Oral Anatomy at the U of T Faculty of Dentistry and was the Life Science representative to the School of

'Rufus' Churcher - The man from Aldershot xxi Nursing. For three years (1975-8) he served as associate dean of the Faculty of Arts and Science, was acting chairman of the Department of Urban and Regional Planning (1980-1), associate chairman of the Department of Zoology (1980-4) and acting director of the Museum Studies Program (1983-5), to name a few. He was a long-time research associate at the Royal Ontario Museum, starting in 1962, and several times served as associate editor or external editor to the Life Sciences Board of Editors at the ROM. He became a fellow of the American Association for the Advancement of Science in 1983 and a fellow of the Geological Association of Canada in 1984. He served as a member of the Board of Management of the Metro Toronto Zoological Society from 1979 to 1986, including chairman of this board from 1981-4. Other positions include chairman of the Canadian Council on Animal Care (1984-5), associate editor of the Journal of Vertebrate Paleontology (1985-9), vice-president of the Society of Vertebrate Paleontology (1988-90), and, finally, president of the Society of Vertebrate Paleontology (1990-2). Rufus takes pride in the graduate students he has supervised or on whose committees he has served. Many have gone on to become palaeontologists or zoologists (e.g., H.N. Bryant, J.A. Burns, G.H. Dalrymple, G. De Iuliis, H. Dompierre, J. Eger, G.R. Hurlburt, H.G. McDonald, P.D. Ross, S. Sampson, T. Skwara, J.E. Storer, M.E. Taylor, J.J. Thomason, and C.G. van Zyll de Jong to name a few, as well, of course, as the co-editors of this volume!). All remark on, and were appreciative of, his approachability as well as his hands-off approach, allowing students to follow their interests. His sense of humour is legendary and often mischievous. The story is told of the innocent wrong-number caller who reached Rufus's office instead of the bus station (they were only one digit different). The conversation went something like this: 'Churcher speaking.' 'Can you tell me when the next bus to Peterborough is?' 'Certainly. Four o'clock.' Needless to say, Rufus had no idea when the next bus to Peterborough was, but he was tired of receiving calls for the bus station, and that particular voice was very trying! And his career, as mentioned, is by no means complete. He continues to publish, so that we were unsure how many papers of his would actually be published by the time this volume went to press! Rufus and Bee are at present in the process of building a retirement home on Gabriola Island, British Columbia, and a research building for Rufus is included in the plans! We wish you all the best, Rufus, and hope that, in some small way, this volume is a tribute to your career.

xxii

K.L. Seymour

C.S. Churcher - A Bibliography (not including abstracts) 1 Churcher, C.S. 1956. The fossil Hyracoidea of the Transvaal and Taungs deposits. Annals of the Transvaal Museum 22:477501. 2 Churcher, C.S. 1959. Fossil Canis from the Tar Pits of La Brea, Peru. Science 130:564-565. 3 Churcher, C.S. 1959. The specific status of the New World red fox. Journal of Mammalogy 40:513-520. 4 Churcher, C.S. 1960. Cranial variation in the North American red fox. Journal of Mammalogy 41 :349-360. 5 Churcher, C.S., and W.A. Kenyon. 1960. The Tabor Hill ossuaries: a study in Iroquois demography. Human Biology 32:249- 273. 6 Lemon, R.R.H., and C.S. Churcher. 1961. Pleistocene geology and paleontology of the Talara region, Northwest Peru. American Journal of Science 259:410-429. 7 Churcher C.S. 1962. Yellow-headed blackbirds breeding at Rainy River, Ontario. Canadian Field-Naturalist 76:122. 8 Churcher, C.S. 1962. Odocoileus salinae and Mazama sp. from the Talara tar seeps, Peru. Royal Ontario Museum, Life Sciences Contributions 57:1-27. 9 Churcher, C.S., and P.F. Karrow. 1963. Mammals of Lake Iroquois age. Canadian Journal of Zoology 41:153-158. 10 Churcher, C.S. 1964. Machairodus latidens from Kent's Cavern, Devon. Pp. 142, 144 in E.M.M. Alexander, Father John MacEnery- Scientist or Charlatan? Transactions of the Devonshire Association for the Advancement of Science, Literature and Art 96. 11 Churcher, C.S. 1965. Camelid material of the Genus Palaeolama Gervais from the Talara Tar-seeps, Peru, with a description of a new subgenus, Astylolama. Proceedings of the Zoological Society, London 145:161205. 12 Kenyon, W.A., and C.S. Churcher. 1965. A flake tool and a worked antler fragment from Late Lake Agassiz. Canadian Journal of Earth Sciences 2:237-246. 13 Churcher, C.S. 1965. Mammals at Fort Albany circa 1700 AD. Journal of Mammalogy 46:354--355. 14 Churcher, C.S., and C.J. van Zyll de Jong. 1965. Conepatus talarae n. sp. from the Talara Tar-seeps, Peru. Royal Ontario Museum Life Sciences Contributions 62:1-15. 15 Churcher, C.S. 1966. The insect fauna from the Talara Tar-seeps, Peru. Canadian Journal of Zoology 44:985-993.

Churcher bibliography

xxiii

16 Churcher, C.S. 1966. Observaciones sobre el status taxon6mico de Epieuryceros Ameghino 1889 y sus especies E. truncus y E. proximus. Ameghiniana 4:351-362. 17 Churcher, C.S. 1966. The affinities of Dinobastis serus Cope 1883. Quaternaria 8:263-275. 18 Churcher, C.S. 1967. Smilodon neogaeus en las barrancas costeras de Mar del Plata, Provincia de Buenos Aires. Publicaciones del Museo Municipal de Ciencias Naturales de Mar del Plata 1:245-262. 19 Churcher, C.S. 1968. Mammoth from the Middle Wisconsin of Woodbridge, Ontario. Canadian Journal of Zoology 46: 219-221. 20 Churcher, C.S. 1968. Pleistocene ungulates from the Bow River gravels at Cochrane, Alberta. Canadian Journal of Earth Sciences 5:1467-1488. 21 Churcher, C.S., and M.B. Fenton. 1968. Vertebrate remains from the Dickson Limestone Quarry, Halton County, Ontario, Canada. Bulletin of the National Speleological Society 30:11-16. 22 Churcher, C.S. 1970. The fossil Equidae from the Krugersdorp caves. Annals of the Transvaal Museum 26:145-168. 23 Churcher, C.S. 1970. The vertebrate faunas of Surprise, Mitchell, and Island Bluffs, near Medicine Hat, Alberta. Geological Survey of Canada, Department of Energy, Mines and Resources, Report of Activities, Paper 70-I(A):158-160. 24 Churcher, C.S. 1970. Two new Upper Miocene giraffids from Fort Ternan, Kenya, East Africa: Palaeotragus primaevus n. sp. and Samotherium africanum n. sp. Pp. 1-106 in L.S.B. Leakey and R.J.G. Savage (eds), Fossil Vertebrates of Africa 2. Academic Press, New York. 25 Churcher, C.S., and A. Macs. Stalker. 1970. A late, postglacial horse from Pashley, Alberta. Canadian Journal of Earth Sciences 7:10201026. 26 Stalker, A. Macs., and C.S. Churcher. 1970. Deposits near Medicine Hat, Alberta, Canada. Geological Survey of Canada, Surveys and Mapping Branch. [Wall-chart] 27 Churcher, C.S. 1971. Les mammiferes. P. 160 in M. Ters, C. Azema, P. Brebion, C.S. Churcher, G. Delibrias, M. Denefle, J. Guyader, A. Lauriat, R. Mathieu, J.P. Michel, P.J. Osborne, A. Rouvillois, and F.W. Shotton, Sur le remblaiement holocene dans l'estuaire de la Seine, au Havre (Seine Maritime), France. Quaternaria 14:151-174. 28 Churcher, C.S. 1972. Imperial mammoth and Mexican Half-Ass from near Bindloss, Alberta. Canadian Journal of Earth Sciences 9:15621567. 29 Churcher, C.S. 1972. Late Pleistocene vertebrates from archaeological sites in the Plain of Korn Ombo, Upper Egypt. Royal Ontario Museum Life Sciences Contributions 82: 1-172.

xxiv

K.L. Seymour

30 Churcher, C.S., and P.E.L. Smith. 1972. Korn Ombo: preliminary report on the fauna of Late Paleolithic sites in Upper Egypt. Science 177:259261. 31 Russell, L.S., and C.S. Churcher. 1972. Vertebrate paleontology, Cretaceous to Recent, interior plains, Canada. 24th International Geological Congress, Montreal, Field excursion A21 :l-46. 32 Russell, L.S., and C.S. Churcher. 1972. Paleontologie des vertebres du Cretace au Recent des plaines interieures au Canada. 24eme Congres Geologique International, Montreal, Livret-guide Excursion A21:l-48. 33 Stalker, A. MacS., and C.S. Churcher. 1972. Glacial stratigraphy of the Southwestern Canadian prairies; the Laurentide Record. 24th International Geological Congress, Montreal, Quaternary Geology, Section 12:110-119. 34 Szabo, B.J., A. Macs. Stalker, and C.S. Churcher. 1973. Uranium-Series ages of some Quaternary deposits near Medicine Hat, Alberta, Canada. Canadian Journal of Earth Sciences 10:1464-1469. 35 Churcher, C.S. 1974. Relationships of the late Pleistocene vertebrate fauna from Korn Ombo, Upper Egypt. In R. Said and B.H. Slaughter (eds), Contributions to the Paleontology of Africa. Proceedings of the 75th Anniversary of the Geological Survey of Egypt, Annals of the Geological Survey of Egypt 4:363-384. 36 Churcher, C.S. 1974. Sivatherium maurusium (Pomel) from the Swartkrans australopithecine site, Transvaal (Mammalia: Giraffidae). Annals of the Transvaal Museum 29:65-70. 37 Churcher, C.S. 1975. Additional evidence of Pleistocene ungulates from the Bow River gravels at Cochrane, Alberta. Canadian Journal of Earth Sciences 12:68-76. 38 Churcher, C.S., and A.V. Morgan. 1976. A grizzly bear from the Middle Wisconsin of Woodbridge, Ontario. Canadian Journal of Earth Sciences 13:341-347. 39 Churcher, C.S. (ed.). 1976. Athlon: Essays on Palaeontology in Honour of Loris Shano Russell. Royal Ontario Museum Life Sciences Miscellaneous Publication, 286 pp. 40 Churcher, C.S., and P.F. Karrow. 1977. Late Pleistocene muskox (Ovibos) from the Early Wisconsin at Scarborough Bluffs, Ontario, Canada. Canadian Journal of Earth Sciences 14:326-331. 41 Wilson, M., and C.S. Churcher. 1978. Late Pleistocene Camelops from the Gallelli Pit, Calgary, Alberta: morphology and geologic setting. Canadian Journal of Earth Sciences 15:729-740. 42 Churcher, C.S., and M.L. Richardson. 1978. Equidae. Pp. 379-422 in V.J. Maglio and H.B.S. Cooke (eds), Evolution of African Mammals. Harvard University Press, Cambridge.

Churcher bibliography

xxv

43 Churcher, C.S. 1978. Giraffidae. Pp. 509-535 in V.J. Maglio and H.B.S. Cooke (eds), Evolution of African Mammals. Harvard University Press, Cambridge. 44 Churcher, C.S. 1979. The large palaeotragine giraffid, Palaeotragus germaini, from Late Miocene deposits of Lothagam Hill, Kenya. Breviora 453:1-8. 45 Churcher, C.S., and M. Wilson. 1979. Quaternary mammals from the Eastern Peace River District, Alberta. Journal of Paleontology 53:7176. 46 Churcher, C.S., and R.R. Dods. 1979. Ochotona and other vertebrates of possible Illinoian age from Kelso Cave, Halton County, Ontario. Canadian Journal of Earth Sciences 16:1613-1620. 47 Churcher, C.S. 1979. Marsupials. Pp. 445-461 in R.W. Fairbridge and D. Jablonski (eds), The Encyclopedia of Paleontology. Dowden, Hutchinson and Ross, Inc., Stroudsburg, Pennsylvania. 48 Churcher, C.S. 1980. Did the North American mammoth migrate? Canadian Journal of Anthropology 1: 103-105. 49 Churcher, C.S., and D.A. Hooijer. 1980. The Olduvai zebra (Equus oldowayensis) from the later Omo Beds, Ethiopia. Zoologische Mededelingen 55:265-280. 50 Churcher, C.S. 1981. Zebras (Genus Equus) from nine Quaternary sites in Kenya, East Africa. Canadian Journal of Earth Sciences 18:330-341. 51 Churcher, C.S. 1982. Oldest ass recovered from Olduvai Gorge, Tanzania, and the origin of asses. Journal of Paleontology 56:1124-1132. 52 Churcher, C.S., and R.L. Peterson. 1982. Chronologic and environmental implications of a new genus of fossil deer from late Wisconsin Deposits at Toronto, Canada. Quaternary Research 18:184-195. 53 Stalker, A. MacS., and C.S. Churcher. 1982. Ice age deposits and animals from the southwestern part of the Great Plains of Canada. Geological Survey of Canada, Miscellaneous Report 31. [Wall-chart] 54 Churcher, C.S. 1983 The origin of the African Buffalo, Syncerus caffer, and the fossil buffaloes of Africa. Pp. 14-21 in M.J. Mloszewski, Behavior and Ecology of the African Buffalo. Cambridge University Press, Cambridge. 55 Churcher, C.S. 1984. The Status of Smilodontopsis (Brown, 1908) and Ischyrosmilus (Merriam, 1918): a taxonomic review of two genera of sabretooth cats (Felidae, Machairodontinae). Royal Ontario Museum Life Sciences Contributions 140:1-59. 56 Churcher, C.S. 1984. Sangamona: the furtive deer. Pp. 316-331 in M.R. Dawson and H.H. Genoways (eds), Contributions to Vertebrate Paleontology: A Volume in Memorial to John E. Guilday. Carnegie Museum of Natural History, Special Publication 8. 57 Churcher, C.S., et al. 1984. Panel discussion. Pp. 407-415 partim in

xxvi

58 59

60 61 62

63

64

65

66 67 68

KL. Seymour

W.C. Mahaney (ed.), Quaternary Dating Methods. Elsevier Science Publishers, Amsterdam. Churcher, C.S. 1984. Faunal Correlations of Pleistocene Deposits in Western Canada. Pp. 145-158 in W.C. Mahaney (ed.), Correlation of Quaternary Chronologies. Geo Books, Norwich, England. Wilson, M.C., and C.S. Churcher. 1984. The Late Pleistocene Bighill Creek Formation and its equivalents in Alberta: correlative potential and vertebrate palaeofauna. Pp. 159-175 in W.C. Mahaney (ed.), Correlation of Quaternary Chronologies. Geo Books, Norwich, England. Mahaney, W.C., R.W. Barendregt, C.S. Churcher, and J.R. Spence. 1985. Gorges Valley Rock Shelters, Mount Kenya Afroalpine Area, East Africa. Nyame Akuma 26:22-25. Churcher, C.S. 1985. Dental functional morphology in the marsupial sabre-tooth Thylacosmilus atrox (Thylacosmilidae) compared to that of felid sabre-tooths. Australian Mammalogy 8:201-220. Churcher, C.S. 1985. Zoological study of the ivory knife handle from Abu Zaidan. Pp. 152-168 in W. Needler, Predynastic and Archaic Egypt in the Brooklyn Museum. The Brooklyn Museum, Wilbour Monographs 9. Churcher, C.S. 1985. 'Bear (p. 151); Bobcat (p. 198); Cougar (p. 431); Fox (p. 684); Lynx (p. 1045); Raccoon (p. 1537); Weasel (p. 1927); Wolf (p. 1956).' Entries in J.H. Marsh (editor-in-chief), The Canadian Encyclopedia. Hurtig Publishers, Edmonton. Hooijer, D.A., and C.S. Churcher. 1985. Perissodactyla of the Omo Group deposits, American Collections. Pp. 97-117 in Y. Coppens and F. Clark Howell (eds), Les faunes Plio-Pleistocenes de la Basse Vallee de l'Omo, (Ethiopie), Tome 1, Perissodactyles-Artiodactyles (Bovidae) 4. Editions du Centre National des Recherches Scientifiques. Churcher, C.S. 1986. Equid remains from Neolithic horizons at Dakhleh Oasis, Western Desert of Egypt. Pp. 41~21 in R.H. Meadow and H.-P. Uerpmann (eds), Equids in the Ancient World. Tiibinger Atlas des Vorderen Orients, Reihe A. (Naturwissenschaften) 19. Churcher, C.S. 1986. A mammoth measure of time: molar compression in Mammuthus from the Old Crow Basin, Yukon Territory, Canada. Current Research in the Pleistocene 3:61-64. Mahaney, W.C., R.W. Barendregt, C.S. Churcher, and J.R. Spence. 1986. Rock shelters in Gorges Valley, Mount Kenya Afroalpine area. Journal of African Earth Sciences 5:321-327. Lundelius, E.L., Jr, C.S. Churcher, T. Downs, C.R. Harington, E.H. Lindsay, G.E. Schultz, H.A. Semken, S.D. Webb, and R.J. Zakrzewski. 1987. The North American Quaternary Sequence. Pp. 211-235 in M.O. Woodburne (ed.), Cenozoic Mammals of North America: Geochronology and Biostratigraphy. University of California Press, Berkeley.

Churcher bibliography xxvii 69 Churcher, C.S. 1987. Micro megafaunal mounds. Nature 325:22. 70 Bryant, H.N., and C.S. Churcher. 1987. All sabretoothed carnivores aren't sharks. Nature 325:488. 71 Churcher, C.S., and J. Pinsof. 1988. Variation in the antlers of North American Cervalces (Mammalia: Cervidae): review of new and previously recorded specimens. Journal of Vertebrate Paleontology 7:373397. 72 Zymela, S., H.P. Schwarcz, R. Griin, A. Macs. Stalker, and C.S. Churcher. 1988. ESR dating of Pleistocene fossil teeth from Alberta and Saskatchewan. Canadian Journal of Earth Sciences 25:235-245. 73 Nielsen, E., C.S. Churcher, and G.E. Lammers. 1988. A woolly mammoth (Proboscidea, Mammuthus primigenius) molar from the Hudson Bay Lowland of Manitoba. Canadian Journal of Earth Sciences 25:933938. 74 Barendregt, R.W., C.S. Churcher, and A. Macs. Stalker. 1988. Stratigraphy, paleomagnetism, and vertebrate paleontology of Quaternary preglacial sediments at the Maser-Frisch site, southeastern Alberta. Bulletin of the Geological Society of America 100:1824-1832. 75 Churcher, C.S. 1988. 'Bear (p. 188); Bobcat (p. 245); Cougar (p. 524); Coyote (p. 529); Fox (p. 832); Lynx (p. 1254); Raccoon (p. 1816); Weasel (p. 2285); Wolf (p. 2322).' Entries in J.H. Marsh (editor-in-chief), The Canadian Encyclopedia, 2nd edition. Hurtig Publishers, Edmonton. 76 Brain, C.K., C.S. Churcher, J.D. Clark, F.E. Grine, P. Shipman, R.L. Susman, A. Turner, and V. Watson. 1988. New evidence of Early Hominids, their culture and environment, from the Swartkrans cave, South Africa. South African Journal of Science 84:828-835. 77 Churcher, C.S. 1989. Fossil vertebrates from near Naro Moru, Western Foothill Zone, Mount Kenya. Pp. 175-186 in W.C. Mahaney (ed.), Quaternary and Environmental Research on East African Mountains. A.A. Balkema, Rotterdam. 78 Churcher, C.S., P.W. Parmalee, G.L. Bell, and J.P. Lamb. 1989. Caribou from the Late Pleistocene of Northwestern Alabama. Canadian Journal of Zoology 67:1210-1216. 79 Churcher, C.S., J.J. Pilny, and A.V. Morgan. 1990. Late Pleistocene vertebrate, plant and insect remains from the Innerkip Site, Southwestern Ontario. Geographie physique et Quatemaire 44:299-308. 80 Churcher, C.S., and M.C. Wilson. 1990. Methods in Quaternary Ecology #12, Vertebrates. Geoscience Canada 17:59-78. 81 Churcher, C.S., and M.C. Wilson. 1990. Vertebrates. Pp. 127-148 in B.G. Warner (ed.), Methods in Quaternary Ecology. Geoscience Canada, Reprint Series 5. [Reprint of #80) 82 Churcher, C.S. 1990. Cranial appendages of Giraffoidea. Pp. 180-194 in G.A. Bubenik and A.B. Bubenik (eds), Horns, Pronghorns, and Ant-

xxviii

83 84

85

86

87 88 89 90

91 92

93 94

95

K.L. Seymour

lers: Evolution, Morphology, Physiology, and Social Significance. Springer-Verlag, New York. Churcher, C.S. 1991. The Egyptian fruit bat Rousettus aegyptiacus in Dakhleh Oasis, Western Desert of Egypt. Mammalia 55:139-143. Churcher, C.S. 1991. The status of Giraffa nebrascensis, the synonymies of Cervalces and Cervus, and additional records of Cervalces scotti. Journal of Vertebrate Paleontology 11 :391-397. Barendregt, R.W., F.F. Thomas, E. Irving, J. Baker, A. MacS. Stalker, and C.S. Churcher. 1991. Stratigraphy and paleomagnetism of the Jaw Face section, Wellsch Valley site, Saskatchewan. Canadian Journal of Earth Sciences 28:1353-1364. Churcher, C.S. 1993. Ostrich bones from the Neolithic of Dakhleh Oasis, Western Desert of Egypt. Pp. 67-71 in K. Heine (ed.), Palaeoecology of Africa and the Surrounding Islands, volume 23. A.A. Balkema, Rotterdam. Churcher, C.S., A.V. Morgan, and L.D. Carter. 1993. Arctodus simus from the Alaskan Arctic Slope. Canadian Journal of Earth Sciences 30:1007-1013. Churcher, C.S. 1993. Western catalpa (Catalpa speciosa) colonising in Toronto, Ontario. Canadian Field-Naturalist 106:390-392. Churcher, C.S. 1993. Equus grevyi. Mammalian Species 453:1-9. Churcher, C.S., and V. Watson. 1993. Additional fossil Equidae from Swartkrans. Pp. 137-150 in C.K. Brain (ed.), Swartkrans: A Cave's Chronicle of Early Man. Transvaal Museum Monograph 8. Churcher, C.S. 1993. Fossil collecting and government regulation. Science 259:581. Churcher, C.S. 1993. Dogs from Ein Tirghi Cemetery, Balat, Dakhleh Oasis, Western Desert of Egypt. Pp. 39-60 in A. Clason, S. Payne, and H.-P. Uerpmann (eds), Skeletons in Her Cupboard: Festschrift for Juliet Clutton-Brock. Oxbow Monograph 34, Oxbow Books, Oxford. Churcher, C.S. 1994. The vertebrate fauna from the Natufian level at Jebel Es-Saa:ide (Saa:ide II), Lebanon. Paleorient 20: 35-58. Churcher, C.S. 1995. Giant Cretaceous lungfish Neoceratodus tuberculatus from a deltaic environment in the Quseir (= Baris) Formation of Kharga Oasis, Western Desert of Egypt. Journal of Vertebrate Paleontology 15: 845-849. Richards, R.L., C.S. Churcher, and W.B. Turnbull. (This volume.) Distribution and size variation in North American Arctodus simus. In K.M. Stewart and K.L. Seymour (eds), Palaeoecology and Palaeoenvironments of Late Cenozoic Mammals: Tributes to the Career of C.S. (Rufus) Churcher. University of Toronto Press, Toronto.

Churcher bibliography xxix In press at time of writing Dompierre, H., and C.S. Churcher. Premaxillary shape as an indicator of the diet of eight extinct Late Cenozoic New World Camels. Journal of Vertebrate Paleontology. Churcher, C.S. A note on the Late Cretaceous vertebrate fauna of Dakhleh Oasis, Western Desert of Egypt. In C.S. Churcher and A.J. Mills (eds), Dakhleh Oasis Project Survey Report. Oxbow Monographs. Churcher C.S. The Holocene faunas of Dakhleh Oasis. In C.S. Churcher and A.J. Mills (eds), Dakhleh Oasis Project Survey Report. Oxbow Monographs. Hollett, A.J., and C.S. Churcher. Notes on the recent faunas of Dakhleh Oasis. In C.S. Churcher and A.J. Mills (eds), Dakhleh Oasis Project Survey Report. Oxbow Monographs. Book reviews 1 Schop£, T.J.M. (ed.) 1972. Models in Paleobiology. Freeman Cooper Co., San Francisco, 250 pp. In Canadian Field-Naturalist 88:393-394; 1974. 2 Johnson, H., and J.E. Storer. 1974. A guide to Alberta vertebrate fossils

from the Age of Dinosaurs. Provincial Museum of Alberta Publication

4, 127 pp. In Canadian Field-Naturalist 89:33€r337; 1975. 3 Boessneck, J., and A. von den Driesch (eds). 1982. Studien an subfos-

silen Tierknochen aus Agypten. Miinchner Agyptologische Studien 40:1-192. In Chronique d'Egypte 59:110-111; 1984. 4 Harris, J.M. (ed.). 1983. Koobi Fora Research Project, volume 2: The Fossil Ungulates - Proboscidea, Perissodactyla, and Suidae: Researches into Geology, Palaeontology and Human Origins. Clarendon Press, New York, 321 pp. In Quarterly Review of Biology 60: 66; 1985. 5 Tchernov, E. (ed.) 1986. Les Mammiferes du Pleistocene inferieur de la

Vallee du Jourdain a Oubeidiyeh. Memoires et Travaux du Centre de Recherche Frarn;ais de Jeursalem 5, Association Paleorient, Paris. 405 pp. In Journal of Vertebrate Paleontology 8:232-233; 1988. 6 Graham, R.W., H.A. Semken, Jr, and M.A. Graham (eds). 1987. Late Quaternary Mammalian Biogeography and Environments of the Great Plains and Prairies. Illinois State Museum, Scientific Papers 22:1-491. In Science 240:1213-1214; 1988. 7 Kessler, D., J. Boessneck, and A. von den Driesch. 1987. Tuna el-Gebel I: Die Tiergalerien. In J. Boessneck and A. von den Driesch. Die Tier-

xxx

K. L. Seymour

knochenfunde. Hildesheimer Agyptologische Beitriige 24:1-221. In Bibliotheca Orientalis 47:67-70; 1990. 8 Boessneck, J. 1987. Die Miinchner Ochsenmumie. Hildesheimer Agyptologische Beitriige 25:1-96. In Bibliotheca Orientalis 47:70-72; 1990. 9 Boessneck, J. 1988. Die Tierwelt des Alten Agypten: untersucht anhand kulturgeschichtlicher und zoologischer Quellen. C.H. Beck, Munich, 197 pp. In Bibliotheca Orientalis 48:507-509; 1991. 10 MacFadden, B.J. 1992. Fossil Horses: Systematics, Paleobiology, and Evolution of the Family Equidae. Cambridge University Press, Cambridge, 369 pp. In Quarterly Review of Biology 69:259; 1994. Selected non-refereed works 1 Churcher, C.S. 1958. The foxes of Ontario. Canadian Camping 1958:186-188. 2 Churcher, C.S. 1967. Man from the early Wisconsin of Alberta. Royal Ontario Museum Newsletter, New Series 25 (June):1-5. 3 Churcher, C.S. 1968. Red fox (Vulpes vulpes). Hinterland who's who. Canadian Wildlife Service R69-4/5:l-5. 4 Churcher, C.S. 1968. Portrait of a palaeontologist. Rotunda 1(2):22-29. 5 Churcher, C.S. 1969. The vertebrate faunas of Surprise, Mitchell and Island Bluffs, near Medicine Hat, Alberta. Field-guide to 19th Field-Conference, Mid-Western Friends of the Pleistocene, 10-11 May. 5pp. 6 Churcher, C.S. 1972. Return to Africa. Rotunda 5(3):40-44. 7 Churcher, C.S. 1980. Dakhleh Oasis Project-preliminary observations on the geology and vertebrate palaeontology of Northwestern Dakhleh Oasis: a report on the 1979 fieldwork. Journal of the Society for the Study of Egyptian Antiquities (Toronto) 10:379-395. 8 Churcher, C.S. 1981 . Dakhleh Oasis Project. Geology and palaeontology: interim report on the 1980 field season. Journal of the Society for the Study of Egyptian Antiquities (Toronto) 11:193-212. 9 Churcher, C.S. 1982. Grevy's - the other zebra. Swara 5(1):12-18. 10 Churcher, C.S. 1982. Dakhleh Oasis Project. Geology and palaeontology: interim report on the 1981 field season. Journal of the Society for the Study of Egyptian Antiquities (Toronto) 12:103-114. 11 Churcher, C.S. 1983 Dakhleh Oasis Project. Palaeontology: interim report on the 1982 field season. Journal of the Society for the Study of Egyptian Antiquities (Toronto) 13:178-187. 12 Churcher, C.S. 1984. Kenya. City and Country Home 3(6): 48-54, 56. 13 Churcher, C.S. 1986. Problems of Africa. Nature (Correspondence) 323:290. 14 Churcher, C.S. 1986. The extinct Cape Zebra. Sagittarius 1(4):4-5.

Churcher bibliography

xxxi

15 Churcher, C.S. 1986. Dakhleh Oasis Project. Palaeontology: interim report on the 1985 field season. Journal of the Society for the Study of Egyptian Antiquities (Toronto) 16: 1-4. 16 Churcher, C.S. 1986. Dakhleh Oasis Project. Palaeontology: interim report on the 1987 field season. Journal of the Society for the Study of Egyptian Antiquities (Toronto) 16: 114-118. 17 Churcher, C.S. 1987. Dakhleh Oasis Project. Palaeontology: interim report on the 1988 field season. Journal of the Society for the Study of Egyptian Antiquities (Toronto) 17: 177-181. 18 Churcher, C.S. 1987. The zooarchaeology of Dakhleh Oasis. Bulletin of the Canadian Mediterranean Institute 7(1):3. 19 Churcher, C.S. 1988 The Neolithic environment of Dakhleh Oasis. Series of Canadian lectures on Archaeology, Reports of the Canadian Club 5:1-7. 20 Churcher, C.S. 1988. An appreciation of Bill Irving. Mammoth Trumpet, March 1988:5. 21 Churcher, C.S. 1993. Romano-Byzantine and Neolithic Diets in Dakhleh Oasis. Canadian Mediterranean Institute Bulletin 13(2):1-2. 22 Churcher, C.S. 1993. Loris Shano Russell. (Eulogy for) Romer Simpson Prize. Society of Vertebrate Palaeontology News Bulletin 157:26-27.

Palaeoecology and Palaeoenvironments of Late Cenozoic Mammals

Diversity bottlenecks, oddball survivors, and negative keys

William A. Akersten

Abstract If a group of related animals undergoes a diversity bottleneck during which most members become extinct, the survivors should be oddballs, atypical of the extinct forms in one or more characteristics which may or may not be reflected in the fossil record. In such cases, modem survivors should not be viewed as detailed analogues for their extinct relatives unless there is corroborating data. Numerous examples demonstrate that the surviving taxa may actually be 'negative keys to the past,' models for what their extinct relatives did not have or did not do. When viewed from the oddball-survivor perspective, comparisons of the distribution patterns (in a geographic, ecologic, morphologic, or other sense) of a group before and after a diversity bottleneck can yield information regarding the nature of the factor(s) which resulted in the bottleneck. Compiling and comparing these patterns for multiple coincident diversity bottlenecks has the potential to produce new insights regarding major extinction events.

Introduction Uniformitarianism has long been the guiding principle of the earth sciences under the assumption that the basic 'laws of nature' have not changed during the history of the earth. We use studies of modem phenomena to interpret past events and freely repeat the phrase 'The present is the key to the past.' This principle is also applicable to interpreting past life from recent related forms, but we must be careful about making overly detailed interpretations. For example, it can be said with considerable assurance that all fossil beavers were herbivores, but palaeontologi-

4 W.A. Akersten cal evidence demonstrates that extending the analogy to an aquatic mode of life and tree cutting would be incorrect for many species. Some extinct forms have been found in clearly terrestrial burrows and few have the chisel-shaped incisors required for gnawing down trees. I argue that, in at least one special case, modern survivors should be atypical of a largely extinct group. If a group of diverse, related fossil taxa undergoes a diversity bottleneck (in other words, most become extinct with few survivors), the survivors would usually make it through because they were different in some significant characteristic(s) from those taxa that became extinct. Hence the term 'oddball survivors.' These would be a subset of 'taxonomic relicts,' a designation that Simpson (1944) used in referring to all survivors of a diversity bottleneck. When a diversity bottleneck has occurred, using the surviving relatives as detailed analogues may often lead to incorrect conclusions. It would be more appropriate to suspect that the survivors are in part 'negative keys'; that in one or more characteristics they serve as models for what most of their extinct relatives did not have or did not do. In addition to the above-mentioned beavers, the palaeontologic literature abounds with examples of demonstrable lifestyle differences between modern forms and most or all of their extinct relatives, especially in groups which have undergone diversity bottlenecks. The surviving tree sloths encompass but a small part of the diversity of extinct Pilosa (e.g., White, 1993), most of which could not have been as fully suspensorial as the extant forms. Fossil coelacanths are known from shallow fresh and marine waters (Lund et al., 1985); the sole living survivor, Latimeria, is a deep-sea form. Aplodontoid rodents were once very diverse and occurred over much of North America and Eurasia (Korth, 1986; Rensberger, 1975; Savage and Russell, 1983); their sole survivor occurs only in a very restricted habitat in a small part of western North America (Carroway and Verts, 1993). Anomaluroid rodents were once very diverse and widespread in Africa (Kingdon, 1974). Survivors include the ricochetal springhare, several gliding forms, and one generalized non-hopping and non-gliding form. The surviving anurans, caecilians, and urodeles are certainly not at all representative of the great Palaeozoic radiation of amphibians. The same principle appears to hold true for invertebrates. The Monoplacophera, a class of primitive molluscs (Runnegar, 1987), were diverse and abundant in shallow Cambrian through Devonian epicontinental seas, although one form has been found in the supposedly deeper-water Burgess Shale. The two known surviving genera live at water depths of 175 to at least 3600 metres (Lemche, 1957; McLean, 1979). Similarly, fossil members of the gastropod family Pleurotomariidae (slit shells) were abundant in shallow Mesozoic marine deposits; living forms are only

Oddball survivors 5 known from deep waters (Cox, 1960). The shelled cephalopods provide another probable example with a great diversity of forms until the major extinction at the end of the Mesozoic, then a gradual tapering off to one modern genus, Nautilus (Pojeta and Gordon, 1987: Fig. 14.66). The several modern species are limited to moderate depths in tropical waters of the southwestern Pacific. Fossil forms were widely distributed and must have been adapted to a variety of environments. Many more examples exist, too many to simply dismiss oddball survivors as aberrations. Since many groups have poor palaeontological records, are inadequately studied, or may differ in factors not discernable from the fossil record, it seems probable that oddball survivors may even be more abundant than they appear to be. Even man may be an oddball survivor.

Theoretical basis Continued existence of a population relies on a host of factors which can be termed survival factors. These include internal (e.g., behavioural, physiologic, reproductive) and external (e.g., temperature, moisture, availability of food, predation, competition) factors. Since the population is vulnerable to the loss of these factors or portions thereof, they can also be thought of as vulnerability factors. Members of a group of related populations at any taxonomic level might be expected to share portions of these factors and differ in others. What would happen if part of one factor changed so that populations dependent on that part of it would not survive? If the changed portion was shared by only one or a few taxa of a large group, the group as a whole would not exhibit significant decline in diversity. If the changed portion was shared by all members of the group, total extinction would result. However, if the changed portion was shared by most members of a group, the result would be a diversity bottleneck. The survivors would be oddballs in the sense that, unlike their extinct relatives, they were not vulnerable to those particular changes in that factor. This single-factor model (Fig. 1) can be thought of as linear in which the factor varies, continuously or discontinuously, along a single axis and various populations occupy segments of the factor. Minor changes or deletions of portions of the factor would only effect one or several populations and there would be no diversity bottleneck. Even a number of scattered small changes would not have a major effect. Only loss of a major portion of the factor, and of those populations which depend on it, would result in a visible diversity bottleneck. While not provable, it seems logical that such a major loss would usually be of one contiguous portion of the factor distribution rather than of several scattered portions. The simplest way to produce a bottleneck in this model, by reducing the

6 W.A. Akersten la: Population distribution before elimination of any portion(s) of the factor. Populations: Factor:

a b

E

CD FG HI

J K L

m

DO

----------------------

Some results of removing one or two contiguous portions of the factor. lb: le: ld:

le: lf:

-

m

a b

-

E CD FG HI

J

K L

-

HI

a b

a b

DO

C

FIGURE 1. Single-factor model of oddball survivorship. Horizontal lines represent factor distributions. Lower-case letters denote oddball populations, uppercase denote typical populations. Any single population occupies a segment of the corresponding factor distribution. (See text for further explanation.)

distribution of contiguous survival factors, would be to eliminate the factors affecting most or all of the 'typical' forms in the central area and possibly some of the less typical forms. Figure 1 illustrates the single-factor model of oddball survivorship, where typical populations are represented by capital letters and oddball populations by lower-case letters. The distribution of each along the horizontal factor axis is relative to their dependence on a particular portion of the factor. While the different populations would actually occupy varying segments along the factor axis, I did not depict these in order to keep the illustration from becoming overly complex. If any portion of the factor was to be deleted, the corresponding populations, such as 'a' through 'G' or 'C' through 'o,' would also disappear. For example, in Figure lb, the portion allowing survival of typical forms is eliminated, with its corresponding populations, and only the oddball extremes remain; note the diversity bottleneck. Figure le depicts the converse: both extremes of the factor are removed, so that all survivors are typical and there is no diversity bottleneck. In Figure ld, removal of the factor continues from the extremes to the centre so that only two typical populations remain and there is also a diversity bottleneck. Thus, it is not impossible for survivors to be typical forms, but this scenario of removing

Oddball survivors 7 both atypical extremes plus most of the typical populations intuitively seems rather unlikely. Figure le shows a major diversity bottleneck in which only two oddball populations survive. Figure 1f is similar except that a single typical population also survives. Populations are more likely to depend on combinations of survival factors which can be conceptualized as occupying multidimensional space. While more complex, the same general principles would apply and the results of deleting portions of the factors would be similar to those presented in the one-dimensional model. This simple model supports the concept that oddball survivors should be quite common, but it also shows that typical survivors, while much less common, may occur. Thus, in any single case, neither oddball nor typical survivors should automatically be assumed, but the former should occur far more often than the latter. If taxonomic relicts resulting from a diversity bottleneck are demonstrably different in some way from their extinct relatives, this difference should be related to a cause of extinction. For example, if oddball survivors are restricted to one particular ecosystem but extinct forms were more widespread, one or more environmentally related factors probably played a major role in loss of the extinct forms.

Some case histories The following examples interpret oddball survivors at different taxonomic levels and over varied geographic areas. The emphasis is on groups which underwent Pleistocene diversity bottlenecks and which have living survivors. The taxonomy of all of these groups is problematic for one reason or another and most groups have significant gaps in their known records, but I believe that they serve to illustrate the principles and potentials of this approach.

Genus Tapirus This genus, including the living tapirs, was quite diverse and widespread during the Neogene of the New World and Eurasia. It is first recorded from the earliest Miocene (Agenian) of Europe (Savage and Russell, 1983; Prothero and Schoch, 1989); it then appears during the later Miocene almost simultaneously in Asia (Vallesian) and North America (Clarendonian). The oldest records from areas in and near the current range of the genus are substantially younger. For South America they are from the Uquian (Savage and Russell, 1983) equivalent to the Plio-Pleistocene (Marshall et al., 1982), and for southeast Asia, the Pleistocene (Hooijer, 1952).

8 W.A. Akersten During the Pleistocene, Tapirus was moderately diverse. Although the taxonomy of this genus is still in a state of flux (Lundelius and Slaughter, 1976; Jefferson, 1989), there appears to have been at least four valid Pleistocene species in North America, ranging as far north as Oregon, Kentucky, and Pennsylvania (Kurten and Anderson, 1980). In Europe, Kurten (1968) mentioned T. arvernensis in the Villefranchian and earlier; he stated that no tapirs were found in Europe after the end of the Villefranchian. However, Savage and Russell (1983) listed T. arvernensis as a constituent of Biharian faunas, and Figure 28.1 of Prothero and Schoch (1989) implied survival of the genus in Europe to the end of the Pleistocene. Tapirus was listed as a member of early Pleistocene 'temperate' Asian faunas and of later Pleistocene faunas from the eastern two-thirds of Asia by Savage and Russell (1983). Prothero and Schoch (1989: Fig. 28.1) showed it as occurring throughout the Pleistocene of Asia. The genus is also known from Pleistocene deposits in and near its current range in South America and southeastern Asia. While many blank and ambiguous areas remain, the history of Tapirus can tentatively be summarized as follows. Tapirus appears to have originated (perhaps itself as a lineage of oddball survivors) in temperate Europe during the earliest Miocene, then spread to temperate North America and Asia by the late Miocene. Fringe populations adapted to tropical Central and South America and southeastern Asia by the late Pliocene to early Pleistocene. During the Pleistocene, the genus was modestly diverse in both temperate and tropical regions until the temperate forms disappeared near the end of the epoch. Three of the four surviving species inhabit tropical regions in Central and South America and southeastern Asia. The fourth, T. pinchaque, inhabits colder high elevations of the Andes (Nowak, 1991). Tapirus displays a pattern of a group originally adapted to one type of ecosystem, with a few populations later moving into marginal regions. Subsequently, the populations adapted to the original ecosystem became extinct and only those in the several marginal areas survived as disjunct species. The differences between the Pleistocene and Recent distributions of Tapirus suggest that the vulnerability factor(s) causing the diversity bottleneck primarily acted on temperate forms. The woolly mountain tapir, T. pinchaque, may represent a remnant of the more typical temperately adapted tapirs. Perhaps the factor(s) which allowed it to survive the diversity bottleneck were different from those which favoured the extant tropic-adapted forms.

Family Elephantidae This family of proboscideans presents a slightly different pattern. Origi-

Oddball survivors 9 nating in the tropical African Miocene, elephants diversified until at least nine lineages (not counting dwarfed island types) were present on four continents from the tropics to the arctic about 1,000,000 years ago (Maglio, 1973). Both modern species (Elephas maximus of southern Asia and Loxodonta africana of Africa) are tropical with disjunct distributions. The diversity bottleneck eliminated one genus, Mammuthus, with two lineages of temperate to arctic forms. Loxodonta atlantica, which seems to have been characteristic of less tropical habitats in northernmost and southernmost Africa, also disappeared. Elephas lost four lineages: E. platycephalus of southern Asia and E. hysudrindicus of the East Indies were undoubtedly tropical, E. namadicus of Eurasia was temperate and E. iolensis primarily inhabited the less tropical areas of northernmost and southernmost Africa. In summary, the Elephantidae had tropical origins and diversified into temperate and even arctic regions until the temperate and arctic forms were more typical of the group than the tropical types. A Pleistocene diversity bottleneck eliminated the arctic, the temperate, and several subtropical to tropical forms while the two survivors are tropical. In this example, the forms which invaded new areas during diversification became extinct, while their oddball survivors inhabit areas similar to those utilized by the stem group.

Subfamily Tremarctinae These bears originated in North America and became widespread and moderately diversified in the New World during the Pleistocene. Their oddball survivor is the spectacled bear, Tremarctos ornatus, of the Andes. In North America three Pleistocene species of this subfamily are recognized: T. floridanus, Arctodus simus, and A. pristinus (Kurten and Anderson, 1980). Valid South American Pleistocene forms appear to include at least three species of Arctodus and one of Tremarctos (Kurten, 1967). The extant spectacled bear is primarily herbivorous and appears to be adapted to a great variety of mountain habitats (Peyton, 1980). Kurten (1966) noted that the dentition of T. floridanus and T. ornatus were virtually identical, but pointed out substantial differences in the post-cranial skeleton. Kurten and Anderson (1980) concluded that T. floridanus was almost exclusively herbivorous. Kurten and Anderson (1980: Fig. 11 .12) showed that it was known almost exclusively from the south-central and southeastern United States into eastern Mexico. They indicated two records from New Mexico, but these were not mentioned by Harris (1985). The various species assigned to Arctodus are generally considered to be more carnivorous (Kurten, 1967; Kurten and Anderson, 1980). The predominantly North American form, A. simus, was very large and widespread

10 W .A. Akersten (Kurten and Anderson, 1980: Fig. 11.13; Richards et al., this volume), mostly in temperate areas. The somewhat smaller A. pristinus is known only from the southeastern United States (Kurten and Anderson, 1980). Distribution of the South American species is not well known. It appears that the tremarctine bears had a temperate origin, diversified into many New World habitats from arctic to tropical, with relatively herbivorous and relatively carnivorous lineages. Their only oddball survivor is a mountain vegetarian restricted to a portion of the Andes Mountains. This pattern is somewhat similar to that of Tapirus, except that the tremarctine bears never migrated out of their continent of origin and that at least one extinct species, A. simus, occurred in a great variety of environments.

Tribe Ovibovini The living muskox, Ovibos moschatus, is the only definite survivor of what appears to have been a major diversification of the Ovibovini. The takin, Budorcas taxicolor, may also be an ovibovine, but its relationships are controversial and the literature on its ecology contains many contradictions (Lent, 1988; Neas and Hoffmann, 1987). For these reasons, I exclude the takin from consideration here; its inclusion would not greatly alter this scenario. Ovibos is highly specialized for low-snow-cover arctic tundra conditions (Lent, 1988; Nowak, 1991). Extinct ovibovines ranged over much of Eurasia, Africa, and North to Central America. Gentry (1971) assigned a number of Miocene and younger Eurasian forms to the Ovibovini. In a subsequent paper, Gentry (1978) briefly discussed fragmentary evidence of a modest Late Cenozoic ovibovine radiation in Africa, ranging into South Africa. In North America, there were at least four lineages of Pleistocene ovibovines: Soergelia, Bootherium (including Symbos as per McDonald and Ray, 1989), Praeovibos, and Ovibos; five lineages if Euceratherium is included within the Ovibovini (Kurten and Anderson, 1980). Of these, Bootherium is by far the best known; specimens have been collected from Alaska and the Yukon to Texas and Louisiana (McDonald and Ray, 1989). According to Kurten and Anderson (1980), Euceratherium has been reported from Maryland to Mexico. Soergelia is rare in the New World, but has been found in Irvingtonian faunas of the Yukon, Kansas, and Texas. Fossil Ovibos occurred in tundra and arctic-steppe faunas . McDonald et al. (1991) revised Praeovibos, recognizing three species from early to middle Pleistocene arctic deposits of Eurasia and eastern Beringia. They considered this genus to be the probable ancestor of both Bootherium and Ovibos, although all three appear to have been contemporaries. If the assignment of numerous Mio-Pliocene Old World forms to the Ovibovini is correct, then this group has undergone a drastic reduction in

Oddball survivors

11

diversity. One might question the use of the term 'diversity bottleneck' for such a long, drawn-out reduction, but some bottles have short necks, others have longer ones. At any rate, the known history of the Ovibovini indicates a temperate origin, expansion and diversification into tropic and arctic environments, and subsequent reduction to a single coldadapted oddball survivor. Whatever factor(s) caused the elimination of the temperate to tropical forms, they did not affect the very cold-adapted lineage. Family Equidae The Miocene through Pleistocene record of horses includes two major diversity bottlenecks. As shown by MacFadden (1992: Fig. 5.14), a dramatic reduction of equid genera occurred in the late Miocene to Pliocene, leaving only the Dinohippus-Equus lineage. Except for the loss of several browsing forms, the available evidence does not seem to show significant changes in geographic or ecologic distributions before or after the bottleneck. However, most members of the surviving lineage were monodactyl and most members of the extinct groups were tridactyl. Perhaps more important was the appearance of the front leg 'stay' apparatus described by Hermanson and MacFadden (1992) in the surviving lineage. Thus, the vulnerability factor for this diversity bottleneck may have been related to morphology. The second diversity bottleneck among the horses consisted of the late Pleistocene extinction of most members of the Dinohippus-Equus lineage. MacFadden (1992) estimated that the genus Equus includes about 25-30 valid species, extinct and extant. The count could be higher because Winans (1989) suggested that any or all of her five North American Equus groups (four of which survived until the late Pleistocene) may actually be multispecific. The number of extant species (including the recently extinct onager) is between seven and ten depending on the degree of lumping and splitting(Nowak, 1991). While this is quite a large number, it is enough of a reduction from the known Pleistocene diversity that modem equids can be viewed as oddball survivors. As a result of this diversity bottleneck, New World horses completely disappeared. Old World forms were greatly reduced in distribution and moderately reduced in diversity. The horse story seems to present a case in which a broadly diverse group went through a major diversity bottleneck, their oddball survivors subsequently rediversified, and then went through another, less severe, bottleneck. Again, temperate forms were among those most severely affected during the Pleistocene bottleneck. The typical way of looking at the extinction of New World horses has been to say, 'Why did horses become extinct in the New World at the end

12 W.A. Akersten of the Pleistocene if they did so well when reintroduced by the Spanish?' When viewed from the oddball-survivor perspective, the question can be rephrased as, 'How did the New World and Old World horses which became extinct in the Pleistocene differ from those oddballs which survived?' The former assumes that the introduced forms were essentially the same as the extinct ones; the latter assumes that they differed in some significant way(s) which are not currently obvious to us.

Oddball survivors and the terminal Pleistocene extinction While the examples given here do not constitute a significant sample of the many forms which underwent a diversity bottleneck at or near the end of the Pleistocene, they show some suggestive similarities in patterns. The geographic/ ecologic distributions of all were greatly reduced and all lost temperate members of the group, though other members may also have disappeared. A comprehensive examination and comparison of survivorship patterns for all forms which underwent a diversity bottleneck near the end of the Pleistocene should provide information about the cause(s) of that extinction event. (It might also be informative to try this approach in exploring the extinction of dinosaurs, sensu stricto, using birds and crocodilians as their oddball survivors or of ammonoids using the extant Nautilus.)

Conclusions When a diverse group of animals undergoes a diversity bottleneck so that relatively few forms survive, the survivors will usually be atypical in one or more ways when compared with the majority of the group; in other words, they will be oddball survivors. Thus, we must refrain from using the survivors of a diversity bottleneck as absolute analogues for their extinct brethren unless corroborated by independent evidence. However, the factor(s) which made the typical forms vulnerable to extinction may not be determinable from the fossil record. Oddball survivors may be generalized or specialized, but most inhabited relatively small areas on the fringe of the geographic, ecologic, morphologic, or other distribution of the group as a whole before the diversity bottleneck. Since these distributions change through time, the oddballs may be similar in morphology, ecology, or distribution to forms typical of the group at an earlier time. If several oddballs survive, they often have disjunct distributions. Many socalled 'relicts' or 'living fossils' are oddball survivors. As appears to have happened in the tapirs and the horses, oddball survivors can rediversify. This may explain the frequently encountered pattern where an obscure or even unknown group gives rise to a great

Oddball survivors

13

diversity of related forms. Determining the patterns of change for a group before and after a diversity bottleneck may yield information about the factor(s) which caused extinction of the majority of forms. If the geographic distribution of oddball survivors differs from that of the extinct forms, a factor related to geography may be implicated. If extinct forms can be shown to be ecologically different from their oddball survivors, an ecologic factor should be suspected. Similarly, consistent differences in morphology between extinct forms and their oddball survivors may be of importance. Of course, these are not necessarily mutually exclusive; a factor may be expressed in more than one way. In cases where a number of different groups have undergone roughly simultaneous diversity bottlenecks, analysing and comparing their patterns of oddball survivorship has the potential to reveal information about the causes of major extinction events.

Acknowledgments I thank the many colleagues with whom I discussed this idea over the years. Comments from an anonymous reviewer were helpful in clarifying several portions of the manuscript. LITERATURE CITED

Carroway, L.N., and B.J. Verts. 1993. Aplodonta rufa. Mammalian Species 431:1-10. Cox, L.R. 1960. General characteristics of Gastropoda. Pp. 84-169 in R.C. Moore and C.W. Pitrat (eds), Treatise on Invertebrate Paleontology, Part I, Mollusca 1. Geological Society of America and University of Kansas Press, Lawrence. Gentry, A.W. 1971. The earliest goats and other antelopes from the Samos Hipparion fauna. Bulletin of the British Museum (Natural History), Geological Series 20:229-296. - 1978. Bovidae. Pp. 540-572 in V.J. Maglio and H.B.S. Cooke (eds), Evolution of African Mammals. Harvard University Press, Cambridge. Harris, A.H. 1985. Late Pleistocene Vertebrate Paleontology of the West. University of Texas Press, Austin, 293 pp. Hermanson, J.W., and B.J. MacFadden. 1992. Evolutionary and functional morphology of the shoulder region and stay-apparatus in fossil and extant horses (Equidae). Journal of Vertebrate Paleontology 12:377-386. Hooijer, D.A. 1952. Fossil mammals and the Plio-Pleistocene boundary in Java. Koninklijke Nederlandse Akademie van Wetenschappen, Series B, Physical Sciences 55:436-443. Jefferson, G.T. 1989. Late Cenozoic tapirs (Mammalia: Perissodactyla) of Western

14

W .A. Akersten

North America. Natural History Museum of Los Angeles County, Contributions in Science 406:1-21. Kingdon, J. 1974. East African Mammals: An Atlas of Evolution in Africa. Volume II, Part B (Hares and Rodents). University of Chicago Press, Chicago, pp. 342704.

Korth, W.W. 1986. Aplodontid rodents of the genus Pelycomys Galbreath from the Orellan (middle Eocene) of Nebraska. Journal of Mammalogy 67:545-550. Kurten, B. 1966. Pleistocene bears of North America, 1. Genus Tremarctos, spectacled bears. Acta Zoologica Fennica 115:1-120. - 1967. Pleistocene bears of North America, 2. Genus Arctodus, short-faced bears. Acta Zoologica Fennica 117:1--{i0. - 1968. Pleistocene Mammals of Europe. Aldine Publishing, Chicago, 317 pp. Kurten, B., and E. Anderson. 1980. Pleistocene Mammals of North America. Columbia University Press, New York, 442 pp. Lemche, H. 1957. A new living deep-sea mollusc of the Cambro-Devonian Class Monoplacophera. Nature 179:413-416. Lent, P.C. 1988. Ovibos moschatus. Mammalian Species 320:1-9. Lund, W.L., R. Lund, and G.A. Klein. 1985. Coelacanth feeding mechanisms and ecology of the Bear Gulch coelacanths. Pp. 492-500 in J.T. Dutro, Jr (ed.), Neuvieme Congres International de Stratigraphie et du Geologie du Carbonifere. Compte rendu, volume 5; Paleontology, paleoecology, paleogeography. Southern Illinois University Press, Carbondale. Lundelius, E.L., Jr. and B.H. Slaughter. 1976. Notes on North American tapirs. Pp. 226-243 in C.S. Churcher (ed.), Athlon, Essays on Palaeontology in Honour of Loris Shano Russell. Royal Ontario Museum, Life Sciences Miscellaneous Publication. McDonald, J.N., and C.E. Ray. 1989. The autochthonous North American musk oxen Bootherium, Symbos, and Gidleya (Mammalia: Artiodactyla: Bovidae). Smithsonian Contributions to Paleobiology 66:1-77. McDonald, J.N., C.E. Ray, and C.R. Harington. 1991. Taxonomy and zoogeography of the musk ox genus Praeovibos Staudinger, 1908. Pp. 285-314 in J.R. Purdue, W.E. Klippel, and B.W. Styles (eds), Beamers, Bobwhites, and Bluepoints: Tributes to the Career of Paul W. Parmalee. Illinois State Museum Scientific Papers 23. MacFadden, B.J. 1992. Fossil Horses: Systematics, Paleobiology, and Evolution of the Family Equidae. Cambridge University Press, Cambridge, 369 pp. McLean, J.H. 1979. A new monoplacopheran limpet from the continental shelf off southern California. Natural History Museum of Los Angeles County, Contributions in Science 370:1-19. Maglio, V.J. 1973. Origin and evolution of the Elephantidae. Transactions of the American Philosophical Society 63:1-149. Marshall, L.G., R.F. Butler, R.E. Drake, and G.H. Curtis. 1982. Geochronology of

Oddball survivors 15 type Uquian (Late Cenozoic) land mammal age, Argentina. Science 216:986-989. Neas, J.F., and R.S. Hoffmann. 1987. Budorcas taxicolor. Mammalian Species 277:17.

Nowak, R.M. 1991. Walker's Mammals of the World. 5th edition, 2 volumes. Johns Hopkins Press, Baltimore, 1629 pp. Peyton, B. 1980. Ecology, distribution, and food habits of spectacled bears, Tremarctos ornatus, in Peru. Journal of Mammalogy 61:639--652. Pojeta, J., Jr, and M. Gordon, Jr. 1987. Class Cephalopoda. Pp. 329-358 in_R.S. Boardman, A.H. Cheetham, and A.J. Rowell (eds), Fossil Invertebrates. Blackwell Scientific Publications, Palo Alto. Prothero, D.R., and R.M. Schoch. 1989. Origin and evolution of the Perissodactyla: summary and synthesis. Pp. 504-529 in D.R. Prothero and R.M. Schoch (eds), The Evolution of Perissodactyls. Oxford Monographs on Geology and Geophysics 15. Rensberger, J.M. 1975. Haplomys and its bearing on the origin of the aplodontoid rodents. Journal of Mammalogy 56:1-14. Richards, R.L., C.S. Churcher, and W.D. Turnbull. (This volume). Distribution and size variation in North American short-faced bears, Arctodus simus. Runnegar, B. 1987. Class Monoplacophera. Pp. 297-304 in R.S. Boardman, A.H. Cheetham, and A.J. Rowell (eds), Fossil Invertebrates. Blackwell Scientific Publications, Palo Alto. Savage, D.E., and D.E. Russell. 1983. Mammalian Paleofaunas of the World. Addison-Wesley Publishing, Reading, Massachusetts, 432 pp. Simpson, G.G. 1944. Tempo and Mode in Evolution. Columbia University Press, New York, 237 pp. White, J.L. 1993. Indicators of locomotor habits in xenarthrans: evidence for locomotor heterogeneity among fossil sloths. Journal of Vertebrate Paleontology 13:230-242. Winans, M.C. 1989. A quantitative study of the North American fossil species of the genus Equus. Pp. 262-297 in D.R. Prothero and R.M. Schoch (eds), The Evolution of Perissodactyls. Oxford Monographs on Geology and Geophysics 15.

Comparison of mammalian response to glacialinterglacial transitions in the middle and late Pleistocene

Anthony D. Barnosky, Tina I. Rouse, Elizabeth A. Hadly, David L. Wood, Felicia L. Keesing, Victor A. Schmidt

Abstract A succession of fossil faunas from Porcupine Cave, Colorado, records the response of mammals as a glacial stage gave way to an interglacial stage nearly 425,000 years ago. As in the Wisconsinan to Holocene glacial-interglacial transition, relative abundances of taxa changed and some populations became extinct. In contrast to the Wisconsinan-Holocene transition, however, no-analogue species assemblages persisted across the climatic boundary, and neither their presence nor species richness is correlated with seasonal equability in terms of temperature (although equability in precipitation may be important). These discoveries suggest that the response of mammals to global warming is more complex than previously thought and that response to different events may vary according to the species involved, the magnitude of the warming event, and orbital parameters governing particular aspects of seasonal equability.

Introduction In the past three decades considerable data have been published to document how mammals respond to the transition from glacial times to interglacial times (Graham, 1976, 1984, 1986, 1988; Graham and Grimm, 1990; Graham and Lundelius, 1984; Graham and Mead, 1987; Graham and Semken, 1976; Grayson, 1987; Rhodes, 1984; Webb, 1991; Webb and Barnosky, 1989). These climatic transitions are thought to be relatively rapid (on the order of hundreds or a few thousand years) (Hays et al., 1976; Imbrie et al., 1984; Imbrie and Imbrie, 1979; Shackleton et al., 1990; Winograd et al., 1992), and thus to be natural analogues for the changes that might be expected with 'greenhouse' warming, although some models

Mammalian response to glacial-interglacial transitions

17

predict greenhouse warming to be even faster and of greater magnitude (Houghton et al., 1990). Most of the published studies that track mammalian response across glacial-interglacial transitions focus on what happened as the last glacial gave way to the Holocene some 10,000 years ago. However, according to the marine oxygen-isotope curve, there have been at least nine times in the past 800,000 years when glacial conditions transformed relatively rapidly into interglacials (Shackleton and Opdyke, 1973). Is the response similar each time? To address this question we examine new data from a glacial-interglacial transition that took place high in the Rocky Mountains of Colorado, United States, between approximately 350,000 and 500,000 years ago. In particular, we examine the general applicability of three hypotheses, all of which arose primarily from studies of the terminal PleistoceneHolocene transition. First, the terminal Pleistocene-Holocene studies universally suggest that mammal species respond more or less individualistically to climatic warming, such that the species composition of communities changes over time-scales of a few hundred to a few thousand years (Graham, 1976, 1984, 1986, 1988; Graham and Grimm, 1990; Graham and Lundelius, 1984; Graham and Mead, 1987; Graham and Semken, 1976; Grayson, 1987; Guilday, 1984; Rhodes, 1984; Webb, 1991; Webb and Bamosky, 1989). More controversial are two other hypotheses: (1) that changes in seasonality drive mammalian community composition and species richness at any given point on the earth's surface (Graham and Mead, 1987; Hibbard, 1960; Lundelius et al., 1983) and (2) that climatic amelioration, in the sense of a glacial interval warming to an interglacial, can cause widespread extinctions such as characterized the terminal Pleistocene-Holocene transition (Graham and Lundelius, 1984; Guilday, 1984; Guthrie, 1984; King and Saunders, 1984). The climaticallyinduced extinction idea has been widely challenged by those who champion the idea that human hunters were responsible for the decimation of especially (but not exclusively) the megafauna at the end of the Pleistocene (Martin, 1967, 1973, 1984, 1990; Mossiman and Martin, 1975). Indeed, a major point in the human overkill hypothesis is that North American glacial-interglacial transitions older than the terminal Pleistocene event (i.e., those without human presence) are assumed not to be associated with extinction or even faunal turnover. In fact, such older terrestrial glacial-interglacial sequences are so little known that testing the general applicability of any of these hypotheses derived from terminal Pleistocene studies has not been attempted until now. A record that allows comparison of an earlier climatic-amelioration event with the Pleistocene-Holocene transition comes from deposits at 2900 metres elevation within Porcupine Cave, Colorado (Fig. 1). A 2.5-m thick stratigraphic section located 30 m within the cave, at the bottom of a

18 A.O. Barnosky et al. shaft known as The Pit, includes alternating glacial sediments and interglacial sediments. The glacial sediments are brown nodular clay, which had to have been formed under a wetter subterranean environment than exists in most parts of the cave today, whereas the interglacial deposits are a loose, flour-like dust that would have required generally drier conditions (A. Barnosky and Rasmussen, 1988). This interpretation concords with information that demonstrates middle Pleistocene glacial times (stages 22 and 20) were wet compared to interglacials (stages 21 and 19) 180 km south of Porcupine Cave (Rogers et al., 1985). The mammals that lived near the entrance to the cave are well represented by thousands of identifiable specimens in both the glacial and interglacial deposits. Bones and teeth accumulated in the sediments as woodrats (genus Neotoma) dragged raptor pellets, bone-laden carnivore scats, and small pieces of large-animal carcasses into the cave. Other bones were supplied by carnivores such as badgers (Taxidea taxus), coyotes (Canis latrans), or wolves (C. lupus) when they consumed their prey in or near the cave. Detailed taphonomic studies of such deposits indicate sampling was robust for rodent-sized mammals that lived within about 5 km of the fossil site (E. Barnosky, 1992). The subsequent analyses are confined to the rodents that are most consistently sampled and most informative: the arvicolines (voles and lemmings) and sciurids (squirrels).

Chronology We focus on the fauna from the uppermost of the day-dust couplets. Figure 2 illustrates that the glacial deposits comprise excavation levels 5 and 4, the transition between the glacial and interglacial deposits is bracketed by levels 4 and 3, and the interglacial deposits include levels 3, 2, and 1. Because material for radiometric dating is lacking, age determination relies primarily on biostratigraphic techniques. Levels 1 to 5 contain the arvicoline rodents Terricola meadensis (extinct), Mictomys meltoni (extinct), Lemmiscus curtatus (with variants showing both 4 and 5 triangles on M 1), and either Microtus montanus or M. longicaudus (these two currently cannot be discriminated by their molars). Detailed biochronological networks (Repenning, 1987, 1992; Repenning et al., 1990; Wood and Barnosky, 1994) indicate that this assemblage would be expected in the Colorado Rockies between 365,000 and 487,000 years ago. At the glacial-interglacial transition, the vole with Allophaiomys pliocaenicus morphology of the M 1 disappears from the record. Elsewhere in North America, voles with this tooth morphology are thought to have gone extinct by 800,000 years ago (Barendregt et al., 1991; Repenning, 1987, 1992; Repenning et al., 1990), although Allophaiomys as a genus survived until perhaps as late as 450,000 years ago (Guilday et al., 1984;

■ Lemmiscus curtatus

m

Cynomys (Leucocrossuromys)

lllcynomys (Cynomys)

Mictom ys borealis

;; .E

0 ai

.a

14 12 10 8 6

E :::,

z

Oto 10

11 to 20

21 to 30

31 to 40

41 to 50

51 to 60

Age class (African Elephant Years)

Port Kennedy Mastodons Age profile (lower dentition)

10 Ill

iii :::,

8

"C

·;

;;

6

0

4

-... .E GI

.0

E :::,

z

o to 10

11 to 20

21to 30

31 to 40

41 to 50

51 to 60

Age class (African Elephant Years)

FIGURE 2. Age profiles for Port Kennedy mastodons suggesting selective mortality of young individuals. Height of column indicates number of individuals rather than percentage of MNI. The same individuals may be counted in the upper-dentition and lower-dentition samples.

Port Kennedy mastodon mortality, Pennsylvania 91 mortality of very young and young mastodons is clear, with the majority of individuals judged to be less than five years old at the time of death. These age profiles differ significantly from other mass accumulations of mastodons in North America such as the death assemblages reported by Saunders (1977) from sites in Missouri. What taphonomic factor, or factors, could be responsible for the selective mortality of young mastodons from the Port Kennedy Cave?

Taphonomic factors

Differential preservation and collecting bias It is possible that the larger teeth of adult mastodons were more likely to be weakened along the median sulcus or between the transverse crests and thus fragmented in the wet conditions that plagued the collecting efforts at Port Kennedy. Indeed, there are a moderate number of mastodon tooth fragments which could not be specifically identified or associated. A piece-by-piece examination, however, demonstrated that fragments of young mastodon teeth are at least as common as fragments of adult teeth. Therefore, the fragments do not explain the significantly large percentage of juvenile teeth. The selective mortality pattern does not seem to be the result of differential preservation and collection bias.

Pitfall trap The morphology of the Port Kennedy Cave (Fig. 1) and nature of the fissure fill suggest that a vertical solution feature in the limestone opened to the surface, forming a sink-hole that acted as a pitfall trap for a geologically short period of time. This mechanism for accumulation of organic remains is possible for some of the taxa found there. For example, the turtles from Port Kennedy are of the type that would be expected to make overland journeys and be susceptible to a pitfall trap. Other taxa in the Port Kennedy fauna may also have been trapped or fallen to their death in the sink-hole. The selective mortality of mastodons is not adequately explained by the pitfall-trap mechanism, however. One could argue that the young mastodons were inexperienced and more likely to wander into the natural trap, but the same would be true for other herbivores at the site. Other moderately abundant taxa such as tapir and peccary (MNI = 10 to 20) show no selective mortality of juveniles. Alternatively, there could have been a size filter on the trap if the opening of the sink-hole was only large enough to accommodate young-mastodon-sized animals. This is an unlikely scenario since there are also a few full-sized mastodons in the sample.

92 E.B. Daeschler

Carnivore accumulation The mastodon remains that were incorporated into the Port Kennedy Cave could have been the result of carnivore predation or scavenging. Direct evidence such as carnivore tooth marks are not found on the mastodon material, although one would not expect such damage on mastodon teeth. Carnivores may have dragged young mastodon remains to hiding places near the sink-hole opening. Dens or hiding places could have been located under ledges of the Triassic shale, or in passages leading to the solution cavity. Some of the mastodon remains would have eventually tumbled or been washed down and become buried in the fissure fill. Unfortunately, our knowledge of the shape of the solution cavity is limited by the two-dimensional sketches in early publications (Fig. 1). If the majority of the mastodon remains found in Port Kennedy Cave represent the prey of a carnivore, which carnivore was the agent of accumulation? Several taxa from the site are possibilities. The felids Smilodon gracilis and Panthera onca, the ursid Arctodus pristinus, or the canid Canis cf. C. armbrusteri are all large enough to prey upon a young mastodon. There is no direct evidence to implicate any of these taxa. It is tempting to suspect Smilodon because of their specialized dentition for mortally wounding proboscideans and other large, slow-footed herbivores (Kurten and Anderson, 1980). Arctodus, on the other hand, is the most abundant large carnivore from Port Kennedy Cave. Interestingly, at Frankstown Cave in Blair County, PA, a limestone fissure-fill contained the remains of six or seven juvenile mastodons and several individuals of Arctodus (Holland, 1908; Peterson, 1926). Canis cf. C. armbrusteri and Panthera onca are both rare at Port Kennedy. It is possible that more than one of these taxa were preying upon the young mastodons, or that other large carnivores not found at the site had a role. None of the carnivores from the site shows an age profile of very young and very old individuals as one might expect from attritional deaths at a denning site (Kurten, 1969).

Discussion The most robust hypothesis to explain the pattern of selective mortality of young mastodons at Port Kennedy Cave is carnivore accumulation. Regardless of which taxon or taxa were responsible for the mastodon remains in the Port Kennedy Cave, it seems reasonable to assert that juvenile proboscidean remains in Pleistocene cave sites may generally indicate the activity of carnivores. A compelling case for a Pleistocene carnivore accumulation of young proboscideans is presented by Rawn-Schatzinger (1992) for the remains of mammoths (Mammuthus cf. M. columbi) and the scimitar cat (Homotherium serum) from Friesenhahn Cave, Texas. At this site a very large num-

Port Kennedy mastodon mortality, Pennsylvania 93 ber of juvenile mammoths (300 to 400 individuals) are associated with an attritional assemblage of the scimitar cat, suggesting that the site was a denning area where the felids returned with their prey. Most of the young mammoths were around two years of age, the age at which juvenile modem elephants begin to separate from the herd for play and, thus, become viable targets for large predators (Rawn-Schatzinger, 1992). Laursen and Bekoff (1978) reported that modem African elephant juveniles fall prey to lions, hyaenas, and wild dogs usually at less than two years of age. The predator-prey dynamics in the Pleistocene woodlands of eastern North America might have been quite different than those in the more open habitats of mammoths or modem East African elephants. Mastodons were believed to have been primarily solitary browsers, filling a niche somewhat analogous to the modem moose (Haynes, 1991). Mastodons may have fed on the abundant nut crop that was produced by some of the hardwoods in the terrain around the Port Kennedy sink-hole. Without the protection of a group and feeding in forests with thick undergrowth where the mother might not see danger, young mastodons would have been susceptible to carnivore predation from the moment they were born. Indeed, the remains from Port Kennedy include several individuals with virtually no wear on the dp2's, suggesting that they were not yet weaned. There are relatively large numbers of other forest herbivores such as ground sloth, peccary, and tapir from Port Kennedy (see Table 1). These herbivores are represented almost exclusively by subadults and adults. The adult body mass of these taxa are similar to or less than a juvenile mastodon. It is not clear whether carnivores played a role in the accumulation of these and other taxa found at the site, but the scenario for the selective mortality of mastodons strengthens the argument for the significance of carnivores in creating other portions of the fossil assemblage from the Port Kennedy Cave.

Acknowledgments The author would like to acknowledge Earle E. Spamer for his collaboration on a historical perspective of the Port Kennedy site. Thanks also go to Jeffrey J. Saunders for his instruction and advice in discussions of mastodon tooth identification and age determination.

LITERATURE CITED

Berta, A. 1987. The sabercat Smilodon gracilis from Florida and a discussion of its relationships. Bulletin of the Florida State Museum, Biological Sciences 31:1-63. Cope, E.D. 1871. Preliminary report on the Vertebrata discovered in the Port

94 E.B. Daeschler Kennedy bone cave. Proceedings of the American Philosophical Society 12:73102. - 1899. Vertebrate remains from the Port Kennedy bone deposit. Journal of the Academy of Natural Sciences of Philadelphia 11:193-267. Daeschler, E., E.E. Spamer, and D.C. Parris. 1993. Review and new data on the Port Kennedy local fauna and flora (Late Irvingtonian), Valley Forge National Historical Park, Montgomery County, Pennsylvania. The Mosasaur 5:23-41. Gidley, J.W., and C.L. Gazin. 1938. The Pleistocene vertebrate fauna from Cumberland Cave, Maryland. Bulletin of the United States National Museum 171:1-99. Guilday, J.E., J.F.P. Cotter, D. Cundall, E.B. Evenson, J.B. Gatewood, A.V. Morgan, A.D. Mccrady, D.M. Peteet, R. Stuckenrath, and K. Vanderwal. 1984. Palaeoecology of an Early Pleistocene (Irvingtonian) Cenote: preliminary report on the Hanover Quarry #1 Fissure, Adams County, Pennsylvania. Pp. 119-132 in W.C. Mahaney (ed.), Correlation of Quaternary Chronologies. GeoBooks, Norwich, England. Hall, E.R. 1936. Mustelid mammals from the Pleistocene of North America; with systematic notes on some recent members of the genera Mustela, Taxidea and Mephitis. Carnegie Institution of Washington Publication 473:41-119. Haynes, G. 1991. Mammoths, Mastodonts and Elephants. Cambridge University Press, Cambridge, 413 pp. Hibbard, C.W. 1955. Notes on microtine rodents from Port Kennedy Cave deposit. Proceedings of the Academy of Natural Sciences of Philadelphia 107:87-97. Hirschfeld, S.E., and S.D. Webb. 1968. Plio-Pleistocene megalonychid sloths of North America. Bulletin of the Florida State Museum, Biological Sciences 12:213-296. Holland, W.J. 1908. A preliminary account of the Pleistocene fauna discovered in a cave opened at Frankstown, Pennsylvania, in April and May, 1907. Annals of the Carnegie Museum 4:228-233. Kurten, B. 1967. Pleistocene bears of North America. 2. Genus Arctodus, shortfaced bears. Acta Zoologica Fennica 117:1-60. - 1969. Cave bears. Studies in Speleology 2(1):13-24. Kurten, B., and E. Anderson. 1980. Pleistocene Mammals of North America. Columbia University Press, New York, 442 pp. Laursen, L., and M. Bekoff. 1978. Loxodonta africana. Mammalian Species 92:1-8. Laws, R.M. 1966. Age criteria for the African elephant Loxodonta a. africana. East African Wildlife Journal 4:1-37. Mercer, H.C. 1899. The bone cave at Port Kennedy, Pennsylvania, and its partial excavation in 1894, 1895, and 1896. Journal of the Academy of Natural Sciences of Philadelphia 11:269-286. Nowak, R.M. 1979. North American Quaternary Canis. University of Kansas Museum of Natural History Monograph 6:1-154. Parris, D.C., and E. Daeschler. 1995. Pleistocene turtles of Port Kennedy Cave (Late Irvingtonian), Montgomery County, Pennsylvania. Journal of Paleontology 69:563-568.

Port Kennedy mastodon mortality, Pennsylvania 95 Peterson, O.A. 1926. The fossils of the Frankstown Cave, Blair County, Pennsylvania. Annals of the Carnegie Museum 16:249-315. Pfaff, K.S. 1990. lrvingtonian Microtus, Pedomys, and Pitymys (Mammalia, Rodentia, Cricetidae) from Trout Cave No. 2, West Virginia. Annals of the Carnegie Museum 59:105-134. Rawn-Schatzinger, V. 1992. The Scimitar Cat Homotherium serum Cope: osteology, functional morphology, and predatory behavior. Illinois State Museum Reports of Investigations 47:1-80. Repenning, C.A., and F. Grady. 1988. The microtine rodents of the Cheetah Room Fauna, Hamilton Cave, West Virginia, and the spontaneous origin of Synaptomys. United States Geological Society Bulletin 1853:1-32. Saunders, J.J. 1977. Late Pleistocene Vertebrates of the Western Ozark Highland, Missouri. Illinois State Museum Reports of Investigations 33:1-118. Van Valkenburgh, B., F. Grady, and B. Kurten. 1990. The Plio-Pleistocene cheetah-like cat Miracinonyx inexpectatus of North America. Journal of Vertebrate Paleontology 10:434--454. Wheatley, C.M. 1871. Notice of the discovery of a cave in Eastern Pennsylvania, containing remains of Post-Pliocene fossils, including those of Mastodon, Tapir, Megalonyx, Mylodon, etc. American Journal of Science and Arts, 3rd ser., 1(4):235-237. APPENDIX Tooth measurements of Mammut americanum from Port Kennedy Cave• ANSP#

Position

Length•• (mm)

Width••• (mm)

18840 18841 18844 18845 257 18842 18843 18828 18846 257 291 18827 18828 18829 18830 257 18827 18829 18831 259 291

rt. dP2 rt. dP2 rt. dP2 (anterior Ioph) rt. dP2 (anterior loph) It. dP2

35.7 31 .3

32.0 29.2 30.5 31.9 37.4 28.8 31.9

It. dP2 (posterior Ioph) It. dP2 (posterior Ioph) rt. dP3 (posterior Ioph) rt. dP3 (anterior Ioph) It. dP3 rt. dP4 (posterior 2 Iophs) rt. dP4 rt. dP4 rt. dP4 rt. dP4 (?unerupted) It. dP4 It. dP4 It. dP4 (anterior 2 Iophs) It. dP4 (posterior 2 Iophs)

rt. Ml rt. Ml

35.8

46.1

43.3

35.0 46.2

60.0 67.0

72.8 66.2 64.6 76.1

66.9

86.6 88.7

52.0 61.4 52.7 53.3 59.3 52.4 52.9 59.8 70.6 67.0

Continued

96 E.B. Daeschler APPENDIX - Concluded ANSP#

Position

Length** (mm)

Width*** (mm)

18826 18827 18828 252 258 18826 18827 256 254 255 254 18832 18833 18835 19451 18832 18837 18832 18834 18838 250 292 293 194 192 193

rt. Ml rt. Ml rt. Ml (unerupted) It.Ml It. Ml It. Ml It. Ml rt.M2 lt.M2 rt. M3 (anterior loph) It. M3 (anterior loph) rt. dp2 rt. dp2 rt. dp2 It. dp2 (anterior lophid) rt. dp3 rt. dp3 (anterior !aphid) It. dp3 It. dp3 It. dp3 (posterior lophid) rt.dp4 rt. dp4 It. ml rt. m3 It. m3 (unerupted) It. m3

87.4 87.8 90.4 90.3 90.2 88.3 85.5 114.5 110.0

68.5 67.1 72.0 70.3 72.5 67.0 65.8 83.2 84.2 93.7 97.2 22.5 24.5 25.7 21.6 36.8 31.9 37.1 35.4 36.9 56.4 54.7 60.0 79.2 86.6 79.0

29.5 29.8 32.8 42.2 41.8 40.2 83.0 74.2 88.6 151.6 191.3

* Because of incompleteness, length and width measurements could not be taken on the following teeth: 18836, rt. dP2; 18847, It. dP3; 292, rt. dp3; 18839, rt. dp3. ** Length = the maximum length of the enamelled crown measured parallel to the median sulcus. *** Width = the maximum width of the enamelled crown measured perpendicular to the median sulcus.

Middle Pleistocene (early Rancholabrean) vertebrates and associated marine and non-marine invertebrates from Oldsmar, Pinellas County, Florida

P.F. Karrow, G.S. Morgan, R. W. Portell, E. Simons, K. Auffenberg

Abstract Two excavations for urban development at Oldsmar, Pinellas County, Florida, revealed, below surface soil and sand, a marine shell bed, a nonmarine black clay and sand bed of middle to late Pleistocene age referred to the Fort Thompson Formation, and underlying Miocene Arcadia Formation. Both early to middle Miocene (late Hemingfordian or early Barstovian) and late Miocene (early Hemphillian) land mammals, particularly horses, were present at the Oldsmar sites, confirming a Miocene age for the Arcadia Formation. The non-marine bed contained a mixture of terrestrial, freshwater, estuarine, and marine vertebrates, including 63 taxa of Pleistocene fish, amphibians, reptiles, birds, and mammals, as well as terrestrial and freshwater mollusks (36 taxa) and ostracodes (10 taxa). Near-coastal, poorly drained marsh, pond, and stream assemblages dominated. The marine bed overlying the non-marine bed contained more than 100 taxa of mollusks, which indicate warm, shallow, quiet waters similar to Tampa Bay today. U /Th and amino-acid age estimates on the bivalve Mercenaria from the marine unit are 206.6 ka (U /Th) and 250 ka (AA) for Oldsmar Pit 1 and 78.3 ka (U /Th) and about 120-210 ka (AA) for Oldsmar Pit 2. These dates place the sites in marine isotope stages 7 and 5, respectively, and imply the presence of an unobserved unconformity between them. The biochronology of the rich vertebrate assemblage from Oldsmar, in particular the association of the small rodents Oryzomys palustris and Sigmodon bakeri, suggests an early Rancholabrean age (late middle Pleistocene, between 300 and 130 ka). The stratigraphic occurrence of the Oldsmar vertebrate fauna below a marine shell unit substantiates

98 P.F. Karrow, G.S. Morgan, R.W. Portell, E. Simons, K. Auffenberg an age greater than 130 ka, as the site formed before the beginning of the last or Sangamonian interglacial, which was the last time sea levels were significantly higher (+6 m) than at present. The Oldsmar site provides another example of using integrated geochronology to determine the age of Florida Late Cenozoic faunas, incorporating data from vertebrate, invertebrate, and microfossil biochronologies, various numerical dating techniques (U /Th and amino acid), and relative sea level.

Introduction Two excavations for commercial and residential development near Oldsmar, Pinellas County, Florida (Fig. 1), exposed a marine shell bed overlying a non-marine bone-bearing unit. These excavations were visited in the course of a study of Quaternary sea-level history in west-central Florida, begun in January 1984. The low-relief terrain of the Florida Gulf Coast, with the water-table near ground surface, dictates that exposures be sought in man-made excavations with pumping systems to lower the water table. The principal aim of this study was to collect mollusks for amino-acid analysis to establish a chronology of former high sea level stands, which represent past interglacial times. At each site studied, collections of mollusks were made from which to infer a correlation to known biostratigraphic units, as well as for future reference because shell pits, the chief source of aggregate for construction in the area, are only temporarily accessible for study. The two excavations described here are located northwest of Tampa Bay on a nearly flat surface which slopes gently southward toward the bay. This report describes the stratigraphy, palaeontology (both vertebrate and invertebrate), and age, as determined by various dating methods, and attempts a palaeoenvironmental interpretation and correlation of the Oldsmar pits based on these data.

Previous work In spite of numerous studies on the Late Cenozoic strata of southern Florida, the stratigraphy and geological history remain incompletely known. There is much uncertainty about relationships between lithostratigraphic units, fossil assemblages, and marine terrace geomorphology. Each has a nomenclature lacking consistent application. Their extent and correlations, locally and more widely, remain inadequately understood. Lyons (1991) presents a recent review of the problems associated with each of the named stratigraphic units of southwestern Florida and the problems

Oldsmar Local Fauna, Florida 99

r-

-----,

'

L..........u..J

0

5 km

....

C

~Okm

FIGURE 1. Location map showing Oldsmar in relation to Tampa Bay

of stratigraphy and chronology are discussed in a collection of papers edited by Scott and Allmon (1992). Early work on marine terraces, largely based upon studies of the Atlantic and Gulf coastal plains, was extrapolated into Florida on the basis of elevation. A series of names was applied, with their ages believed to increase with elevation. White (1970) provided a description of terraces near Tampa Bay, from which it can be concluded that the sites at Oldsmar were excavated into the lowest and youngest terrace, referred to as the Pamlico terrace.

100 P.F. Karrow, G.S. Morgan, R.W. Portell, E. Simons, K. Auffenberg The extensive marine shell beds near Fort Myers were studied by DuBar (1958) and a regional summary was presented by him in 1974 (DuBar, 1974). Although little detailed work has been published on the Quaternary palaeontology of the Tampa area, a major work has appeared recently on the early Pleistocene vertebrate and invertebrate faunas of the Leisey Shell Pits near Ruskin in Hillsborough County (see Hulbert et al., 1995). Study of vertebrate faunas of the region first achieved prominence with the discovery of the 'bone bed' at Seminole Field, near St Petersburg in Pinellas County. This rich assemblage, described by Simpson (1929), is similar to others found along the Atlantic coast of Florida, in particular the Vero LF (local fauna) in Indian River County (Hay, 1916, 1917; Weigel, 1962) and the Melbourne LF in Brevard County (Gazin, 1950; Ray, 1958). In these sites the bones lay above marine shell beds and were regarded as late Pleistocene (Rancholabrean Land Mammal Age). All three of these sites postdate the last interglacial and thus are younger than 120 ka. Webb (1974) and Kurten and Anderson (1980) reviewed the status of Pleistocene vertebrate occurrences in Florida. The Leisey sites near Ruskin on the eastern shore of Tampa Bay (Fig. 1), have yielded rich vertebrate assemblages of early lrvingtonian (early Pleistocene) age from between marine shell beds (Hulbert and Morgan, 1989; Webb et al., 1989). Much light could be shed on the confusing state of southern Florida stratigraphy if reliable and applicable dating methods were available. Several methods have been tried, with varying degrees of success. Coral has been generally accepted to be the most reliable material for uraniumseries (U /Th) dating, but has been surprisingly hard to find at sites along the southwest Gulf Coast of Florida. Muhs et al. (1992) have obtained a few uranium-series ages on corals from sites in southern Florida. Aminoacid analysis on the bivalve Mercenaria was used by Mitterer (1974, 1975) to derive a young chronology for southern Florida shell beds. These results were later recalculated and time-stretched by Wehmiller and Belknap (1978). Multi-component studies at the Leisey sites have used palaeomagnetic, strontium-isotopic, sea-level, and biochronologic analyses to derive an age of 1.5-1.1 million years for the early Irvingtonian bone bed (Jones, 1992; Webb et al., 1989).

Site descriptions Oldsmar I (FK-9) The site is located in the NW¼ Section 24, T28S, R16E (Latitude 28° 2'30" north and Longitude 82° 39' west), which is north of the CSX Railroad

Oldsmar Local Fauna, Florida 101 and Florida Route 584, south of Douglas Road, east of Dunbar Avenue, and west of Burbank Road in Oldsmar, eastern Pinellas County (Fig. 2). It is about 1 kilometre (km) northeast of Safety Harbor, a small bay in northwestern Tampa Bay. The site is situated on the USGS Oldsmar 7.5 minute quadrangle topographic map. The land surface is flat and is at an elevation of about 3 metres (m) above sea level. There is a gradual slope to the present bay shore. First visited in January 1984, this excavation remained open into 1985 but was flooded by 1987. The general stratigraphy along the south wall at the site (Fig. 3) consists of about 1 m of surface sand, over 1.5 m of marine shell bed, over about 0.5 m of black sand and gyttja containing mostly nonmarine mollusks and bones, over 0.5 m of buff sand, over green clayey sand of the marine Miocene Arcadia Formation (Hawthorn Group). Samples of the black layer were examined for pollen but no identifiable remains were recovered (B. Leyden and B.G. Warner, pers. comm.).

Oldsmar 2 (FK-28) The site is located in NW¼ Section 23, T28S, R16E (Latitude 28° 2'30" north, Longitude 82° 41' west), south of Florida Route 584, north of the CSX Railroad and east of Shore Blvd., in eastern Pinellas County (Fig. 2). This site is also at a surface elevation of about 3 m and less than 1 km from the shore of Safety Harbor. It is about 2 km west of Oldsmar 1 and is also situated on the Oldsmar 7.5 minute quadrangle topographic map. The land surface slopes gradually southward to Tampa Bay. This site was first visited in March 1984, with excavation most active in 1985. The stratigraphy as exposed in the southwest corner (Fig. 4) consists of about 1.5 m of sand over about 5 m of marine shell, over 0.5 m of black sand containing freshwater mollusks and bones, over the Miocene Arcadia Formation.

Vertebrate fauna Vertebrate fossils were recovered from rain-washed spoil piles and from excavated vertical and horizontal surfaces in situ or nearly so. To obtain specimens of microvertebrates, samples of sediment from the non-marine layer were washed through 1.5 mm and 1.0 mm screens and the remaining matrix was picked under a low-power binocular microscope. Some bulk samples required soaking in a dispersing agent to loosen the black organic clayey sediment. Vertebrate specimens are housed in the Vertebrate Paleontology Division of the Florida Museum of Natural History (FLMNH). Selected specimens are shown in Figures 5 and 6. The excavations at Oldsmar exposed a rather complex stratigraphic

102 P.F. Karrow, G.S. Morgan, R.W. Portell, E. Simons, K. Auffenberg

"t,

a::

Cl)

>

~ j:::

200 l----.l--....;.__-l--

!

!

~

(!)

- ; __ :

:

!

--+--t" ···: ·: --

!

!

I

1---'----'- --'----'--'----'----''----.I...---'---'----'---+--+ ---'--i

g.

C:

-E..

Arkalon:

rn2 length:

rn2 anterior

width:

m3 length:

SW:

14.0 15.2

14.0 15.2

NE: SE: SW:

27.4 25.6 26.7

26.0-28.7 23.6--27.2 26.7

1.0 1.37

NE: SE: SW:

18.0 16.6 18.9

17.3--18.6 15.2-18.7 18.9

0.53 1.34

NE: SE: SW:

21.3 19.1 22.3

20.4-22.5 16.3--21.3 22.3

0.78 2.13

3.8 5.37

2.9 8.06

3.7 11.14

1 1

simus (LaB): simus (PtC): simus (other):

17.4 16.5 16.7

16.7-18.1 15.8--17.3 15.2-18.0

0.46 0.56 0.94

2.6 3.4 5.62

7 6 1

yukonensis: yukon (Arkalon): ss indet: simus (LaB): simus (PtC): simus (other):

30.3 28.2 30.3 31.5 28.9 29.4

28.2-32.5 27.9-28.5 29.2-32.3 30.2-33.6 28.5-29.7 26.3--33.4

1.48

4.88

1.44 1.04 0.44 2.32

4.74 3.3 1.5 7.89

7 6 1

yukonensis: yukon (Arkalon): ss Indet.: simus (LaB): simus (PtC): simus (other):

20.5 20.3 21.0 21.8 19.8 19.9

18.7-22.8 20.2-20.4 19.5-22.7 20.6--22.7 19.5-20.0 18.4-21.8

1.6

7.83

1.51 0.56 0.19 1.0

7.18 2.6 0.9 5.05

7 4 1

yukonensis: yukon (Arkalon): ss Indet.: simus (LaB): simus (PtC): simus (other):

22.1 23.2 20.3 22.4 19.7 20.3

19.9-23.5 23.1-23.3 20.0-20.6 19.0-24.2 19.0-20.3 18.7-21.6

1.37

6.2

1.5

6.8

0.79

3.91

10

5 9 6 1 5 15 7 11

5 1 5 15 6 9 5 1 2 10 4 11

Continued

z

0

'""1 ..+

:,-"

> :3 '""1 n·

(D

~

:I ::i,.. ~ ..... C

;:,.. ;:::

V, V,

§" ;:::

V,

N

,i:.....

APPENDIX 2 - Continued

N

~

N

Arctodus pristinus Measurement m3 width:

Loe. NE: SE: SW:

x 16.5 15.2 15.0

Arctodus simus OR 16.0-17.0 13.5-16.3 15.0

SD 0.39 1.24

CV

2.3 8.15

N 7 4 1

Population yukonensis: yukon (Arkalon): ss Indet.: simus (LaB): simus (PtC): simus (other):

:;.::I

x

OR

SD

CV

0.9

5.28

1.13

6.4

N

17.1 18.0 17.8 17.7 16.1 16.1

16.5-18.6 17.9-18.2 17.5-18.1 15.4-19.1 15.9-16.2 14.5-17.2

0.85

5.29

11

5 1 2 11

3

r

~

n

::r....

1).)

0.. ;fl (')

~

(')

::r....i:::n ::r(t) ,....

Humerus, total length:

yukonensis: simus:

594.0 451.5

555.0-633.0 436.0-497.0

39.0 18.43

6.57 4.08

3 10

Humerus, greatest distal width across epicondyles:

yukonensis: ss Indet.: simus:

150.0 160.0 130.7

123.2-178.0 160.0 113.5-156.0

21.79

14.52

15.84

12.12

10 1 8

9 ...,

Ulna, greatest length:

yukonensis: yukon (Arkalon): simus:

553.1 535.0 436.8

536.0-591 .0 535.0 375.0-475.0

25.46

4.6

24.13

5.53

4 1 14

-

Ulna, smallest diameter from bottom of semilunar notch to margo dorsalis:

yukonensis: yukon (Arkalon): simus:

65.5 67.0 45.1

57.5-70.5 67.0 42.0-58.0

4.02

6.13

4.47

9.91

Radius, greatest length:

yukonensis: yukon (Arkalon): simus:

480.8 476.0 377.1

436.0-511 .0 476.0 355.0-394.0

27.94

5.81

14.02

3.72

10 1 12

5 1 8

:E i:::

g. i:::

Radius, long diameter of proximal end:

Arkalon:

Metacarpal I, length:

NE:

Metacarpal I, width over epicondyles:

NE:

1

yukonensis yukon (Arkalon): sirnus:

73.0 64.1 55.0

68.0-$2.6 64.1 50.0-63.0

yukonensis: simus:

91.7 78.3

yukonensis: simus:

Metacarpal II, length: Metacarpal II, width over epicondyles: Metacarpal III, length

NE:

Metacarpal III, width over epicondyles:

NE:

Metacarpal IV, length:

NE:

Metacarpal IV, width over epicondyles:

NE:

53.5

70.0 18.7

96.0 22.9

100.0 25.3

53.5

70.0 18.7

96.0 22.9

100.0 25.3

1 1

1 1

1 1

5.42

7.43

4.17

7.58

6 1 7

83.3--100.0 73.5--86.8

4.59

5.86

2 8

25.9 20.6

24.6-27.2 17.4-25.7

2.42

11.76

2 8

yukonensis: sirnus:

124.4 108.2

115.3--135.0 97.0-126.0

9.92 10.55

7.98 9.75

3 6

yukonensis: simus:

33.1 26.6

31.7-34.0 23.5-29.8

1.22 2.51

3.67 9.42

3 6

yukonensis: sirnus:

137.8 111.4

133.6-142.0 104.0-127.0

7.78

6.98

2 8

yukonensis: simus:

33.9 26.4

32.3--35.4 24.8-30.0

6.52

2 8

yukonensis: simus:

120.1 116.8

107.0-133.2 107.0-131.0

yukonensis: simus:

29.3 27.5

25.9-32.8 23.5-33.7

1.72

9.53 3.69

z

0

>-1

r+

~

>

!3ro

::t r,

8.16

2 6

Ill

13.41

2 6

i=l ..... 0

Continued

:::s

::i,. i::,..

:.::

"' "'§" :.::

"' N

"'"

C;)

APPENDIX 2 - Continued

N

Arctodus pristinus

x

"'"'""

Arctodus simus

::,::i

Population

x

OR

Metacarpal V, length:

yukonensis: simus:

112.9 112.6

112.9 98.0-130.0

Metacarpal V, width over epicondyles:

yukonensis: simus:

27.4 27.9

27.4 23.0-34.2

Femur, greatest length:

Rock Ck: yukonensis: simus:

558.8 678.8 504.4

557.5-560.0 651.0-723.0 490.0-524.0

Rock Ck: yukonensis: simus:

112.3 142.1 107.4

Rock Ck: yukonensis: simus:

Measurement

Loe.

OR

SD

CV

N

SD

CV

N

12.01

1 6

15.98

1 6

31.5 10.71

4.64 2.12

1 4 9

112.0-112.5 134.0-152.0 99.0-115.0

8.02 4.35

5.65 4.05

9

388.3 487.1 378.4

387.0-389.5 462.0-524.0 352.0-404.0

25.85 17.94

5.31 4.74

4 7

40.6 48.0 44.0 37.2

40.3-41 .0 46.9-50.5 44.0 31.2-43.0

1.31

2.73

3.92

10.53

348.3 445.5

346.5-350.0 441.0-450.0

r

~

I")

Femur, greatest d istal width over epicondyles: Tibia, greatest length:

NE:

Tibia, least width of shaft:

NE:

Fibula, greatest length:

355.0

31.0

355.0

31.0

1

1

Rock Ck: yukonensis: yukon (Arkalon): simus: Rock Ck: yukonensis:

13.51 4.46

1 4 1

1 6 1 8 1

1

::,-Ill

'"I

0-

JI'

n n ::,--

0 C: '"I

I")

::,--

... l'b

,

~

~

>-I C:

'"I

::s

O"'

-2.

Fibula, greatest diameter of distal end: Calcaneum, greatest length:

NE:

Calcaneum, greatest width, obliquely across sustentaculum:

NE:

Metatarsal I, length:

NE:

Metatarsal I, width over epicondyles: Metatarsal II, length:

NE:

NE:

99.0

65.0

66.0

17.7

85.0

99.0

65.0

66.0

17.7

80.~90.0

1

1

1

1

2

Rock Ck: yukonensis:

40.3 49.8

39.0-41.7 46.~51.5

Rock Ck: yukonensis: ss Indet.: simus:

109.0 129.9 134.0 111.3

107.5-110.5 114.1-136.0 134.0 101.~127.0

Rock Ck: yukonensis: simus:

72.4 89.2 77.0

71.7-73.2 81.5-92.8 67.~7.2

Rock Ck: yukonensis: ss Indet.: simus:

71.8 84.0 80.2 73.6

71.4-72.2 84.0 80.2 71 .~75.2

Rock Ck: yukonensis: ss Indet.: simus:

20.6 24.6 25.0 19.1

20.2-21.0 24.6 25.0 16.7-21.3

Rock Ck: simus:

88.2 93.7

87.9-88.5 86.~101.3

1 2

9.21

7.09

7.76

6.98

1 5 1 9

5.2 7.88

5.83 10.23

1 4 9

2.65

1 1 1 4

9.4

1 1 1 5

1.95

1.79

1 2

Continued

z

....0:r > >;

3

(I)

::::1. I") Ill

::,

::i,..

M ..... C)

i::...

;::

"'"'

§"

;::

"' N

,j::,.

c.n

APPENDIX 2 - Concluded

N

Arctodus pristinus

Arctodus simus

Measurement

Loe.

x

OR

Metatarsal II, width over epicondyles:

NE:

22.7

22.4-23.0

Metatarsal III, length:

NE:

Metatarsal III, width over epicondyles:

NE:

Metatarsal IV, length:

NE:

SD

CV

N

Population

2

Rock Ck: simus:

x 25.4 24.2

:,:;

OR 25.2-25.6 23.6-24.8

SD

0.6

CV

2.48

N 1 3

r"'

~

l"l

::r Ill >-t

0..

!!'

Metatarsal IV, width over epicondyles:

NE:

Metatarsal V, length:

NE:

Metatarsal V, width over epicondyles:

""'a-,

NE:

98.0

24.2

98.0

24.0

98.0

22.0

90.0-106.0

24.2

98.0

24.0

98.0

22.0

2

1

1

1

1

1

Rock Ck: yukonensis: simus:

97.9 119.5 107.1

97.4-98.5 115.5--124.1 94.0-121.2

Rock Ck: yukonensis: simus:

27.1 30.4 25.4

26.4-27.8 25.3--35.8 23.2-30.5

Rock Ck: yukonensis: simus:

107.5 125.1 112.2

107.2-107.8 118.2-132.3 105.0-122.0

Rock Ck: yukonensis: simus:

28.0 32.9 24.7

27.9-28.1 31.6-35.1 23.5--25.5

Rock Ck: yukonensis: yukon (Arkalon): simus:

110.8 129.2 108.1 109.6

110.1-111.6 120.7-135.0 108.1 85.0-129.0

Rock Ck: yukonensis: yukon (Arkalon): simus:

26.7 30.6 24.0 24.5

26.3-27.1 30.2-31.4 24.0 18.3--30.8

3.03 9.02

1 6 5

13.64 11.49

1 6 5

7.25

1 1 4

1.08

4.36

1 2 3

7.54

5.84

14.41

13.15

0.67

2.19

4.05

16.53

3.63 9.66 4.14 2.92

8.14

3 1 6 3 1 6

(")

yi

(")

::r c:: ::r (t)

>-t l"l ,>-t

~

~

'"'.I

c::>-t

g. c::

::::

Origin of the vertebrate fossil sites near Medicine Hat, Alberta A. Macs. Stalker

Abstract Quaternary beds exposed along the South Saskatchewan River near Medicine Hat, in southwest Alberta, have yielded vertebrate fossils from nine age levels. The factors that produced this remarkable sequence of beds include some common to most of the Prairies and others peculiar to the Medicine Hat region. The former embrace those that shielded the fossil beds from subsequent glacier erosion. They include the deep preglacial and interglacial valleys that protected debris and fossils contained in them; the upslope advance of glaciers that dammed proglacial lakes in those valleys, again helping protect fossils; and the thinning of the ice sheets owing to that upslope advance and the nearness of glacier limits, which combined to cause glacier deposition to predominate over glacier erosion. The latter includes features found near Medicine Hat that shielded the fossil beds from river erosion. A major river crossed that region in preglacial, interglacial, and postglacial times. However, as during each glaciation its valley was largely filled with debris, the river generally took a new course through the area following each glaciation and abandoned its previous channel. Fossil beds in those abandoned channels had a good chance of preservation, for not only did their depth below prairie level protect them from subsequent glaciers, but the diversion of the river from them tended to shelter the fossil beds from subsequent stream erosion. As a result of these factors, fossil beds of many ages are preserved in the Medicine Hat area. Those beds are now exposed where the modern river valley crosses those abandoned channels.

Introduction In this paper, the geological factors responsible for the presence of multi-

248

A. MacS. Stalker

'. - -L__ ___ - -

""" ~-~--~-~---~--~-~-~~~-~~ --~ AA

1 10"Jr

Tp 14

- - - ~;II..=~\-

Tp U

Tp 13

Tp 12

--

Tp 12

.... ____ _ .__

___,

110"0 "

_..__._

lllf

__ __ -'--

.

~....._-"-"---"----"---'----""'---' R4

,....

110",o,

Fig. 1. Medicine Hat area showing locations of fossil sites mentioned in text, edges of present South Saskatchewan Valley, and assumed courses of some interglacial valleys. TI!! = boundaries of present South Saskatchewan Valley;::::!:!:= assumed limits of interglacial valley.

pie Quaternary vertebrate fossil beds near Medicine Hat, in southeast Alberta, are described. These fossil beds are remarkable for both their abundant fossils and for the range of ages represented. Thus, in a composite stratigraphic section prepared for the region, Stalker and Wyder (1983, Table 2) show nine levels of vertebrate fossil beds, some separated by glacial till, and there is no reason to believe all have been found. The oldest predate glaciation of the region and may be as old as one million years. The youngest are late postglacial and bones are still being buried and preserved on some river slip-off slopes. Altogether the Medicine Hat fossil beds appear to encompass more than half the Quaternary Period. The writer's interest in these beds began about 1963, when he discovered bones in cliffs along South Saskatchewan River north of Medicine Hat. The study of these beds intensified greatly in 1965 when Dr C.S. Churcher became involved. This began a happy collaboration that continued for more than twenty years. Dr Churcher studied and identified the

Medicine Hat fossil sites, Alberta 249 bones and teeth, whereas the author examined the geology of the sites. Both took part in recovery of fossils. The results have greatly advanced knowledge of Canadian fauna, climate, and events during the Ice Age and have contributed substantially to the author's stratigraphic studies. Figure 1 of this report shows the area examined and some of the fossilproducing localities. The fossiliferous units are described in the Appendix, which is based on Table 2 of Stalker and Wyder (1983). More detailed information about the geology and stratigraphy of the area can be found in their report. A chart prepared by Stalker and Churcher (1982) shows taxa recognized up until 1976; identifications made since 1976 have not been published.

Discussion Over the years, Quaternary vertebrate fossils have been found in various parts of Canada, notably the Yukon, near Toronto, and in caves and swamps in other parts of the country (see Harington, 1978). More recently, the most productive sites have been on the southwestern Prairies, with the first major discoveries being along the South Saskatchewan River near Medicine Hat (Churcher, 1970). Since then, other important sites have been found, for instance in Alberta along the North Saskatchewan River near Edmonton (Bums and Young, 1994; Churcher, 1968), the Red Deer River near Empress (Churcher, 1972), and the preglacial Red Deer Valley near Camrose (as yet unpublished). In Saskatchewan sites have been found along the South Saskatchewan River north of Swift Current (Barendregt et al., 1991; Stalker and Churcher, 1982) and at Saskatoon (Skwara and Walker, 1989; SkwaraWoolf, 1981). However, the Medicine Hat sites differ from these others in the multiplicity of their beds, for most of the others yield only one age of fossils. They also have been the most productive, though this may be partly due to the amount of work done there. Two main questions are addressed here. The first is why so many of the most productive sites in Canada are on the Prairies; the second is what factors led to the multiplicity of fossil beds at Medicine Hat.

General features of the Prairie sites Common features Six factors were chiefly responsible for the numerous fossil sites on the Prairies. These were: (1) the presence in preglacial, interglacial, and postglacial times of numerous deep valleys, (2) the attraction and lure of those valleys for animals, (3) the existence of means for burial and preservation of bones and teeth, (4) the uphill advance necessary for glaciers from the east and

250 A. Macs. Stalker north to reach the region, (5) the dominance of glacial deposition over glacial erosion in the region, and (6) the general semi-aridity of the region and the alkaline nature of its deposits. These factors are examined in order. (1) The valleys. Today the southwest Prairies display many large river valleys, which typically are between 30 and 100 metres deep, up to 2 km wide, and steep-walled. Most of these were incised from near prairie level following the last glaciation, but earlier ones formed during interglacial and interstadial times were probably much similar. On the other hand, preglacial valleys had a much longer time to develop and so were much larger and their walls had gentler slopes. They typically were 10 to 30 km broad from rim to rim, though probably not much deeper than most of the present valleys. All those valleys, no matter what their origin, provided low areas that offered some protection from glacier erosion to any bones and materials that had been laid down in them. Further, if the river adopted another course during the next interglacial, those deposits also were shielded from subsequent river erosion.

(2) The animals. Reasons for the enticement of animals into the valleys are best studied along the present South Saskatchewan River near Medicine Hat and then extrapolated to the earlier valleys. This river enters the Medicine Hat region from the west, then at the City of Medicine Hat it swings abruptly north to flow out of the area (Fig. 1). Its valley is typically 1 to 2 kilometres broad and as much as 100 m deep. It meanders strongly through the area, for it is crossing readily eroded Quaternary deposits laid down in preglacial and interglacial valleys. To the west and north, where it crosses more resistant bedrock, its course is straighter and its banks typically steeper. In many places in the area the walls of this valley are steep, especially at the main fossil sites. Elsewhere, however, gentle slip-off slopes and tributary gullies provide easy access to the river for animals. Further, where opposite each other, these gentle slopes enabled animals to ford the river readily at low water. On the other hand, the steep valley walls found in the bedrock areas west and north of the area prevented such easy access, helping cause further concentration of animals near Medicine Hat. Besides this ease of access, much of the valley supported shrubs and trees - generally lacking on the nearby Prairies - along with local good grazing. The river valley thus furnished provender, water, and some protection for browsing and grazing animals, food and homes for aquatic animals, and shade for all during hot summer days. These items were especially important when drought diminished grazing capacity of the adjoining prairie. The South Saskatchewan Valley, as a result, provided a good milieu for

Medicine Hat fossil sites, Alberta

251

concentration of animals. Similar conditions are found in many of the other river valleys on the Prairies and undoubtedly were present in many of the interglacial and interstadial valleys. If so, those earlier valleys, like the present ones, were good congregating grounds for animals. (3) Fossil burial. Animals enticed into the valley by the above factors (see 2, above) undoubtedly had a high mortality rate. Some, of course, died from natural causes. More importantly, however, the concentration of animals provided good hunting for predators, there was always the peril of flash floods, some animals undoubtedly drowned while fording the river and others may have drowned by breaking through river ice in winter. Those that ended up in the river were carried downstream until they grounded on gravel bars or floodplains. The bones of some of those that died on the floodplain were picked up by the river during high water; others were scattered directly on the floodplain. In all these instances, the bones were subject to burial by alluvium during the next high water. These processes tended to disarticulate and scatter the bones, much as they are now found in the fossil beds, and the burial helped protect them. (4) Uphill glacier advance. Any glacier expanding over the Prairies from the Precambrian Shield had to advance uphill, in the case of the Medicine Hat area a rise of about 500 m. As drainage was mostly downslope towards the ice-sheet, valleys were blocked, proglacial lakes developed, and lake and delta deposits laid down. These deposits (Appendix, units IV, VIII, and XII) helped shield any previous deposits in the valleys, including fossiliferous beds, from subsequent glacier erosion. (5) Glacier erosion versus deposition. The uphill advance of the glaciers, combined with the normal downslope of the surface of a glacier towards its margin, meant that glaciers were markedly thinner here than farther north or east. Further, they were nearing their limits of advance, which presumably meant that they carried much debris. These factors tended to reduce their erosive power in the region and to make glacier erosion subordinate to glacial deposition. Once again, this helped preserve any materials, including bones and teeth, present in the valleys. (6) Climate. The semi-arid climate of the region diminished growth of vegetation, with consequent reduction in production of bone-rotting humic acids. Further, the soils and deposits of the region tend to be fairly alkaline. These factors, combined with the scant amount of moisture, slowed decomposition of bones and teeth. The contrast in this with the wetter and forested regions farther east, north, and west is extremely strong. Unhappily, these same factors that helped preserve the bones were largely responsible for the lack of pollen in most of the beds, which renders determination of the past environments more difficult.

252 A.MacS.Stalker

The Medicine Hat region Factors responsible for the multiple fossil beds All the factors described above (1 to 6) as common to the Prairies operated in the Medicine Hat region. There were, however, two additional factors which combined to cause a multiplicity of fossil beds. The first was (7) the replication of factors 1 to 6 time and again owing to the presence, right through the Quaternary except during glaciations, of a major river. The second helped preserve the fossil beds once they were formed. It involved (8) repeated local diversions of the river. (7) River presence. In the Medicine Hat region, a major river was present in preglacial time and during most or all of the Quaternary interglacials and interstadials. The former presence of the preglacial river is shown by the broad valley it carved into bedrock, a valley wide enough to encompass most of the area shown on Figure 1, and by the preglacial gravels and other deposits it laid down in parts of that broad depression. The interglacial valleys are similarly shown by their stream deposits and also by abandoned, partly filled channels. This presence of a major river in preglacial time and during each interglacial allowed the repeated operation of the factors (1 to 6) described above. It was, therefore, the major cause behind the recurrence of fossil beds. Like the valleys described above under (1), (2), and (3), the successive valleys at Medicine Hat generally provided ideal milieus for animals and subsequent protection from glacier and stream erosion for their bones and enclosing material. This facilitated the formation, over a long time, of a succession of fossil beds. In most other parts of the Prairies a river, once diverted from its interglacial valley by a glacier, tended not to return to its former course during the next interglacial. This is the main reason why, though one or two fossil beds might be produced at some locations to be exposed by a modern river valley, the repetition of fossil beds found at Medicine Hat has not been encountered elsewhere. (8) Abandoned channels. The features discussed above (1 to 7) indicate how bones were buried and preserved from glacier erosion, but fail to show how they were preserved from subsequent river erosion. For instance, the South Saskatchewan River has destroyed any fossil beds that were in the course of its present valley and is currently eating into those found along the banks. However, in the Medicine Hat area, sections of the river were diverted to new courses following each glaciation. This process produced numerous abandoned valley segments, which were never very far from previous courses. Figure 1 shows some of these abandoned channels, and undoubtedly there are others.

Medicine Hat fossil sites, Alberta

253

The importance of those abandoned channels cannot be overemphasized, for they preserved their valley fills, along with any fossil beds contained in them, from both subsequent glacier and river erosion. The present South Saskatchewan River exposes the fossil beds where it now crosses those abandoned channels. Similar local concentrations of abandoned channels are not common elsewhere on the Prairies. This was a contributing factor to the paucity of sites with repeated fossil beds in other parts of the Prairies.

Miscellaneous In all studies of fossils there is always the possibility of redeposition of bones from earlier deposits, though it must be rare that bones would survive two sequences of river deposition. It could be argued that the bones in the lag gravel of Unit VIII were redeposited, though their presence involved more the removal of other fine material. In most of the units studied, however, the possibility of redeposition appears remote. There are notable differences in some characteristics of the bones between most of the units. These involve features such as mineralization, colour, and breakage. Further, of course, the change of taxa with time also indicates that the bones in some of the units were not redeposited.

Acknowledgments First and foremost, I wish to thank Dr C.S. Churcher for the knowledge he imparted to me about vertebrate palaeontology during many pleasant field seasons and for the great help and the many courtesies he extended to me. Many other persons also gave great help, among them Dr Hope Johnson and Dr Luke Lindoe of the cities of Redcliff and Medicine Hat, respectively. Thanks are due to the City of Medicine Hat and the members of the Medicine Hat Museum for their valuable aid over the years. I especially wish to thank the ranchers of the region for their many kindnesses, for allowing entry to their properties and access to the river and for their interest and encouragement in the work. I also express my thanks to Dr R.W. Barendregt, of the Department of Geography, University of Lethbridge, and another critical reader, for their careful study of the manuscript and valuable comments. Finally, I express my gratitude to my daughter, Mary, for drafting Figure 1.

LITERATURE CITED

Barendregt, R.W., F.F. Thomas, E. Irving, J. Baker, A. MacS. Stalker, and C.S. Churcher. 1991. Stratigraphy and paleomagnetism of the Jaw Face section,

254 A. Macs. Stalker Wellsch Valley site, Saskatchewan. Canadian Journal of Earth Sciences 28:1353-1364. Burns, J.A., and R.R. Young. 1994. Pleistocene mammals of the Edmonton area, Alberta. Part 1: the carnivores. Canadian Journal of Earth Sciences 31:393--400. Churcher, C.S. 1968. Pleistocene ungulates from the Bow River gravels at Cochrane, Alberta. Canadian Journal of Earth Sciences 5:1467-1488. - 1970. The vertebrate faunas of Surprise, Mitchell and Island Bluffs, near Medicine Hat, Alberta. Geological Survey of Canada, Department of Energy, Mines and Resources, Report of Activities, Paper 70-l(A):158-160. - 1972. Imperial mammoth and Mexican half-ass from near Bindloss, Alberta. Canadian Journal of Earth Sciences 9:1562-1567. Dyck, W., J.G. Fyles, and W. Blake, Jr. 1965. Geological Survey of Canada, Radiocarbon Dates IV. Geological Survey of Canada, Paper 65-4:1-23. Harington, C.R. 1978. Quaternary vertebrate faunas of Canada and Alaska and their suggested chronological sequence. Syllogeus 15:1-105. Lowden, J.A., and W. Blake, Jr. 1975. Geological Survey of Canada, Radiocarbon Dates XV. Geological Survey of Canada, Paper 75-7:1-32. Lowden, J.A., J.G. Fyles, and W. Blake, Jr. 1967. Geological Survey of Canada, Radiocarbon Dates VI, Geological Survey of Canada, Paper 67-2:1-42. Lowden, J.A., I.M. Robertson, and W. Blake, Jr. 1971. Geological Survey of Canada, Radiocarbon Dates XL Geological Survey of Canada, Paper 71-7:1-70. Skwara, T., and E.G. Walker. 1989. Extinct muskox and other additions to the Late Pleistocene Riddell Local Fauna, Saskatoon, Canada. Canadian Journal of Earth Sciences 26:881-893. SkwaraWoolf, T. 1981. Biostratigraphy and paleoecology of Pleistocene deposits (Riddell Member, Floral Formation, late Rancholabrean), Saskatoon, Canada. Canadian Journal of Earth Sciences 18:311-322. Stalker, A. MacS., and C.S. Churcher. 1982. Ice age deposits and animals from the southwestern part of the Great Plains of Canada. Geological Survey of Canada, Miscellaneous Report 31. [Wall-chart] Stalker, A. Macs., and J.E. Wyder. 1983. Borehole and outcrop stratigraphy compared, with illustrations from the Medicine Hat area of Alberta. Geological Survey of Canada, Bulletin 296:1-28.

APPENDIX

Description of the fossil beds The fossil beds are described in chronological order. Non-fossil units, such as tills, are omitted. The descriptions and unit numbers are derived from Table 2 of Stalker and Wyder (1983), where descriptions of the nonfossiliferous beds can also be found. Units III, IV, IX, X, XIV, and XXI are

Medicine Hat fossil sites, Alberta

255

thought to have been deposited under cool conditions, probably as an approaching glacier blocked drainage and raised river base levels. It is, indeed, very likely that most of the fossil beds were laid down under cool or cold conditions, commonly as glaciation was beginning. This may have caused a bias in the types of animal taxa recorded from the area. Preglacial beds

Unit II This unit occurs at all sections shown on Figure 1 except Reservoir Gully and Surprise Bluff West. In some (e.g., Bain Bluff) it is recorded only in boreholes. The unit lies directly on bedrock. It consists mainly of gravel composed of quartzites and hard sandstones, with minor silt bands especially in its younger parts. It represents floodplain and bar deposits laid down intermittently as the preglacial river swung slowly back and forth while deepening its valley. The deposits range in height from well above the present river to below its bed, with the higher gravels typically being the older. The age of this unit is estimated by Stalker and Churcher (1982) at about 600,000 years, but it undoubtedly spans a long time and its higher beds could be as much as a million years old. The coarse river gravels typically contain only a few poorly preserved fossils, probably because pounding by stones in the river destroyed most bones during deposition and easy seepage of water through the gravels later on helped to rot others. Fossils are more common and better preserved near the silt units, where currents were gentler during deposition and subsequent groundwater seepage less. Most of the fossils were found at Mitchell Bluff. Unit III This unit consists of a lower clay band and an upper carbonaceous band, in places nearly a peat. Both are thin and unoxidized. The former is found mainly at Island and Mitchell Bluffs, where it overlies the gravels of Unit II. It consists mostly of clay particles tom by the river from local bedrock shales. This subunit has produced abundant bones and teeth, probably owing to periodic inundation of a river floodplain with resultant scattering and burying of bones of animals that had died on it. The upper peaty band is found at most sites that reach river level. Where the lower band is absent, it overlies the preglacial gravels. It contains leaves, twigs, and logs to 15 cm or so in diameter, but few bones. Neither band tells much about climatic conditions during its deposition.

256

A. MacS. Stalker

Unit IV Found at most sites except Reservoir Gully, this unit extends from Unit III to the first occurrence of stones brought by ice from the Canadian Shield. It is 70 m thick at Bain Bluff, where it consists predominantly of lake clay and silt. At Island and Mitchell Bluffs it consists largely of strongly oxidized, deltaic silt and sand, with minor clay and poorly sorted gravel. It commonly has a cap of poorly sorted lag gravel. Its lower part contains numerous sticks and logs, which decrease in number upward.

Ice Age beds These beds span the time from first glaciation of the area, marked by first appearance of stones from the Canadian Shield, until the last glacier had melted. Till intervenes between several of the fossiliferous beds. Unit VIII This predominantly sand and silt unit occurs at most sites, except for Bain Bluff and Lehr Valley. It is up to 14 m thick and typically is strongly oxidized. Locally, it starts with a lag gravel formed during river erosion of the till laid down by the preceding glacier. This lag contains a few vertebrate fossils, which may have been reworked from underlying deposits, perhaps partly from Unit I. The rest of the unit contains more, though scattered, bones. Unit IX Like unit VIII, this unit is found at most sites apart from Bain Bluff and Lehr Gully. It is typically a poorly sorted gravel with tight silt and clay matrix, along with beds of sand and silt. All are strongly oxidized. The unit appears to be of alluvial and deltaic origin. Though only about 3 m thick, it is a prolific producer of fossils. As a result, it has been excavated extensively and has received more study than any other unit. Where excavated at Island and Mitchell Bluffs it yields about three bones per cubic metre. Much the same fossil abundance is shown wherever the present river exposes this unit, for a width of at least one-half km over the 15 or so kilometres from Island Bluff to Surprise Bluff East. UnitX Unit X overlies Unit IX without any intervening glacial deposits. It is up

Medicine Hat fossil sites, Alberta

257

to 12 m thick and represents channel and delta deposits, with well-developed cross bedding and cut-and-fill structures throughout. Most of it is a pebbly sand and grit whose grains are composed of clay and silt fragments ripped from pre-existing, mostly varved, lake deposits; perhaps the lake deposits of Unit IV. The top part is a fine gravel with compact silt and clay matrix, which much resembles till. Fossils are found in all parts of this unit, but are much rarer than in unit IX. Most are small and many of the rodent fossils came from it. Unit XII This intertill unit is an enigma. It has been found only at Bain, Island, and Mitchell Bluffs and reaches a thickness of about 3 m. The lower part consists of poorly sorted gravel and sand that resembles glacial outwash, whereas the upper part is a quiet lake deposit of silt and clay. The latter has yielded a few fossils, mostly of large animals, at Island and Mitchell Bluffs. Unit XVI This unit is best displayed at Galt Island Bluff, where it contains scattered bones for more than a kilometre along the river-bank. It is up to 25 m thick and consists largely of alluvial and lacustrine silt and clay. One especially prolific area was excavated. This is the lowest unit to provide finite radiocarbon dates, as all samples from lower ones gave readings beyond the range of carbon dating. Radiocarbon ages of 37,900 ± 1,100 (GSC-1442) and 38,700 ± 1,100 (GSC-1442-2) were obtained from wood found just above the main fossil deposits in the excavation at Galt Island (Lowden and Blake, 1975:17-18). At Evilsmelling Bluff scattered small pieces of wood, in what apparently is the same unit, yielded dates of 24,490 ± 200 (GSC-205, Dyck et al., 1965:8), 25,000 ± 800 (GSC-1370, Lowden et al., 1971:288), and 28,630 ± 800 (GSC-543, Lowden et al., 1967:13). These latter dates were in proper stratigraphic order. At the Garbage Dump Site in Medicine Hat, accelerator dates (not published) of approximately 27,210 (TO-102) and 40,620 (TO-680) were obtained from what is assumed to be the same unit. Unit XXI This unit is best displayed at Reservoir Gully, where it is overlain and underlain by till. There it consists of some 20 m of alluvial sand, gravel, and silt. The sand and silt beds, which form the upper part of the unit, have yielded many bones. Most came from about two-thirds of the way up the unit.

258

A. Macs. Stalker

Postglacial beds Unit XXIII This unit embraces all the postglacial deposits found in the region. It consists mostly of alluvial sand, gravel, and silt, commonly with a cover of wind-blown sand or loess. In places it reaches a thickness of 25 m. It has yielded a few bones, mainly from its lower alluvial beds, but rare bones may be found almost anywhere in it.

A preliminary report on the Carnivore of Porcupine Cave, Park County, Colorado

Elaine Anderson

Abstract Porcupine Cave, a late Irvingtonian site in south-central Colorado, contains a rich and diversified fauna that includes land snails, fish, amphibians, reptiles, birds, and at least 60 species of mammals. At an elevation of 2900m, it is the highest Pleistocene faunal site in North America. The cave was sealed for more than 365 ka; and the only entrance today is manmade. Fossils occur throughout the cave and have been recovered from a fissure and nine different rooms; the Pit and Velvet Room are stratified and document major environmental changes from relatively humid glacial to warmer, dry interglacial conditions. Woodrats, Neotoma spp., were probably the major agent of accumulation. Carnivores are numerous, both in species (22) and specimens, and include the earliest known occurrence of Mustela nigripes and the first lrvingtonian records in the western United States of Martes diluviana, Gulo cf. G. schlosseri, Brachyprotoma obtusata, and Miracinonyx cf. M. inexpectatus. In addition, badgers, weasels, mink, skunks, otters, foxes, coyotes, wolves, and bobcats are present. At least 26 species of rodents and lagomorphs provided a prey base for the carnivores.

Introduction In 1981 Larry Rasmussen, an amateur caver, discovered bones in Porcupine Cave, showed them to his father, Don Rasmussen, a geologist and palaeontologist, and together they began the exploration for fossils in the cave. Don brought the fossils to the attention of Anthony Barnosky, then at the Carnegie Museum of Natural History. He, Don, and other personnel conducted a reconnaissance of the cave in 1985, and Barnosky' s subse-

260 E. Anderson

1/)

z

3 Q.

0

~__...SOUTH PARK BASIN

~ w

a:

(!)

0

< cc

0 ..J 0

u

COLORADO FIGURE 1. Map of Colorado showing the location of Porcupine Cave

quent analysis of the arvicoline rodents showed that extinct lrvingtonian species were represented. Preliminary excavation began in 1986 and was continued in 1987. In 1992 and 1993, crews from the Denver Museum of Natural History and the Western Interior Paleontological Society, with the help of many volunteers, worked in the Velvet Room and the Badger Room. The history of the cave goes back much further. Porcupine Cave is located on the southwest rim of South Park, a vast open basin surrounded by high mountains, in Park County, Colorado, at an elevation of 2900m (Fig. 1). The ridge on which the cave is located is composed of eastwarddipping strata of the Ordovician Manitou Dolomite and Harding Sandstone. The ridge was first formed by erosion of folded Palaeozoic strata during the Eocene, then buried by volcanoclastics from the Thirty-nine Mile Volcanic Field during the late Eocene and Oligocene and then reexposed in the late Neogene. Although there are no known prePleistocene strata in the cave, initial cave formation probably began during the Palaeozoic, continued over long periods of time by the action of water and sulphur-rich solutions, and was reactivated during the Tertiary. By early Pleistocene, there were several natural entrances; woodrats, Neotoma spp., lived in the cave and hauled bones in; badgers, Taxidea

Porcupine Cave Carnivora, Colorado 261

taxus, and other small- to medium-sized carnivores may have had limited access to certain parts of the cave close to the surface (such as the Badger Room). Different areas of the cave were sealed off at different times during the early and medial Pleistocene; by about 365 ka, all outside entrances were sealed, thus preserving the remains of a high-altitude late Irvingtonian fauna. The cave remained sealed until the late 1800s when miners, following iron-mineralized rocks filling a fissure in the Manitou Dolomite, blasted it open. The only entrance today is through a 12m-long timbered adit built by the miners. By the 1920s cavers were exploring the cave, but over 30 years elapsed before it was mapped. Since its discovery, no one apparently noticed or reported bones which occur throughout the cave. The only other known lrvingtonian fauna in Colorado is at Hansen Bluff in the San Luis Valley, Alamosa County (Rogers et al., 1985). The fauna is especially rich in arvicoline rodents and salamanders. Only one bone referable to a carnivore (Canis cf. C. priscolatrans) was found. The fauna ranges in age between 690 and 900 ka.

Methods Carnivores and other fossil vertebrates have been collected from ten localities within the cave (Tables 1, 2; Fig. 2). The methods used in collecting the fossils are given in Barnosky and Rasmussen (1988) and will not be repeated here. Brief descriptions of the main fossil localities follow. Badger Room: Located near the man-made entrance, the sediments are part of the distal terminus of an alluvial fan deposited when there was an ancient cave entrance 3m to the west. Sediments are not stratified and have been reworked by woodrats and possibly badgers. Remains of rabbits are numerous. Many bones show signs of rodent-gnawing. Pit: Located 30m from the cave entrance at the bottom of an 8-lOm shaft are 2.5-3m of stratified sediments. Fossils occur in all layers but are most numerous in fine-grained strata at the top of the deposit. The upper layers of interbedded fine-grained sediments indicate drier, probably interglacial, intervals; Cynomys and Spermophilus are common, Marmota is absent. Deeper layers with brown mudstone nodules and other layers of calcite flowstone indicate wetter, probably glacial, intervals; Marmota is abundant (see Barnosky and Rasmussen, 1988, for details). Velvet Room: Located about 30m from the cave entrance and 20m south of the Pit. Originally separated from the rest of the cave by sediments in the southern part of the Gypsum Room, the room was discovered in the mid-1980s. Sediments are stratified in the two areas excavated: the first in

N

°'

N

trJ

► :::,

0..

--, /. -

~r -

-,·' MINE ENTRANCE •~ ,\

__ /

~

•

•\ ......c• -.,

/

I

,

! ',

-J...-

,., I

: CRAMPED ~~•' QUARTERS •' DM NH

CR YS TA L ROOM DM NH

CM NH - ":: FERRET ROOM

f

NORTH

PORCUPINE CAVE 10

METERS

Ma p of Po(

;;f

~

w w

"-1

338 R.W. Graham, J.O. Farlow, J.E. Vandike

10cm

FIGURE 3. Photographs of plaster casts of P. atrox tracks from CTC; (A) ISM 494015; (B) ISM 494016 (track patterns are mirror images of the actual tracks because they were photographed from inverted plaster casts)

Panthera atrox tracks 339 represents an isolated hindfoot print (see below). Based upon the previous discussion and the tracks that we have collected from other cats, it appears that the lobate pattern is frequently not preserved for a variety of reasons, including substrate conditions and differential pressure on foot pads. Claw marks are not apparent on either ISM 494015 or ISM 494016 (Fig. 3) or any of the other tracks observed in CTC. As previously noted, claw marks were recorded only when it appeared that the animal had slipped on the mud banks of the cave. Canid claw marks are more blunt than the marks left by the sharp claws of felids (Morse, 1992) and the sharp claw marks in CTC were characteristic of those of cats. The two tracks, ISM 494015, and 494016, may represent forefoot and hindfoot impressions, respectively. This identification is suggested by the larger size of ISM 494015, as the forefeet of living big cats are broader than the hindfeet. The distinction between right and left feet is a little more problematic for cat tracks. For F. concolor it has been observed that the toes are asymmetrically arranged around the interdigital pad with the third digit serving as the leading toe and the fifth digit placed more posteriorly than the others (Morse, 1992:40). This would suggest that track ISM 494015 may be a left forefoot and that track ISM 494016 may be a right hindfoot. We believe that the CTC tracks were made by the largest Pleistocene North American felid, the American lion (P. atrox). The two tracks from CTC are larger than any of the tracks that we have studied, including the living lion and tiger (Fig. 4 and Table 1). The American lion (P. atrox) was the largest of all the North American cats, including two sabretooths, in a diverse felid guild (Table 2). It has been estimated that the Pleistocene American lion (P. atrox) was larger (Table 2) than the Recent African lion (P. lea), perhaps weighing at least half again as much (Anderson, 1984:59; Anyonge 1993). The CTC tracks are approximately 10% larger than the African lion and Siberian tiger tracks that we collected at the Indianapolis Zoo (Fig. 4, Table 1). Even though the female tigers were young (1.5 years old), their paws and tracks would have been nearly full adult size (Sankhala, 1977:172). Furthermore, we selected some large tracks in the enclosure which may have been made by the older (5 years) males. Based upon the size of the CTC tracks, we believe that they can only be referred to P. atrox. Comparison of casts of the articulated foot bones of P. atrox from Rancho La Brea also suggest that the tracks from CTC are of the appropriate size for this species. We have compared the CTC tracks with the dried paws of a Recent specimen of P. onca paraguensis (= P. o. palustris) from South America preserved in the collections at the University of Connecticut. This is the largest living subspecies of P. onca (for discussion of subspecies see Seymour, 1989). The dried paws are significantly smaller than the tracks from CTC.

340 R.W. Graham, J.O. Farlow, J.E. Vandike 130------------------------~

P1T

120

,...._

...

E

PAT

a '-' no ~

•

PTI

•

PTI

~

"C

03 100

i:i.,

•

P~I

30 kg) Felis concolor + Miracinonyx trumani Panthera onca + Homotherium serum + Smilodon fatalis + Panthera atrox

Cougar, Mt. lion American cheetah Jaguar Scimitar cat Sabertooth American lion

36---100 75-100 60-120 146---231 347-442 344-523

The maximum width and length of the jaguar hindfoot and forefoot are 69.6 x 74.5 mm and 93.2 x 75.7 mm, respectively. These measurements should be considered as minimum estimates, since the paws had been dried. Rabinowitz (1986) reported the average widths and lengths for the hindfoot and forefoot of 70 jaguars as 90 x 90 mm and 100 x 90 mm, respectively. It has been estimated that the Pleistocene subspecies (P. o. augusta) was 15-20% larger than the extant species (Seymour, 1989, 1993). The two tracks from CTC are significantly larger than the fossil track from Berome Moore Cave (Fig. 2C) which has been referred to P. o. augusta (Table 1). The CTC tracks are also significantly larger than a track of P. o. augusta from Craighead Caverns which had a maximum transverse diameter of 70-75 mm for the rear of the interdigital pad (Simpson, 1941:5). We did not have samples of tracks from the extant African cheetah (Acinonyx jubatus) or the Pleistocene North American cheetah-like cat (Miracinonyx trumani). However, Acinonyx weighs between 35 and 72 kg and is much smaller than either the African lion or Siberian tiger (Table 2). Miracinonyx is comparable in size to Acinonyx, but unlike Acinonyx, Miracinonyx had retractile claws (Adams, 1979; Van Valkenburgh et al. 1990).

Panthera atrax tracks 343 To our knowledge, there are no tracks that have been attributed to either of the two sabretooth cats, Hamatherium or Smiladan. Comparison of the tracks from CTC with measurements for the feet of Hamatherium illustrated in Rawn-Schatzinger (1992) indicated that the CTC tracks are significantly larger than the feet of this extinct species. The claws of Hamatherium may not have been fully retractile (Rawn-Schatzinger, 1992) and, therefore, its foot print would have been more like those of other cursorially adapted taxa, such as the cheetah and canids (e.g., Van Valkenburgh, 1987). Also, it has been suggested by Rawn-Schatzinger (1992:6566) that Hamatherium had a semiplantigrade hindfoot, which should have left a distinctive track. Martin et al. (1988), however, reconstructed both the forefeet and hindfeet of Hamatherium as fully digitigrade. We compared casts of the articulated foot bones of Smiladan from Rancho La Brea to the CTC tracks. The feet of Smilodan were significantly smaller than the tracks and also smaller than casts of the feet of P. atrox. The feet of Smiladan were fully digitigrade and the claws were completely retractile (Gonyea, 1976). It has been noted (Cox and Jefferson, 1988) that the terminal phalanx of the fifth digit of a recently discovered articulated forefoot of Smilodan was significantly smaller than the corresponding element in other large felids. In previous reconstructions of the forefoot of Smiladan the largest terminal phalanx was placed on digit I and the terminal phalanges of the other four digits were of uniform size (Merriam and Stock, 1932: PL 22, Fig. 6). It is possible that this difference would not affect the forefoot track, since the claws were retractile. Bones and teeth of P. atrax have been found as fossils from Alaska to Peru and from California to Florida (Kurten and Anderson, 1980; Harington, 1971). In fact, many researchers believe that P. atrax is conspecific with the African lion (Panthera lea) and the Eurasian cave lion (Panthera lea spelaea) (Kurten and Anderson, 1980). This would make P. lea one of the most widely distributed species throughout the Quaternary and Recent times. However, until now, fossil tracks have never been attributed to P. atrax. Recently, 29 tracks of P. l. spelaea were found at the tracksite BottropWelheim in Germany (von Koenigswald, 1994), but these tracks appear to be narrower than the P. atrax tracks from CTC (von Koenigswald, pers. comm.). Finally, the preservation of these tracks on the surface of unconsolidated clay deposits in a cave with an active stream documents longterm stability of this cave-floor surface, because P. atrax has been extinct for at least 10,000 years.

Acknowledgments We thank Jerry Vineyard, Paul Wightman, Gene Gardner, Bill Palmer, and Sharon Vandike for their assistance in exploring CTC. Susan Gordon and

344 R.W. Graham, J.O. Farlow, J.E. Vandike Lynne Villers gave us the opportunity to cast tracks of African lions and Siberian tigers at the Indianapolis Zoo. Julianne Snider, John Keltner, and Mary Ann Graham assisted Jim Farlow in making casts of these tracks. Christopher Shaw and Robert Dubos made collections available to us for study. Pete Bostrum recast the tracks from CTC. Linda Prescott and Tina Boehle cast the cougar track at the Prairie Wildlife Park Zoo, Peoria, and Gregory Wessel loaned us a cast of the jaguar track from Berome Moore Cave. Julianne Snider created the layout for Figures 2 and 3 and drafted Figure 1. Rickard Toomey assisted with Figure 4. Marlin Roos took the photographs of the plaster casts in Figure 3. We thank all of them for their assistance in this project. We also thank two anonymous reviewers for their comments.

LITERATURE CITED

Adams, D.B. 1979. The cheetah: native American. Science 205:1155-1158. Anderson, E. 1984. Who's who in the Pleistocene: a mammalian bestiary. Pp. 4089 in P.S. Martin and R.G. Klein (eds), Quaternary Extinctions - A Prehistoric Revolution. University of Arizona Press, Tucson. Anyonge, W. 1993. Body mass in large extant and extinct carnivores. Journal of Zoology, London 231:339-350. Brown, R.W., M.J. Lawrence, and J. Pope. 1984. The Larousse Guide to Animal Tracks, Trails and Signs. Larousse and Company, Inc., New York, 320 pp. Burt, W.H., and R.P. Grossenheider. 1976. A Field Guide to the Mammals. Houghton Mifflin Company, Boston, 289 pp. Cox, S.M., and G.T. Jefferson. 1988. The first individual skeleton of Smilodon from Rancho La Brea. Current Research in the Pleistocene 5:66--67. Farlow, J.O ., and M.G. Lockley. 1993. An osteometric approach to the identification of makers of early Mesozoic tridactyl dinosaur footprints. Pp. 123-131 in S.G. Lucas and M. Morales (eds), The Nonmarine Triassic. New Mexico Museum of Natural History and Science Bulletin 3. Gonyea, W.J., 1976. Behavioral implications of saber-toothed felid morphology. Paleobiology 2:332-342. Halfpenny, J., and E. Biesiot. 1986. A Field Guide to Mammal Tracking in North America. Johnson Books, Boulder, 161 pp. Harington, C.R. 1971. A Pleistocene lion-like cat (Panthera atrox) from Alberta. Canadian Journal of Earth Sciences 8:170-174. Koenigswald, W. von. 1994. Recent discoveries - the tracksite Bottrop-Welheim. European Quaternary Mammal Research Association Newsletter 1:58-59. Kurten, B., and E. Anderson. 1980. Pleistocene Mammals of North America. Columbia University Press, New York, 442 pp.

Panthera atrox tracks 345 Martin, L.D., C.B. Schultz, and M.R. Schultz. 1988. Saber-toothed cats from the Plio-Pleistocene of Nebraska. Transactions of the Nebraska Academy of Sciences 16:153-163. Merriam, J.C., and C. Stock. 1932. The Felidae of Rancho La Brea. Carnegie Institution of Washington Publication 422:1-231. Morse, S.C. 1992. Cougar tracks. Pp. 40-41 in K. Hansen, Cougar-The American Lion. Northland Publishing, Flagstaff. Murie, O.J. 1974. A Field Guide to Animal Tracks. Houghton Mifflin Company, Boston, 375 pp. Nowak, R.M. 1991. Walker's Mammals of the World, 5th edition, 2 volumes. Johns Hopkins University Press, Baltimore, 1181 pp. Oesch, R.D. 1969. Fossil Felidae and Machairodontidae from two Missouri caves. Journal of Mammalogy 50:367-368. Rabinowitz, AR. 1986. Jaguar. Struggle and Triumph in the Jungles of Belize. Arbor House, New York, 368 pp. Rawn-Schatzinger, V. 1992. The scimitar cat Homotherium serum Cope: osteology, functional morphology, and predatory behavior. Illinois State Museum Reports of Investigations 47:1-80. Robbins, L., R.C. Wilson, and P.J. Watson. 1981. Paleontology and archeology of Jaguar Cave, Tennessee. Eighth International Congress of Speleology Proceedings 1:377-380. Sankhala, K. 1977. Tiger! The Story of the Indian Tiger. Simon and Schuster, New York, 220 pp. · Seymour, K.L. 1989. Panthera onca. Mammalian Species 340:1-9. - 1993. Size change in North American Quaternary jaguars. Pp. 343-372 in R.A Martin and AD. Barnosky (eds), Morphological Change in Quaternary Mammals of North America. Cambridge University Press, Cambridge. Simpson, G.G. 1941. Discovery of jaguar bones and footprints in a cave in Tennessee. American Museum Novitates 1131:1-11. Vandike, J.E. 1983. Of caves and cat tracks. Missouri Life (July/ August) 83:49-54. Van Valkenburgh, B. 1987. Skeletal indicators of locomotor behavior in living and extinct carnivores. Journal of Vertebrate Paleontology 7:162-182. Van Valkenburgh, B., F. Grady, and B. Kurten. 1990. The Plio-Pleistocene cheetahlike Miracinonyx inexpectatus of North America. Journal of Vertebrate Paleontology 10:434-454.

Pleistocene mammals of Dublin Gulch and the Mayo District, Yukon Territory

C.R. Harington

Abstract Pleistocene mammal bones from near the base of an organic silt/ colluvium unit overlying glacial till at Dublin Gulch, Yukon Territory, represent the following taxa (in order of abundance of specimens): small horse (Equus lambei), bison (probably steppe bison, Bison priscus), Dall sheep (Ovis dalli), caribou (Rangifer tarandus), moose (Alces cf. A. alces), American lion (Panthera leo atrox), and possibly mammoth (cf. Mammuthus sp.). A radiocarbon date of 31,450 ± 1,750 BP on a horse bone indicates that this fauna is of middle Wisconsinan age. Species represented in the Dublin Gulch fauna evidently lived in a nearly treeless environment like the present low arctic tundra, but drier.

Introduction In 1916, during the First World War, placer scheelite deposits at Dublin Gulch in the Mayo District of the Yukon (Fig. 1) attracted considerable attention because of the increasing demand for scheelite and other tungsten minerals. Such minerals collected in sluice boxes as a by-product of placer mining for gold in the area over the past century (Cairnes, 1916; Cockfield, 1919). Another important by-product of gold placer mining in the Yukon is fossil bone. However, despite intensive, long-term placer mining in the Mayo District, only one site - Dublin Gulch - is known to have produced many Pleistocene mammal fossils. I received two lots of bones from Dublin Gulch in 1977 from Father H . Huijbers of the Kluane Museum of Natural History via Dr A.M. Pearson, then Commissioner of the Yukon. The first lot, consisting of seven specimens representing small horse,

135•

138 ·

.&. Mount Cameron

...

~,fou nt Haldane

i

McOuesten River

' '

::,

Ste,.,, Ort

R·

30"

Ive,.

=-

0

""

[

:

....

(ll

36,830 BP (AA-11834). If this date truly indicates a minimum age for this specimen, this is the oldest absoh,tely dated record of caribou in the southeastern United States and establishes the presence of caribou in this region before the late Wisconsinan glacial stade. The caudal (occipital) and dorsal (frontal) surfaces are nearly complete, the lateral (temporal) surfaces are partially complete, and the ventral side is lacking in its entirety. The pedicles are present; they are nearly round in cross-section and measure 55.4 mm across at their maximum diameter. This cranium is similar to those of modern Rangifer in all respects, and is readily differentiated from other cervids, but it is noticeably larger than most skulls of modern male caribou. Measurements of this specimen include minimum distance across cranium posterior to antler pedicles 118.5 mm; minimum distance across pedicles - 133.0 mm; depth of occiput, opisthion to nuchal line - 88.5 mm; breadth of occiput- 143.0 mm; pedicle length (left side)- 56.5 mm; pedicle breadth (left side)- 58.0 mm. 9. Continental Shelf, off Currituck Beach, North Carolina (ca. 36°18' N, 75°36' W); partial antler (Fig. 8). This specimen (USNM 476397), identified tentatively as a fragment of the first tine of a left antler of Rangifer, was dredged from the Continental Shelf at about 13 miles (21 km) ESE of Currituck Beach, at Corolla, Currituck County, North Carolina, by Ernest Riccio in 1976. This fragment compares well with USNM 246317 and 242244, males from Alaska. The dominant consideration in determining the side represented by this specimen was that at the proximal end the narrower arc is ventral and the more strongly concave side is medial. 10. New Bern, North Carolina (35°08'30" N, 77°05'00" W; Askin Quadrangle, USGS 7.5' series); partial antler (Fig. 9). The basal portion of a right antler (USNM 306478) was collected by Donnie Bailey, date unknown, in the Superior Stone Company (now Martin Marietta) Quarry at New Bern, Craven County, North Carolina. This site also has yielded a tooth of the extinct woodland musk ox, Bootherium bombifrons (McDonald and Ray, 1993). The beam extends from the burr to immediately above the bez tine and is 203 mm in length; the brow tine is lacking and about 133 mm of the bez tine remains. The main beam is subcircular at the base, then becomes flattened at the level of the bez tine. The base measures 35.9 mm in diameter and 34.8 mm X 40.0 mm immediately distal to the bez tine.

420 J.N. McDonald, C.E. Ray, F. Grady

3cm

FIGURE 8. Fragment of left antler, USNM 476397, in lateral aspect, of Rangifer tarandus dredged from 13 miles (21 km) ESE of Currituck Beach, NC

Rangifer tarandus in eastern United States 421

FIGURE 9. Incomplete right antler, USNM 306478, in medial aspect, of Rangifer tarandus from New Bern, NC

422 J.N. McDonald, C.E. Ray, F. Grady 11? Myrtle Beach, South Carolina (33°47 45" N, 78°45 53" W; Hand Quadrangle, USGS 7.5 series); partial antler (Fig. 10). This partial (?)right antler (USNM 467795) was found at the slosh line on the west side of the Intracoastal Waterway directly opposite the boundary between Briarcliff Mall and Briarcliff Woods in the Myrtle Beach area of Horry County, South Carolina. The specimen was found at the base of a cliff which, in vertical section, from the upper surface down, consisted of approximately 4.6 m (15 ft.) of gray sand, 2.4 m (8 ft.) of gray clay, 25-50 mm (1-2 in.) of peat, 25 mm (1 in.) of fine gravel, and 0-75 mm (0-3 in.) of black peaty fine sand. Beneath the sand, continuing to below the water level, was a layer of coquina (D. Bohaska, pers. comm., 8 April 1994). The antler fragment was found lying on the fine gravel, presumably having eroded from the peat. The specimen was received from John F. Arthur in 1988. Collagen from a fragment of this antler yielded an AMS radiocarbon date of 27,900 ± 775 BP (AA-11833). This specimen consists of the proximal beam; it is subrounded at the burr but becomes rapidly palmate distally. The brow tine is lacking and, based on the fracture scar, appears to have been unusually small. There are two tines at the bez-tine level, both palmate; one 78 mm in length is directed rostrally and the other ca. 30 mm in length is directed caudally. The diameter of the pedicle contact is 34.2 mm and the beam, above the brow tine, measures 33.0 mm X 44.5 mm. This specimen is referred tentatively to Rangifer. We have been unable to locate antlers known to be Rangifer that show the combination of unusual characters represented in USNM 467795: the extreme palmation, the rostral and caudal expressions of the bez tine, and the seemingly unusually small brow tine. We have, however, observed extreme palmation (USNM 258444), rostrally and caudally directed tines at nearly the same level on an antler (USNM 236577), and small accessory tines at the brow-tine level directed laterally or caudally (USNM 222491). Antler 2267 from Angmagssalik, Greenland (illustrated in Degerb0l, 1957: Pl. VII) has both a palmate beam from near the burr and a caudally directed tine unusually close to the bez-tine level, and constitutes the Rangifer antler most like USNM 467795 that we were able to locate. Our opinion is that USNM 467795 most nearly resembles the antlers of Rangifer, and that the combination of characters present, although problematic, could reasonably combine and be expressed in Rangifer antlers, given the extreme variation known to occur in antlers of this species. At the same time, we allow that this specimen might eventually be shown to belong to another taxon, and thus record it here as ?Rangifer. 1

1

1

Rangifer tarandus in eastern United States 423

FIGURE 10. Incomplete left(?) antler, USNM 467795, in lateral(?) aspect, of ?Rangi-

fer from Intracoastal Waterway, SC

424 J.N. McDonald, C.E. Ray, F. Grady

Discussion We are of the opinion that all specimens described herein are properly referred to Rangifer tarandus, but not with equal confidence. The teeth and cranium are securely referred to Rangifer. Although the morphology of antlers of Rangifer is generally predictable, the details of these antlers are notoriously plastic and variable in expression owing to several factors attributable to age, sex, individual physiology, and environmental influences (Banfield, 1961; Groves and Grubb, 1987; Lister, 1987; Spiess, 1979). In all the antler specimens described above, we found identical or nearly identical examples of most characters and their combinations, the most important of which we considered to be the caudolaterally directed proximal part of the beam, the presence of a brow tine, the subcircular to tearshaped cross-section of the beam, occasional palmation of the proximal beam or tines, and the repetition of the general tine patterns observed in specimens of known taxonomic identity. In most specimens we found at least three characters present that could readily permit assigning the specimen to Rangifer. Only in the case of USNM 467795 from Myrtle Beach, South Carolina, were we unable to find a nearly identical fit between the characters represented in the specimen and a specimen of known taxonomic identity. The antlers described herein are, therefore, more nearly consistent with those known for Rangifer than for any other cervid genus, yet we allow that one or more might eventually be shown to belong to some other taxon. We have least confidence in USNM 467795, yet we feel that this antler is more nearly like those of Rangifer than of any other cervid known to us. The records described in this paper extend the southeastern range of Rangifer in North America onto the Gulf and Atlantic coastal plains (including their seaward expressions as the Continental Shelf). The Catalpa Creek record extends the range nearly 160 km (100 mi.) southwest of its previous projected location, whereas the Atlantic Coastal Plain records extend the range eastward from 160 km (100 mi.) in the north (Georges Bank) to nearly 362 km (225 mi.) in the south (Myrtle Beach). One of the most distinctive and immediately recognizable features of the distribution of Rangifer in the eastern United States is the lobate shape of the range (Churcher et al., 1989), a shape that mirrors that of the Wisconsinan Laurentide glacial system. This correspondence of shape, and the fact that Rangifer is a boreal ungulate, strongly suggests that Rangifer probably reached its maximum southeasterly distribution during glacial time, an idea that has been accepted for many years. The fact that three of the specimens described in this paper came from the Continental Shelf or the surf along the Atlantic Coast indicates that Rangifer occupied the region during some glacial stage(s) when sea level was lower and the Continen-

Rangifer tarandus in eastern United States 425 tal Shelf was emergent. The radiocarbon dates presented in this paper correspond to this expectation, but they also caution that the remains of terrestrial ungulates recovered from the coastal plains, including the Continental Shelf, do not necessarily date from the last glacial stade. Recent attention to terrestrial vertebrate fossils of Quaternary age from the Atlantic Coastal Plain has shown that large boreal ungulates - Bootherium, Cervalces, and Rangifer - were, in fact, widely distributed as far south as South Carolina (McDonald and Ray, 1993; McDonald and Ray, unpub. data.). Ovibos, a boreal ungulate with extremely narrow ecological requirements, has been reported from the Continental Shelf east of New Jersey (McDonald and Ray, 1993). Bison, the boreal ungulate occupying the widest niche, is known in the southeastern United States from as far south as southern Florida (Robertson, 1974). Reconstructions of regional environments during the last glacial maximum postulate that the climate was cold and dry because of zonal air flow, the rain-shadow effect of the Appalachian Mountains, the southward penetration of the effects of the Labrador Current, and the associated seasonal patterns of surface winds (Delcourt and Delcourt, 1986; Webb et al., 1987). Regional vegetation, consisting primarily of boreal forest, similar to that found in southern Manitoba and east-central Ontario today, occurred as far south as central South Carolina (Delcourt and Delcourt, 1985, 1986). A mixed-forest ecotone bordered this boreal forest on the south. Northern Mississippi, according to this model, would have been vegetated by mixed or boreal forest. In lowland areas, aeolian-driven erosion and dune formation likely would have contributed to patchiness of vegetation, as would browsing by large herbivorous mammals such as ground sloths, mastodons, and mammoths. The Appalachian Mountains would have had a more extremely alpine vegetation ranging from unvegetated patches through tundra to alpine evergreen forest and woodland. Boreal vertebrates have long been recognized as members of the Appalachian biotic communities during glacial stages, but the emerging pattern of ungulate distribution at lower elevations in the southeastern United States also fits well with the model of regional environments during glacial stages, and lends support to the idea that Rangifer was well adapted to tundra, woodland, and forested environments during the Wisconsinan glaciation. These data also demonstrate that the large boreal ungulates were able to occupy lower-elevation habitats in the southeastern United States, and were not restricted to the Appalachian Mountains during glacial times as has been intimated by previously documented patterns of distribution. One implication of this is that, if the larger boreal mammals were able to inhabit lower elevations, then so too might smaller boreal and alpine

426 J.N. McDonald, C.E. Ray, F. Grady mammals or other vertebrate groups, all of which has important implications for our understanding of Quaternary zoogeography of the United States east and south of the Appalachian Mountains. We accept the fact that climatic and botanic conditions certainly might have influenced the narrower distribution of some vertebrate taxa, but we also feel that the distribution of many boreal taxa will change when more fossiliferous sites from the Piedmont or Coastal Plain, such as those already reported upon by Voorhies (1974) and Frazier (1980), have been located and thoroughly examined. Much more work needs to be done with well-documented rich faunas in this poorly known region where only Florida may be regarded as being reasonably well known.

Acknowledgments We wish to thank John F. Arthur, Donnie Bailey, John Hall, John M. Kaye, and Charles W. Totten for donating the specimens described above to the National Museum of Natural History, and Eleanor Daly and Peter J. Harmatuk for assisting with the transfer of the Catalpa Creek, Mississippi, and New Bern, North Carolina, specimens, respectively. E. Ray Garton co-directed excavations at New Trout Cave. Tomm and Joan Reinbold located, and assisted with excavations at, Wormhole Cave, and Tom Kaye located, and assisted with excavations at, Organ Cave. Vivian Blackford kindly permitted extended study of the partial cranium from near the North Carolina/Virginia border. Terry J. Powell graciously permitted study of the Rangifer material in the Sunrise Children's Museum. Mary Parrish prepared Figure 1 and Connie Barut Rankin prepared Figure 3. Victor E. Krantz and Peter Kroehler took the photographs for, and David Bohaska prepared, Figures 2 and 4 through 10. Two anonymous reviewers made helpful comments that improved this paper.

Addendum After this manuscript had been submitted to the editors, A.E. Saunders reminded us of an isolated tooth (Charleston Museum No. PV3411) from the Chandler Bridge locality, Dorchester County, South Carolina, that had been referred tentatively to Rangifer by one of us (CER) some years previously. The tooth is a well-preserved, little-worn left p4 that can be matched closely by those of large male individuals of modem Rangifer tarandus in the collections of the US National Museum. We are reluctant to add a positive extralimital locality record on the basis of one tooth, but place it on record here to emphasize further the need to be alert for additional materials of Rangifer from this and other southerly localities.

Rangifer tarandus in eastern United States 427 LITERATURE CITED

Banfield, A.W.F. 1961. A Revision of the Reindeer and Caribou, Genus Rangifer. National Museum of Canada, Bulletin 177, Biological Series Number 66, 138 pp. Churcher, C.S. 1984. Sangamona: The furtive deer. Pp. 316-331 in H.H. Genoways and M.R. Dawson (eds), Contributions in Quaternary Vertebrate Paleontology: A Volume in Memorial to John E. Guilday. Carnegie Museum of Natural History, Special Publication 8. Churcher, C.S., P.W. Parmalee, G.L. Bell, and J.P. Lamb. 1989. Caribou from the Late Pleistocene of northwestern Alabama. Canadian Journal of Zoology 67:1210-1216. Daly, E. 1992. A List, Bibliography and Index of the Fossil Vertebrates of Mississippi. Mississippi Office of Geology Bulletin 128, 47 pp. Degerb0l, M. 1957. The extinct reindeer of East-Greenland, Rangifer tarandus groenlandicus, subsp. nov., compared with reindeer from other Arctic regions. Acta Arctica 10, 58 pp. Delcourt, H.R., and P.A. Delcourt. 1985. Quaternary palynology and the vegetational history of the southeastern United States. Pp. 1-37 in V.M. Bryant, Jr, and R.G. Holloway (eds), Pollen Records of Late-Quaternary North American Sediments. American Association of Stratigraphic Palynologists Foundation, Dallas. - 1986. Late Quaternary vegetational change in the central Atlantic states. Pp. 23-35 in J.N. McDonald and S.O. Bird (eds), The Quaternary of Virginia A Symposium Volume. Virginia Division of Mineral Resources Publication 75. Frazier, M.K. 1980. A Late Quaternary fossil vertebrate assemblage from Lowndes County, Mississippi (Abstract). Journal of the Mississippi Academy of Science, 25 Supplement 44. - 1985. Paleontology study. Pp. C88--Cl06 in G.R. Muto and J. Gunn (eds), A study of Late-Quaternary environments and early man along the Tombigbee River, Alabama and Mississippi. Benham-Blair and Affiliates, Inc., Oklahoma City. Gallagher, W.B., D.C. Parris, B.S. Grandstaff, and C. DeTample. 1989. Quaternary mammals from the Continental shelf off New Jersey. The Mosasaur 4:101-110. Grady, F. 1984. Wormhole Cave: Another significant bone site for Pendleton County, West Virginia. Potomac Caver 27(12):213-214. - 1988. Vertebrate fossils. Pp. 51-55 in P. Stevens (ed.), Caves of the Organ Cave Plateau. West Virginia Speleological Survey Bulletin 9. Grady, F., and E.R. Garton. 1982. Pleistocene fauna from New Trout Cave. Capital Area Cavers Bulletin 1:62-69. Groves, C.P., and P. Grubb. 1987. Relationships of living deer. Pp. 21-59 in C.M.

428 J.N. McDonald, C.E. Ray, F. Grady Wemmer (ed.), Biology and Management of the Cervidae. Smithsonian Institution Press, Washington, DC. Guilday, J.E., H.W. Hamilton, E. Anderson, and P.W. Parmalee. 1978. The Baker Bluff Cave deposit, Tennessee, and the Late Pleistocene faunal gradient. Bulletin of the Carnegie Museum of Natural History 11, 68 pp. Guilday, J.E., H.W. Hamilton, and A.D. Mccrady. 1966. The bone breccia of Bootlegger Sink, York County, PA. Annals of Carnegie Museum 38:145-163. Guilday, J.E., H.W. Hamilton, and P.W. Parmalee. 1975. Caribou (Rangifer tarandus L.) from the Pleistocene of Tennessee. Journal of the Tennessee Academy of Science 50:109-112. Guthrie, R.D., and J.V. Matthews, Jr. 1971. The Cape Deceit fauna - Early Pleistocene mammalian assemblage from the Alaskan Arctic. Quaternary Research 1:474-510. Hay, O.P. 1920. Descriptions of some Pleistocene vertebrates found in the United States. Proceedings of the United States National Museum (2328):58 - 1923. The Pleistocene of North America and Its Vertebrate Animals from the States East of the Mississippi River and from the Canadian Provinces East of Longitude 95°. Carnegie Institution of Washington Publication 322,499 pp. Jacobi, A. 1931. Das Rentier: Eine zoologische Monographie der Gattung Rangifer. Ergii.nzungsband zum Zoologischen Anzeiger 96, vii + 264 pp. Jillson, W.R. 1936. Big Bone Lick: An Outline of Its History, Geology and Paleontology. Big Bone Lick Association Publications 1, 164 pp. - 1949. The Discovery of Pleistocene Vertebrates at Lower Blue Licks. Transylvania Series 6, 30 pp. - 1968. The Extinct Vertebrata of the Pleistocene in Kentucky. Roberts Printing Co., Frankfort, 122 pp. Kaye, J.M. 1974. Pleistocene sediment and vertebrate fossil associations in the Mississippi Black Belt: a genetic approach. Ph.D. dissertation (Geology), Louisiana State University, 116 pp. Kurten, B., and J.M. Kaye. 1982. Late Quaternary carnivora from the Black Belt, Mississippi. Boreas 11:47-52. Leidy, J. 1869. The extinct mammalian fauna of Dakota and Nebraska, including an account of some allied forms from other localities, together with a synopsis of the mammalian remains of North America. Journal of the Academy of Natural Sciences of Philadelphia, Series 2, no. 7, 472 pp. Lister, A.M. 1987. Diversity and evolution of antler form in Quaternary deer. Pp. 21-59 in C.M. Wemmer (ed.), Biology and Management of the Cervidae. Smithsonian Institution Press, Washington, DC. Lundelius, E.L., Jr, T. Downs, E.H. Lindsay, H.A. Semken, R.J. Zakrzewski, C.S. Churcher, C.R. Harington, G.E. Schultz, and S.D. Webb. 1987. The North American Quaternary sequence. Pp. 211-235 in M.O. Woodburne (ed.), Cenozoic Mammals of North America: Geochronology and Biostratigraphy. University of California Press, Berkeley.

Rangifer tarandus in eastern United States 429 McDonald, J.N. 1984. The Saltville, Virginia, Locality: A Summary of Research and Field Trip Guide. Virginia Division of Mineral Resources, Charlottesville, 46 pp. - 1986. Valley-bottom stratigraphy of Saltville Valley, Virginia, and its paleoecological implications. National Geographic Society Research Reports 21:291296. McDonald, J.N., and C.S. Bartlett, Jr. 1981. An associated musk ox skeleton from Saltville, Virginia. Journal of Vertebrate Paleontology 2:453-470. McDonald, J.N., and C.E. Ray. 1993. Records of musk oxen from the Atlantic Coastal Plain of North America. The Mosasaur 5:1-18. Martin, R.A., and J.M. Sneed. 1989. Late Pleistocene records of caribou and elk from Georgia and Alabama. Georgia Journal of Science 47:117-122. Oldale, R.N., F.C. Whitmore, Jr, and J.R. Grimes. 1987. Elephant teeth from the western Gulf of Maine, and their implications. National Geographic Research 3:439-446. Pewe, T.L. 1975. Quaternary Geology of Alaska. United States Geological Survey Professional Paper 835, v + 145 pp. Pewe, T.L., and D.M. Hopkins. 1967. Mammal remains of pre-Wisconsin age in Alaska. Pp. 266-270 in D.M. Hopkins (ed.), The Bering Land Bridge. Stanford University Press, Stanford. Pewe, T.L., R.D. Reger, and J.A. Westgate. 1989. Fairbanks area. Pp. T102:3T102:16 in T.L. Pewe and R.D. Reger (eds), Quaternary Geology and Permafrost along the Richardson and Glenn Highways between Fairbanks and Anchorage, Alaska. 28th International Geological Congress, Field Trip Guidebook T102, American Geophysical Union, Washington, DC. Ray, C.E., B.N. Cooper, and W.S. Benninghoff. 1967. Fossil mammals and pollen in a late Pleistocene deposit at Saltville, Virginia. Journal of Paleontology 41:608622. Roberts, R. 1990. Old bones come up with clues to the Ice Age. SeaFrontiers 36(3):16-21. Robertson,J.S.,Jr. 1974. Fossil Bison of Florida. Pp. 214-246 in S.D. Webb (ed.), The Pleistocene Mammals of Florida. University of Florida Press, Gainesville. Schultz, C.B., F.C. Whitmore, Jr, L.L. Ray, and E.C. Crawford. 1963. Paleontological investigations at Big Bone Lick State Park, Kentucky: a preliminary report. Science 142:1167-1169. Sher, A. 1986. Olyorian Land Mammal Age of northeastern Siberia. Paleontographia ltalica 74:97-112. Spiess, A.E. 1979. Reindeer and Caribou Hunters. Academic Press, New York, 312 pp. Steadman, D.W. 1988. Vertebrates from the Late Quaternary Hiscock Site, Genesee County, New York. Bulletin of the Buffalo Society of Natural Sciences 33:95113. Voorhies, M.R. 1974. Pleistocene vertebrates with boreal affinities in the Georgia Piedmont. Quaternary Research 4:85-93.

430

J.N. McDonald, C.E. Ray, F. Grady

Webb, S.D. 1992. A cranium of Navahoceros and its phylogenetic place among New World Cervidae. Annals Zoologica Fennici 28:401-410. Webb, T., III, P.J. Hartlein, and J.E. Kutzbach. 1987. Climatic change in eastern North America during the past 18,000 years; comparisons of pollen data with model results. Pp. 447-462 in W.F. Ruddiman and H.E. Wright, Jr, (eds), North America and Adjacent Oceans during the Last Deglaciation. The Geology of North America, vol. K-3. Geological Society of America, Boulder, CO. Westgate, J.A., B.A. Stemper, and T.L. Pewe. 1990. A 3 m. y. record of PliocenePleistocene loess in interior Alaska. Geology 18:858-861.

Dental evolution and size change in the North American muskrat: Classification and tempo of a presumed phyletic sequence

Robert A. Martin

Abstract The simultaneous occurrence of two muskrat species is examined and it is tentatively concluded that muskrat samples available for study represent remains of a single species that can be treated as a phyletic sequence. Fossil muskrats originally referred to two genera and six species are combined under the living species, 0ndatra zibethicus Linnaeus. A modified form of the nomenclature introduced by Krishtalka and Stucky (1985) is used to recognize five chronomorphs: 0 . z. /minor, 0 . z. /meadensis, 0. z. /idahoensis, 0. z. /annectens, and 0 . z. /zibethicus. Size and character evolution within the muskrat phyletic sequence were considerable and the patterns of change displayed support the model of episodic phyletic evolution as discussed by R. Martin and Barnosky (1993). Dental features did not change in unison over time intervals up to approximately 300,000 years; instead they reflect an overall mosaic character response. Body size in muskrat evolution, as estimated by changes in M1 length, displayed a pattern of modest change through much of late Pliocene and Pleistocene time, followed by a rapid increase during the last 0.6 million years and a period of dwarfing during the latest Pleistocene and Holocene. Developmental heterochronic processes appear to have been important in muskrat dental and size evolution and the patterns displayed conform to McNamara's (1990) 'stepped anagenetic paedomorphocline. ' Despite the apparent lack of uniformity in character response and some overlap in populations with derived and underived features, over millions of years the North American muskrat changed in a generally consistent fashion, allowing the recognition of intermediate forms that can be used effectively as biostratigraphic markers.

432

R.A. Martin

Introduction The muskrat, 0ndatra zibethicus, one of North America's most widespread and familiar rodents, has a long and well-documented fossil history. At approximately 0.84 kg, it is the largest of three living North American aquatic arvicoline rodents, the other two being Neofiber alleni (roundtailed muskrat) and Microtus richardsoni (water vole). The muskrat is represented in the palaeontological literature by a single lineage spanning three and three-quarter million years, composed of the following taxa: Pliopotamys minor - P. meadensis - 0ndatra idahoensis - 0. annectens - 0. nebracensis - 0. zibethicus. The direct ancestor of Pliopotamys is currently unknown. More than 100 muskrat specimens have been reported from deposits throughout North America. A few sites have associated dates and are additionally correlated by palaeomagnetic data and biostratigraphy. The Upper Pliocene Glenns Ferry Formation, from which the Hagerman Local Fauna (LF) was recovered, has allowed examination of morphological variation in muskrats in a set of superposed strata spanning considerable time (R. Martin, 1993; Zakrzewski, 1969). As I will demonstrate (and others have assumed for many years; e.g., Semken, 1966), it is highly likely that there has been only one muskrat species in North America since the late Pliocene. Consequently, the muskrat fossil record offers the opportunity to examine some of the more central and fascinating questions in evolutionary biology and mammalian classification. Below, I review the muskrat fossil record and investigate the possibility of simultaneous occurrence of two or more species. These sections are followed by a new classification and an analysis of morphological and size change through time. Finally, this information is examined and synthesized in the framework of modern evolutionary theory.

Materials and methods Measurement methods follow Rabeder (1981) and Barnosky (1987). Greatest length of the first lower molar (M 1) is the occlusal, not girth, length (see Semken, 1966). It is often unclear from published accounts which dentine tract height has been measured or is being discussed. Terminology such as 'the dentine tract on the labial (buccal) side of the posterior loop of Mi' do not foster communication. The German palaeontologist Gernot Rabeder (1981) introduced a lexicon of names for tracts on arvicoline molars associated with the Linea sinuosa and those names are followed here. Crown height (C) was measured in the same manner as hyposinuid height (H), but was extended to the occlusal surface (Fig.I). The H/C ratio was used to examine the relationship between dentine-

Phyletic evolution in muskrats 433

Asd

crown height (C) hyposlnuld hei1ht (H)

FIGURE 1. Labial (buccal) view of Ondatra left M1, showing measurement of crown height and hyposinuid height (Asd = anterosinuid)

tract and crown-height evolution (see discussion in text) and has a range from 0-1.00. When H/C = 1.00, H = C and the hyposinuid opens at the occlusal surface. Body mass of extinct muskrats was estimated from a least squares regression of log 10 body size on log 10 M1 length in living arvicolines published by R. Martin (1993). I did not endeavour to study all of the fossil muskrat material from North America, but I have, at one point or another through the years, examined samples of most of the fossils mentioned in the text and included in Table l. I recently restudied the controversial Vera material and have casts of the Kentuck Ondatra. Samples of the living muskrat listed in R. Martin and Tedesco (1976) and two new samples from Georgia and Kentucky (Table 1) formed the basis of my discussion of Recent populations. Large fossil muskrats from Bell Cave, Alabama, and Yarbrough Cave, Georgia, reported in this paper are currently in the collections at the Red Mountain Museum (Birmingham, AL) and Murray State University (Murray, KY), respectively. Specimens of Ondatra from the early Pleistocene Java LF of South Dakota (R. Martin, 1989; R. Martin and Tedesco, 1976) were restudied and form the basis for comments regarding 0. annectens. The latter specimens are catalogued in the collections of the South Dakota School of Mines and Technology (Rapid City, SD).

The fossil record The genus Pliopotamys was named by Hibbard (1937) and includes a pair

434

R.A. Martin

TABLE 1 Average length and width of first lower molar in fossil and Recent samples of the muskrat, Ondatra zibethicus Chronomorph

Locality

zibethicus zibethicus zibethicus zibethicus zibethicus zibethicus zibethicus zibethicus zibethicus zibethicus zibethicus zibethicus zibethicus zibethicus zibethicus zibethicus annectens annectens annectens an nee tens an nee tens annectens annectens idahoensis idahoensis meadensis idahoensis idahoensis idahoensis idahoensis meadens is meadensis meadensis meadensis minor minor

British Columbia New Jersey Nebraska Louisiana *Georgia *Kentucky *Kingston Sr Cave Ichetucknee R. *Yarbrough Cave *Bell Cave (Z 1/2) *Bell Cave (Z 3) Doby Springs Hay Springs Anderson/Flohr Kanopolis Mullen 3 Cudahy Cumberland Cave *Patterson Ranch Courtland Canal Hamilton Cave Kentuck Java *Inglis IA Borchers Boyle Ditch Grandview White Rock Mullen 2 Seneca Dixon Sand Draw Deer Park Broadwater Sand Point Hagerman

Age

N

0 10 0 17 0 10 31 0 0 10 0 7 0.010 1 0.011 23 0.016 2 0.015 1 (0.020) 3 (0.25) 11 (0.35) 7 (0.35) 2 (0.50) 1 (0.50) 6 0.60 5 (0.70) 1 (0.70) 2 (1.2) 1 (1.5) 3 (1.75) 4 (1 .75) 8 (1 .9) 5 1.9 1 (2.0) 1 (2.3) 10/14 (2.5) 7/11 (2.5) 3 (2.5) 7 (2.6) 7/9 (2.7) 3 (3.0) 1 (3.0) 22 (3.5) 4 3.75 9

xL

xW

xL/xW xM

7.0 7.4 7.1 7.0 7.45 7.43 8.8 7.7 8.7 8.7 8.3 6.5 6.72 6.5 6.0 6.3 5.7 5.1 5.63 5.92 4.9 6.0 5.7 4.65 5.0 4.4 4.6 4.8 5.2 5.04 4.8 4.8 4.8 4.63 4.10 4.15

2.9 3.2 3.1 3.1 3.25 3.30 3.2 2.9 3.3 3.5 3.4

2.41 2.31 2.29 2.26 2.29 2.25 2.75 2.66 2.64 2.49 2.44

2.54 2.6

2.65 2.50

2.4 2.3 2.1 2.49 2.44

2.63 2.48 2.43 2.32 2.43

2.4 2.5 2.16 2.3 2.2 2.2 2.1 2.4

2.50 2.28 2.15 2.17 2.00 2.09 2.29 2.17

2.3 2.2 2.3 2.17 1.78 2.19

2.09 2.18 2.09 2.13 2.30 1.90

0.77 0.94 0.81 0.77 0.96 0.95 1.75 1.08 1.68 1.68 1.42 0.59 0.66 0.59 0.44 0.53 0.37 0.25 0.35 0.42 0.21 0.44 0.37 0.18 0.23 0.15 0.17 0.20 0.26 0.24 0.20 0.20 0.20 0.17 0.11 0.12

Notes: Age is given in millions of years, with 'O' indicating a Recent sample; most dates are from Lundelius et al. (1987) and those not determined by radiocarbon dating or defined by a dated ash are noted in parentheses. N = number of specimens; where N differs for width measurement, both are given. xL = mean length; xW = mean width at occlusal surface; xM = mean mass in kg, as calculated from equation (1) in text;•= new data; Sr = Saltpeter. Other data taken from the literature.

Phyletic evolution in muskrats 435 of primitive muskrat species with cementum either lacking or poorly developed in the molar re-entrant angles. There are additional features that characterize the earliest muskrats, P. minor and P. meadensis, and these will be discussed in another section of the paper. The earliest record of P. minor is from the Hagerman LF of Idaho, dated at approximately 3.75 ma (million years ago). According to the latest North American correlations (Barnosky, 1985; Lundelius, et al., 1987), P. minor, known only from the Hagerman and Sand Point LF's of Idaho, has not been found after approximately 3.5 ma. P. meadensis was the common muskrat between 3.0 and 2.5 ma and is known from the Dixon, Deer Park, Sand Draw, Grand View, and other LFs. Its range apparently extended into the now arid southwest, as it is tentatively identified from the Tusker LF of Arizona (Tomida, 1987). Muskrat molars intermediate between P. meadensis and 0 . idahoensis are recorded from the Boyle Ditch LF of Wyoming (Barnosky, 1985) and White Rock LF of Kansas (Eshelman, 1975) at roughly 2.0-2.5 ma. 0. idahoensis, recovered from deposits in Kansas and Idaho (Eshelman, 1975; Shotwell, 1970; Wilson, 1933), dominated the time from approximately 2.5 ma to the earliest Irvingtonian Borchers LF at about 1.9 ma. To the north, at approximately 1.75 ma, 0. annectens appears in the Java LF of South Dakota (R. Martin, 1989; R. Martin and Tedesco, 1976). 0ndatra annectens has a long and relatively stable history on the Great Plains until about 0.5 ma, when a somewhat more advanced form, 0. nebracensis, appears in the Kanopolis LF of Kansas (Hibbard et al., 1978). The extant muskrat, 0. zibethicus, is recorded in the literature from latest lrvingtonian local faunas such as Berends of Oklahoma (Starrett, 1956) around 0.25 ma. 0ndatra hiatidens was described by Cope (1871) from the Port Kennedy Cave site in Pennsylvania. The age of this fauna is unknown, but is considered by Lundelius et al. (1987) as middle Irvingtonian, perhaps slightly older than the Cumberland Cave LF, which dates to about 0.7 ma.

The number of muskrat species Hibbard and Dalquest (1966) were the first authors to report cooccurrence of two muskrat species in the same deposit. A single M1 fragment from a large, advanced muskrat was associated with material referable to 0. annectens in the Vera LF from the Jones Ranch site, a Cudahy equivalent in Texas. I have examined the broken M1 (UMMP 46340) and concur with Hibbard and Dalquest (1966) that it does represent a species advanced over 0 . annectens. The specimen is coloured differently than the other 0ndatra fossils from Vera and I am suspicious that contamination has occurred. Dalquest (1990, letter) also considered this likely and Nelson

436

R.A. Martin

and Semken (1970) noted the possibility of this specimen being an 'artifact.' Semken's (1966) chronocline of M1 width versus length showed considerable size variation in the Kentuck, Kansas, 0ndatra material. Semken (1966) in fact referred one M1 (UMMP 24508) to 0. annectens and the remaining material to 0. nebracensis. Van der Meulen (1978) suggested that the presence of two size groups in the Kentuck LF reflected deposition during a glacial-interglacial cycle and that all the material could be from a single species. I examined casts of three of the four M1's discussed by Semken (1966): UMMP 24508 (referred to 0. annectens) and two specimens referred to 0 . nebracensis, UMMP 50931 and UMMP 34712. These specimens actually represent three different wear stages. UMMP 50931 is very slightly worn, UMMP 24508 is moderately worn, and UMMP 34712 is heavily worn. In size and overall morphology, UMMP 24508 and UMMP 34712 can be duplicated in the sample from the Java LF of South Dakota that has been identified as 0. annectens (R. Martin, 1989; R. Martin and Tedesco, 1976). Both faunas have produced Microtus pliocaenicus and are considered roughly contemporaneous (Lundelius et al., 1987; R. Martin, 1987), although R. Martin (1989) suggested on faunal evidence that the Java LF was somewhat older than Kentuck. The dentine tract of the posterior loop of UMMP 50931 is slightly higher than a tooth in comparable wear from the Java LF, but the size range represented by the Kentuck sample (range = 1.1 mm; observed range = 5.3 - 6.4 mm) is typically found among samples of extant 0 . zibethicus (R. Martin and Tedesco, 1976). From these observations, it probably is not necessary to invoke a glacial/ interglacial cycle to explain the differences among muskrats from the Kentuck site. The variation expressed is consistent with a single population of advanced 0 . annectens. Nelson and Semken (1970) identified two size classes of muskrats in the American Museum of Natural History (AMNH) collections from the Hay Springs LF of Nebraska. It now seems clear that the 1890s AMNH field crews combined materials from a series of terrace fills in Nebraska that could range in age from Illinoian through Kansan time (Nelson and Semken, 1970). Cope's (1871) 0ndatra hiatidens, redescribed by Hibbard (1955), was supposedly a small, middle Pleistocene species lacking cementum in the re-entrant angles. I recently examined these specimens at the American Museum of Natural History and the material I saw, including the holotype M 1, was from juvenile individuals, which explains their small size and lack of cementum. Another small muskrat M 1 was recovered from the Irvingtonian Fyllan Cave LF of Texas (A. Winkler, unpublished), but this molar was broken and heavily weathered. Theoretically, these samples could represent a small southern and eastern species living simultaneously with 0 . annectens, the latter being most commonly

Phyletic evolution in muskrats 437 encountered in Irvingtonian faunas on the Great Plains. In various publications (e.g., Nelson and Semken, 1970; Repenning, 1987; Stephens, 1960), there has been brief mention of the possibility of two muskrat lineages and the small muskrats noted above certainly create a question in this regard. However, over 150 muskrat molars have been reported from numerous deposits throughout the continental United States, and with the exception of the molar fragments discussed above, there has been no verified report of sympatry between two muskrat species. Consequently, for the purposes of this study it is assumed that the history of muskrats in North America represents the evolutionary story of a single species.

A novel classification of extinct and extant muskrats R. Martin (1993) synonomized the fossil and Recent muskrats under the living species Ondatra zibethicus Linnaeus, invoking a modified nomenclatural method introduced originally by Krishtalka and Stucky (1985). The philosophical basis for this decision and subsequent modifications to the Krishtalka/Stucky system were discussed in detail by R. Martin (1995) and will not be repeated here (although some concerns about this system are addressed in the Discussion). An informal trinomial is used to recognize temporal populations, or chronomorphs. Throughout the text the full trinomial will often be abbreviated only by its informal identifier (e.g., / annectens). In the classification below and after each new taxonomic combination (chronomorph), the name of a local fauna is provided. The fossil muskrat sample from each LF is considered representative of the chronomorph. A key to the identification of each chronomorph is given following the classification.

Ondatra zibethicus (Linneaus), 1766 Ondatra zibethicus I minor - Hagerman LF = Pliopotamys minor (Wilson), 1933 Ondatra zibethicus I meadensis - Dixon LF = Pliopotamys meadensis Hibbard, 1937 Ondatra zibethicus I idahoensis - Grandview LF ? = Ondatra hiatidens (Cope), 1871 = Ondatra idahoensis (Wilson), 1933 Ondatra zibethicus I annectens - Conard Fissure LF = Ondatra annectens (Brown), 1908 = Ondatra kansasensis Hibbard, 1944 Ondatra zibethicus /zibethicus - living muskrat = Ondatra nebracensis (Hollister), 1911 = Ondatra triradicatus Starrett, 1956 = Ondatra zibethicus floridanus Lawrence, 1942

438

R.A. Martin

The chronomorphs may be defined as follows (unless specifically stipulated, characters are restricted to adult specimens):

0ndatra zibethicus /minor: size small (x Length [L] M 1 < 4.2 mm), usually five triangles plus an anteroconid (ACD) on M 1, 'mimomyan' atoll (= islet) on ACD of M 1 in juveniles, dentine tracts absent or poorly developed, cementum absent from re-entrant angles, M 1 with three roots 0ndatra zibethicus /meadensis: size small, but larger than 0. z. /minor (x L M 1 between 4.4 and 4.8 mm), usually five triangles plus an ACD on M 1, mimomyan atoll present on ACD of M 1 in juveniles, dentine tracts considerably better developed than in 0. z. /minor, cementum in re-entrant angles absent or poorly developed, M 1 with three roots 0ndatra zibethicus I idahoensis: size small, but generally larger than 0. z. /meadensis (x L M 1 between 4.6 and 5.2 mm); cementum present in reentrant angles, mimomyan atoll rare on ACD of M 1 even in juveniles, dentine tracts moderately developed, usually five triangles plus ACD on M 1, M 1 with three roots 0ndatra zibethicus I annectens: larger than 0. z. / idahoensis, but smaller than 0. z. /zibethicus (x L M 1 between 4.9 and 6.0 mm), dentine tracts moderately to well developed, five to seven triangles plus ACD on M 1, M 1 with two or three roots 0ndatra zibethicus /zibethicus: generally includes the largest muskrats (x L M 1 between 6.0 and 8.8 mm), at least seven triangles plus ACD on M 1, cementum in re-entrant angles, dentine tracts well developed, M 1 with two roots in young animals, a third root often developing in older individuals Measurements, bivariate plots and qualitative features published by Sarnosky (1985), Eshelman (1975), L. Martin (1979, 1993), R. Martin and Tedesco (1976), Semken (1966), Wilson (1933), and Zakrzewski (1969) can be used to identify Quaternary muskrat chronomorphs and further refinement can be attained by the use of the graphs on size change in R. Martin (1993) and in the next section of this paper.

Character evolution in the North American muskrat Muskrats are aquatic animals, although presumably derived from a large terrestrial vole. It is unfortunate that there have been no attempts to infer habitat preference from postcranial material, although such material may be minimal. This is an important study that needs to be attempted. The following evaluation, therefore, is restricted to dental material. In a series of papers describing most of the late Pliocene and Pleis-

Phyletic evolution in muskrats 439 tocene muskrats of North America, Hibbard (1959), R. Martin (1993), Semken (1966), and Zakrzewski (1969) identified and elaborated on those features that have come to be used at both the generic and specific level in muskrat evolution, although some of these characters had been considered briefly by Cope (1871), Hinton (1929), Hollister (1911), and Shotwell (1970). The trends in these features through time are as follows: 1 2 3 4 5 6

Size: small to large Number of triangles on M1: five to seven Height of dentine tracts on M1: low to high Enamel atoll in anteroconid of M 1: present to absent Cementum in re-entrant angles: absent to present Number of roots on M 1: three to two

These features are illustrated in Figure 2. They are considered individually below.

Size As demonstrated for other cricetid rodents (R. Martin, 1984, 1986, 1993), size changes in the dentition of arvicolines can be correlated with changes in body mass. This relationship is described by the exponential form of the least squares equation: W = 0.71L 3·59

(1)

where W = body mass in grams and L = M1 length in mm (R. Martin, 1993). The Pearson correlation coefficient for these data is 0.94. The Recent muskrats Nelson and Semken (1970) measured for their study, with the exception of a dwarf population on Isle Royal, Michigan, averaged longer M 1's than the samples reported here. From their scattergram of length against width of M1, the average length of M1 for Recent 0. zibethicus appears to be about 7.5 mm, with 10 specimens over 8.0 mm (largest 8.4 mm). I calculated another mass-estimating equation using the average value 7.5 mm for Recent 0 . zibethicus M 1 length and the formula derived is W = 0.80L 3·47

(2)

When the latest Pleistocene southeastern muskrat data are calculated according to this formula, they average 1.46 kg, slightly less than esti-

440

R.A. Martin

MODERN /zibethicus

Wr = 1 H/C = 0.90

Wr = 2 H/C = 1.00

Wr = 3 H/C = 1.00

JAVA (1.7S ma) /annectens Wr = 3 H/C = 1.00 Wr = 1 H/C = 0.62

Wr = 2 H/C = 0.86

SAND DRAW (3.0 ma) /meadensis Wr = 1 H/C = 0.29

FIGURE 2. Variation in the first lower molar among select chronomorphs of the muskrat, Ondatra zibethicus (from R. Martin, 1993) (see text for details)

Phyletic evolution in muskrats 441 mated by equation (1). Further calculations of mass in this paper utilize equation (1). For comparative purposes only, if we assume constant and gradual size increase in 0 . zibethicus since the late Pliocene, the calculated value is 0.0004 g/year (1500 g/3.75 my). However, we know from other sources that muskrat size change was not gradual, but concentrated in the last 600,000 years (L. Martin, 1979, 1993; R. Martin, 1993). If we use only radiometrically dated deposits, muskrats increased from a body size of about 0.12 kg (as /minor, Hagerman LF, 3.75 ma) to about 0.23 kg (as /idahoensis, Borchers LF, 1.9 ma) and then to approximately 0.37 kg (as / annectens, Cudahy LF, 0.60 ma). Between Cudahy time and the time represented by deposits with large muskrats from the southeastern U.S., change occurred much more rapidly. At 0.016 ma (Yarbrough Cave), we see muskrats with an average body size of 1.68 kg. In the last half million years, muskrat body mass increased by 4.5 times the increase seen in the previous 3.25 million years. Figure 3 tracks change in body mass through the entire muskrat sequence, whereas Figure 4 presents size change limited to the last 0.60 my (million years) in order to illustrate better the rapid dwarfing at the end of the Pleistocene. The gigantism of southern late Pleistocene muskrats is interesting, as Nelson and Semken (1970) found no difference in absolute measurements between extant 0. zibethicus subspecies from different latitudes. Why the mean size of Wisconsinan muskrats, at least in the southeast, diminished so rapidly during the last 15,000 years is unknown, but the correlation in time with the spread of Palaeolndian in North America may be more than coincidental. This hypothesis could be tested by comparing the size distributions of muskrats taken from Indian middens and natural accumulations during the same time period.

Characters of the M1 Unlike the remainder of the dentition, the genetic field governing development of the M1 appears to allow considerable variation in geometric form. This is the typical condition in arvicolines and it is not surprising that the first lower molar is the focus of most evolutionary work with arvicoline rodents. Primitive arvicolines, with thick enamel and minimal number of triangles on M1, often display an enamel(= 'mimomyan') atoll on the anteroconid. This feature is occasionally seen in M1's from juvenile /minor, /meadensis, and (rarely) /idahoensis, but not in moderately worn molars of these chronomorphs or in samples from younger deposits. Dentine tracts are occasionally seen on the M 1 in /minor (Zakrzewski, 1969) and increase in height through time. Most published measurements (e.g., Barnosky, 1985) are provided for the labial tract of the posterior loop

442 RA. Martin 2000

y = 1165.9 - 1555.0x + 766.17x"2 - 116.00x"3 R"2 = 0.751

-s• • CII

:I > "Cl

1000

0

ID

•

••

•

• • o+-------.---.....----------r---~-----.---..... -----0

Tlme(ma)

FIGURE 3. The evolution of body mass in the North American muskrat, Ondatra zibethicus (ma= millions of years ago). Mass estimates were generated from equation (1) in the text. The late Pleistocene dwarfing event clearly noticeable on Fig. 4 is only minimally defined here because of scale.

9

8

'

e e

-.... :I

s:.

7

a,

.,

C

..I

6

5 --~-.....--.....----r--~-----.------.------.-~----..-----.- --~---. 0 .1

0 .2

0.3

0.4

0 .5

0.6

0.7

Time(ma)

FIGURE 4. Change in size of M 1 in Ondatra zibethicus during the last 0.60 million years (ma= million years ago)

Phyletic evolution in muskrats 443 of M1 (the 'hyposinuid' of Rabeder, 1981), but it would be worthwhile to investigate carefully the evolution of other tracts on M1 and other molars. Additionally, it would be interesting to examine the ratio of hyposinuid height (H) to crown height (C) because, although dentine tract height is obviously correlated with crown height, it may not track it exactly. I calculated this ratio for the Java LF sample and compared it with variation in a small (N =17) sample of Recent muskrats. I also estimated the ratio from a few illustrations in the literature. The results are therefore only preliminary. It is crucial that the measure be compared only among similar wear stages. Consequently, I set I = minimal wear, 2 = moderate wear, and 3 = heavy wear. In the sample of Recent /zibethicus I examined, H/C = 1.00 at all wear stages, although Stephens (1960) illustrated an M1 from a juvenile with H/C = 0.90 (Fig. 2) and Viriot (pers. comm.) has identified at least one juvenile specimen of 0. zibethicus with H/C less than 1.00. At wear stages 1 and 2 H/C in /meadensis is 0.29 and 0.28, respectively. The type specimen of/ idahoensis has an H/C value of 1.00 at a wear stage between 2 and 3, as does the Java / annectens. However, the Java / annectens do not present the modern configuration, as H/C ranges between 0.57 and 0.84 at wear stage 1, with one specimen at wear stage 2 with a ratio of 0.86. From illustrations published by Hibbard (1955), Cape's 0. hiatidens from Port Kennedy has H/C = 0.39 at wear stage 1/2 and 0.83 by wear stage 2. The relationship between wear and dentine tract height is illustrated in Figure 2. Perhaps the most important observation of change in hyposinuid to crown height ratio is that the H/C ratio increases at progressively earlier wear stages through time. H/ C ratios should prove to be helpful characters for correlation of sites, although it must be recognized, with this as well as with all characters, that less-advanced populations may temporarily exist simultaneously with those in a more advanced state. An increase in number of triangles isolated from the anteroconid also characterizes muskrat M1 evolution (Fig. 2). The occlusal pattern of fossil muskrats needs to be re-examined, as it also demonstrates considerable variation with wear. Unlike the condition in beavers, in which a lessderived occlusal pattern is seen in teeth from juveniles (R. Martin, 1969), a more complex pattern is seen in little-worn teeth of 0ndatra. In the sample from the Java LF of South Dakota, for example, teeth in stage 1 may have seven well-separated triangles, but as the teeth wear, the anteroconid broadens and the separation becomes minimal. Similar changes were also documented by Viriot et al. (1993) in Recent 0. zibethicus. Moderately to well-worn teeth from Java have only five distinct triangles plus an anteroconid with rudimentary sixth and seventh triangles. As can be seen in Fig. 2, the minimally worn dental pattern of the less-derived sample (e.g., /meadensis) approximates in complexity the pattern of the late

444 R.A. Martin wear stage of its descendant (e.g., / annectens). Rabeder (1981) also recognized this pattern in a number of Old World arvicolines.

Roots on M 1 The number of roots on arvicoline molars has received considerable attention since the time of Hinton (1929) and they can be important taxonomic features. However, one must be careful of ontogenetic variation. In a sample of 654 M 1's from the extant muskrat, Stephens (1960) showed that the number of roots increased with age. The predominant number of roots on M1 in first-year muskrats was two (85.5%), but this had shifted to 43.8% in three-year animals, the majority of these older animals evidencin? three roots. The earliest muskrats characteristically had three roots on M ; partial reduction in number appeared later, during the / annectens lineage segment.

Cementum in re-entrant angles Although Shotwell (1970) preferred dentine tracts, presence or absence of cement in the re-entrant angles was used by Hibbard (1959) and other workers to identify muskrat dental samples as either Ondatra or Pliopotamys. Thin, poorly developed interstitial cement was present in two M1's from the Dixon LF, tentatively dated at about 2.6 ma (Eshelman, 1975; Lundelius et al., 1987). Complete cementum is recorded in many molars from a variety of faunas dated between 1.9 and 2.5 ma (Barnosky, 1985; R. Martin, 1993). Cementum was not added instantaneously and completely in Ondatra evolution, but we lack the refinement to determine how long the process took. The maximum time for cementum development appears to have been about 0.10 my, but it could have been added much more quickly.

Enamel histology The underived, or primitive, enamel microstructure (the Schmelzmuster of von Koenigswald, 1980) for arvicoline rodents is a thick radial enamel pattern, in which enamel prisms are parallel to each other and crossed by interprismatic substance at right angles (L. Martin, 1993). In later arvicolines the prisms are twisted into other configurations, one of which is called lamellar enamel (see L. Martin [1993] for description). The chronomorphs /minor, /meadensis, and /idahoensis display only radial enamel. Lamellar enamel appears on the apices of triangles in / annectens and replaces radial enamel entirely in /zibethicus (L. Martin, 1993).

Phyletic evolution in muskrats 445

Dental microevolution: The Hagerman samples Zakrzewski (1969) provided an analysis of character change in 0. z. I minor through 83.8 m of sediments in the Glenns Ferry Formation of Idaho. The lowest unit from which a sample of 0ndatra was measured, Locality 20765, was 7.6 m above the Peters Gulch Ash, dated at 3.75 my (Lundelius et al., 1987). The highest unit from which a measured sample was listed, Locality 19216, was correlated directly with the Horse Quarry. A date of 3.48 my was associated with the Devil's Gulch Lava, approximately 45.7 m below the Horse Quarry. Extrapolating from these data, the 53.3 m between localities 20765 and 19216 represents approximately 315,000 years. Concentrating on molars considered by Zakrzewski (1969: Fig. 10) to represent adult individuals, /minor first lower molars changed from an average of 4.00 to 4.25 mm in length from the time represented by Loe. 19216 to the time represented by Loe. 20765. This represents a modification in average mass from about 103 to 128 g. If gradual, it equates to a rate of 0.000079 g/year. This difference is well within the limits of intraspecific variation in extant 0 . zibethicus, although it is slower by a factor of ten than the average value of mass increase in muskrats calculated in the previous section of this paper for the entire 3.75 my. Means of other characters, such as( 1) height of dentine tract on labial side of posterior loop of M 1, (2) occlusal width of M 1, and (3) height of enamel crown do not change in any fashion suggestive of long-term trends. The overall pattern appears to be stochastic meandering around a mean value, or stasis. However, Zakrzewski (1969) also documented a tendency for the anterior loop of M1 to be somewhat more tightly closed from T4-T5 in Mi's from Loe. 19216 than from the lower Loe. 20765. This is a pattern one would expect for a tooth adding triangles and in 0 . z. /meadensis the trend continues, with TS separated entirely from the anterior loop and T4 contiguous to, but often closed from, TS. In summary, mean length of the M 1 (and therefore body size) appears to increase through 315,000 years of sediments in the Glenns Ferry Formation. This slight increase in size is accompanied by a tendency towards a more complex M 1• Width of the first lower molar and dental features normally associated with increased hypsodonty remain static during the same period.

Heterochrony in muskrat dental evolution All structural changes in organismal evolutionary progressions represent

446 R.A. Martin modifications in epigenetic systems of one sort or another. When enough information is available about the ontogeny of ancestral and descendant species, it is sometimes possible to identify the specific developmental patterns involved. That is, we may be able to determine if the structures being studied evolved through retention of juvenile characters into adulthood (paedomorphosis) or through the extension of the juvenile period in one way or another, seen as the retention of adult ancestral features in juvenile descendants (peramorphosis) and an addition of novel or more pronounced features (sometimes including larger size) during development. Time series of fossils that demonstrate such patterns are known as heterochronoclines (McNamara, 1990) and may more specifically be either paedomorphoclines or peramorphoclines (McNamara, 1982). Rodent lineages would seem to be ideal for study of the influence of heterochronic forces and a few palaeontologists have begun to recognize developmental effects in such taxa as Mimomys (Chaline and Sevilla, 1990) and Allophaiomys and Kislangia (Agusti et al., 1993). Other lineages, including muskrats, should prove equally as interesting. For example, in the line leading to Castoroides, the North American giant beaver, adult dental features of ancestral taxa (Dipoides) are expressed in juvenile descendants (R. Martin, 1969), an example of peramorphosis. On the other hand, muskrats evidence paedomorphosis in a number of features. As can be seen in Figure 2, there is an overall increase in the number of triangles through time and the juvenile ancestral pattern resembles the late (adult) wear pattern of its descendant. In addition, muskrat molars become progressively higher crowned and manifest higher dentine tracts earlier in development in descendant morphs. This is reflected by the higher H/C ratios observed in early wear stages through time. Root development and evolution generally tracks these crown-pattern modifications. In each morph root development progresses from a basically rootless condition to one in which the roots are rather elongate. This is especially well illustrated by Pankakoski (1980: Fig. 1) for the first upper molar in / zibethicus. The high crowns of descendant morphs appear to be associated with a slowing of root development, resulting in a rootless condition maintained longer during development. Crown cementum, absent in the adults of ancestral morphs (/minor and /meadensis), appeared early during development in subsequent morphs (/idahoensis, I annectens) and became thicker as development progressed in all morphs. These ontogenetic changes follow the example provided by McNamara (1990: Fig. 3.2) of a stepped anagenetic paedomorphocline, in which descendants display more juvenile characters than their ancestors and the changes occur through time in an episodic fashion . McNamara (1990) defined three processes that can result in a pattern of paedomorphosis:

Phyletic evolution in muskrats 447 (1) neoteny, a reduction of developmental rate, (2) progenesis, an early cessation of development, and (3) postdisplacement, a delayed onset of structural growth. Because muskrat molars and body size increased through time, these processes are said to be 'global,' or to have 'global effects'; that is, all growth fields are affected to some degree (McNamara, 1990; McKinney, 1990). Without further study it is not possible to pinpoint one or more of the above processes as most dominant in muskrat dental and size evolution but, as McKinney (1990) noted, more than one process may be operating simultaneously. Peramorphosis was identified in the evolution of Kislangia first lower molars by Agusti et al. (1993), although von Koenigswald (1993) concluded that heterochronic dental processes may be more complicated than can be described by simple models of peramorphosis and paedomorphosis.

Discussion

Nomenclature of phyletic series Darwin (1859) understood that both cladogenesis and phyletic change were necessary to explain life's full diversity. It seems clear from his writings that he accepted phyletic evolution as a major avenue of morphological change. What is not clear is how he would have applied Linnaean nomenclature to this system if given a full set of fossils within a lineage. He commented (Darwin, 1859:424--425) that it would be absurd to classify a bear and a kangaroo together if it were known that a kangaroo species descended from a bear, but he also notes in the same section that living species displaying different morphs (sometimes as larval stages) are often classified in the same species. Others have since commented on this issue, including those who make cogent arguments for breaking phyletic series into constituent species (e.g., Gingerich, 1985; Rose and Bown, 1986) and those who do not view phyletic change as a speciation process, preferring instead to designate intermediate morphs with some sort of intraspecific sobriquet (Krishtalka and Stucky, 1985; R. Martin, 1993; Wiley, 1978). If a complex pattern of episodic mosaic evolution dominates in the evolution of character states, then it is likely none of the current nomenclatural systems will suffice to represent adequately all ancestor-descendant relationships, at least not at the species level and below, and particularly not in a geographically widespread species with a patchy distribution. Selander (1970) showed that mice in the same barn did not freely interbreed, living instead in distinct demes, so one can imagine the effects of environmental and topographic influences on the potential for population fragmentation and character distributions in a rodent species with almost a continental distribution. It is interesting that 0. zibethicus did not

448

R.A. Martin

speciate throughout its approximately 3.75-million-year lifespan, but, as noted in an earlier treatment (R. Martin, 1992), this may have been due to its aquatic lifestyle. In any case, as Table 1 indicates, assuming that the chronology is even roughly valid, there appears to be some temporal overlap in character states, leading one to believe that evolution did not proceed at the same rate for all characters in all populations. Krishtalka and Stucky (1985) and R. Martin (1993) presented a system in which intermediate populations in a phyletic series were identified by informal taxonomic identifiers known either as either 'lineage segments' (Krishtalka and Stucky, 1985) or 'chronomorphs' (R. Martin, 1993). R. Martin (1993) preferred the chronomorph designation because he recognized the mosaic nature of character expression and expected there to be temporal overlap in populations displaying these features. These concepts need to be further explored and more carefully defined in order to avoid areas of potential confusion. For instance, it must be recognized that chronomorphs, or lineage segments, are populations defined on the basis of their characters and are therefore only recognizable to the extent that the character suite used to define them remains intact. In the above classification I listed a series of character states to define five chronomorphs. But we can legitimately expect those characters to vary somewhat independently of one another and it will not be surprising at all to find other samples that do not exactly conform with one of the five identified populations. For example, Barnosky (1985) reported a sample of Pliopotamys from the Blancan Boyle Ditch LF of Wyoming. The population expressed a suite of features that mostly identified it as P. meadensis, but the molars also displayed thin, intermittent cement, a feature normally associated with more advanced populations collectively known as 0. idahoensis. Barnosky wisely referred the Boyle Ditch sample to 'Pliopotamys near meadensis,' noting its intermediate morphology between P. meadensis and 0 . idahoensis. This sample represents an interesting problem. It does not exactly fit into one of the five chronomorphs as defined above, so how should it be treated? There are four possibilities: (1) as a new species, (2) as a new chronomorph, (3) as a member of an existing chronomorph, or (4) it, along with all the intermediate samples in the muskrat phyletic sequence, should bear no names; no formal or informal identifier of any sort (that is, all named intermediates in phyletic sequences should be abandoned). My objection to (1) has been discussed in detail elsewhere (R. Martin, in press). Phyletic change can be viewed as an intraspecific evolutionary process that may, in fact, result in considerable morphological change, but not to the extent of the bear /kangaroo scenario proposed hypothetically by Darwin. Evolution does not appear to progress in this extreme fashion. Recent studies and summaries of morphological change in the mammalian fossil record (e.g., Avery, 1982; Barnosky, 1990, 1993;

Phyletic evolution in muskrats 449 Krishtalka and Stucky, 1985; R. Martin, 1993; R. Martin and Barnosky, 1993; R. Martin and Prince, 1990; Nadachowski, 1991; Rose and Bown, 1986) suggest that microevolution is the stuff also of large-scale morphological change and that, at least in arvicolines, a complex mosaic pattern of episodic phyletic evolution characterizes morphological change in many lineages. I also reject option (2) and favour option (3). There is no a priori reason why a new chronomorph should not be recognized, but we should be careful that such new entities contribute significantly to our knowledge of the taxon before adding another name to the literature. It is for this reason that each chronomorph should be defined on as many characters as possible. The Boyle Ditch LF muskrat population may indeed represent a temporally as well as morphologically intermediate population between /meadensis and /idahoensis, but it may also represent a population that happens to have cementum in the re-entrant angles existing contemporaneous with other /meadensis populations. Davis (1987) showed that there was a progression of dental complexity in Microtus pennsylvanicus Mi's from those with five to those with six triangles in late Pleistocene faunas from the Great Plains, while eastern populations today retain a preponderance of five triangle M/s. R. Martin (1993) estimated that this process took about 0.25 ma, which in a sense represents the kind of temporal overlap one might expect in chronomorphs. It would not be justified to place Boyle Ditch at a later date than other faunas with / meadensis unless independently corroborated by additional evidence (e.g., absolute dates, palaeomagnetic data, or morphological change in other lineages). Since the majority of features of the Boyle Ditch LF sample indicate alliance with /meadensis, then it seems logical to refer it to that chronomorph. Alternatively, if the Boyle Ditch sample is shown to be chronologically intermediate to faunas with / meadensis and / idahoensis, there may still be little practical or philosophical basis for naming it as a new chronomorph, since only one feature has changed to bridge the gap between the two informal taxa. There is a compelling argument that can be developed to eliminate all names in a phyletic sequence, as indicated by option (4), perhaps representing the variation of intermediate samples with complex pie diagrams or some other quantitative graphical expression. Personally, I would have little objection to this outcome, but it would create some concern on the part of biostratigraphers using fossil taxa to age rock units when more refined dating techniques are inapplicable. It is also for this reason that many palaeontologists will likely be hesitant to embrace chronomorph or lineage segment nomenclature; it is admittedly more cumbersome than naming intermediate populations as distinct species. Nevertheless, we have reached a point where the fossil record is dense enough to address classical

450 R.A. Martin questions in evolutionary biology and it would be helpful if the nomenclature reflected biological reality more than it does at the present time.

Character change and palaeoecology in muskrats The muskrat fossil record offers tremendous potential for investigators concerned with both the tempo and mode of character variation in a lengthy phyletic sequence with a good representation of fossil materials. Pankakoski (1980, 1986) and Viriot et al. (1993) have applied modern statistical and digital imaging techniques to examining ontogenetic changes in muskrat skull and dental characters. Boyce (1978), Galbreath (1954), Stephens (1960), and Pankakoski (1986) have also provided baseline information on individual variation in molar morphology and body size as well as other characters. While we cannot legitimately use variation in living 0 . zibethicus to determine the taxonomic status of intermediates in the muskrat phyletic series (see R. Martin, 1993:228-234), we can use this information to determine the extent and therefore the significance of morphological change in the lineage. Zakrzewski's work (1969) reflects the limits of morphological change on the 300-ka (thousand year) scale and we see during this period a classic mosaic response, with slight changes in M1 size and a tendency towards closure of the anterior loop, but stasis in other features. Changes of this magnitude have also been recorded in Microtus pennsylvanicus M1's over a period of a few thousand years during the latest Pleistocene (Barnosky, 1990, 1993; R. Martin and Prince, 1990). These modifications would be somewhat suggestive of overall stasis were it not for additional data from longer periods of time in a variety of arvicoline lineages, including muskrats. R. Martin (1993) summarized information from the muskrat sequence and integrated it with reports on morphological change in other rodent lineages and clades. Phyletic change is common in arvicolines and has been recognized in various lineages of the genera Mimomys and Arvicola (Chaline and Laurin, 1986; Chaline and Sevilla, 1990; Heinrich, 1987; Stuart, 1982) as well as in the North American lineage leading from M. pliocaenicus to M. ochrogaster (R. Martin, 1995). Lineages are often characterized by similar dental modifications; increased size, increased number of triangles, increased hypsodonty, addition of crown cementum, loss of enamel atolls, and loss of roots. R. Martin and Tedesco (1976) also provided evidence in support of a mechanism for molar shape change in the muskrat sequence. In early stages, lower-first-molar length approximately doubled for every increase in width, but as the molars, as well as the animals, grew larger, length increased more rapidly, resulting in a longer, narrower M1 in the latest members of the species and an increase in length/ width ratios through

Phyletic evolution in muskrats 451 2.8

=

y 2.5029 • 0.12912x R'2

=0.550

2.6

e§. ~

•• :E C

2.4

2.2

2.0

• 1.8

0

2

3

Tlme(ma)

FIGURE 5. Mean length/width ratios of the first lower molar in Ondatra zibethicus (all chronomorphs) plotted as a function of time (data from Table 1)

time (Fig. 5). Similar trends in geographically disjunct, evolving arvicoline sequences attest to the limited epigenetic pathways available for dental change and perhaps morphological change in general. They also attest to the fact that significant morphological change can come about through phyletic evolution, represented not only by changes in dental anatomy, but also by extraordinary change in body size. Changes in muskrat body size can be viewed as a form of complex stairstep evolution, with only modest increase during much of late Pliocene through middle Pleistocene time, followed by a rapid 'punctuation' in size leading to the largest muskrats during late Wisconsinan time. A rapid decrease in size appears to have followed leading to Recent muskrats, the agent of which is unknown (larger samples of extant southeastern muskrats need to be measured to further corroborate and refine the dynamics of this change). The earliest muskrats (/minor) were about the size of the living water vole, Arvicola terrestris, perhaps averaging a bit over 100 g. Recent individuals are larger in mass by a factor of ten. As indicated by Martin (1993: table 2), this body-size difference results in significant shifts in almost any biological parameter one cares to examine. These changes, in home-range size, population density, metabolic rate, and so on are at least as critical as the dental modifications that accompanied overall size change. The habitat specificity of early muskrats remains a mystery, although Zakrzewski (1969) presented circumstantial evidence favouring an

452

R.A. Martin

aquatic lifestyle. He was impressed with the replacement patterns of arvicolines through the Glenns Ferry Formation and suggested that the demise of Cosomys primus might have been due to competition with 0. z. /minor and that both species may have been aquatic. Their larger size, seen in extant arvicolines only in aquatic taxa (Arvicola, Neofiber, Microtus richardsoni), indirectly supports this hypothesis. Perhaps one of the more amazing and enlightening aspects of this study was the recognition that despite a mosaic character response on scales up to 300,000 years (the Hagerman samples) and despite the indication from Table 1 of chronomorph overlap (that is, some overlap of underived and derived character states), on a scale of millions of years, the North American muskrat evolved in a reasonably consistent manner that allows use of intermediate populations (chronomorphs) and their character complexes for biostratigraphic purposes. It is almost as if two evolutionary processes were contending simultaneously: one that over the short term tracks local environmental events and responds in a mosaic fashion and another that works on a longer scale and acts to maintain species genetic cohesion. While stasis and minimal shape changes may characterize molluscan lineages (Stanley and Yang, 1987), at least arvicoline rodent species appear to change continuously throughout their life-spans, albeit at different rates. In conclusion, the muskrat fossil record is probably one of the best-documented phyletic series available for study. Samples have been examined, at least in a preliminary fashion, at all temporal scales and both stasis and episodic (stairstep) change can be identified depending on the scale examined. Muskrats increased in size at a very modest rate for millions of years and then increased rapidly to their present size in the last 600 ka. Dental modifications were only loosely related to size change, as cementum was added, for example, well before the final burst towards larger size. Occlusal pattern and some attendant character changes display paedomorphosis, identifying developmental heterochrony as an important evolutionary influence. It can certainly be debated whether or not changes in molar anatomy were 'significant,' as it is difficult to determine their effects on fitness, but the changes extend beyond variation in Recent populations. Overall size change was so dramatic that it appears clear the beginning and end products in muskrat evolution were somewhat different biological entities.

An afterword: Phyletic evolution and character change Recent studies of rodents, pampatheriid armadillos, moose, and mammoths (Goodwin, 1993; Hulbert and Morgan, 1993; Lister, 1993; R. Martin, 1993) identified phyletic evolution as an agent of considerable morphological change. But is phyletic change, even in its more complex manifes-

Phyletic evolution in muskrats 453 tations as a mosaic character response, a satisfactory mode for all shape changes? What, if anything, is the contribution of speciation to the process of morphological change? At least with mammals, this question has yet to be carefully addressed, in part because of a dearth of appropriate case studies. Many of us familiar with living mammals can point to generic clades (e.g., Peromyscus, Microtus, Sigmodon, or Reithrodontomys, noting that the constituent species appear to be minor variations on the same theme (as I have done, e.g., R. Martin, 1986, 1993). They demonstrate no consequential character differences, but the fact is that they do differ slightly in morphology, be it among hard or soft parts, and there are no carefully designed studies comparing the extent of these differences to those among fossil representatives of the same groups. We are beginning to accumulate an inventory of Quaternary mammalian case histories dominated by phyletic evolution and I would like to suggest that these sorts of comparisons may prove most helpful in identifying the primary evolutionary agents of morphological change.

Acknowledgments I greatly appreciate the loans of fossil and Recent Ondatra material arranged by M. McGhee, P. Gingerich, and G. Gunnell. G. Morgan of the Florida State Museum kindly provided unpublished data for Florida Irvingtonian muskrats. Gifts of Pliopotamys material and correspondence from C. Repenning also proved of value. I thank H. Semken, L. Viriot, and an anonymous reviewer for helpful comments. LITERATURE CITED

Agusti, J., C. Castillo, and A. Galobart. 1993. Heterochronic evolution in the late Pliocene-early Pleistocene arvicolids of the Mediterranean area. Quaternary International 19:51-56. Avery, D.M. 1982. Micromamrnals as palaeoenvironmental indicators and an interpretation of the late Quaternary in the southern Cape Province, South Africa. Annals of the South African Museum 85:183-374. Barnosky, A.D. 1985. Late Blancan (Pliocene) microtine rodents from Jackson Hole, Wyoming: biostratigraphy and biogeography. Journal of Vertebrate Paleontology 5:255-271. - 1987. Punctuated equilibrium and phyletic gradualism: some facts from the Quaternary mammalian record. Current Mammalogy 1:109-144. - 1990. Evolution of dental traits since latest Pleistocene in meadow voles (Microtus pennsylvanicus) from Virginia. Paleobiology 16:370-383. - 1993. Mosaic evolution at the population level in Microtus pennsylvanicus.

454 RA. Martin Pp. 24-59 in R.A. Martin and A.O. Barnosky (eds), Morphological Change in Quaternary Mammals of North America. Cambridge University Press, Cambridge. Boyce, M.S. 1978. Climatic variability and body size variation in the muskrats (Ondatra zibethicus) of North America. Oecologia 36:1-19. Chaline, J., and B. Laurin. 1986. Phyletic gradualism in a European PlioPleistocene Mimomys lineage. Paleobiology 12:203-216. Chaline, J., and P. Sevilla. 1990. Phyletic gradualism and developmental heterochronies in a European Plio-Pleistocene Mimomys lineage. Pp. 85-98 in 0. Fejfar and W.-O. Heinrich (eds), International Symposium on Evolution, Phylogeny and Biostratigraphy of Arvicolids. Geological Survey, Prague, Czechoslovakia. Cope, E.D. 1871. Preliminary report on the Vertebrata discovered in the Port Kennedy Bone Cave. Proceedings of the American Philosophical Society 12:73102. Darwin, C. 1859. On the Origin of Species. (A Facsimile of the First Edition, 1964.) Harvard University Press, Cambridge, 513 pp. Davis, L.C. 1987. Late Pleistocene/Holocene environmental changes in the Central Great Plains of the United States: the mammalian record. Pp. 88-143 in R.W. Graham, H.A. Semken, Jr, and M.A. Graham (eds), Late Quaternary Mammalian Biogeography and Environments of the Great Plains and Prairies. Illinois State Museum Scientific Papers 22. Eshelman, R.E. 1975. Geology and palaeontology of the early Pleistocene (Late Blancan) White Rock fauna from north-central Kansas. University of Michigan Museum of Paleontology, Papers on Paleontology 13 (C.W. Hibbard Memorial Volume 4): 1-60. Galbreath, E.C. 1954. Growth and development of teeth in the muskrat. Transactions of the Kansas Academy of Sciences 57:238-241. Gingerich, P.O. 1985. Species in the fossil record: concepts, trends and transitions. Paleobiology 11 :27-41. Goodwin, H.T. 1993. Patterns of dental variation and evolution in prairie dogs, genus Cynomys. Pp. 107-133 in R.A. Martin and A.O. Barnosky (eds), Morphological Change in Quaternary Mammals of North America. Cambridge University Press, Cambridge. Heinrich, W.-D. 1987. Some aspects of the evolution and biostratigraphy of Arvicola (Mammalia, Rodentia) in the central European Pleistocene. Pp. 165-182 in 0. Fejfar and W.-D. Heinrich (eds), International Symposium on Evolution, Phylogeny and Biostratigraphy of Arvicolids. Geological Survey, Prague, Czechoslovakia. Hibbard, C.W. 1937. An upper Pliocene fauna from Meade County, Kansas. Transactions of the Kansas Academy of Science 40:239-265. - 1955. Notes on the microtine rodents of the Port Kennedy Cave deposit. Proceedings of the Academy of Natural Sciences Philadelphia 107:87-97. - 1959. Late Cenozoic microtine rodents from Wyoming and Idaho. Papers of the Michigan Academy of Science, Arts and Letters 44:3-40.

Phyletic evolution in muskrats

455

Hibbard, C.W., and W.W. Dalquest. 1966. Fossils from the Seymour Formation of Knox and Baylor Counties, Texas, and their bearing on the late Kansan climate of that region. University of Michigan Museum of Paleontology Contribution 21 :1--{i6.

Hibbard, C.W., R.J. Zakrzewski, R.E. Eshelman, G. Edmund, C.D. Griggs, and C. Griggs. 1978. Mammals from the Kanopolis local fauna, Pleistocene (Yarmouth) of Ellsworth County, Kansas. University of Michigan Museum of Paleontology Contribution 25:11-44. Hinton, M.A.C. 1929. Monograph of the Voles and Lemmings (Microtinae) Living and Extinct, Volume 1. British Museum (Natural History), London, 488 pp. Hollister, N. 1911. A systematic synopsis of the muskrats. North American Fauna 32:1-47.

Hulbert, R.C., Jr, and G.S. Morgan. 1993. Quantitative and qualitative evolution in the giant armadillo Holmesina (Edentata: Pampatheriidae) in Florida. Pp. 134177 in R.A. Martin and A.D. Barnosky (eds), Morphological Change in Quaternary Mammals of North America. Cambridge University Press, Cambridge. Koenigswald, W. von. 1980. Schmelzstruktur und morphologie in den molaren der Arvicolidae (Rodentia). Abhandlungen der Senkenbergishen Naturforschenden Gesellschaft 539:1-129. - 1993. Heterochronies in morphology and schmelzmuster of hypsodont molars in the Muroidea (Rodentia). Quaternary International 19:57--{il. Krishtalka, L., and R.K. Stucky. 1985. Revision of the Wind River faunas, early Eocene of central Wyoming, Part 7. Revision of Diacodexis (Mammalia, Artiodactyla). Annals of the Carnegie Museum of Natural History 54:413-486. Lister, A.M. 1993. Evolution of mammoths and moose: the Holarctic perspective. Pp. 178-204 in R.A. Martin and A.D. Barnosky (eds), Morphological Change in Quaternary Mammals of North America. Cambridge University Press, Cambridge. Lundelius, E.L. Jr, C.S. Churcher, T. Downs, C.R. Harington, E.L. Lindsay, G.E. Schultz, H.A. Semken, S.D. Webb, and R.J. Zakrzewski. 1987. The North American Quaternary sequence. Pp. 211-235 in M.O. Woodburne (ed.), Cenozoic Mammals of North America. University of California Press, Berkeley. McKinney, M.L. 1990. Trends in body-size evolution. Pp. 75-118 in K.J. McNamara (ed.), Evolutionary Trends. University of Arizona Press, Tucson. McNamara, M.J. 1982. Heterochrony and phylogenetic trends. Paleobiology 82:130-142. - 1990. The role of heterochrony in evolutionary trends. Pp. 59-74 in K.J.

McNamara (ed.), Evolutionary Trends. University of Arizona Press, Tucson. Martin, L.D. 1979. The biostratigraphy of arvicoline rodents in North America. Transactions of the Nebraska Academy of Sciences 7:91-100. - 1993. Evolution of hypsodonty and enamel structure in Plio-Pleistocene rodents. Pp. 205-225 in R.A. Martin and A.D. Barnosky (eds), Morphological Change in Quaternary Mammals of North America. Cambridge University Press, Cambridge.

456

R.A. Martin

Martin, R.A. 1969. Taxonomy of the giant Pleistocene beaver Castoroides from Aorida. Journal of Paleontology 43:2033-2041. - 1984. The evolution of cotton rat body mass. Pp. 179-183 in H.H. Genoways and M.R. Dawson (eds), Contributions in Quaternary Vertebrate Paleontology: A Volume in Memorial to John E. Guilday. Carnegie Museum of Natural History, Special Publication 8. - 1986. Energy, ecology and cotton rat evolution. Paleobiology 12:370-382. - 1987. Notes on the classification and evolution of some North American fossil Microtus. Journal Vertebrate Paleontology 7:270-283. - 1989. Arvicolid rodents of the early Pleistocene Java local fauna from north-central South Dakota. Journal of Vertebrate Paleontology 9:438-450. - 1992. Generic species richness and body mass in North American mammals: support for the inverse relationship of body size and speciation rate. Historical Biology 6:73-90. - 1993. Patterns of variation and speciation in Quaternary rodents. Pp. 226-280 in R.A. Martin and A.D. Barnosky (eds), Morphological Change in Quaternary Mammals of North America. Cambridge University Press, Cambridge. - 1995. A new middle Pleistocene species of Microtus (Pedomys) from the southern United States, with comments on the taxonomy and early evolution of Pedomys and Pitymys in North America. Journal of Vertebrate Paleontology 15:171-186. Martin, R.A., and A.D. Barnosky. 1993. Quaternary mammals and evolutionary theory: Introductory remarks and historical perspective. Pp. 1-12 in R.A. Martin and A.D. Barnosky (eds), Morphological Change in Quaternary Mammals of North America. Cambridge University Press, Cambridge. Martin, R.A., and R.H. Prince. 1990. Variation and evolutionary trends in the dentition of Microtus pennsylvanicus from three levels in Bell Cave, Alabama. Historical Biology 4:117-129. Martin, R.A., and R. Tedesco. 1976. Ondatra annectens (Mammalia, Rodentia) from the Java local fauna of South Dakota. Journal of Paleontology 50:846-850. Nadachowski, A. 1991. Systematics, geographic variation, and evolution of snow voles (Chionomys) based on dental characters. Acta Theriologa 36:1-45. Nelson, R.S., and H.A. Semken, Jr. 1970. Paleoecological and stratigraphic significance of the muskrat in Pleistocene deposits. Bulletin of the Geological Society of America 81:3733-3738. Pankakoski, E. 1980. An improved method for age determination in the muskrat, Ondatra zibethica (L.). Acta Zoologica Fennica 17:113-121. - 1986. Skull morphology of Finnish muskrats: geographic variation, age differences and sexual dimorphism. Annals Zoologi Fennici 23:1-32. Rabeder, G. 1981. Die arvicoliden (Rodentia, Mammalia) aus dern Pliozan und dern alteren Pleistozan von Niederosterreich. Beitrage Palaontologie und Geologie Oesterreich 8:1-373. Repenning, C.A. 1987. Biochronology of the microtine rodents of the United

Phyletic evolution in muskrats 457 States. Pp. 236-268 in M.O. Woodburne (ed.), Cenozoic Mammals of North America. University of California Press, Berkeley. Rose, K.D., and T.M. Bown. 1986. Gradual evolution and species discrimination in the fossil record. University of Wyoming Contributions to Geology, Special Papers 3:119-130. Selander, R.K. 1970. Behavior and genetic variation in natural populations. American Zoologist 10:53-66. Semken, H.A., Jr. 1966. Stratigraphy and paleontology of the McPherson Equus beds (Sandhal local fauna), McPherson County, Kansas. University of Michigan Museum of Paleontology Contributions 20:121-178. Shotwell, J.A. 1970. Pliocene mammals of southeast Oregon and adjacent Idaho. University of Oregon Museum of Natural History Bulletin 17:1-103. Stanley, S.M., and X. Yang. 1987. Approximate evolutionary stasis for bivalve morphology over millions of years: a multivariate, multilineage study. Paleobiology 13:13-26. Starrett, A. 1956. Pleistocene mammals of the Berends fauna of Oklahoma. Journal of Paleontology 30:1187-1192. Stephens, J.J. 1960. Stratigraphy and paleontology of a late Pleistocene basin, Harper County, Oklahoma. Bulletin of the Geological Society of America 71:1675-1702. Stuart, A.J. 1982. Pleistocene Vertebrates in the British Isles. Longman, London, 212 pp. Tomida, Y. 1987. Small Mammal Fossils and Correlation of Continental Deposits, Safford and Duncan Basins, USA. National Science Museum, Tokyo, 141 pp. van der Meulen, A.J. 1978. Microtus and Pitymys from Cumberland Cave, Maryland, with a comparison of some New and Old World species. Annals of the Carnegie Museum of Natural History 47:101-145. Viriot, L., J. Chaline, A. Schaaf, and E. LeBoulenge. 1993. Ontogenetic change of Ondatra zibethicus (Arvicolidae, Rodentia) cheek teeth analyzed by digital image processing. Pp. 373-391 in R.A. Martin and A.D. Barnosky (eds), Morphological Change in Quaternary Mammals of North America. Cambridge University Press, Cambridge. Wiley, R.H. 1978. The evolutionary species concept reconsidered. Systematic Zoology 27:17-26. Wilson, R.W. 1933. A rodent fauna from later Cenozoic beds of southwestern Idaho. Carnegie Institution of Washington Publication 440:117-135. Zakrzewski, R.J. 1969. The rodents from the Hagerman local fauna, upper Pliocene of Idaho. University of Michigan, Museum of Paleontology Contributions 23:1-36.

Early Rancholabrean mammals from Salamander Cave, Black Hills, South Dakota

Jim I. Mead, Carol Manganaro, Charles A. Repenning, Larry D. Agenbroad

Abstract Salamander Cave (Wind Cave National Park, Black Hills, South Dakota) is a small cavern with a natural trap entrance. The Horse Room contains a small fauna produced by the infilling from a now-sealed entrance. Uranium-series analysis of speleothems and horse bone indicate that the Horse Room is recording a local faunal community approximately 252,000 years old. Sixteen taxa are recognized. Extinct taxa include the rodents Mictomys cf. M. meltoni, Microtus paroperarius, and Terricola meadensis, along with Canis cf. C. dirus, Equus spp., and Camelops sp. Extralimital species include Cynomys (Leucocrossuromys) sp. and Lepus cf. L. americanus. The fauna may mark the youngest co-occurrence of the extinct rodent species. The Black Hills offer a desirable location to examine evolutionary changes and speciation following immigration because of their unique 'middle ground' location in North America.

Introduction In the United States the Pleistocene has been divided into two landmammal ages, the Irvingtonian (older) and the Rancholabrean (younger). The Rancholabrean is certainly one of the better known and represented land-mammal ages in North America; however, it is far from being understood in its entirety. The beginning of the Rancholabrean is defined, in part, on the first appearance of Bison (bison), which has not been constrained by precise isotopic dating. Lundelius et al. (1987) estimate the beginning of the Rancholabrean, and therefore the end of the Irvingtonian, in a range between 200,000 and 550,000 yr BP. Most of what is

Early Rancholabrean mammals in Black Hills 459 known about the Rancholabrean is from the late Rancholabrean, notably from the last (i.e., Wisconsinan) glacial episode (approximately 80,00011,000 yr BP). The faunas dating from the various glacial and interglacials before the Wisconsinan, but still within the Rancholabrean are still inadequately known. Compounding the problem is the fact that at present there are few isotopic dating techniques that produce accurate ages between 100,000 and 500,000 yr BP. Adding to this problem, few species are diagnostic of the Rancholabrean (other than Bison). One of the more recently published and better-known Rocky Mountain faunas dating to the middle Pleistocene is Porcupine Cave, in the Rocky Mountains of Colorado (Bamosky and Rasmussen, 1988; Wood and Barnosky, 1994). Levels 1 to 5 have palaeomagnetic data indicating ages between 487,000 and 365,000 years old, and certainly younger than 750,000 yr BP (Barnosky and Rasmussen, 1988; Barnosky et al., this volume). Here we present a previously unpublished vertebrate fauna dating to the middle-later Pleistocene, from Salamander Cave, Black Hills, South Dakota. The Black Hills today consist of an isolated mountainous island rising out of the surrounding vastness of the Great Plains in western South Dakota and eastern Wyoming. The area is relatively small, extending approximately 200 km in a northwest-southeast direction, with its widest point being roughly 100 km. The nearest mountain ranges to the Black Hills are the Laramie and Big Horn mountains, approximately 250 km to the southwest and west, respectively, in Wyoming. Isolation as a mountainous area and size serve to make the Black Hills a unique region for palaeontological studies, but few such studies have been conducted. Today the Black Hills are positioned as a floral and fauna} ecotone between eastern and western communities. Faunal descriptions and palaeoclimatic reconstructions have been analysed for the Holocene material recovered from Beaver Creek Shelter, an archaeological site in Wind Cave National Park (WICA) (Abbott, 1989; Benton, 1991; Martin et al., 1993). Late Pleistocene faunal studies not associated with cultural remains are rare for the western portion of the Great Plains, and usually date to the Wisconsinan Glacial (Agenbroad et al., 1990; Graham and Mead, 1987; Graham et al., 1987). The Rancholabrean mammalian record of the Black Hills is not well known in comparison to that of the eastern portion of South Dakota and the region surrounding the hills (Kurten and Anderson, 1980; Lundelius et al., 1983). The Mammoth Site (near the town of Hot Springs) is the only previously described late Rancholabrean-age locality in the Black Hills (Agenbroad, 1984). The Black Hills do hold great promise for the recovery of Rancholabrean and even lrvingtonian faunas because of the numerous caves. Caves in the Black Hills have perhaps the most complex evolutionary history of any caves in

460

J.I. Mead, C. Manganaro, C.A. Repenning, L.D. Agenbroad

the world (Palmer and Palmer, 1989). Many of them are multi-level networks of passages in the Pahasapa Limestone of Mississippian age. The geologic structure and the geomorphic setting of these caves are presented in Palmer and Palmer (1989). Although the geology and speleology of the numerous caves in the Black Hills are fairly well known, the Pleistocene palaeontology within the passageways is just now being recognized. This report will concentrate on one cave found within the boundary of Wind Cave National Park (WICA) in the southern Black Hills.

Salamander Cave Salamander Cave is located at 1365 m elevation near the northern boundary of WICA. The small cave is a solution cavity in the Pahasapa Limestone, which was presumably formed in the manner described by Palmer and Palmer (1989) and Ford (1989) for the other caves in the region. Present entrance into the cave is along the top of a ridge of partially exposed limestone. The immediate area is a series of small, more-or-less parallel ridges and valleys. Most of the ridges have minimal soil development, with the adjacent valleys containing some fairly deep soil and alluvial accumulations. The ridges and valleys today are covered by open ponderosa pine (Pinus ponderosa) forest. Short-grass prairie occurs within approximately 1 to 2 km, in the valleys below the series of hills and ridges. The existing opening to Salamander Cave is rectangular in shape, measuring approximately 1.6 by 0.75 m, forming a natural trap with a drop of 6.0 m to the top of an active talus cone. The cave contains several rooms, of which only two (Entrance and Porcupine rooms) are being infilled via the present entrance. Some of the more common species being entrapped today include Ambystoma tigrinum (tiger salamander; hence the name of the cave), Peromyscus spp. (deer mice), and Erethizon dorsatum (porcupine). The Porcupine Room is being actively infilled on one side via the present entrance. However, on the opposite side of the room there is an older, more consolidated and cemented talus cone that used to be infilling the room. Here there are loose fragments of Equus (not to be discussed here). From these first two rooms a small corridor leads down into the back chamber, named the Horse Room (Fig. 1). The infilling pattern in the Horse Room indicates that an entrance distinct from the present entrance provided access into this portion of the cave. Mapping of the cave indicates that the Horse Room and its presumed palaeo-entrance is now below the valley, adjacent to the ridge containing the present pit-fall entrance. A miner's pit within the Horse Room dated AD 1912 permitted access to subfloor deposits. The miner chopped through a thin travertine flowstone layer that capped most of the talus cone filling the room. Below the

Early Rancholabrean mammals in Black Hills 461 SALAMANDER CAVE

A

Schematic Profile

Horse Room

Notto Scale

B Porcupine Room

% %

~ ~

~

~

~-

~

~

~

iii

~

~

~

"I

~

~ ~

% %

% % %

%

"" %

~

~

"I "I

. ~ J?t~' 1i'

Horse Room

r,.

1i'

Test Pit

Entrance Room

Plan View To Scale

FIGURE 1. Schematic profile (A) and plan view (B) of Salamander Cave, Wind Cave National Park, Black Hills, South Dakota

462

J.I. Mead, C. Manganaro, C.A. Repenning, L.D. Agenbroad

travertine cap is a mud, cobble, and bone unit. It is possible that some faunal remains exposed by the miner's pit were removed. In 1991 the Quaternary Studies Program (QSP), Northern Arizona University (NAU), engaged in an agreement with WICA to conduct preliminary investigations and palaeontological assessment of Salamander Cave (NAU QSP locality 9148; WICA Accession 136). This is the report of the fossils recovered from the deposits in the Horse Room. In 1991 the Horse Room was visited and a test pit measuring 1 m wide, 2 m long, and up to 1 m deep was excavated incorporating the miner's pit. Follow-up visits to the test pit were conducted to remove an additional 1/4 m 2 of material for fine screening and sampling for pollen. Further discussion in this report of the faunal remains from Salamander Cave refer to only those specimens from the Horse Room. The stratigraphy appears simple, based on the profile exposed by our test excavations (Fig. 2). The floor of the Horse Room is covered by a thin travertine layer with 'popcorn' speleothems forming the surface (Layer 1). This capping unit covers a lower unit containing mud, cobble-sized angular roof spall, and faunal remains (Layer 2). The faunal remains are mixed within the mud and cobbles, indicating that they are part of the adjacent talus cone. The larger boulders of the talus cone provided voids which were filled by the jumbled mass of mud and bones. Below the mud-faunal unit is Layer 3, another layer of botryoidal popcorn speleothems, apparently deposited directly on Pahasapa limestone bedrock.

Chronology Three samples from the test-pit excavations were submitted for uraniumseries analyses. All analyses were conducted by Richard Ku, Department of Geological Sciences, University of Southern California. Popcorn speleothems from Layers 1 and 3 were processed for dates. Activity ratio of 23°Th/ 234U for the sample from Layer 1 (1.433 ± 0.044) is in excess of the secular equilibrium value of 1.00, and therefore probably indicates detrital contamination. No 23°Th or 231 Pa ages are calculated for Layer 1. The popcorn travertine sample from Layer 3 contains the following data: 238 U dpm/g 1.28 ± 0.03; 234U/ 238U 1.101 ± 0.033; and 23°Th/ 234U 0.977 ± 0.031. These data imply an estimated 23°Th age of 323,000 ± 64,000 yr BP. A first phalanx of Equus, recovered from the bone-bearing unit (Layer 2), was used for uranium-series analysis. An age of 252,000 ± 30,000 yr BP was determined on the 23°Th/ 231 Pa ratio. The bone sample had a high U concentration. The excess 23°Th and 231 Pa indicate that the sample had not acted as a closed system with respect to U /Th/Pa. It seems likely that preferential loss of U over its Th and Pa daughters might have occurred; hence, no 23°Th/ 234U and 231 Pa/ 235U ages could be computed.

Stratigraphic Profile

Present surface

200

Layer

Analysis

{''°Thi

l

2'4U

Age

1 433 ± 0.044 } No

2 30-fh/or 231 Pa

''°Thl"'Pa

Age Estimate

} 252,000± 30,000

222

a:

259

~

>~ 300

Conservative

·200,000 years. The possibilities and similarity in values between the 23°T / 234U (1.379 ± 0.011) and 231 Pa/ 235U ratios (1.390 ± 0.052) in the bone suggest that the occurrence of U loss might be fairly recent. If so, the 234Th/ 231 Pa ratio should be relatively insensitive to the U loss and a dosed-system age can be calculated from this ratio. The calculated age of 252,000 yr BP should be meaningful. The two dates from Layers 2 and 3 are in stratigraphic accordance. A conservative view (used here by Mead, Manganaro, and Agenbroad) based upon the radiometric dates and the standard deviations would imply that the bones in Layer 2 began deposition sometime between 282,000 and 259,000 years ago. The bones are probably no older than 323,000 ± 64,000 yr BP and no younger than 252,000 ± 30,000 yr BP (Fig. 2). Utilizing two standard deviations of the age estimate of Layer 3 sample would imply that the time for initial deposition of the bone could have been as early as 451,000 yr BP. Repenning (here) believes, as will be discussed below, that the various arvicoline species from the Horse Room fauna are like those of the Cudahy fauna and show an evolutionary stage approximately 600,000 ± 100,000 years old.

Fauna The fauna from the Horse Room is listed in Table 1. Repenning identified the arvicolines; Mead, Manganaro, and Agenbroad produced the rest of the identifications. Skeletal element descriptions are followed by parentheses containing the number of identified specimens (NISP), if more than one, and the National Park Service specimen number recorded in the collections of WICA. All specimens are currently reposited in the Laboratory of Quaternary Paleontology, Quaternary Studies Program, Northern Arizona University, Flagstaff. Left is 'L'; Right is 'R.' Superscript numbers refer to upper dentition; subscript numbers refer to mandibular dentition. Lagomorpha Leporidae - Rabbits and hares

Lepus sp. - jackrabbit Specimens: Premaxilla (2819), LP 2 (2820), LF4-M 2 (single tooth; 2822), LP3 (2818), RP4-M2 (single tooth; 2823), RM 3 (2817), R femur proximal half (2113), R femur distal half (2; 2439), R tibia proximal fragment (2420, 2440), L tibia diaphysis (2442), calcaneum (2484) Identifications and comparisons: All the leporid skeletal elements are larger than those of most living species of Sylvilagus, and therefore are

Early Rancholabrean mammals in Black Hills 465 TABLE 1 The fauna from the Horse Room, Salamander Cave, Black Hills, South Dakota (+ = extinct taxon; ¥ = extralimital taxon today) Mammalia Lagomorpha Lepus sp. Lepus cf. L. americanus ¥ Rodentia Cynomys (Leucocrossuromys) sp.¥ Cynomys sp. Mictomys meltoni/kansasensis + Microtus paroperarius + Neotoma sp. Peromyscus sp. Spermophilus (Spermophilus) sp. Spermophilus sp. Terricola meadensis + Camivora Mustela sp. Canis cf. C. dirus + Perissodactyla Equus spp. + Artiodactyla Came/ops sp. + Antilocapra americana

referred to Lepus. Crenulations on the P3 also indicate Lepus and not Sylvilagus. In general size the teeth resemble those of the large species of Lepus and are larger than those of L. americanus (snowshoe hare). Only Lepus townsendii (white-tailed jackrabbit) is known to live in the Black Hills region today. It is questionable whether or not L. californicus (blacktailed jackrabbit) currently lives in the Black Hills, but it does occur nearby (Jones et al., 1983; Turner 1974). The smaller species, L. americanus, lives no closer to the Black Hills than northern and eastern North Dakota (Jones et al., 1983). The specimens from Salamander Cave did not lend themselves to identification below generic level. The specimens from the Horse Room were not compared to Lepus othus.

Lepus cf. L. americanus - snowshoe hare Specimens: Lr4-M2 (single tooth; 2485), RP4-M2 (single tooth; 2821) Identifications and comparisons: These two cheek teeth were smaller than those of the other Lepus discussed above. The size of the teeth is similar to those of L. americanus and larger than those of typical Sylvilagus spp. Until a larger sample of the fossil forms and of the varying sizes of living species can be assembled for measuring, we are provisionally refer-

466 J.I. Mead, C. Manganaro, C.A. Repenning, L.D. Agenbroad ring our Horse Room specimens to this distinctly smaller Lepus species. Its distribution today is discussed above. Rodentia Geomyidae - Pocket gophers

Thomomys sp. -pocket gopher Specimens: Lp4 (2815), RF4 (2828), RM1"2 (2; 2829), LM1· 2 (2830), RM 1 (2816) Identifications and comparisons: There are no characters on the isolated teeth that allow for species identification. Only Thomomys talpoides (northern pocket gopher) lives in the Black Hills today (Jones et al., 1983). Its grassland relative, Geomys bursarius (plains pocket gopher), lives immediately outside of the Hills proper (Jones et al., 1983; Turner 1974). Muridae - Mice and rats Cricetinae - New World mice and rats

Neotoma sp. - Woodrat Specimens: LM 2 (2470) Identifications and comparisons: Lophids are simple; the apex of the buccal edge of the mesolophid is oriented anteriorly. There is a minute accessory cusp at the bottom of the posterobuccal re-entrant (least height is 0.23 mm; a feature of N. cinerea and some other species). Buccal and lingual postero- and antero-re-entrants occur at the base of the enamel crown (minimum height of crown is 0.18 mm). Occlusal length is 3.06 mm. Occlusal width of anterolophid is 1.33 mm and across the posterolophid is 1.66 mm. Height of the tooth from the base of the crown to the occlusal surface at the mesolophid is 1.91 mm (following measurements in Zakrzewski, 1991, 1993). Until more Neotoma specimens can be recovered, we are not ready to identify the species of woodrat that lived in the local community 252,000 years ago. Today only Neotoma cinerea lives in the Black Hills (Turner, 1974). This is a large species of woodrat with distinctive molars. Neotoma floridana lives south, well into southern Nebraska. East of the Black Hills, along the Nebraska and South Dakota border, there occurs an isolated population of this species, possibly a relictual location from a middle Holocene warmer-than-present period (Jones et al., 1983). The LM2 from Salamc1nder Cave is similar in lophid occlusal pattern to that found on N. cinerea. The line that bisects the mesolophid in N. cinerea and the specimen from Salamander Cave is not perpendicular to a line that bisects the molar in an anteroposterior direction (as is found in N . alleni) (Zakrzewski, 1993). Zakrzewski (1993) has reviewed the morphological changes of molars

Early Rancholabrean mammals in Black Hills 467 observed in the various species of Neotoma of Hemphillian, Blancan, Irvingtonian, Rancholabrean, and Holocene ages. Dental characters observed on the specimen from Salamander Cave were compared to those found in the descriptions of various species. N. amplidonta from the Java local fauna, South Dakota, of Irvingtonian age, has a relatively large accessory cusp filling the posterobuccal fold; all the folds extend down the lingual side of the enamel crown almost the entire height (Zakrzewski, 1985). The Salamander Cave specimen does not appear to be N. amplidonta, based on these characters.

Peromyscus sp. - deer mouse Specimens: L mandible (2827), R mandible (2; 2825, 2826) Identifications and comparisons: Preservation of the edentulous mandibles did not permit identification beyond generic level. Peromyscus leucopus (white-footed mouse) and P. maniculatus (deer mouse) currently live in the Black Hills (Jones et al., 1983; Turner 1974). Arvicolidae - Voles and lemmings

Terricola meadensis (= Pitymys meadensis)- Meade vole Specimens: M 1(4; 2831), M3 (2832) Identifications and comparisons: Terminology and characters for identifying this extinct species are provided in Barnosky and Rasmussen (1988), Martin (1987), and Repenning (1983, 1992). Pitymys ochrogaster (prairie vole) is found living in the Black Hills today (Turner, 1974). Mictomys meltoni or kansasensis - vole Specimens: M1_2 in mandible (2833) Identifications and comparisons: Terminology and characters for identifying this species are provided in Barnosky and Rasmussen (1988) and are discussed in von Koenigswald and Martin (1984). Specimens referred to these extinct species have been found in Porcupine Cave, CO (Barnosky and Rasmussen, 1988), Hansen Bluff, CO (Rogers et al., 1985), Cumberland Cave, MD (van der Meulen, 1978), and the type locality of Cudahy, KS. The specimens from the Horse Room are not as advanced (i.e., the lingual re-entrants of the lower teeth do not extend toward the buccal side of the tooth as far as in more advanced forms and the anterior projection of the cap is too broad) in comparison with those from Snowville, UT (Repenning, 1987) or Porcupine Cave. Microtus paroperarius - Hibbard' s vole Specimens: M1_2 in mandible (2835), M1(4; 2834), M3 (2) Identifications and comparisons: Terminology and characters for identifying this species are provided in Barnosky and Rasmussen (1988) and Repenning (1992). The specimens from the Horse Room are similar to

468

J.I. Mead, C. Manganaro, C.A. Repenning, L.D. Agenbroad

those morphotypes known from the Cudahy fauna and do not contain the primitive morphotypes found in Hansen Bluff, CO, or Cumberland Cave, MD. Microtus paroperarius is an extinct species. Barnosky and Rasmussen (1988) recovered Microtus paroperarius in association with the similar M. montanus/longicaudus at Porcupine Cave. M. longicaudus (longtailed vole) and M. pennsylvanicus (meadow vole) live in the Black Hills today (Turner, 1974). Neither of these species were identified from the specimens recovered in the Horse Room. Repenning (in Rogers et al., 1985) did not recover either M. montanus or M. longicaudus at Hansen Bluff. Living Microtus montanus/longicaudus have a morphotype of the first lower molar that is duplicated by some of the individual morphotypes of the extinct M. paroperarius. The distinction between these species depends upon a large sample to recognize the great morphologic range of first lower molar of M. paroperarius from the more uniform morphotype range of the living species. The sample from the Horse Room fauna is not large (five first lower molars). Accordingly, the identification of species is not secure. However, three distinct first-lower-molar morphotypes are present in this small sample; this is sufficient to identify the species M. paroperarius if it is assumed that only one species is represented. In addition, a single second upper molar is present in the sample and it has a fifth (posterior) dentine field which is usually attributed to M. pennsylvanicus, but which is also found in a minority (2%) of second upper molars of M. paroperarius from the type locality, Cudahy, Kansas. We are not aware of the occurrence of this fifth dentine field in M. montanus/longicaudus. M. paroperarius thus appears to be present, but it is not demonstrable that other species of Microtus are not. The specimen from the Horse Room may be the northernmost record for the species. Sciuridae - squirrels

Cynomys (Leucocrossuromys) sp. - white-tailed prairie dog Specimens: R mandible (3; 2091, 2104, 2433), L mandible (2; 2453, 2481), RP2 (2824), radius (2; 2431, 2482) Identifications and comparisons: Living prairie dogs belong to two subgenera, Cynomys, the black-tailed prairie dog group, and Leucocrossuromys, the white-tailed group. Cynomys (Leucocrossuromys) contains the species C. leucurus, C. parvidens, C. gunnisoni (Gunnison's prairie dog), and the extinct Rancholabrean form, C. niobrarius (including C. churcherii; Burns and McGillivray, 1989; Dalquest, 1988) (Goodwin, 1993). C. (Cynomys) includes C. ludovicianus and C. mexicanus. Only C. (C.) ludovicianus lives in the immediate area of the Black Hills today, however; C. leucurus lives to the west in Wyoming (Jones et al., 1983; Turner 1974). Teeth within the subgenus Leucocrossuromys have a stylid that abuts

Early Rancholabrean mammals in Black Hills 469 the ectolophid of the M3. Cynomys teeth from the Salamander Cave possess this stylid. This stylid does not occur in prairie dogs in the subgenus C. (Cynomys). Semken (1966) found that the stylid is not present 10% of the time in C. (Leucocrossuromys). However, the stylid's presence is unique to the subgenus, causing assignment of our specimens to C. (Leucocrossuromys). Although we cannot assign the teeth from Salamander Cave to a species, it is still of great importance that they are placed within Leucocrossuromys, which does not exist in the region today (Jones et al., 1983). Goodwin (1993) indicates that two species of Cynomys lived during the latest Irvingtonian and earliest Rancholabrean (or Sheridanian; 500,000 to 200,000 BP): (1) C. (C.) new species B, and (2) C. (L.) niobrarius. Based purely on this simple scheme, the prairie dogs from the Horse Room in Salamander Cave most closely resemble C. niobrarius (Goodwin, 1993). Cynomys (Cynomys) is a characteristic inhabitant of the short- and mixed-grass prairies of central North America, including the Black Hills. In contrast, Cynomys (Leucocrossuromys) today lives farther west of the Black Hills, where it typically inhabits xeric sites with mixed stands of shrubs and grasses (Armstrong, 1972). The northern limits of modem prairie dogs seems environmentally significant.

Cynomys sp. - prairie dog Specimens: RM3 (2461), LM3 (2462) Identifications and comparisons: These specimens are readily identifible to Cynomys, however, it was not possible to assign them to a subgenus. cf. Spermophilus (Spermophilus) sp. - ground squirrel Specimens: L maxilla (2111), RM 1-2 (2483) Identifications and comparisons: Specimens were assigned to this subgenus because they had a complete metaloph, an abrupt change in the direction of the parastyle which joins the protocone, and a rectangular rather than quadrate outline (Hall, 1981). The size of the individuals indicates that they probably belong to Spermophilus richardsonii (Richardson's ground squirrel). Until more specimens can be recovered we will not identify this specimen beyond the subgeneric level. Ground squirrels that live today in the region of the northern Great Plains include (1) Spermophilus (Spermophilus) richardsonii (to the south of the Black Hills) and S. elegans (Wyoming ground squirrel; to the north and east of the Black Hills); (2) S. (Ictidomys) tridecemlineatus (thirteen-lined ground squirrel; within and around the Black Hills) and S. spilosoma (spotted ground squirrel; immediately to the south of the Black Hills); and (3) S. (Poliocitellus) franklinii (Franklin's ground squirrel; to the east of the Black Hills) (Jones et al., 1983; Turner, 1974).

470

J.I. Mead, C. Manganaro, C.A. Repenning, L.D. Agenbroad

Spermophilus sp. - ground squirrel Specimens: L mandible (2438) Identifications and comparisons: This specimen could not be identified beyond the generic level. It is much smaller than S. richardsonii and does not belong to that subgenus. Carnivora Mustelidae - weasels, skunks, and allies

Mustela sp. - weasel Specimens: LP4 (2471) Identifications and comparisons: The P4 is referrable to the genus Mustela. The specimen is not similar to the larger, more robust premolars of the skunks (Mephitis or Conepatus), nor to the shape of the smaller P4 of Spilogale. The P4 of Brachyprotoma (extinct skunk) was not examined, but it is similar to that of Spilogale (Youngman, 1986). The length of the P4 (2.90 mm) from Salamander Cave resembles those found in Mustela frenata (long-tailed weasel; 2.50-3.98 mm), is much longer than P4 lengths in M. erminea (ermine) and M. nivalis (least weasel; both less than 2.20 mm), and is shorter than lengths in M. vison (mink; 4.40-5.08 mm) and the larger species, M. nigripes (black-footed ferret) and M. macrodon (extinct sea mink). Without more indicative fossil teeth or postcranial elements, and without more measurements from a greater number of modern comparative specimens, we will not identify the Salamander Cave specimen beyond the generic level and just indicate the apparent size of the individual. M. erminea, M. frenata, M. nigripes (now extirpated), and M. vison are found in the Black Hills historically (Turner, 1974). Canidae - Wolves, dogs, and foxes

Canis cf. C. dirus - dire wolf Specimens: L maxilla fragment with M1-2 (2445), R maxilla fragment with M 1-2 (2446, 2447), RI 2 (2114), LP1 (2840), metatarsal (2474), L radius proximal fragment (2472), R tibia distal fragment with attached fibula fragment (2452) Identifications and comparisons: Remains of Canis from Salamander Cave are restricted to upper molars and fragments of certain postcranial elements. Based on the size of the specimens, the remains certainly belong to a large member of the genus and not something the size of C. latrans (coyote). Specimens were compared to C. armbrusteri (Armbruster's wolf), C. dirus (dire wolf), and C. lupus (timber wolf). C. armbrusteri is distinctly larger than C. rufus (red wolf) and similar in size to C. lupus, but usually smaller than C. dirus. The M 1 from Salamander Cave has a prominent paracone, metacone,

Early Rancholabrean mammals in Black Hills 471 and hypocone (dissimilar to that found in Canis armbrusteri). The anterior ridge of the hypocone connects with the protocone, and does not extend around it as it does on C. armbrusteri. There is a fairly prominent buccal cingulum (2.0 mm wide at the metacone) on the Salamander Cave specimen, as it is on C. armbrusteri (Nowak, 1979). These qualitative characters would imply that the Salamander Cave specimen is not C. armbrusteri, but belongs to C. dirus. The specimens from Salamander Cave have a prominent hypocone, a pronounced buccal cingulum, and an enlarged metacone and paracone; all characters found in C. dirus and not in C. lupus (Nowak, 1979). The radius and the tibia from Salamander Cave are identical in size and robustness to a female C. lupus zoo specimen (QSP 5936). The Salamander Cave canid specimens seem to belong to a C. dirus, but not an overly large animal; in fact, the individual was about the size of a modern large female C. lupus, but with robust teeth. However, Nowak (1979) cautions that some specimens of C. armbrusteri approach those of C. dirus in size. Kurten (1984) has illustrated that C. dirus dirus and C. lupus were similar in limb proportions, but that C. d. guildayi was shorter. Measurements of the molars follow the method of von den Driesch (1976) and are presented in Table 2. Measurements to represent the Salmander Cave specimens were taken from WICA 2446 for the M1 and WICA 2447 for the M2• Other measurements were taken from the literature (Table 2). The combined measurements imply that the Salamander Cave specimen is relatively large and robust, and probably represents Canis dirus, although some of the metrics are within the range of the other species. For the above reasons, we feel that the identification of the Canis material from Salamander Cave should be referred to C. dirus, realizing that more material is needed (such as M1 and more complete cranial remains) to distinguish it conclusively from C. armbrusteri. Size and dental morphology would seem to dictate that the Salamander Cave specimens are not C. lupus. If these specimens do represent C. dirus, then they are the first record of species in South Dakota and probably the oldest in North America north of Mexico. C. armbrusteri is not known from South Dakota and typically is found in lrvingtonian age deposits. Perissodactyla Equidae - Equine horses and asses

Equus spp. - extinct horse Specimens: L premaxilla fragment with 12•3 (2043), LP2 (2030), L mandible fragment (2459), LP2 (2037), RP2 (2039), Ll3 (2045), Rl3 (2089), RP3-M2 (2; 2036, 2038), cervical vertebra (2; 2015, 2458), lumbar vertebra (2457), R ilium and acetabulum (2019, 2454), L scapula fragments (2; 2035, 2042,

472 J.I. Mead, C. Manganaro, C.A. Repenning, L.D. Agenbroad TABLE 2 Measurements of the molars in Canis armbrusteri, C. dirus, and C. lupus. K=Kurten (1984), N=Nowak (1979), QSP=Quaternary Studies Program specimen 5936, R=Ruddell (1992); measurements in mm; OR=Observed Range Breadth Locality or species

OR

1. Ml Salamander Cd,RLBK Cd,SJCK CddsK CddwK Cd,RLBR CIQSP

23.52 22.80-28.30 22.70-26.70 24.50-28.50 24.40-27.90 19.11-27.69 19.99

2. M2 Salamander Cd,RLBK Cd,SJCK CddsK CddwK ClN CdN CaN

16.24 13.20-17.00 13.60-16.20 14.30-17.80 14.40-16.70 11.20-16.30 13.10-17.00 14.10-17.10

Length Mean

25.40 25.11 26.08 26.25

OR

18.08 17.30-22.00 17.60-20.90 18.50-21.30 18.70-20.50 14.95--23.14 16.42

Mean

19.48 19.38 19.99 19.99

14.98 14.85 15.94 15.60 13.80 15.15

Locality or species: Salamander = Salamander Cave; RLB = Rancho La Brea; SJC = San Josecito Cave; Cd = Canis dirus; Cdds = Canis dirus dirus of Sangamon age; Cddw = Canis dirus dirus of Wisconsinan age; Cl = Canis lupus; Ca = Canis armbrusteri.

2092, 2094, 2448), L radius (2449, 2478), R radius (2044), L ulna (2456), R ulna (2025), R os carpale 3 (2033), L femur (2014), L os tarsale 3 (2; 2086, 2477), R calcaneum (3; 2026, 2027, 2064), L astragalus (2028), R astragalus (2; 2016, 2020), L metatarsal (2018), R metatarsal (2017), 1st phalanx (5; 2029, 2031, 2032, 2450, 2480), 2nd phalanx (2479) Identifications and comparisons: The fossil record of equine horses is excellent - maybe too good. More than 40 species have been proposed for Equus, sensu lato, for the Blancan, lrvingtonian, and Rancholabrean of North America (Kurten and Anderson, 1980); there is obvious and continued confusion as to valid species and species-groups. Kurten and Anderson (1980) list a number of potential species, predominantly based on the work of Lundelius (1972). Under their system, E. tau (pygmy onager) is the smallest of the horses (hemionine) of the lrvingtonian and Rancholabrean (Dalquest, 1979). E. conversidens (Mexican horse) is another small horse of the stout-limbed group, although larger than E. tau (Dalquest, 1979). E. scotti (Scott's horse) was a large Pleistocene horse, approximately

Early Rancholabrean mammals in Black Hills 473 the size of a large modern horse, with heavy limbs. A species similar to E. scotti, but slightly smaller, was E. niobrarensis (Niobrara horse); most of the records for this horse seem to be of Irvingtonian age according to Kurten and Anderson (1980). Winans (1989) provided a multivariate quantitative analysis of characters of skull, mandible, and metapodials for Equus. Her work indicates that the genus can be divided into no more than five groups, but it was not stated that these are to be considered species: E. simplicidens, E. scotti (containing E. niobrarensis), E. laurentius, E. francisci, and E. alaskae (containing E. conversidens and E. niobrarensis alaskae). Under her analysis, only the first species group did not live during the Pleistocene. Winans (1989) determined that her discriminant analysis correctly identified nearly all skulls to a species group; however, postcranial elements were misidentified 25% of the time. Harris and Porter (1980) provided a cluster analysis of measurements of single elements of Equus remains from Dry Cave, New Mexico. E. conversidens, E. niobrarensis, E. occidentalis (western horse), and E. scotti were considered to be valid species. Measurements of skeletal elements from Salamander Cave follow those of Harris and Porter (1980) and von den Driesch (1976). Our measurements are compared with those in Harris and Porter (1980). Length of the two metatarsi from Salamander Cave are 290 mm (2017) and 279 mm (2018). These both fit best into the measurement grouping confined to E. niobrarensis (263-288 mm; E. conversidens, 262-274 mm; E. scotti, 304 mm). Mid-diaphysis width (41.5 and 38.5 mm, respectively) and the width of the proximal end (58.0 and 53.5 mm, respectively) of the two metatarsi from Salamander Cave, Horse Room, also compare well with the groupings of E. niobrarensis. Three astragali were recovered from Salamander Cave, one left and two right. All three measured to be similar to the Equus niobrarensis from Dry Cave. Two measurements of Harris and Porter (1980; numbers 5 and 18; illustrating height and breadth of the astragalus) were used to make this comparison. Three calcanea (all rights) from Salamander Cave were analysed using measurement numbers 1, 2, 6, and 14 of Harris and Porter (1980). Two were small and similar to the Dry Cave E. conversidens measurements, whereas the very large specimen was similar to E. niobrarensis. Five first phalanges were recovered from Salamander Cave, and were compared to the measurement numbers 1, 6, 7, 11, 12, and 13 of Harris and Porter (1980). Two size grouping are determined based on these comparisons. The two larger first phalanges compare with horses the size of E. niobrarensis to E. scotti from Dry Cave. The three smaller first phalanges are similar in size to E. conversidens from Dry Cave. Although not in complete agreement with Harris and Porter (1980), the

474

J.I. Mead, C. Manganaro, C.A. Repenning, L.D. Agenbroad

Salamander Cave specimens are from two different-sized horses, possibly corresponding to two different species or, alternatively, possibly representing individual variation of one species. Comparing the Salamander Cave specimens to those from Dry Cave, it would appear that there is a small horse, similar to Equus conversidens, and a large horse approximately the same size of (and sometimes larger than) E. niobrarensis, but not as large as E. scotti. Using Winans's (1989) analysis, both of these two larger species found at Dry Cave would be considered the same species of large horse (E. scotti-group). Artiodactyla Camelidae - camels

Camelops sp. - extinct camel Specimens: L femur (2451), distal fragment of metapodial (2067), metapodial distal epiphysis (2476) Identifications and comparisons: The metapodial is identifiable as camelid because of the divergent distal ends (a character of the family) and its large size. Although fragmented, the general size and the robustness of the metapodial and femur from Salamander Cave is consistent with those of Camelops (from Rancho La Brea) and larger than those of Hemiauchenia (which tends to be more slender and more gracile). Fragmentation of the specimens from the Horse Room did not permit further identification. The specimens were not compared to Palaeolama or Titanotylopus. Antilocapridae - pronghorns

Antilocapra americana - pronghorn Specimens: R tibia minus the proximal epiphysis (2444) Identifications and comparisons: The adult tibia fragment has the distal articular surface well preserved. Identification to Antilocapra follows from the characters outlined in Lawrence (1951), which readily distinguish the Salamander Cave specimen from Odocoileus and Ovis. The size of the tibia is identical to those found in Antilocapra, which is smaller and less rugose than those elements found in Cervus and Navahoceros . Based on the length of Salamander Cave specimen (196.7 mm minimum length), the antilocaprid specimen from Salamander Cave is well within the range of the lengths of Antilocapra and the extinct Stockoceros (Skinner, 1942). 'As a group, the tibiae of Stockoceros average 6 mm longer than those of Antilocapra. This is an outstanding difference in the skeletons of the two species, since Antilocapra is the larger animal' (Skinner, 1942:213). When viewed distally, the intermediate ridge between the articular grooves is smoothly rounded on Stockoceros. This same area on Antilocapra and WICA 2475 is not rounded, but has the synovial fossa (Skinner, 1942). Antilocapra americana is common today to the west of and within the Black Hills.

Early Rancholabrean mammals in Black Hills 475

Discussion and conclusions

Chronology The age of the fauna from the Horse Room at Salamander Cave is of concern. The Th/Pa age of 252,000 yr BP seems reliable. Simplistically this would place the 252,000 yr BP fauna in Rancholabrean I (early Rancholabrean) dating between 400,000 ± 25,000 and 150,000 ± 25,000 yr BP of Repenning (1987). As mentioned earlier, there is some question as to whether or not the Horse Room fauna is of Irvingtonian or Rancholabrean age. Unfortunately there is no Bison in the fauna to dictate a Rancholabrean age. As Repenning (1987:251) has stated, 'there is no record of Bison in association with Rancholabrean I microtines.' The beginning of the Rancholabrean in the USA can be characterized by the introduction of several new species of Microtus. East of the Rocky Mountains, M. pennsylvanicus replaces M. paroperarius, and west of the mountains the monopoly of the region held by M. californicus (California vole) since early Irvingtonian is broken by the first appearance of M. montanus (montane vole) and M. mexicanus (Mexican vole) (Repenning, 1987). Terricola meadensis survived into Rancholabrean I. Mictomys survives into Rancholabrean I; however, it has not previously been reported from the Great Plains region. Also, at this time it is known from the Appalachian Mountains and northern Utah as a form intermediate between M. meltoni and the living M. borealis (Repenning and Grady, 1988). Based on the recovered arvicolines from Salamander Cave, the fauna could be of late lrvingtonian (Irvingtonian II; 900,000-400,000 yr BP) or of Rancholabrean I age (Repenning, 1987). All three arvicolines fit best in a Cudahy or younger type of fauna (i.e.,

p.. 0

-a....!:I"" rJl

....3 :;:l

::i,..,

lS.: C

2 ~

~

01

"""

'I

548 H.B.S. Cooke

E~ FIGURE 3. Dentitions of Antidorcas recki from Pit 3. A, UC 69519, RP3-M3 and root of P2. B, UC 69520A, LM 1_3, and broken roots of LP2_4. C, UC 69523, Lr4-M 3. D, UC 69522, Lr4-M3. E, LM3 of male cranium UC 69521. Natural size

high in very early wear. The other maxilla fragment, UC 69522, is in moderately advanced wear and M 1 is badly broken. None of the specimens preserves the area in which p 2-3 would occur.

Lower dentition The best mandible is UC 69519, which contains RPcM3 and the roots of RP2 and RP3 . Although P2 was small, it appears to have been a functional tooth, whereas in A. marsupialis it is commonly absent or, if present, is very small and non-functional. In state of wear and in the details of morphology of the jaw itself, this mandible is very similar to the holotype of 'Gazella wellsi' (Cooke, 1949) from the Vaal River. Comparative measurements are set out in Table 2. Three other partial mandibles also occur and have the same characteristics but are slightly smaller (see Table 2). The premolar/molar length ratios in all these dentitions is close to 30%. None of the lower teeth displays the central constriction of the fossettes that Vrba (1973) regards as typical of A. bondi. The lingual walls of the molars do not show the rather strongly undulating aspect seen in A. bondi and are almost as flat as in the living springbok (Fig. 3).

Discussion In 1973 Vrba described two species of Antidorcas from Swartkrans, followed in 1976 by an account of the fossil Bovidae, including Sterkfontein and Kromdraai. Most of the antilopine material is rather fragmentary, consisting largely of partial upper and lower dentitions. Swartkrans

TABLE 2 Comparative measurements (mm) of lower cheek teeth of Antidorcas recki from Bolt's Fann Pit 3

Antidorcas marsupialis

'G.wellsi' holotype•

UC69519 right

UC69520A left

UC69520B left

P2

length breadth height

absent

-

3.Se

3.0e

P3

length breadth height

5.0 3.4 4.0+

-

6.0 4.0e 8.5+

5.0e

4.3e broken 6.5+

P4

length breadth height

9.3 4.8 9.0+

-

7.5 4.3 8.5+

6.0e

4.4e 3.Se 6.5+

UC69520C right

VJ l'l) X

i:::

length breadth height

12.9 6.8 14.0

10.5 6.0 11.0e

11.6 6.5 12.0e

10.0 5.8 15.0+

9.4 5.0e 9.0+

-

length breadth height

15.7 6.8 20.0+

13.1

11.9 6.2 18.0e+

10.7 6.2 12.0+

-

7.1 14.0e+

4.3 7.5 17.0e+

length breadth height

21.9 6.3 25.0+

20.0 7.0 28.0e+

21.0 7.0 28.0e+

18.7 6.7 24.5+

18.4 6.8 14.0+

19.5 6.5 25.0+

length P2-P4

14.5 (P,-PJ

-

17.0e

14.0e

12.0e

-

....

length M1-M,

50.5

46.6

47.0

40.8

38.5

-

~ U1 ~ ~

550 H.B.S. Cooke Member 2 has abundant Antidorcas bondi, easily recognized by the hypsodonty of the molars, and Vrba also assigned some different remains to A. australis. The latter is otherwise a southern species not found at the northern breccia sites, but A. bondi is recorded from Kromdraai A and B, from Sterkfontein extension, and doubtfully at the type site. Antidorcas recki is firmly identified at Kromdraai A and tentatively at the other sites, except Swartkrans Member 2, and it is possible that the Swartkrans material regarded as A. australis may be a variant of A recki. The only really good specimen assigned to A recki is an almost complete skull from Kromdraai A with well-preserved palate and frontal area but lacking the dorsal cranium and with the horns broken away through the frontals. Vrba (1976:35) suggests that the horn-cores must have been 'of a size and positioning on the skull very close to that observed in, for instance, extant female A. marsupialis.' Schwartz's type specimen is clearly male and the general form of the horns is shown by a frontlet, by a young cranium, and by isolated cores from Olduvai, although all are a little smaller and shorter than in the holotype (Gentry and Gentry, 1978; Fig. 4). The material originally assigned to Gazella wellsi by Gentry (1966) includes two crushed skulls with horn-cores and several horn-cores less sharply bent back than in the type of A. recki, which was one of the reasons for their initial placement in G. wellsi. Gentry (1978) decided that three Olduvai cores and one from Kanjera must be females whose horns 'differ from the male by being less backwardly bent, probably less outwardly divergent in their distal portions and, in M.14513, in tapering to a point over a much shorter distance.' Vrba (1976) reported one good horn-core, as well as a juvenile and a horn-core tip from the Sterkfontein extension site, and some fragmentary cores from Kromdraai A and B, although dentitions are better represented. Antidorcas bondi is well represented at Swartkrans and there are both resemblances to, and differences from, A. recki that were effectively analysed by Vrba (1973). The only good cranium of A. bondi is SK 3001, and it differs from A. recki in that it has a shorter braincase and the horn-cores lie more nearly in line with the face and are straighter, lacking the rather marked flexure shown by A. recki (Fig. 4). The core is robust, although smaller than in the type of A. recki or in the Bolt's Farm male, or in specimens from Kromdraai. A good frontlet of another specimen (SK 14126A) has horn-cores that are almost straight, unribbed, and round (12.8 x 12.7 mm), and are considered by Vrba (1973:295, pl. 19) as representing a female. These horns are very similar to those of the Bolt's Farm female but are a little less divergent. Gentry (1966) demonstrated a fair range of variation in the horn-cores of A. recki (including 'Gazella wellsi') at Olduvai and suggested that the cores that were morphologically similar, but somewhat smaller, shorter,

Sexual dimorphism in Antidorcas recki 551

A

-==--,=--•m

FIGURE 4. A, Comparative outline drawings of three Antidorcas crania (thin line= type specimen of A. recki from Olduvai Gorge; bold line= Bolt's Farm male A. recki, UC 69521; dashed line shows cranium, SK 3001, of Antidorcas bondi from Swartkrans). B-D are outlines of three horn-core variants in A. recki from Olduvai Gorge, after Gentry, 1966 (B, from Bed I surface, BMNH M.14513; C, from Bed I locality FLKN l , 1960, 067 /250; D, frontlet from Upper Bed II, 1935, SHK II, BMNH M.21462).

and less backwardly bent, might represent females. However, in 1978 Gentry and Gentry stated that this interpretation could not be maintained and they suggested that one specimen (BMNH.M.22362) had been ascribed erroneously to a female Gazella in 1966 (p. 65, pl. 7A) and might rather be a female of Antidorcas recki. This core is rounded in cross-section, only 2-3 mm larger than in the Bolt's Farm female, very gently curved, and so similar in morphology that the two are virtually identical. Thus, at three of the sites where A. recki occurs, there appear to be two kinds of cores and, in view of the undoubted association at Bolt's Farm, it now seems clear that both A. recki and A. bondi exhibit a degree of sexual dimorphism very similar to that shown by A. marsupialis, or perhaps even more marked.

Acknowledgments The writer is indebted to the Museum of Paleontology at the University of California for permitting him to study and describe this material, as

552 H.B.S. Cooke well as to staff and students for their assistance. It is a privilege to be able to make this contribution to a volume dedicated to an old friend, colleague, and fellow student of African fossil mammals, Rufus Churcher.

LITERATURE CITED

Broom, R. 1936. A new fossil anthropoid skull from South Africa. Nature 138:486488. - 1937. On some new Pleistocene mammals from limestone caves of the Transvaal. South African Journal of Science 33:750-768. - 1938. The Pleistocene anthropoid apes of South Africa. Nature 142:377-379. - 1939. A preliminary account of the Pleistocene Carnivores of the Transvaal caves. Annals of the Transvaal Museum 19:331-338. - 1948. Some South African Pliocene and Pleistocene mammals. Annals of the Transvaal Museum 21 :1-38. Churcher, C.S. 1956. The fossil hyracoidea of the Transvaal and Taungs deposits. Annals of the Transvaal Museum 22:477-501. - 1970. The fossil Equidae from the Krugersdorp caves. Annals of the Transvaal Museum 26:145-168. Cooke, H.B.S. 1949. Fossil mammals of the Vaal River deposits. Geological Survey of South Africa Memoir 35 (Pt. 3):1-109. - 1985. Ictonyx bolti, a new mustelid from cave breccias at Bolt's Farm, Sterkfontein area, South Africa. South African Journal of Science 81:618-619. - 1991. Dinofelis barlowi (Mammalia, Carnivora, Felidae) cranial material from Bolt's Farm, collected by the University of California African Expedition. Palaeontologia africana 28:9-21 . - 1993a. Undescribed suid remains from Bolt's Farm and other Transvaal cave deposits. Palaeontologia africana 30:7-23. - 1993b. Fossil proboscidean remains from Bolt's Farm and other Transvaal cave deposits. Palaeontologia africana 30:25-34. Gentry, A.W. 1966. Fossil Antilopini of East Africa. British Museum of Natural History Bulletin (Geology series) 12:43-106. Gentry, A.W., and A. Gentry. 1978. Fossil Bovidae (Mammalia) of Olduvai Gorge, Tanzania, Part I. British Museum of Natural History Bulletin (Geology series) 29:290-446. Jones, T.R. 1937. A new fossil primate from Sterkfontein, Krugersdorp, Transvaal. South African Journal of Science 33:709-728. Leakey, L.S.B. 1965. Olduvai Gorge 1951-1961, Vol.1: A Preliminary Report on the Geology and Fauna. Cambridge University Press, Cambridge, 109 pp. Oakley, K.P. 1960. The history of Sterkfontein, with a comment by B.D. Malan. South African Journal of Science 56:110.

Sexual dimorphism in Antidorcas recki 553 Schwarz, E. 1932. Neue diluviale Antilopen aus Ostafrika. Stuttgart: Zentralblad fiir Mineralogie, Geologie und Palaontologie 1932B:1-4. - 1937. Die fossilen Antilopen von Oldoway. Wissenshaftliche Ergebnisse der Oldoway-Expedition, 1913, N.F. 4:8-104. Shaw, J.C.M. 1937. Evidence concerning a large fossil Hyrax. Journal of Dental Research 16:37-40. - 1938. The teeth of the South African pig (Notochoerus capensis syn. meadowsi) and their geological significance. Transactions of the Royal Society of South Africa 26:25-37. - 1939. Growth changes and variations in wart-hog third molars and their palaeontological importance. Transactions of the Royal Society of South Africa 27:5194.

Vrba, E.S. 1973. Two species of Antidorcas Sundevall at Swartkrans (Mammalia, Bovidae). Annals of the Transvaal Museum 28:287-352. - 1976. The fossil Bovidae of Sterkfontein, Swartkrans, and Kromdraai. Transvaal Museum, Memoir 21:166 pp.

A review of Dietrich's hipparions from South Serengeti (Tanzania) and a comparison with similar materials Ann Forsten

Abstract The morphology and taxonomy of the Pliocene, so-called caballoid, hipparions of Africa are discussed, with special regard to those from South Serengeti (Tanzania). Two main taxonomic, possibly ancestor and descendant, categories are distinguished. The origin of the African caballoid hipparions is discussed from the point of view of the evolution of this group of horses in the Old World in general.

Introduction The African caballoid hipparions have been discussed in numerous papers, both as members of faunas from particular sites and synthetically: for example, by Arambourg (1970), Bone and Singer (1965), Churcher and Richardson (1978), Cooke (1950), Dietrich (1942), Eisenmann (1976a, 1977, 1983, 1985), Haughton (1932), Hendey (1978), Hooijer (1975, 1976, 1987), Hooijer and Churcher (1985), Joleaud (1933), Pomel (1897), and van Hoepen (1930, 1932). Eisenmann (1977, 1979) discussed the evolution of the caballoid enamel pattern and the emergence and increase in size of the ectostylid. The term 'caballoid' derives from the enamel pattern of the (chiefly lower) cheek teeth of these hipparions (Fig. 1), which in some features paralleled that in the horse, Equus caballus L. Thus, the protocone of the uppers tended to be long, the entoflexid between the angular metaconid and metastylid of the lowers U-shaped, and the longitudinal enamel crests straight. The straight longitudinal enamel crests on the occlusal surface of the upper and lower cheek teeth had clearly functional significance (Rensberger et al., 1984). In many African caballoid hipparions the

South Serengeti hipparions 555

FIGURE 1. Right lower P3---4 and M1_2 of a hipparionid (top; from Sefve, 1927: Fig.14) and caballoid (bottom, from Teilhard and Young, 1931: Fig. 14, ectostylid added), hipparion

ectostylid was exceptionally well developed and large in the permanent lowers. The caballoid hipparions, which occurred in Eurasia, Africa, and North America and had partly overlapping time ranges, were dentally advanced over the conventional, 'hipparionid,' hipparions (Fig. 1), in which the protocone varied in length, the entoflexid between the rounded metaconid and metastylid was V-shaped, and the longitudinal enamel crests convex. Although variably present, the ectostylid in the hipparionid permanent lowers was seldom large. The still-unresolved question is: When did caballoid hipparions first appear in Africa and what was their origin? Lower cheek teeth with an incipiently caballoid stamp occurred in Africa as early as at Lukeino (dated < 5.6 Ma [Hill et al., 1992; Hooijer, 1975]), Lothagam (ca. 5.8 Ma [Hill et al., 1992]), and Kanapoi (3.7 Ma [Hooijer and Maglio, 1974; Vrba, 1987: Tab. 1]), all in Kenya. Teeth with a hipparionid pattern still dominated at Sahabi (Libya) (ca. 5 Ma [Bernor et al., 1987; Boaz et al., 1987]), Kaperyon (Kenya) (ca. 5 Ma [Bishop et al., 1971]), Langebaanweg 'E' Quarry (South Africa) (ca. 4.5 Ma [Hendey, 1978; Hooijer, 1976; Vrba, 1987]), and Early Kaiso (Uganda) (Nyabrogo,

556 A. Forsten Nyawiega: ca. 4-7 Ma [Cooke and Coryndon, 1970; Hooijer, 1975]). Hipparionid or incipiently caballoid teeth are found at some still younger sites, e.g., Denen Dora Member (Hadar Fm., Ethiopia) (3.1-3.0 Ma [Eisenmann, 1983: Pl. 5.1; Haileab and Brown, 1992]), Kubi Algi Fm., zone B (Kenya) (ca. 3 Ma [Eisenmann, 1983]), and Hasuma Tuff (Koobi Fora Fm., Kenya) (ca. 2.9 Ma [Brown and Feibel, 1991; Eisenmann, 1983: Pl. 5.21). From about 4 Ma caballoid teeth dominated in the hipparion samples from fossil sites in Africa; among the earliest occurrences are Chemeron (dated ca. 5.6-4.1 Ma [Hill et al., 1992; Hooijer, 1975: Pl. 4:1 and 21) and left Pi-P4 (L 24197) from the Pelletal Phosphate Member, horizon 3aN, of the Langebaanweg 'E' Quarry(< 4 Ma [Hendey, 1978; Hooijer, 1976: Pl. 7:41). The ectostylid, the large size of which characterized the late PliocenePleistocene African hipparions and set them apart from their Eurasian relatives, was clearly associated with the caballoid enamel pattern. However, the ectostylid was already frequent in some samples of incipiently caballoid hipparions, for instance, from Denen Dora, Kubi Algi B, and Hasuma Tuff (Eisenmann, 1983), while it was rare in some caballoid forms, such as Dietrich's (1942) samples from Vogel River (=Laetoli; dated 3.7 Ma [Vrba, 1987]) and Garussi (both Tanzania), and from Langebaanweg Baard's Quarry (South Africa) (ca. 2 Ma [Bone and Singer, 1965; Hendey, 1978]). In the taxonomically heterogeneous Vogel River sample the presence of the ectostylid is strongly associated with the stratigraphically ?late, smaller teeth (Fig. 2a, b): M 1_i occlusal length < 2.5 cm : ectostylid presence xi 9.36, 1-P 1%; P3-4 occlusal length < 2.6 cm : ectostylid presence xi 9.36, 1-P 1%. The ectostylid tended to be large in the late, small, and pronouncedly caballoid hipparions also from other localities in Africa (Eisenmann, 1977: Figs. 2-4).

The South Serengeti hipparions The hipparion material from South Serengeti (Tanzania), kept in the Humboldt Museum in Berlin and described by Dietrich (1942), derives from several sites of different age: Vogel River (=Laetoli), Garussi, Gadjingero, Deturi, and Marambu. Dietrich (1942) identified two hipparion taxa in this material, Hypsohipparion albertense (Hopwood), believed to derive chiefly from the Grey Tuff older fauna, and Stylohipparion sp. from a younger fauna. He believed the former to be the larger, to lack an ectostylid, and to have 11_3 present and grooved. Stylohipparion sp. he believed to be smaller and to sport an ectostylid. Both taxa are very hypsodont, although not quite as high-crowned as Dietrich alleged, the maximum crown height measured being 8.20 cm (left M1, Vo 70). Both have incipiently caballoid to caballoid teeth (Dietrich, 1942: Taf. XIII, Figs. 89, 90, 91, and 92; see also Bone and Singer, 1965:345).

South Serengeti hipparions

557

+

- :!:-

+

1.50

♦

+

+

+ ♦

♦

+

+ 3.00

2.50

FIGURE 2a. Occlusal breadth (enamel to enamel) plotted to occlusal length of P34 in the sample from South Serengeti (+ = specimens with ectostylid; - = specimens lacking ectostylid)

1.50

+

+ +

♦•• +

+

••

+

2.50

3.00

FIGURE 2b. Occlusal breadth (enamel to enamel) plotted to occlusal length of M 1_2 in sample from South Serengeti (symbols as in Fig. 2a)

558 A. Forsten An analysis of dental size versus presence of the ectostylid in Dietrich's samples shows that, although the ectostylid is strongly associated with the smaller teeth, it is not wholly absent in the larger ones (Fig. 2a, b). In addition, the local samples from Vogel River and Garussi probably comprise both of Dietrich's hipparion taxa, although the faunas seem to differ in age. Equus is common from Garussi (Dietrich 1942: Taf. XIV, Figs. 98, 99 except M 1, and 100), indicating an age less than 2.5 Ma, which marks the early appearance of Equus in the Old World (Lindsay et al., 1980). Equus is rare or absent from Vogel River (there is possibly a milk dentition, VO 313; Dietrich, 1942: Taf. XIII, Fig. 93b), dated 3.7 Ma (Vrba, 1987); Vogel River thus predates the appearance of Equus. However, the fauna may be mixed (Bone and Singer, 1965:319; Dietrich, 1942). Deturi may also have two hipparions (there are few specimens), while Gadjingero (=Gadjuniro= ?Galudjiro) seems to comprise only Stylohipparion sp. From Gadjingero there is a lower, very flat symphysis (Gadj. 18/2 39 =Gadj. 10) lacking both I3 and C 1. From Galudjiro there is a juvenile skull fragment (Galudj. 2/39) lacking a preorbital fossa. The lack of a fossa and the flat lower symphysis lacking or with reduced I3 characterize Eurygnathohippus cornelianus van Hoepen from the Notochoerus scotti zone and KBS Tuff (ca. 2.5-1.9 Ma [Brown and Feibel, 1991]) of the Koobi Fora Fm. (Eisenmann, 1983). Hipparion ethiopicum Joleaud from the same and later beds of the Koobi Fora Fm. (Eisenmann, 1983) and (part) from Olduvai Bed II (Tanzania)(< 1.6 Ma [Vrba, 1987]) is probably a synonym (Hooijer, 1975; Leakey, 1965). The Laetoli lower cheek teeth studied by Hooijer (1979, 1987) also indicate a grouping into early, large specimens lacking an ectostylid (from Laetoli proper) and late, small ones with an ectostylid (from the Ndolanya Beds). Although the total number of teeth in Hooijer's sample is low, there seems to be a significant association between small tooth size and development of the ectostylid: P3-4 xi 8.73, 1-P 1-0.1%; M 1_i xi 3.77, 1-P a. 5%. A right lower dentition (Laetoli 75,491) from an unknown locality at Laetoli (Hooijer, 1979:30-31, Pl. 2, Figs. 1-2) has large, non-caballoid cheek teeth but ectostylids. Plotted on tooth length and ectostylid length, these teeth correspond to those from the Tulu Bor and Hasuma Tuff lower members of the Koobi Fora Fm. (Eisenmann, 1977: Fig. 2), with an age of 3.35 Ma, resp. ca. 2.9 Ma (Brown and Feibel, 1991; Haileab and Brown, 1992), and to those from Shungura Members B & C (Ethiopia) (3.3-2.5 Ma [Vrba, 1987: Tab. 1; data Eisenmann, 1985; Hooijer, 1975]). The teeth from the lower Koobi Fora Fm. were described as Hipparion hasumense Eisenm., characterized as having 'rather caballine' lowers with ectostylids (Eisenmann, 1983: Pl. 5.1). The teeth figured by Eisenmann resemble the large teeth from Vogel River, which lack or have small ectostylids. When plotted on tooth length and ectostylid length, the large teeth from Vogel River

South Serengeti hipparions 559 correspond almost exactly to those from the lower Koobi Fora Fm. (Eisenmann, 1977: Fig. 2). The Vogel River large lowers also correspond to those from the Kada Hadar Member (upper Hadar Fm., Ethiopia) (ca. 3.0-2.5 Ma [Eisenmann, 1977: Fig. 3; Haileab and Brown, 1992)). The latter sample was described as Hipparion afarense Eisenm. and the teeth characterized as caballine (Eisenmann, 1976a). Later Eisenmann (1983:168) seems to have doubted the assignment of lower cheek teeth to this species. However, the jaw AL 340-8 from Denen Dora (middle Hadar Fm.), belonging to a skull with the same number (Eisenmann, 1976a:580-582) and referred to H. afarense by Eisenmann (1982), has lower cheek teeth in situ although much worn; the teeth have ectostylids. The jaw AL 340-8 has a very long and flat symphysis, resembling that in VO MB 9 /10 38 from Vogel River (Dietrich, 1942: Taf. XVI, Fig. 112); neither jaw has reduced incisors. The lower cheek teeth from the Hadar Fm., with its successive members Sidi Hakoma, Denen Dora, and Kada Hadar, differ from those from other African localities in not showing a decrease in tooth size and increase in ectostylid size with time. While the lower members, Sidi Hakoma and Denen Dora, appear to comprise both large teeth with small ectostylids and small teeth with large ectostylids, all the teeth from the upper member, Kada Hadar, seem large and have relatively small ectostylids (Eisenmann, 1977: Fig. 3). Did the small caballoid hipparion with large ectostylids become extinct in the upper Hadar Fm.? The Vogel River large teeth resemble even more closely those from the Members B and C of the Shungura Fm., identified as Hipparion sp. (Eisenmann, 1985; Hooijer and Churcher, 1985: Fig. 2:7). The size bimodality of the teeth in Dietrich's (1942) hipparion material from South Serengeti is evident also in the few limb bones: the bones from Gadjingero (distal MT III, fragment of a proximal phalanx, calcaneum, and right astragalus) are 'small,' while those from Vogel River and Garussi (distal MT III, proximal phalanges, and astragalus) are either 'small' or 'large.' In a scattergram (Fig. 3) these size categories among the limb bones correspond to two size categories of limb bones from Olduvai, which clearly represent two hipparion taxa (see also Eisenmann, 1985:33), the larger being the more common (Fig. 3). The teeth from Olduvai cannot be differentiated on size and all are 'small,' with well-developed ectostylids (Hooijer, 1975). Small bones from the Metridiochoerus andrewsi zone of the Koobi Fora Fm. (ca. 1.9-1.6 Ma [Brown and Feibel, 1991]) and from the Members E, F, G, H, and L of the Shungura Fm. (2.3 Ma and younger [Vrba, 1987)), identified as Hipparion sitifense (Hooijer, 1975) and as Hipparion sp. B (Eisenmann, 1983, 1985), group together with the 'small' bones in Dietrich's samples and with those from Olduvai, as do the metapodials from Koobi Fora referred by Eisenmann (1983: Tab. 5.9) to Hipparion cf. H.

560

A. Forsten

5.00

0

A

0

•o

A A

0

A

(C)

o••

V 0

0 ., D

'o0

A

0

0

00 0

4.00

'

0 (f)

.

,.,

3.50

D

0

4.00

4.50

FIGURE 3. MC III and MT III distal protuberance width plotted to distal articular width in caballoid hipparions from Olduvai (0), Afar (A), Koobi Fora, Metridiochoerus andrewsi zone (KF), Gadjingero (Gd), Vogel River (V), and Shungura B, C, D, F, G, and uncertain level(?) (data Eisenmann, 1983, 1985; Hooijer, 1975; and own)

South Serengeti hipparions 561

ethiopicum. Large bones from the Hasuma Tuff, from Kubi Algi Fm. zone A, and Shungura C and G (data Eisenmann, 1983; Hooijer, 1975) group together with the large bones from South Serengeti and Olduvai (Fig. 3). It should be stressed that 'small' here is relative only to the 'large' bones; in reality the 'small' metapodials are as massive as those of the African Vallesian hipparions of the H. primigenium group.

Taxonomy Which is the correct (sub)generic and specific name(s) to be used for the hipparions from South Serengeti? The very fitting (sub)generic name Stylohipparion (van Hoepen, 1932) has often been used for the African caballaid hipparions with a large ectostylid. However, it was suggested (Dietrich, 1942), and subsequently shown (Leakey, 1965), that typical 'Stylohipparion' teeth are associated with the peculiar lower symphysis with reduced 13, described as Eurygnathohippus (van Hoepen, 1930). Since both Stylohipparion and Eurygnathohippus were erected on material from Uitzoek (South Africa) (0.7 Ma [Hendey, 1974]), it can safely be stated that Eurygnathohippus has priority over Stylohipparion for the advanced caballaid hipparions of Africa sporting a large ectostylid. The name Hypsohipparion was erected for an incipiently caballoid hipparion believed to lack ectostylids in the permanent lowers (Dietrich, 1942). However, the ectostylid has been shown to be variable in the type sample from Vogel River and in African hipparions in general, whether hipparionid, incipiently caballoid, or caballoid. The Old World caballoid hipparions, which comprise well-defined groups of species of which Eurygnathohippus is one, may in the future need their own subgeneric/ generic denomination. Joleaud's (1933) specific name ethiopicum is younger than the names given by Pomel (1897: ambiguum, massaesylium, and libycum) and van Hoepen (1930, 1932: steytleri, cornelianum, hipkini) to morphologically similar forms. The name ethiopicum has the advantage that it was bestowed on material from a rich and still productive unit, Shungura Fm. in Ethiopia. The earliest name given a caballoid hipparion with a large ectostylid, Pomel's (1897) libycum, is based on two lower teeth and a much-rolled distal MT III from the St Pierre Sandstone, Algeria. While the teeth seem 'large' (no measurements are given), the metapodium seems 'small' (judging from Pomel's Pl. II, Figs. 11 and 12), even discounting its wear. Lower cheek teeth from Lac Ichkeul and Ain Brimba (Tunisia) and Beni Foudda (=Ain Boucherit, Algeria), referred to as Stylohipparion libycum, are large with relatively small ectostylids (Arambourg, 1970); also, the two hipparion metapodials from Lac Ichkeul are large, particularly the MC III. The teeth from Ain Hanech (Algeria) and Ain Akrech (Morocco) seem smaller, with larger ectostylids (Arambourg, 1970).

562 A. Forsten The teeth from Uitzoek, described by van Hoepen (1930, 1932) as probably Hipparion steytleri and Stylohipparion hipkini, are 'small,' but the symphysis of Eurygnathohippus cornelianus seems large (Cooke, 1950:424). The name hipkini was synonymized with steytleri (Cooke, 1950; Dietrich, 1942), which has page priority over cornelianum if they represent the same species. Hipparion baardi Bone and Singer, originally described as a subspecies of H. albertense Hopwood, from Langebaanweg, Baard' s Quarry (Bone and Singer, 1965:389), has caballoid teeth but ?lacks an ectostylid (L 944, a M 1_2, looks like it has an ectostylid; Bone and Singer 1965: PL XII:E). The lower cheek teeth (measurements in Hendey, 1978: Tab. 2) correspond in size to the larger teeth from the Serengeti, but tend to be broader. More recently, Eisenmann (1976a, 1983) erected H. afarense with caballoid teeth and small ectostylids, from the Kada Hadar Member of the Hadar Fm., and H. hasumense with incipiently caballoid teeth and small ectostylids, from the lower Koobi Fora Fm. The two species appear rather similar in size and preserved morphology (H. afarensis is mainly characterized on its skull, unknown in H. hasumense). Both also resemble Dietrich's (1942) large hipparion from the Serengeti. Limb bones from these sites are large to very large. An interesting detail regarding the limb bones from the Denen Dora Member: two humeri (L 155-6AV and 155-lc) have a well-developed proximal internal tubercle, indicating that the stay-apparatus had evolved. Hermanson and MacFadden (1992) recently discussed the evolution and function of the stay-apparatus in the monodactyl horses. Other species names erected and used for African hipparions from about this period are albertense (Hopwood, 1926), considered a nomen vanum by Hooijer (1975), and namaquense Haughton (1932). While namaquense has caballoid lowers with a (weak) ectostylid, lowers from beds which yielded the holotype upper tooth of albertense are hipparionid and lack an ectostylid (Cooke and Coryndon, 1970). Hipparion albertense, based on inadequate material, dentally resembles H. turkanense Hooijer and Maglio from Lothagam, but since non-caballoid, these hipparions need not interest us here. The African caballoid hipparions comprise two morphological (and possibly ancestral-descendant, taxonomic) categories: - large, incipiently caballoid to caballoid forms with relatively weak or absent ectostylid. The incisors are not, or are barely, reduced. The earliest name for this category (?species) may be libycum Pomel (1897), with the possible synonyms H. baardi Bone and Singer (1965), afarense Eisenmann (1976b), and hasumense Eisenmann (1983). - mostly rather small, caballoid forms with pronounced ectostylids; 13

South Serengeti hipparions 563 (and C 1?) reduced even to absence. The earliest name for this category seems to be H. steytleri van Hoepen (1930), with the possible synonyms H. ethiopicum Joleaud (1933) and namaquense Haughton (1932). There is no sharp time difference between these categories and both occur in the lower Hadar Fm (from 3.35 Ma) and from Vogel River (from 3.7 Ma; although not from Laetoli). At some localities one or the other category dominated or occurred alone. Neither is there a sharp morphological difference (both categories have a broad and flat lower symphysis and more or less grooved incisors); an ancestor-descendan t relationship is possible (Eisenmann, 1985). The small category with pronounced ectostylids seems to have increased to dominance with time, except in the Hadar Fm. In the Notochoerus scotti zone of Koobi Fora, Shungura G, and Oldovai Bed I it occurred together with Equus, but appears to have been present already in Shungura E/F together with the larger form, according to Hooijer (1987) even in Shungura C. Its appearance and increase may, thus, have coincided with a climatic shift at this time (Vrba, 1992).

Origin Caballoid hipparions appeared in different parts of the world, that is, in Africa, Eurasia, and North America. Since the earliest caballoid hipparions seem to have appeared in the Hemphillian of North America, ca. 8.24.3 Ma (Lindsay et al., 1984), possibly even in the late Clarendonian (F:AM 111727 from Hans Johnson Quarry, Nebraska; MacFadden, 1984: Fig. 75, not Neohipparion trampasense), the Old World forms were originally believed to have been immigrants from the New World (Zhegallo, 1971, 1978). However, in spite of remarkably similar cranial and adaptive dental characters (e.g., lack of or faint development of a preorbital fossa, enhanced hypsodonty, straight upper tooth crowns, straight longitudinal enamel crests, elongated protocone, pronounced and grooved styles, strong pli caballinid, and often bifid M3 talonid), the Old World forms may have evolved autochthonously via forms with incipiently caballoid teeth, such as occur in Asia and Africa in the late Miocene (see also Eisenmann and Sondaar, 1989; MacFadden, 1984). Incipiently caballoid teeth, combined with crenellated enamel and occasionally well-developed ectostylids (e.g., BMNH M 26211), occur in the upper Dhok Pathan (Hasnot and Bhandar Bone Bed) of Pakistan and India, in the 'Pontian' of Hsinan and Lushi (Henan, coll. Paleontological museum, Uppsala, and V 57206, Institute of Vertebrate Paleontology and Paleoanthropology, Beijing), Wudu County (Gansu, coll. V 5654, Beijing), and from several localities from the Turolian to early Villafranchian in Shanxi (Nanhaohsia, Houxian, Nan Ho, and Xia Gou; Tedford and Galiano: mimeographed list of

564 A. Forsten Chinese fossil localities represented in the American Museum of Natural History; Huang et al., 1974; Tung et al., 1975), all in China, and in Africa mainly before ca. 4 Ma (see above). Hasnot has been dated ca. 5.8 Ma or 7.4-4.6 Ma (Flynn et al., 1990); Hsinan was dated~ Ma by Zhegallo (1978: Fig. 80), but has a typical 'Pontian' (i.e., Turolian) equivalent, fauna lacking Pliocene elements (Kurten, 1952). Caballoid teeth have been found from Kirgis Nur-2, levels 37-40 (Mongolia) (Zhegallo, 1978: Fig. 75), dated > 5.5 Ma (Pevzner and Vangengeim, 1986: Fig. 3), from the Tatrot (Pakistan and India), for instance, BMNH M 15396, dated< 5.1 Ma, resp. 2.5-2.9 Ma (Barry et al., 1982; Johnson et al., 1982), and from Esekartkan (Kazakhstan), dated 3.7 Ma (Vangengeim and Pevzner, 1991: Tab. p. 130). The earliest dated find of a caballoid hipparion in Europe is from Las Higueruelas (Spain), 3.8-3.5 Ma (Alberdi and Bonadonna, 1987), representing immigration from Asia. Already in the Turolian, occasional cheek teeth in otherwise conventional hipparionid dental samples may show caballoid traits, especially when little worn. Pronouncedly caballoid teeth typically characterize Old World Pliocene hipparions, however. In addition to the characters listed above, Eurasian and African caballaid hipparions show such features in common as large size and massive build, more or less grooved incisors, and angular metaconid-metastyl ids. The incipiently caballoid hipparions of Asia lacked some of the advanced characters of the skull, dentition, and feet later evolved by all or some of the lineages, such as loss of a preorbital fossa, long protocone, and extension and flattening of the volar V-scar of the proximal phalanx and the occasional presence of a trapezoidal facet on MC III. The limb characters are also lacking in the African caballoid hipparions and are only incipient in the Asiatic proboscidipparions , indicating branching-off of these lineages before the derived characters of the feet developed. Each Old World lineage then evolved its own particulars: in Africa an ectostylid in the lower cheek teeth, reduction of I3 (and C 1?), broadening and flattening of the lower symphysis, and certain cranial traits (Eisenmann, 1982), in Proboscidohipparion strong retraction of the nasal opening and reduction of the nasal bones (Sefve, 1927), and in the H. rocinantis-houfenense lineage modernization of the foot (Forsten, 1975; Vekua, 1972; Villalta, 1952). It has been suggested that Hipparion from the upper Dhok Pathan of India and Pakistan represented the ancestor of the Pliocene African caballoid hipparions via forms with conventional, hipparionid (or incipiently caballoid) cheek teeth (Forsten, 1968). Since incipiently caballoid hipparions seem to have occurred in the Old World both in Asia and Africa and at the same time (late Miocene), the characters in common between the later, mainly Pliocene, pronouncedly caballoid forms in the two areas could indicate a 'common evolutionary theme' without close phylogenetic relationships. The many advanced morphological similarities would

Hipparion

incip. caballoid libycum Hippaion

Eurygnathohippus

rocinantis

Proboscidipparion

16

11.12

w

,::

;.

FIGURE 4. Tentative phylogeny of the Old World caballoid hipparions. Character states: 1. Hipparionid lower cheek teeth with rounded-rounded angular loops of the double knot; ectostylid variable, mostly weak. 2. Slightly curved upper tooth crowns. 3. Preorbital fossa variable to absent; nasal opening variable, nasals long. 4. Distance from palatal rear border to vomer notch longer than distance from vomer notch to foramen magnum. 5. V-scar of proximal phalanx variable, mostly short. 6. No trapezoidal facet on MC III. 7. Increased angularity of loops of double knot. 8. Caballoid lower cheek teeth with loops of double knot angular; entoflexid U-shaped; upper crowns straight. 9. I variable to grooved. 10. Faint to no preorbital fossa; distance from palatal rear border to vomer notch equal to or shorter than distance from vomer notch to foramen magnum. 11. Ectostylid increased in frequency and size. 12. Flattened lower symphysis. 13. Ectostylid present, large. 14. Reduced 13 (and C1?). 15. V-scar of proximal phalanx variable to large; trapezoidal facet on MC III incipient to present. 16. V-scar large and flattened. 17. Nasal opening deeply retracted, nasals reduced

l

~!:;" .;·

11

g· '"I

CJ)

Ul

~

566 A. Forsten in that case have to be explained as parallelisms caused by similar ecological conditions to which these hipparions adapted independently in each area. The dental characters, especially the straight longitudinal enamel crests, are readily explained as ecological adaptations. The loss of the preorbital fossa would have to be similarly explained. If, as seems probable, the rims of the fossa served for attachment of the levator muscle of the upper lip (Studer, 1911; Zhegallo, 1978), which during contraction could swell into the fossa, a large and deep fossa, furnishing large attachment and expansion room for the muscle, could be associated with browsing and the need for mobile lips at selective feeding. In contrast a faint fossa could be associated with more unselective grazing as in the extant Equus. There is partial support for such an idea in that Turolian hipparions with a weak fossa tended to have a rather broad snout and low plication count in their upper cheek teeth, both features interpreted as adaptations to grazing (Antonius, 1919; Bunnell and Gillingham, 1985). In the caballoid hipparions, however, the snout tended to be relatively narrow and the plication count medium to high. The origin of the African caballoid hipparions is still unsolved, but Figure 4 shows a tentative phylogeny of the Old World forms. What were the ecological conditions which in the Old and New World alike gave rise to the caballoid hipparions, conditions which seem to have set in earlier in the New World? The Miocene saw a world-wide climatic change to a cooler and drier, more seasonal climate, but the exact timing in different areas is poorly known. The vegetation changed from forest and woodland to open woodland and from predominantly C 3 to C4 plants, possibly following a lowering of atmospheric CO2 (Ceding et al., 1993). These changes seem to be associated with the evolution of the caballoid hipparions.

LITERATURE CITED

Alberdi, M.T., and F.P. Bonadonna. 1987. Evaluation of lower and middle Villafranchian chronostratigraphy. Annales Instuti Geologici Hungarici 70:85-91. Antonius, 0. 1919. Untersuchungen iiber den phylogenetischen Zusammenhang zwischen Hipparion und Equus. Zeitschrift fiir induktive Abstammungs- und Vererbungslehre 20:273-295. Arambourg, C. 1970. Les vertebres du pleistocene de l' Afrique du Nord. Archives du Museum national d'Histoire Naturelle 10:126 pp. Barry, J.C., E.H. Lindsay, and L.L. Jacobs. 1982. A biostratigraphic zonation of the Middle and Upper Siwaliks of the Potwar Plateau of northern Pakistan. Palaeogeography, Palaeoclimatology, Palaeoecology 37:95-130. Bernor, L.R., K. Heissig, and H. Tobien. 1987. Early Pliocene Perissodactyla from Sahabi, Libya. Pp. 233-254 in N.T. Boaz, A. El-Arnauti, A.W. Gaziry, J. de Hein-

South Serengeti hipparions 567 zelin, and D. Dechaut (eds), Neogene Paleontology and Geology of Sahabi. Alan R. Liss Inc., NY. Bishop, W.W., C.R. Chapman, A Hill, and J.A Miller. 1971. Succession of Cainozoic vertebrate assemblages from the Northern Kenya Rift Valley. Nature 233:389-394. Boaz, N.T., A El-Arnauti, AW. Gaziry, J. de Heinzelin, and D. Dechaut (eds) 1987. Neogene Paleontology and Geology of Sahabi. Alan R. Liss Inc., NY, 401 pp. Bone, E.L., and R. Singer. 1965. Hipparion from Langebaanweg, Cape Province and a revision of the genus in Africa. Annals of the South African Museum 48:273397. Brown, F.H., and C.S. Feibel. 1991. Stratigraphy, depositional environment and palaeogeography of the Koobi Fora Formation. Pp. 1-30 in J.M. Harris (ed.), Koobi Fora Research Project, Vol. 3. Clarendon Press, Oxford. Bunnell, F.L., and M.P. Gillingham. 1985. Foraging behaviour: dynamics of dining out. Pp. 53-79 in R.J. Hudson and R.G. White (eds.), Bioenergetics of Wild Herbivores. CRC Press Inc., Florida, 314 pp. Cerling, Th.E., Yang Wang, and J. Quade. 1993. Global ecologic change in the late Miocene. Expansion of C4 ecosystems. Nature 361:344-345. Churcher, C.S., and M.L. Richardson. 1978. Equidae. Pp. 379-422 in V. Maglio and H.B.S. Cooke (eds), Evolution of African Mammals. Harvard University Press, Cambridge, MA Cooke, H.B.S. 1950. A critical revision of the Quaternary Perissodactyla of Southern Africa. Annals of the South African Museum 31:393-479. Cooke, H.B.S., and S.C. Coryndon. 1970. Pleistocene mammals from the Kaiso Formation and other related deposits in Uganda. Pp. 197-224 in L.S.B. Leakey and R.J.G. Savage (eds), Fossil Vertebrates of Africa, Vol. 2. Academic Press, London and NY. Dietrich, W.D. 1942. Altestquartare Saugetiere aus der siidlichen Serengeti, Deutsch-Ostafrika. Palaeontographica 94, Abt. A:43-133. Eisenmann, V. 1976a. Nouveaux cranes d'hipparions (Mammalia, Perissodactyla), Plio-Pleistocene d' Afrique orientale (Ethiopie et Kenya): Hipparion sp., Hipparion cf. ethiopicum, et Hipparion afarense nov. sp. Geobios 9:577-605. - 1976b. Equidae from the Shungura Formation. Pp. 225-233 in Y. Coppens, F.C. Howell, G.L. Isaac, and R.E.F. Leakey (eds), Earliest Man and Environment in the Lake Rudolf Basin. University of Chicago Press, Chicago and London. - 1977. Les hipparions africains: valeur et signification de quelques caracteres de jugales inferieures. Bulletin du Museum national d'Histoire Naturelle 438:6987. - 1979. Le genre Hipparion (Mammalia, Perissodactyla) et son interet biostratigraphique en Afrique. Bulletin du Societe Geologique de France 21:277-281. - 1982. La phylogenie des Hipparion (Mammalia, Perissodactyla) d' Afrique d'apres ces caracteres craniens. Proceedings of the Koninklije Nederlandse Akademie van Wetenschappen, Ser. B, 85:219-227. - 1983. Family Equidae. Pp. 156-214 in J.M. Harris (ed.), Koobi Fora Research

568

A. Forsten

Project, Vol. 2. The Fossil Ungulates: Proboscidea, Perissodactyla and Suidae. Clarendon Press, Oxford. - 1985. Les equides des gisements de la Vallee de l'Omo en Ethiopie. Cahiers de Paleontologie -Travaux de Paleontologie Est-Africaine: pp. 14--55. Eisenmann, V., and P.Y. Sondaar. 1989. Hipparions and the Mio-Pliocene boundary. Bollettino della Societa Paleontologica Italiana 28:217-226. Flynn, L.J., D. Pilbeam, L.L. Jacobs, J.C. Barry, A.K. Behrensmeyer, and J.W. Kappelman. 1990. The Siwaliks of Pakistan: time and faunas in a Miocene terrestrial setting. Journal of Geology 98:589-604. Forsten, A. 1968. Revision of the Palearctic Hipparion. Acta Zoologica Fennica 119:1-134. - 1975. Adaptive evolution of the equid foot. Zeitschrift fiir Saugetierkunde 40:304--308. Haileab, B., and P.H. Brown. 1992. Turkana Basin-Middle Awash Valley correlations and the age of the Sagantole and Hadar Formations. Journal of Human Evolution 22:453- 468. Haughton, S.H. 1932. The fossil Equidae of South Africa. Annals of the South African Museum 28:407-427. Hendey, Q.B. 1974. Fauna! dating of the late Cenozoic of Southern Africa, with special reference to the Carnivora. Quaternary Research 4:149-161. - 1978. The age of the fossils from Baard's Quarry, Langebaanweg, South Africa. Annals of the South African Museum 75:1-24. Hermanson, J.W., and B.J. MacFadden. 1992. Evolutionary and functional morphology of the shoulder region and stay-apparatus in fossil and extant horses (Equidae). Journal of Vertebrate Paleontology 12:377-386. Hill, A., S. Ward, and B. Brown. 1992. Anatomy and age of the Lothagam mandible. Journal of Human Evolution 22:439-451. Hooijer, D.A. 1975. Miocene to Pleistocene hipparions of Kenya, Tanzania and Ethiopia. Zoologische Verhandelingen 142:3-80. - 1976. The late Pliocene Equidae of Langebaanweg, Cape Province, South Africa. Zoologische Verhandelingen 148:3-39. - 1979. Hipparions of the Laetolil Beds, Tanzania. Zoologische Mededelingen 54:15-33. - 1987. Hipparion teeth from the Ndolanya Beds. Pp. 312-315 in M.D. Leakey and J.M. Harris (eds), Laetoli, a Pliocene site in northern Tanzania. Clarendon Press, Oxford. Hooijer, D.A., and Churcher, C.S. 1985. Perissodactyla of the Omo Group Deposits. Cahiers de paleontologie -Travaux de paleontologie Est-Africaine: 99-117. Hooijer, D.A., and Maglio, V. 1974. Hipparions from the late Miocene and Pliocene of northwestern Kenya. Zoologische Verhandelingen 134:3-34. Hopwood, AT. 1926. Fossil mammalia. Geological Survey of Uganda, Occasional Papers No. 2:13-36. Huang Wan-po, Tung Yung-sheng, Teng Wei-sen, Yuan Chein-chung, and Yii

South Serengeti hipparions 569 Cheng-chi. 1974. Pliocene stratigraphy of Hohsien, Shansi. Vertebrata Palasiatica 12:54-59. [Chinese, with English summary]. Johnson, N.M., N.D. Opdyke, G.D. Johnson, E.H. Lindsay, and R.A.K. Tahirkeli. 1982. Magnetic polarity stratigraphy and ages of Siwalik Group rocks of the Potwar Plateau, Pakistan. Palaeogeography, Palaeoclimatology, Palaeoecology 37:17-42. Joleaud, L. 1933. Un nouveau genre d'equide quaternaire de l'Omo (Abyssinie): Libyhipparion ethiopicum. Bulletin du Societe Geologique de France 3:7-23. Kurten, B. 1952. The Chinese Hipparion fauna. Commentatione s Biologicae 13(4):1-82. Leakey, L.S.B. 1965. Olduvai Gorge 1951-1961, Vol. 1. A Preliminary Report on the Geology and Fauna, pp. 1-109. Cambridge University Press, Cambridge. Lindsay, E.H., N.D. Opdyke, and N.M. Johnson. 1980. Pliocene dispersal of the horse Equus and the late Cenozoic mammalian dispersal events. Nature 287:135-138. - 1984. Blancan-Hemphillian land mammal ages and late Cenozoic mammal dispersal events. Annual Review of Earth and Planetary Sciences 12:445-488. MacFadden, B.J. 1984. Systematics and phylogeny of Hipparion, Neohipparion, Nannippus, and Cormohipparion (Mammalia, Equidae) from the Miocene and Pliocene of the New World. Bulletin of the American Museum of Natural History 179:1-195. Pevzner, M.A., and E.A. Vangengeim. 1986. Sootnoshenie kontinental'noj skaly pliozena zapadnoj Evropy so stratigrafizheskimi skalami Sredizemnomor 'ja i vostoznogo Paratetisa. Izvestija Akademija Nauk SSSR, Ser. geol., 3:3-17. [In Russian]. Pomel, A. 1897. Les equides. Carte geologique de l' Algerie, Paleontologie Monographies 12:1-44. Rensberger, J.M., A. Forsten, and M. Fortelius. 1984. Functional evolution of the cheek tooth pattern and chewing direction in Tertiary horses. Paleobiology 10:439- 452. Sefve, I. 1927. Die Hipparionen Nord-Chinas. Palaeontologica Sinica, Ser. C, 4:191. Studer, Th. 1911. Eine neue Equidenform aus dem Obermiodin von Samos. Verhandlungen Deutschen Zoologischen Gesellschaft 1911: 192-200. Teilhard de Chardin, P., and C.C. Young. 1931. Fossil mammals from the late Cenozoic of northern China. Palaeontologia Sinica, Ser. C, 12(1):1-66. Tung Young Sheng, Huang Wan-po, and Qiu Zhu ding. 1975. Hipparion fauna in Anlo, Hohsien, Shansi. Vertebrata Palasiatica 13:34-47. [In Chinese]. Vangengeim, E.A., and M.A. Pevzner. 1991. Villafrank SSSR: bio- i magnitostratigrafija. Paleogeografija i biostratigrafija miozena i antropogena, Moskva. Rotaprint, pp. 124-145. [In Russian]. van Hoepen, E.C. 1930. Fossiele perde van Cornelia, O.V.S. Paleontologiese Navorsing van die Nasionale Museum, Bloemfontein 11:13-24. [In Afrikaans].

570 A. Forsten - 1932. Die stamlyn van die sebras. Paleontologiese Navorsing van die Nasionale Museum, Bloemfontein, 11:25-37. [In Afrikaans]. Vekua, A.K. 1972. Kvabebskaja fauna Akcagyl'skih pozvonocnyh. Nauka, Moskva, 360 pp. [In Russian]. Villalta, J.F. de. 1952. Contribuci6n al conocimiento de la fauna de mamiferos f6siles del plioceno de Villarroya (Logrono). Boletin del Instituto Geologico y Minero de Espana 64:1-201. Vrba, E.S. 1987. A revision of the Bovini (Bovidae) and a preliminary revised checklist of Bovidae from Makapansgat. Palaeontologia africana 26:33-46. - 1992. Mammals as a key to evolutionary theory. Journal of Mammalogy 73:1-28. Zhegallo, V.I. 1971. Gippariony iz neogenovyh otloshenij zapadnoj Mongolii i Tuvy. Trudy sovmestnoj Sovetsko-Mongol'skoj naucno-issledovatel'skoj geologiceskoj ekspedicii 3:98-119. [In Russian]. - 1978. Gippariony Zentralnoj Azii. Trudy sovmestnoj naucno-issledovatel'skoj geologiceskoj ekspedicii 7:5-153. [In Russian] .

A fossil Budorcas (Mammalia, Bovidae) from Africa A.W. Gentry

Abstract A new species of the bovid genus Budorcas is described from the Pliocene of the Hadar Formation, Afar, Ethiopia. The only extant species in Budorcas is B. taxicolor, the takin of Tibet and China. The new African species had different specializations in the area of its horn insertions. Budorcas has long been placed in the tribe Ovibovini of subfamily Caprinae, but recent studies on fossil forms leave the position of Budorcas and the contents of the Ovibovini in doubt.

Introduction

Extant Ovibovini The tribe Ovibovini of the subfamily Caprinae is customarily thought to contain two extant species, the muskox Ovibos moschatus (Zimmermann, 1780) of the far north of North America and the takin Budorcas taxicolor (Hodgson, 1850) of Tibet and China. In the Pleistocene the muskox had a wide additional distribution across northern Eurasia. The two species differ in many characters, suggesting a long evolutionary history independent of one another. Ovibos moschatus has expanded horn bases on top of the skull and horn-cores strongly turned downwards at the sides of the skull, while Budorcas taxicolor has horn-cores somewhat compressed anteroventrally to posterodorsally, inserted high on the top of the skull, turning outwards immediately above their bases and then backwards in their distal parts. The torsion of B. taxicolor horn-cores is anticlockwise on the right side from the base up, while in 0. moschatus it is slightly clockwise. Juvenile muskoxen show that the horn insertions are originally

572 A.W. Gentry wide apart before the growth of the basal expansions, but takin horns are inserted closer together. There are also other differences: 0. moschatus, for example, has a rectangular rather than a triangular basioccipital, very small incisors, and a strong metastyle on M3, while Budorcas taxicolor has stronger longitudinal ridges behind the anterior tuberosities of the basioccipital, foramina ovalia opening in a more lateral plane, and mastoids small and exposed well within the occipital surface. A small number of characters are shared by muskox and takin, including a fairly large size and stocky build, horn-cores inserted well behind the orbits and strongly divergent, the front of the zygomatic arch having great depth below the back of the ventral rim of the orbits, an expanded area for the lachrymal, absence of ethmoidal fissure, no contact of the premaxilla with the nasals, and the vomer fused to the palate rather far posteriorly. These characters are found in only a few other bovids and are therefore potential defining characters of the Ovibovini. In both the cladistic and phenetic classifications of extant bovids proposed by Gentry (1992), 0vibos and Budorcas are linked together as a unit within the subfamily Caprinae. The large size and stocky build of the two species have suggested a relationship to cattle, buffalo, and bison (Hodgson, 1850), but the prevailing view has been of relationship to the Caprini (sheep and goats) within the subfamily Caprinae. Similarities which the two species share with Caprinae include the infraorbital foramen high and posterior, no lateral flanges at the front of the nasals, a short premolar row, P4 with the metaconid transverse crest taking its origin behind the protoconid, P4 with the metaconid and entoconid connected lingually, P4 with a paraconid-metaconid fusion to create a closed lingual wall anteriorly, molars without basal pillars, upper molars with strong mesostyles, and flat labial walls of the metacones.

Extinct Ovibovini The Ovibovini are evidently a relict group, and have an extensive fossil record in addition to that of 0vibos moschatus in the Pleistocene. Praeovibos priscus Staudinger, 1908 and other species of this genus are found in Eurasia and northernmost North America around the end of the lower Pleistocene and during the middle Pleistocene (McDonald et al., 1991). They are obviously closely related to 0vibos but have horn-core bases less expanded and more raised above the frontals' surface than in male 0. moschatus. Above their bases the horn-cores are also less tightly tucked in against the sides of the skull. Bootherium bombifrons (Harlan, 1825) and its synonyms, including Symbos cavifrons (Leidy, 1852a, b), is a North American middle and upper

A fossil Budorcas from Africa 573 Pleistocene species, again a close relative of Ovibos. It has been monographed by McDonald and Ray (1989). Its main differences from Ovibos are the fusion of the expanded horn bases on the top of the males' skulls, and the horn-cores less tucked in against the sides of the skull. Boopsis sinensis Teilhard de Chardin, 1936 from the Chinese middle Pleistocene could well be related to these extinct genera, although its horn-cores are less downturned. B. breviceps Teilhard de Chardin and Trassaert, 1938 from the Yushe Basin zone III is similar and possibly of later Pliocene age (Qiu, 1989: Tab. 1). Euceratherium collinum Furlong and Sinclair, 1904 existed throughout the Pleistocene in North America, with a more southerly and westerly distribution than Ovibos or Symbos (Carranza-Castaned a and Miller, 1987). Its horn-cores, which can be quite long, curve upwards and outwards from their insertions, then the distal parts curve a little downwards and finally forwards. Soergelia elisabethae Schaub, 1951 from the middle Pleistocene of Siissenborn, Germany, and possibly other Soergelia species in Eurasia back to the late lower Pleistocene (Moya Sola, 1987), have shorter horn-cores inserted more widely apart than in Euceratherium. North American representatives of Soergelia have been claimed (Kurten and Anderson, 1980). Other ovibovines occur in the Eurasian Pliocene to lower Pleistocene and have horn-cores generally less curved and without a downward component in their distal curvature. Among these are Pliotragus ardeus (Deperet, 1884), Megalovis latifrons Schaub, 1923 (see also see Schaub, 1928, 1944) and Hesperoceridas merlae (Villalta and Crusafont-Pairo, 1956). Despite some small differences, they are all rather similar and could well constitute a single species or evolving lineage. An earlier relative for them would be Kabulicornis ahmadi Heintz and Thomas, 1981 from the lower Pliocene of Afghanistan. The only fossils related to Budorcas taxicolor are the Chinese horn-cores and tentatively associated metacarpal of Budorcas teilhardi Young, 1948 (= Budorcas sp. of Teilhard de Chardin and Trassaert, 1938:84, Figs. 60-61, Pl. 4, Figs. la, lb) from the Yushe Basin zone III possibly of later Pliocene age, and perhaps pieces of the 'Ovibovine gen.indet.' of Teilhard de Chardin and Piveteau (1930:76) from Nihowan, of latest Pliocene age. Budorcas teilhardi differed from B. taxicolor in its longer horn-cores, which were less compressed and retained a forward curvature distally. Back in the upper Miocene (Vallesian and mainly Turolian) faunas of Europe and Asia there are a group of ovibovines centred on Urmiatherium polaki Rodler, 1889, which has some very specialized characters like a united frontals' boss anterior to extremely short horn-cores, and a strengthened basicranium. The much smaller Parurmiatherium rugosifrons Sickenberg, 1933 from Samos is extremely similar. Plesiaddax Schlosser,

574 A.W. Gentry 1903 is also similar but more primitive in its outwardly directed short horn-cores, non-united frontals' boss, and smaller facets on the back of the basioccipital. All these forms can well be regarded as one genus. Criotherium Major, 1891 and Tsaidamotherium Bohlin, 1935b are other related genera. Teilhard de Chardin and Trassaert (1938:91) have already noted that this group look more specialized than later ovibovines. They will be referred to again in the discussion. It was not until 1970 that an ovibovine was recognized in Africa. The later Pliocene Makapania broomi Wells and Cooke, 1956 from Makapansgat, South Africa, is very like Megalovis latifrons of the European Villafranchian (Gentry, 1970). It differs by having longer and more divergent horncores in which the limited degree of compression is anteroposterior rather than dorsoventral. The frontals between the horn insertions are also more elevated. Makapania broomi also probably occurred at the Sterkfontein Type Site(= Sterkfontein Member 4) as shown by teeth identified by Vrba (1976:48). Subsequently, other ovibovine remains were noted in Africa, such as those at Langebaanweg, South Africa (Gentry, 1980), and Olduvai Gorge, Tanzania (Gentry and Gentry, 1978:445, PL 41). The most striking find was a skull from the Hadar Formation, illustrated and briefly described by Gentry (1981). This skull is strongly different from the named fossils hitherto mentioned, and appears to be related to the takin. It will be named, described, and discussed in this paper as a new species of Budorcas.

Systematics Family Bovidae Gray 1821 Subfamily Caprinae Gray 1821 Budorcas Hodgson 1850 Type species: Budorcas taxicolor Hodgson 1850 Generic diagnosis: Large members of the Caprinae. Horn-cores moderately long, compressed anteroventrally to posterodorsally and with the anteroventral surface flatter than the posterodorsal one, an approach to an anterior keel, inserted close together and behind the orbits, directed transversely in their lower parts and then curving backwards, sinuses within frontals, skull wider at orbits than across the occipital, orbital rims strongly projecting, frontals elevated between horn bases, midfrontals' suture not complicated, fronto-parietal suture not indented forwards, temporal lines on cranial roof not approaching closely posteriorly, nasals transversely domed, anterior end of zygomatic arch deep below the back part of the orbits, back of M3 below front of orbit, large jugal, infraorbital foramen high and posterior, median indentation at back of palate passing behind level of lateral ones, basioccipital narrowed anteriorly giving it a

A fossil Budorcas from Africa 575 basically triangular shape, occipital surface facing backwards with little indication of a partially lateral-facing component on each side, no contact of mastoid with parietal. Moderately hypsodont cheek teeth, premolar row short, molars without basal pillars, upper molars with pointed lingual lobes but neither styles nor ribs pronounced. Remarks: The above diagnosis is largely an enumeration of skull characters which can be seen to be shared between Budorcas taxicolor and the new fossil species.

Budorcas churcheri, sp.nov. 1981 Ovibovini sp. aff. 'Bos' makapaani Broom. Gentry:17, Pis. 3, 4 Holotype: Skull AL136-5, housed in Addis Ababa, Ethiopia. Cast in Natural History Museum, London, M42014 (Fig. 1). Collected by the International Afar Research Expedition in 1973. Locality: Western side of the Afar Depression, Ethiopia, at about 11 °1o'N., 40°3o's. Horizon: Middle or upper Hadar Formation, later Pliocene. For information on the Hadar Formation, see Tiercelin (1986) and Walter and Aronson (1993). Name: The species is named for C.S. (Rufus) Churcher, whom this volume honours. Diagnosis: A species of Budorcas differing from B. taxicolor by more massive and more compressed horn-cores situated on more elevated frontals, which cause the cranial roof to be very steeply inclined and the braincase to widen from back to front. The large size and elevation of the horn-cores give them the appearance of being less posteriorly inserted than in B. taxicolor. Their course is also transverse from the very base instead of turning outwards immediately above vertically inserted bases. Nasals probably less shortened than in B. taxicolor, lachrymal probably smaller, preorbital fossa small and shallow as opposed to vestigial or absent in B. taxicolor, infraorbital foramen probably less high and less posterior, auditory bulla less inflated, a prominent central longitudinal groove on the basioccipital instead of a shallow and very narrow one, sharp longitudinally converging ridges in front of the anterior tuberosities of the basioccipital, mastoid larger. (For illustration of B. taxicolor see Fig. 2.) Remarks: The skull AL136-5 has preserved about half the left horn-core and the base of the right one. The top of the skull between the orbits has been destroyed. Part of the lateral edge of the left supraorbital pit appears to be still present. The zygomatic arch behind the orbits is missing. The middle of the nasals and the face is still present, but the muzzle is missing. Of the dentition the only surviving element is much of the left M 3 in early middle wear.

576 A.W. Gentry

C

>

0

-

CJ)

I!). t:i:,

.::

.:,.. 0

2

V,

::r0

FIGURE 1. Holotype skull, AL136-5, of Budorcas churcheri in (a) dorsal, (b) palatal, (c) left lateral, and (d) rear views (Included in upper corner of c is cross-section from base of left horn-core, with anterior direction to the left)

9

> [ n

$l,)

(JI

~

(JJ

~

> ~

C)

~

~

FIGURE 2. Skull of extant Budorcas taxicolor in dorsal, palatal, and left lateral views (scale = 50 mm)

A fossil Budorcas from Africa 579 The generic diagnosis lists the characters which Budorcas churcheri shares with the extant B. taxicolor, and from which it can be seen that the Hadar skull could not be placed in any other extant bovid genus. Budorcas churcheri is more specialized than B. taxicolor in the elevation of its horn insertions and the large size of the horn-cores, but B. taxicolor looks specialized with its anteriorly elevated nasals. Table 1 quantifies some of the different proportions of B. churcheri from B. taxicolor: larger horn-cores, wider horn insertions, longer cranial roof (linked with elevation of horn insertions), somewhat smaller cranial and occipital dimensions, and wider anterior tuberosities of basioccipital relative to posterior tuberosities. The last difference suggests that the anterior narrowness of the basioccipital in B. taxicolor may be secondary. In previous discussions of African fossil Ovibovini (Gentry, 1981; Gentry and Gentry, 1978) mention was made of 'Bos' makilpaani Broom (1937:510, figured). This South African species was found in deposits of unknown age in a cave 'near Makapansgat about ten miles from Pietpotgieters Rust' which later became known as Buffalo Cave (Cooke, 1952:33). The frontlet and horn-cores have never been adequately described. The likely position of the sutures on the frontlet indicates that the convex edge of the horn-cores is anterior or anterodorsal and not posterior as in Bos. 'Bos' makilpaani and a similar horn-core at Olduvai Gorge (Gentry and Gentry, 1978, Pl. 41) are smaller and shorter than the horn-cores on the Hadar skull. Their most probable taxonomic position is as a second, and probably later, African species of Budorcas. More fossils or further study of the Buffalo Cave material are needed to add substance to this suggestion. The Langebaanweg ovibovine remains described in Gentry (1980) are likely to be the earliest in Africa and have correspondingly primitive characters in that the horn-cores are little divergent, their insertions are not posterior relative to the orbits, and they have a narrower basioccipital. The steep inclination of the cranial roof is like Budorcas churcheri rather than Makilpania broomi.

Discussion Classification of Budorcas The genera and species which have generally been included in the tribe Ovibovini arrange themselves into three informal systematic groups: 1 Ovibos moschatus and its large circle of Plio-Pleistocene relatives mentioned in the introduction: Praeovibos, Bootherium, Boopsis, Euceratherium, Soergelia, Pliotragus, Megalovis, Hesperoceridas, Makilpania, and Kabulicornis; 2 Budorcas taxicolor and a few congeneric fossil remains; 3 Urmiatherium, Criotherium, and other genera of upper Miocene age.

580 A.W. Gentry TABLE 1 Measurements (in mm) on the holotype skull of Budorcas churcheri and on adult skulls of both and unknown sexes of B. taxicolor Hadar Budorcas taxicolor Form. No. skull Range

Mean

(A)

(D)

(B)

(C)

(A) x 100 (D)

(%)

Maximum (anterodorsal to posteroventral) diameter of horn-core at its base

86.4

45.2~5.4

2

55.3

156

Diameter at right angle to above

60.2

35.5-41.4

2

38.5

156

Minimum width across lateral surfaces of horn pedicels

136.0

98.3-125.9

12

115.9

117

Skull width across posterior side of orbits

166.4

158.6-198.8

12

179.5

93

Width across lateral edge of supraorbital pits

c.82.0

72.7-97.6

12

84.1

98

Length from back of frontals to top of occipital

71.3

47.7~3.5

11

55.8

128

Minimum width across braincase

83.4

86.9-102.5

11

95.0

88

104.2

112.7-141.8

10

130.0

80

Occipital height from top of foramen magnum to occipital crest

44.9

54.9-80.5

8

67.8

66

Minimum width across temporal lines on cranial roof

44.3

48.7-58.2

10

51.9

85

Distance from rear occlusal edge of M3 to back of occipital condyle

154.0

143.0-184.4

7

162.3

95

Width across anterior tuberosities of basioccipital

27.7

23.7-34.2

6

27.2

102

Width across posterior tuberosities of basioccipital

42.5

41.6-59.0

8

51.3

83

Occlusal length left M3

24.7

25.5-35.4

10

28.2

88

Occlusal width left M

17.0

14.1-21.4

10

17.2

99

Skull width across mastoids immediately behind external auditory meatus

3

It was noted in the introduction that Budorcas and Ovibos show a large number of differences but not a great many similarities. Bouvrain and Bonis (1984) gave an account of the Vallesian (upper Miocene) European ovibovine Mesembriacerus Bouvrain, 1975 and its relationships. They laid emphasis on specialized basicranial-cervical characters to define the tribe Ovibovini. These characters, which are developed in the Upper Miocene ovibovines and carried to their culmina-

A fossil Budorcas from Africa 581 tion in Urmiatherium, are enlargement and strengthening of bone at the base of the paraoccipital processes, additional facets on the back of the posterior tuberosities of the basioccipital, enlarged occipital condyles lying within the plane of the occipital surface instead of standing proud of it, and modifications to atlas and axis vertebrae (Sickenberg, 1933; Bohlin, 1935a). Similar strengthening of the skull and vertebrae in the area around the occipital condyles and basioccipital can also be seen in Ovibos and some of the Plio-Pleistocene forms like Soergelia (Kahlke 1969: Pl. 24, Figs. 2, 4) and suggests a tribal relationship with the upper Miocene ovibovines. Being found in no other bovids, these characters ought to be ideal for a neat definition of Ovibovini. Unfortunately they do not occur in Budorcas and so, on this basis, Budorcas would cease to be an ovibovine (Fig. 3, top). Also Bouvrain and Bonis (1984:220) themselves note the non-occurrence of the basicranial-cervical specializations in some of the Plio-Pleistocene ovibovines like Makapania and Megalovis. The exclusion of Budorcas from the Ovibovini goes against longstanding classificatory practice with extant bovids, recently corroborated by Gentry (1992). Not only are the basicranial-cervical synapomorphies not found in Budorcas, but other synapomorphies which Budorcas does share with Ovibos are not always found in the upper Miocene ovibovines. For example, Mesembriacerus is primitive in retaining both an ethmoidal fissure and a large shallow preorbital fossa, both Mesembriacerus and Criotherium still have basal pillars on the lower molars, and Criotherium has kept non-diverging horn-cores (Fig. 3, top). Finally, some of the tooth characters shared between living ovibovines and caprines, mentioned near the beginning of this paper, have to become parallels once Mesembriacerus in particular is admitted to the Ovibovini. The approach of Bouvrain and Bonis also contrasts with Gentry (1971), who had widened the scope of Ovibovini to include the rather primitive Turolian genus Palaeoryx Gaudry 1861a, b (this paper, Fig. 3). Possible synapomorphies of Palaeoryx with Ovibovini were rather few: large size, slight face lengthening, horn-core insertions beginning to move behind the orbits, horn-cores slightly divergent, some deepening of the front part of the zygomatic arch, and basioccipital being wide anteriorly. Even in this modest list, the divergent horn-cores are not shared with Criotherium. The teeth of Palaeoryx are even more primitive than Mesembriacerus, as for example in their longer premolar rows. Some zoologists such as Erdbrink (1988:154) continue to include Palaeoryx and related genera in the Hippotragini, but adequate discussion of this important viewpoint is beyond the scope of the present paper. A third possibility is that Urmiatherium, Ovibos, and their allies may be related to the specialized middle Miocene bovid Hypsodontus, which may

582 A.W. Gentry llllllIAfflJfllIUII

OYIJJOB

alIOfflJfllIUII

JIESllllalIACEllUS

B,C,D

B,C

OVIBOVINI

OYIJJOB AND IIPPIR NIOCE1111 GDlllA

.BIJDOllCAS

PALAJ!IJllU

C,E

CAPRINI

E,F

OVIBOVINI

OYIJJOB AND

UPl'IIII NIOCEIIII GBDIIA

OTBDl BOVIDAE

PAL.UDllYJ'

BlJ1JOllCAS

CAPRINI

A

CAPRINAE

B,:r I

R,I, J

FIGURE 3. Relationships of Budorcas to Ovibovini. Top: according to Bouvrain and Bonis (1984); middle: according to Gentry (1971); bottom: if Ovibos, Urmiatherium, and their allies are related to Hypsodontus. Advanced characters mentioned in the text and shown on these diagrams: A, basicranial and cervical specializations; B, no basal pillars [ectostylids] on lower molars; C, divergent horn-cores; D, no ethmoidal fissure and reduction of preorbital fossa; E, basioccipital wide anteriorly; F, shortened premolar rows; G, large size, lengthened face, horn-cores inserted behind orbits, zygomatic arch deepened anteriorly; H, hypsodonty; I, lower molars with flattened lingual walls; J, horn-cores with torsion

A fossil Budorcas from Africa 583 itself be remote from other bovids and not a member of the Caprinae (Fig. 3, bottom). (See Gentry, 1990, and Kohler, 1987, for further details of Hypsodontus.) Species in Hypsodontus had forwardly curving horn-cores with clockwise torsion on the right side. Their cheek teeth were hypsodont, premolar rows short, P2's sometimes absent, molars without basal pillars, and lower molars with flattened lingual walls. Some of these precociously advanced caprine-like tooth characters, particularly evident in H. pronaticornis Kohler (1987) and far in advance of other contemporaneous middle Miocene bovids, suggest a possible relationship to the upper Miocene ovibovines. The large size of some Hypsodontus species, and some horn-core features (Solounias, 1990:436), also offer support to the hypothesis. Such a relationship would remove Urmiatherium and its allies, including Ovibos, from affinity with Caprinae or Budorcas. Other possibilities exist, given the extensive parallelism in bovid evolution (Gentry, 1992); for example, that the basicranial-cervical specializations emphasized by Bouvrain and Bonis (1984) could have evolved in Ovibos in parallel to the more extreme state of the upper Miocene Urmiatherium group. A full investigation is required on these major questions concerning the relationship of Budorcas to Ovibos, the contents of the tribe Ovibovini, the relationships of that tribe to Caprini, and the relevance of Palaeoryx and Hypsodontus to these relationships. However, the purpose of this paper is to record the presence of a fossil species of Budorcas in Africa, so no further investigations will be made here. Perhaps Ovibovini on the Bouvrain and Bonis (1984) definition and Caprini are two monophyletic groups and various other upper Miocene genera like Palaeoryx are related to the ancestry of both (a different view of Palaeoryx than that of Gentry, 1971). Budorcas might then be a descendant of one of these upper Miocene associates of Ovibovini and Caprini. Its linkage with Ovibos in analyses of living forms would have to be explained by the absence of any survivors from its cladistically nearer relatives.

Zoogeography and ecology of Budorcas Budorcas churcheri has zoogeographic importance in substantiating the former occurrence of Budorcas in Africa. Whatever the taxonomic relationship of 'Bos' makapaani may be, and whatever the relationship of Budorcas to Ovibos, both these latter genera are relicts of once more-widespread lineages. It is difficult to postulate a likely ecological pattern of life for extinct relatives of Budorcas and Ovibos. The muskox, living under an extreme climatic regime in the Arctic, can offer few clues. The takin dwells in the wooded river valleys of mountainous country and shows altitudinal migration between seasonal feeding grounds (Neas and Hoff-

584 A.W. Gentry mann, 1987). Its anteriorly elevated nasals may be an adaptation for breathing in very cold winter conditions, similar to the more specialized Saiga tatarica. They are unlikely to yield information relevant to previous species of the genus in Africa. The leaping ability of the takin, shared with the muskox and the Caprinae generally, suggests that B. churcheri too would have had a preference for hilly, rocky terrain.

Acknowledgments Long ago Y. Coppens and D.C. Johanson invited me to work on Bovidae from the Afar, and B.T. Gray gave much practical help. I thank the Keeper of Palaeontology in London for use of the facilities of his department, and the referees of this paper for helpful suggestions.

LITERATURE CITED

Bohlin, B. 1935a. Cavicornier der Hipparion-Fauna Nord Chinas. Palaeontologia Sinica, Pekin C, 9, fasc. 4:1-166. - 1935b. Tsaidamotherium hedini, n.g., n.sp. Hyllningsskift Sven Hedin [an anniversary volume issued by Geografiska Annaler], Stockholm: 66--74. Bouvrain, G. 1975. Un nouveau bovid du Vallesian de Macedoine (Grece). Compte Rendu des Seances de l' Academie des Sciences Paris, serie D, 280:13571359. Bouvrain, G., and L. de Bonis. 1984. Le genre Mesembriacerus (Bovidae, Artiodactyla, Mammalia): un ovibovine primitif du Vallesian (Miocene superieur) de Macedoine (Grece). Palaeovertebrata Montpellier, 14:201-223. Broom, R. 1937. Notices of a few more new fossil mammals from the caves of the Transvaal. Annals and Magazine of Natural History London 10, 20:509-514. Carranza-Castaneda, 0., and W.E. Miller. 1987. Rediscovered type specimens and other important published Pleistocene mammalian fossils from Central Mexico. Journal of Vertebrate Paleontology 7:335-341. Cooke, H.B.S. 1952. Quaternary events in South Africa. Pp. 26--36 in L.S.B. Leakey, (ed.), Proceedings of the Pan-african Congress of Prehistory 1947. Oxford University Press, Oxford. Deperet, C. 1884. Nouvelles etudes sur les Ruminants pliocenes et quaternaires d' Auvergne. Bulletin de la Societe Geologique de France, Paris 3, 12:247-284. Erdbrink, D.P.B. 1988. Protoryx from three localities east of Maragheh, N.W. Iran. Proceedings, Koninklijke Nederlandse Akademie Wetenschappen, Amsterdam B 91:101-159. Furlong, E.L., and W.J. Sinclair. 1904. Preliminary description of Euceratherium collinum. University of California Publications, American Archaeology and Ethnology 2:18.

A fossil Budorcas from Africa 585 Gaudry, A. 1861a. Resultats des fouilles entreprises en Grece sous les auspices de l' Academie. Compte Rendu des Seances de l' Academie des Sciences Paris 52:238-241. - 1861b. Note sur les Antilopes trouvees a Pikermi (Grece). Bulletin de la Societe Geologique de France, Paris 2, 18:388--400. Gentry, A.W. 1970. Revised classification for Makapania broomi Wells and Cooke (Bovidae, Mammalia). Palaeontologia africana, Johannesburg 13:63-67. - 1971. The earliest goats and other antelopes from the Samos Hipparion fauna. Bulletin of the British Museum (Natural History) Geology, London 20:229-296. - 1980. Fossil Bovidae (Mammalia) from Langebaanweg, South Africa. Annals South African Museum, Cape Town 79:213-337. - 1981. Notes on Bovidae (Mammalia) from the Hadar Formation, and from Amado and Geraru, Ethiopia. Kirtlandia, Cleveland, Ohio 33:1-30. - 1990. Ruminant artiodactyls of Pasalar, Turkey. Journal of Human Evolution, London 19:529-550. - 1992. The subfamilies and tribes of Bovidae. Mammal Review, London 22:1-32. Gentry, A.W., and A. Gentry. 1978. Fossil Bovidae (Mammalia) of Olduvai Gorge, Tanzania. Bulletin of the British Museum (Natural History) Geology, London 29:289--446; 30:1-83. Gray, J.E. 1821. On the natural arrangement of vertebrose animals. London Medical Repository 15:296-310. Harlan, R. 1825. Fauna Americana: being a description of the mammiferous animals inhabiting North America. Philadelphia, Anthony Finley, 318 pp. Heintz, E., and H . Thomas. 1981. Un nouveau Bovide, Kabulicornis ahmadi gen. nov., sp. nov., dans le gisement pliocene de Pul-e Charkhi, bassin de Kabul, Afghanistan. Bulletin du Museum National d'Histoire Naturelle, Paris 4 ser, 3, C:31--44. Hodgson, B.H. 1850. On the takin of the eastern Himalaya: Budorcas taxicolor mihi. Journal of the Asiatic Society of Bengal, Calcutta 19:65-75. Kahlke, H.-D. 1969. Die Soergelia-Reste aus den Kiesen von Siissenborn bei Weimar. Palaontologische Abhandlungen Berlin A. Palaozoologie 3:531-545. Kohler, M. 1987. Boviden des turkischen Miozans (Kanozoikum und Braunkohlen der Turkei 28). Paleontologia i Evolucio, Sabadell 21:133-246. Kurten, B., and E. Anderson. 1980. Pleistocene mammals of North America. Columbia University Press, New York, 442 pp. Leidy, J. 1852a. [Remarks on two crania of extinct species of ox.] Proceedings, Academy of Natural Sciences of Philadelphia 6, 3:71. - 1852b. Memoir on the extinct species of American ox. Smithsonian Contributions to Knowledge, Washington 5, 3:1-20. McDonald, J.N., and C.E. Ray. 1989. The autochthonous North American musk oxen Bootherium, Symbos, and Gidleya (Mammalia: Artiodactyla: Bovidae). Smithsonian Contributions to Paleobiology, Washington 66:1-77. McDonald, J.N., C.E. Ray, and C.R. Harington. 1991. Taxonomy and zoogeogra-

586 A.W. Gentry phy of the musk ox genus Praeovibos Staudinger, 1908. Pp. 285-314, in J.R. Purdue, W.E. Klippel, and B.W. Styles (eds), Beamers, Bobwhites, and Blue-Points: Tributes to the Career of Paul W. Parmalee. Illinois State Museum Scientific Papers, vol. 23. Springfield, IL. Major, C.I.F. 1891 . Considerations nouvelles sur la Faune des Vertebres du Miocene superieur dans l'Ile de Samos. Compte Rendu des Seances de I' Academie des Sciences Paris 113:608-610. Moya Sola, S. 1987. Los bovidos (Artiodactyla, Mammalia) del yacimiento del Pleistoceno inferior de Venta Micena (Orce, Granada, Espana). Paleontologia i Evolucio, Memoria Especial, Sabadell 1:181-235. Neas, J.F., and R.S. Hoffmann. 1987. Budorcas taxicolor. Mammalian Species, New York 277:1-7. Qiu, Z.X. 1989. The Chinese Neogene mammalian biochronology: its correlation with the European Neogene mammalian zonation. Pp. 527-556, in E.H. Lindsay, V. Fahlbusch, and P. Mein (eds), European Neogene Mammal Chronology. Plenum Press, New York and London. Rodler, A. 1889. Ober Urmiatherium polaki n.g., n.sp. Denkschriften der Kaiserlichen Akademie der Wissenschaften, Wien 56:315-322. Schaub, S. 1923. Neue und wenig bekannte Cavicornier von Seneze. Eclogae Geologicae Helvetiae, Basie 18:281-295. - 1928. Die Antilopen des Toskanischen Oberpliodins. Eclogae Geologicae Helvetiae, Basie 21 :260-266. - 1944. Die oberpliocaene Siiugetierfauna von Seneze (Haute-Loire) und ihre verbreitungsgeschichtlich e Stellung. Eclogae Geologicae Helvetiae, Basie 36:270289. - 1951. Soergelia n.gen., ein Caprine aus dem thiiringischen Altpleistocaen. Eclogae Geologicae Helvetiae, Basie 44:375-381. Schlosser, M. 1903. Die fossilen Siiugetiere Chinas, nebst einer Odontographie der recenten Antilopen. Abhandlungen der Bayerischen Akademie der Wissenschaften, Miinchen 22:1-221. Sickenberg, 0. 1933. Parurmiatherium rugosifrons, ein neuer Bovide aus dem Unterplioziin von Samos. Palaeobiologica, Wien 5:81-102. Solounias, N . 1990. A new hypothesis uniting Bose/aphus and Tetracerus. Annales Musei Goulandris 8:425-439. Staudinger, W. 1908. Praeovibos priscus, nov. gen. et nov. sp., ein Vertreter einer Ovibos nahestehenden Gattung aus dem Pleistociin Thiiringens. Centralblatt fur Mineralogie, Geologie und Paliiontologie, Stuttgart 16:481-502. Teilhard de Chardin, P. 1936. Fossil mammals from locality 9 of Choukoutien. Palaeontologia Sinica, Pekin C, 7, fasc. 4:1-61. Teilhard de Chardin, P., and J. Piveteau. 1930. Les mammiferes fossiles de Nihowan (Chine) . Annales de Paleontologie, Paris 19:1-134. Teilhard de Chardin, P., and M. Trassaert. 1938. Cavicornia of south-eastern Shansi. Palaeontologia Sinica, Pekin NS C, 6:1-98.

A fossil Budorcas from Africa 587 Tiercelin, J.J. 1986. The Pliocene Hadar Formation, Afar depression of Ethiopia. Pp. 221-240, in L.E. Frostick, R.W. Renaut, I. Reid, and J.J. Tiercelin (eds), Sedimentation in the African Rifts. Geological Society Special Publications 25. Blackwell Scientific Publications, Oxford. Villalta, J.F., and M. Crusafont-Pairo. 1956. Un nuevo Ovicaprino en la fauna villafranquiense de Villaroya (Logrono). Actes 4me Congres Internationale Association Quaternaire Recherches, Rome: 426-432. Vrba, E.S. 1976. The fossil Bovidae of Sterkfontein, Swartkrans and Kromdraai. Transvaal Museum Memoirs 21:1-166. Walter, R.C., and J.L. Aronson. 1993. Age and source of the Sidi Hakoma Tuff, Hadar Formation, Ethiopia. Journal of Human Evolution 25:229-240. Wells, L.H., and H.B.S. Cooke. 1956. Fossil Bovidae from the Limeworks Quarry, Makapansgat, Potgietersrus. Palaeontologia africana 4:1-55. Young, C.C. 1948. Budorcas, a new element in the proto-historic Anyang fauna of China. American Journal of Science 246:157-164. Zimmermann, E.A.W. von. 1780. Geographische Geschichte des Menschen, und der vierfiissigen Thiere, Band 2. Leipzig, 432 pp.

Basicranial anatomy of the giant viverrid from 'E' Quarry, Langebaanweg, South Africa

Robert M. Hunt, Jr

Abstract Crania of large viverrid carnivorans (Order Carnivora, Family Viverridae) from the Upper Siwaliks of southern Asia and from Langebaanweg in South Africa indicate the presence of two evolving lineages of giant hypercarnivorous civets (Viverra leakeyi Petter, Vishnuictis durandi Pilgrim) in the Plio-Pleistocene of the Old World. Recent discovery of a cranium from Langebaanweg in 1979 and restudy of the Siwalik skulls in the British Museum demonstrate that both lines possessed typical viverrid basicranial and bullar anatomy. A third lineage of these large civets is documented by earlier discoveries at Olduvai (Tanzania) of teeth of a large hypocarnivorous viverrid (Pseudocivetta ingens Petter). Thus, PlioPleistocene faunas of Asia and Africa incorporated diverse large viverrids no longer viable in present environments yet once important as predators and omnivores in the late Cenozoic of these regions.

Introduction In the late Cenozoic of Africa and Asia, giant viverrid carnivorans are known from dentitions and partial skulls, some discovered in the midnineteenth century yet never adequately described or illustrated. The basicranial anatomy of these animals has not been reviewed to determine if the auditory bulla and middle-ear structure are preserved, and to what extent these features provide insight into phylogenetic relationships. Giant viverrids probably attained body weights of 20 to 40 kg and have been primarily reported from East and South Africa, the Siwaliks of Asia, and the Zhoukoudian travertine of China. Rare European dental remains of these large carnivores also were described by Kretzoi and Fejfar (1982)

Basicranial anatomy of 'E' Quarry viverrid 589 from Villafranca d'Asti (Italy) and Ivanovce 1 (Czechoslovakia). They were first made known by fossil crania from the Siwalik Hills given by Falconer and Cautley to the British Museum and mentioned by Falconer in 1868. Subsequently, Bose (1880), Lydekker (1884), and Pilgrim (1932) described these crania. Very little information on basicranial structure was provided in these studies; referral to Viverridae rested on dental traits. The Siwalik crania remained the only evidence of these giant viverrids until the 1960s, when research in the East African rift led to the discovery of Pliocene viverrids from Laetoli in Tanzania. Petter (1963) reported both herpestids and viverrids from Laetoli, and described the partial upper dentition of a single individual of a large viverrid which she named Viverra leakeyi. This was the first record of a species of Viverra from Africa, the genus today restricted to Asia. Since Petter's original description of the Laetoli viverrid, V. leakeyi has been found in the Omo Basin (Howell and Petter, 1979; Petter, 1987; Petter and Howell, 1977) of East Africa and at Langebaanweg in South Africa (Hendey, 1974). None of the African discoveries included basicranial material. The East African fossils were primarily fragmentary dental remains, whereas the Langebaanweg sample was more representative but lacked a complete cranium. Hendey (1974) described from 'E' Quarry at Langebaanweg fragmentary cranial, postcranial, and dental remains of a large civet-like camivoran. He referred these fossils to Viverra leakeyi, noting similarities to Petter's holotype, the upper teeth from Laetoli. Since Hendey's study of this material, new specimens of the giant Langebaanweg viverrid have been discovered, including a magnificent complete cranium and associated mandibles (South African Museum LBW 1979 PQ-L51590) found in 1979. My colleague C.S. Churcher and his student Grant Hurlburt are describing this cranium. I thank them for the invitation to discuss the basicranial anatomy of the animal and to comment on its relevance to the Siwalik crania in the Falconer collection of the British Museum (Natural History), London, which I have recently been able to examine through the courtesy of Dr Alan Gentry.

The Langebaanweg viverrid cranium The cranium and mandible (Fig. 1) are nearly complete, and reveal a gracile, dolichocephalic camivoran much different from the skull form of the Siwalik giant viverrids. The cranium in fact is remarkably similar to that of living species of Viverra, particularly in its elongate rostrum and sloping forehead. The size of the animal is indicated by its basilar length, 20.2 cm. Despite breakage and bone missing in the orbital region, the skull includes nearly the entire dentition and a well-preserved basi-

590 R.M. Hunt, Jr

AGURE 1. Cranium and associated mandible of the giant viverrid Viverra leakeyi from 'E' Quarry, Langebaanweg, South Africa (South African Museum no. PQL51590) (basilar length of skull, 20.2 cm; scale in mm; the basicranial anatomy of this skull is shown in Fig. 3)

cranium. It was found at 'E' Quarry in the Pelletal Phosphorite Member of the Varswater Formation (Hendey, 1974; Tankard, 1974), considered Pliocene in age. The dentition of this carnivoran is sectorial, having a well-developed shearing upper carnassial, a triangular Ml, reduced and smaller M2 (no M3), and laterally compressed yet robust premolars. The canines are conical, slightly recurved, stabbing teeth. Based upon the cranium and limited postcranial material, a restoration of the Langebaanweg Viverra has been created to suggest a carnivorous predator able to actively pursue prey in a variety of environments (Fig. 2). The nature of the dentition is important, distinguishing the Langebaanweg carnivoran from the large crushing-toothed viverrid Pseudocivetta ingens, described by Petter (1965, 1967) from Olduvai Gorge in Tanzania, and later found in the Omo Basin (Petter and Howell, 1977). The strong contrast between the dentitions of these two animals demonstrates their separate identity and thus the presence of at least two large viverrids in Africa in the Plio-Pleistocene.

Basicranium of the Langebaanweg viverrid Although only a fragment of the auditory bulla is present, the basicra-

Basicranial anatomy of 'E' Quarry viverrid 591

FIGURE 2. Restoration of the giant African viverrid Viverra leakeyi based on the fossil material from Langebaanweg, South Africa

nium is otherwise nearly intact (Fig. 3A, B). The middle-ear cavity, petrosal, and surrounding bones of the basicranium are unequivocally viverrid in form, and confirm Hendey's (1974) inference as to the viverrid affinity of this material. Although the bulla is almost entirely lost from the skull, the form of the squamosal, alisphenoid, basioccipital, and exoccipital demonstrate that the viverrid bipartite bulla was fully developed as in living Viverridae. The bulla was enormous and therefore susceptible to breakage. Only part of the ectotympanic is preserved in the right auditory region, and is similar in form to the modem viverrid ectotympanic that houses the anterior chamber of the bulla. The anterior crus of the ectotympanic rests in a deep pit in the squamosal, and the posterior crus although broken away must have contacted the post-tympanic process of the squamosal. Only limited development of an external auditory bony meatus occurs in the fossil just as in living viverrids. Posterior to the ectotympanic fragment, the expanded paroccipital process of the exoccipital bone and the conspicuous impressions on the margins of the basioccipital indicate an expanded viverrid caudal entotympanic of considerable volume as seen in Viverra and Civettictis. The height of the paroccipital process when compared to the height of the ectotympanic fragment shows that the caudal entotympanic chamber undoubtedly extended forward over the smaller subordinate anterior chamber of the bulla, producing a bulla configuration as in living viverrids.

592 R.M. Hunt, Jr

FIGURE 3. A. Basicranium and left auditory region of the giant civet Viverra leakeyi in ventrolateral view (same individual as Figure 1). The auditory bulla is missing from the left side, revealing the petrosal, middle ear, and surrounding bones. The petrosal is of the aeluroid type, with a prominent ventral promontorial process (V) and a lateral groove for the internal carotid artery (ic). The paroccipital process (P) is expanded to receive the enlarged caudal entotympanic chamber of the auditory bulla. Note the expanded exoccipital (ex) to accommodate the inflated caudal entotympanic element. The lateral margin of the basioccipital also is impressed by the medial wall of the caudal entotympanic, creating the small depression (*) posterior to the ventral promontorial process. A prominent flange (f) of the petrosal extends in a posterior direction from the promontorium to the posterior lacerate foramen (F). The condyloid foramen (cf) is situated behind the posterior lacerate foramen. The middle lacerate foramen (L) occurs just dorsal to the basioccipital (B0)-basisphenoid (BS) suture. B. Ventral view of the basicranium (same individual as Figure 3A), partial ectotympanic (T) present in the anterior part of the right auditory region. A prominent pit (er) for the anterior crus of the ectotympanicis evident in the squamosal when the bulla is removed in the left auditory region. The mastoid process (M) is only moderately developed. (Other abbreviations as in Figure 3A; scale in mm; stereopairs)

Basicranial anatomy of 'E' Quarry viverrid 593 A typical aeluroid petrosal of viverrid type is present in the Langebaanweg cranium (Fig. 3A). The left petrosal is most complete and displays a prominent ventral promontorial process diagnostic for aeluroid Carnivora (compare Hunt, 1989: Fig. IA). Only the tip of the process has been lost; this tip protruded ventrad in life between basioccipital and the bulla's medial wall as in living civets. The process lies just posterior to the basisphenoid-basioccipital suture and is grooved on its lateral face for the internal carotid artery. Posterior to the promontorium the petrosal is extended backward as a broad flange that forms the anterior border of the posterior lacerate foramen. In some modern viverrids such as Civettictis the flange is present but is not as developed, leaving a space behind the petrosal that is completely filled in the Langebaanweg skull by bone. The condyloid foramen is recessed within the space for the posterior lacerate foramen, and the postglenoid foramen is lost, indicating that venous drainage has been redirected through the posterior lacerate pathway (internal jugular vein) as in living viverrids and other aeluroids. The mastoid process is only moderately developed. In all respects, the auditory region and the inferred form of the bulla are typical of living Viverridae.

The Siwalik viverrid crania The Siwalik crania were originally described by Lydekker (1884) and Pilgrim (1932), although Falconer included mention of some specimens in his palaeontological memoir of 1868. Here I consider as giant Siwalik viverrids only the lineage named Vishnuictis by Pilgrim (1932:101-108). Pilgrim included in this lineage a cranium and associated mandible of moderate size from the Dhok Pathan 'stage' which he made the genoholotype, Vishnuictis salmontanus. He also referred to Vishnuictis the largest viverrid crania known from the Siwalik Hills, presumed to have come from the Pinjor 'stage' of Plio-Pleistocene age, placing them in a new species, V. durandi (Fig. 4). Both species of Vishnuictis are represented by crania extremely similar to each other in form and different in this respect from the Langebaanweg cranium. As noted by Pilgrim, the Siwalik Vishnuictis have expanded frontal regions, marked postorbital constrictions, relatively short rostra, with deep maxillae. They are much more like the living Civettictis in skull form than like Viverra. It seems probable that the two species are a lineage, V. salmontanus the earlier form, V. durandi the descendant. Two crania from the Pinjor Siwaliks were initially described as belonging to Vishnuictis durandi: the holotype, BM(NH) M1338 (Fig. 4B), and a referred rostrum without basicranium, BM(NH) M37150 (Fig. 4C). The holotype preserves the basicranium and auditory bulla, although damaged to a degree, and retains a partial dentition, whereas the referred

594 R.M. Hunt, Jr

A

C

FIGURE 4. Crania in dorsal view of the large Siwalik civet Vishnuictis durandi Pilgrim (A, M37131; B, M1338; C, M37150) and the smaller Siwalik viverrid Viverra bakeri Bose (D, M40183) conserved in the palaeontological collections of the British Museum (Natural History), London. Basicranial anatomy and teeth are preserved in M1338, the holotype cranium of Vishnuictis durandi. Note the expanded frontal region and short, deep rostrum in Vishnuictis. (Scale in mm)

rostrum lacks teeth. As Pilgrim suggested, they surely belong to a single species; in fact, the holotype lacks the rostrum which is contributed by the referred M37150, allowing a complete reconstruction of the cranium to be made. During my study of the Siwalik collection of Falconer and Cautley at the British Museum, I encountered a third partial cranium (Fig. 4A), BM(NH) M37131, that also can be referred to V. durandi. There is no locality data with this skull, but the preservation is similar to the other two crania. It is a posterior cranium complete as far forward as the frontal region, but lacks the rostrum and teeth entirely. These three crania have estimated basilar lengths of 18-20 cm, and thus are similar in length to the Langebaanweg viverrid. There are, however, significant differences that preclude referral to the same species, and indicate that these are separate lineages. The Siwalik crania of Vishnuictis have markedly expanded frontal regions, so much so that the frontal expansion slopes downward to the smaller braincase. In the Langebaanweg skull the frontal region is not expanded and the forehead slopes upward to the braincase (Fig. 1). The Langebaanweg viverrid also has a long narrow rostrum and shallow maxillae, in contrast to the robust massive rostra and deep facial region of

Basicranial anatomy of 'E' Quarry viverrid

595

the Siwalik viverrids. In the dentition both Siwalik Vishnuictis and Langebaanweg Viverra are hypercarnivorous, sectorial forms in the construction of the carnassial and molars. But the holotype of V. durandi, which is the only cranium to preserve the teeth of this species, shows a more sectorial carnassial-molar region than seen in the Langebaanweg species. One might suppose that the dentition and skull form of Vishnuictis durandi are derived features evolved from an animal like the Pliocene Langebaanweg viverrid, but this is unlikely given the existence in the Dhok Pathan sediments of the smaller yet already specialized Vishnuictis salmontanus. Consequently, it seems that the Siwalik Vishnuictis and Langebaanweg Viverra lines are distinct and separate, and indicate the presence of two large hypercarnivorous viverrid lineages in the Plio-Pleistocene of the Old World. In two of the Siwalik crania (M1338, M37131) the basicranium is preserved. The left auditory bulla survives in the holotype skull, although somewhat damaged. It displays a typical viverrid bulla configuration in which an expanded caudal entotympanic chamber anteriorly overgrows the smaller ectotympanic element. In M37131 the bulla is also preserved on the left side and is similar in what remains to that of M1338. Thus, the Siwalik giant viverrids possess typical viverrid auditory regions and, despite the differences in skull form and teeth between the Siwalik and Langebaanweg lineages, the auditory bulla pattern is similar and definitely of the viverrid type. It was possible to examine another smaller viverrid from the Siwalik Hills during my visit to the British Museum. The holotype skull (Fig. 4D) was presumed by Pilgrim (1932) to have also come from the Pinjor 'stage' of the Upper Siwaliks (Siwalik Hills). This is the cranium of Viverra bakeri, BM(NH) M40183, originally described by Bose in 1880, and later discussed by Lydekker (1884). Basilar length of skull is about 14 cm, much smaller than in Vishnuictis durandi. Lydekker, Pilgrim, and later workers agreed that Viverra bakeri is a different lineage relative to Vishnuictis, and this is supported by the form of the upper teeth, particularly the molars which show additional cuspation and some enlargement, indicating a possible trend towards a more crushing dentition. Matthew (1929) suggested that the teeth of V. bakeri are adapting to a less carnivorous diet and I would concur, regarding this species as a small viverrid evolving towards hypocarnivory but still retaining a moderately sectorial dentition. Its basicranium is too poorly preserved to determine its structure, but the general form and the dentition indicate this is also a viverrid.

Acknowledgments I thank Rufus Churcher for his kind invitation to study the basicranial

596 R.M. Hunt,Jr anatomy of the Langebaanweg viverrid, and also extend my appreciation to Dr Margaret Avery of the South African Museum, Cape Town, for loan of the cranium during completion of this study. My appreciation to Kathlyn Stewart and Kevin Seymour for the opportunity to contribute to this volume. Pauline Denham produced the reconstruction of Viverra leakeyi from Langebaanweg that appears as Figure 2. I also am grateful for the kind hospitality provided during my work at the British Museum (Natural History) by Dr Alan Gentry and his associates. My thanks to F. Clark Howell and Lars Werdelin for their useful reviews of this study. LITERATURE CITED

Bose, P.N. 1880. On undescribed fossil Carnivora from the Siwalik Hills. Quarterly Journal of the Geological Society, London 36:119-136. Falconer, H. 1868. Palaeontological Memoirs, C. Murchison (ed.), Vol. I: 1-590; Vol. II: 1-675, London. Hendey, Q.B. 1974. The Late Cenozoic Carnivora of the Southwestern Cape Province. Annals of the South African Museum 63:1-369. Howell, F.C., and G. Petter. 1979. Diversification et affinites des carnivores pliocenes du groupe de l'Omo et de la formation d'Hadar (Ethiopie). Bulletin de la Societe Geologique de France 21:289-293. Hunt, R.M., Jr. 1989. Evolution of the aeluroid Carnivora: significance of the ventral promontorial process of the petrosal, and the origin of basicranial patterns in the living families. American Museum Novitates 2930:1-32. Kretzoi, M., and 0. Fejfar. 1982. Viverriden (Carnivora, Mammalia) im europaischen Altpleistozan. Zeitschrift fiir geologische Wissenschaft, Berlin 10:979-995. Lydekker, R. 1884. Siwalik and Narbada Carnivora. Palaeontologia Indica 2:178351. Matthew, W.D. 1929. Critical observations on Siwalik mammals. Bulletin of the American Museum of Natural History 56:437-560. Petter, G. 1963. Etudes de quelques Viverrides (Mammiferes, Carnivores) du pleistocene inferieur du Tanganyika (Afrique orientale). Bulletin de la Societe Geologique de France 5:265-274. - 1965. Mammalian fauna other than Bovidae. Order Carnivora, Family Viverridae. Pp. 22-23 in L.S.B. Leakey (ed.), Olduvai Gorge 1951-1961, Vol. I. Cambridge University Press, Cambridge. - 1967. Petits carnivores Villafranchiens du Bed I d'Oldoway (Tanzanie). Colloques Internationaux du Centre National de la Recherche Scientifique no. 163:529-538. - 1987. Small carnivores (Viverridae, Mustelidae, Canidae) from Laetoli. Pp. 194234 in M.D. Leakey and J.M. Harris (eds), Laetoli: A Pliocene Site in Northern Tanzania. Oxford University Press, Oxford.

Basicranial anatomy of 'E' Quarry viverrid 597 Petter, G., and F.C. Howell. 1977. Diversification des Civettes (Camivora, Viverridae) dans les gisements pleistocenes de l'Omo. Comptes Rendus de l'Academie des Sciences, Paris 284:283-286. Pilgrim, G.E. 1932. The Fossil Camivora of India. Palaeontologia Indica 18:1-232. Tankard, A.J. 1974. Petrology and origin of the phosphorite and aluminium phosphate rock of the Langebaanweg-Saldanha area, Southwestern Cape Province. Annals of the South African Museum 65:217-249.

The identification of Equus skulls to species, with particular reference to the craniometric and systematic affinities of the extinct South African quagga

Richard G. Klein, Kathryn Cruz-Uribe

Abstract The quagga was a partially striped zebra that once ranged widely throughout the interior of South Africa. Overhunting extinguished freeranging populations by about 1870, and the last-known captive specimen died in 1883. Some skins and skeletal parts survive in museums, but there are arguably no more than four or five well-authenticated adult skulls. To these some authorities have proposed adding five subfossil skulls that were recovered on the farm Koffiefontein, western Orange Free State, in the early 1900s. However, discriminant analysis employing 15 craniometric ratios that individually tend to differ significantly among Equus species suggests that the Koffiefontein skulls are not a coherent group and that they are more likely to represent donkeys and/ or horse than quagga. Since biomolecular data link the quagga closely to the common plains (or Burchell's) zebra, a priori the Koffiefontein skulls could have been expected to have been most like plains zebra specimens. However, the same analyses that suggest the Koffiefontein skulls are more like those of donkey or horse suggest that three known quagga skulls (in the Natural History Museum, London, the Yale Peabody Museum, and the Academy of Natural Sciences of Philadelphia) are more like those of mountain zebra or horse. We conclude that the craniometric and systematic affinities of the quagga remain open to investigation.

Introduction The taxonomy and phylogeny of the equids are disputed, but most authorities agree that there is but one extant genus, Equus, and that this contains at least six living species: (1) Equus caballus (horses, wild and

Identification of Equus skulls to species 599 domestic), (2) Equus asinus (donkeys, asses, or burros, wild and domestic), (3) Equus hemionus (onagers, kulans, and kiangs, sometimes known collectively as Asiatic wild asses or 'half-asses'), (4) Equus burchelli (plains or Burchell's zebras), (5) Equus zebra (mountain zebras), and (6) Equus grevyi (Grevy's zebras.) Equus evolved in North America roughly 3.5 million years ago and thrived there until the end of the last glaciation, between 12,000 and 10,000 years ago (Grayson, 1989; Kurten and Anderson, 1980; MacFadden 1992; Martin, 1984). Thereafter, it was extinct in the Americas until domestic forms were introduced by Europeans, beginning in the late fifteenth century. In historic times then, wild forms of Equus were limited to Eurasia and Africa. Wild horses, often called E. przewalskii to distinguish them from domestic E. caballus, with which they are completely interfertile, occurred in the steppes of central Asia and eastern Europe, from Mongolia westwards to Poland and Hungary. Wild asses, sometimes called E. africanus to distinguish them from domestic E. asinus, inhabited dry, mainly hilly areas in an arc extending from the northern Sudan in the southeast to eastern Morocco in the northwest. Onagers, kiangs, and kulans were native to the deserts and dry steppes of southwestern and central Asia, from Syria and Iraq in the southwest to western Manchuria in the northeast. Plains zebras were common in the grasslands and savannas of eastern and southern Africa, from the southern Sudan, southern Ethiopia, and southern Somalia in the north to the Transvaal and northern Cape Province of South Africa in the south. Mountain zebras inhabited hilly areas of southwestern Africa, in an arc roughly paralleling the coast from southern Angola through Namibia to the south-central Cape Province of South Africa. Finally, Grevy' s zebras were restricted to arid regions of northern Kenya, southern Somalia, and southern and eastern Ethiopia (Corbet, 1978; Ellerman et al., 1953; Honacki et al., 1982). Where different wild species overlapped, as, for example, plains zebra and Grevy's zebra in northern Kenya and plains zebra and mountain zebra in Namibia and Angola - hybridization did not occur (Ansell, 1971). The different species of Equus do interbreed in captivity, mostly with human encouragement (Clutton-Brock, 1981, 1992), but the hybrids are almost invariably sterile, and the species distinctions therefore remain valid. Many authorities (for example, Ansell, 1971; Ellerman et al., 1953; Meester et al., 1986; and Roberts, 1951) recognize yet a seventh historic species, Equus quagga, known colloquially as the quagga. Large herds of quagga occurred in the Orange Free State and in the Cape Province of South Africa between the Orange River and the mountains of the Cape Fold Belt until the middle of the last century (Skead, 1980; Skinner and Smithers, 1990; Fig. 1). Indiscriminate shooting for sport, meat, and hides extinguished free-ranging populations sometime in the 1870s (Harper,

600 R.G. Klein, K. Cruz-Uribe 1945), and the last known individual died in 1883 in the Amsterdam Zoo (Comrie-Greig, 1985; Rau, 1974). Unlike the well-known plains zebras further north, quagga were not completely covered by alternating black and white stripes. Instead, the contrast between stripes and interstripes diminished progressively from the shoulder backwards, resulting in rear quarters that were a more or less uniform brown or rusty colour (Rau, 1974, 1978, 1986; Skinner and Smithers, 1990). This distinctive pelage, other less well-documented physical characters, and possible range overlap with fully striped zebras in the Orange Free State underlie the placement of the quagga in its own species. Based on a detailed analysis of morphological character states in fossil and living equids, Bennett (1980) concluded that the quagga not only merited species status, but that it was actually more closely related to the horse than to other zebras. However, Rau (1974, 1978, 1983, 1986) argued that the singularity of quagga pelage has been overdrawn and that more southerly plains zebras progressively approached the quagga both in muting of the stripes and in the tendency for striping to disappear on the rear quarters. The southernmost population of undoubted plains zebra (E. burchelli burchelli in conventional terms), now also extinct, broadly resembled the quagga and may have graded smoothly into it. Rau concluded that the quagga was probably only the most extreme colour variant along a north-south gradient within plains zebras. Groves (1985) accepted Rau's analysis and suggested that the putative overlap between quagga and typical plains zebra in the Orange Free State was based on a misreading of Harris (1840). Since the quagga was named first, Rau and Groves joined Antonius (1951), Klingel (1969), Mohr (1964), Pocock (1904), and other authorities in lumping plains zebras with quagga in Equus quagga. Close kinship between plains zebra and quagga is strongly supported by similarities in mitochondrial DNA (Harley, 1988; Higuchi et al., 1984, 1987) and in proteins (Lowenstein, 1985; Lowenstein and Ryder, 1985) extracted from plains zebra and surviving quagga tissues. In theory, skeletons might also illuminate quagga relationships, but this prospect is limited by the tendency of European collectors to use the term 'quagga' for both quagga and ordinary plains zebra, followed by flawed museum curation of many supposed quagga specimens (Rau, 1974, 1978). Based on a detailed morphometric analysis of 13 reputed adult quagga skulls, Eisenmann (1980) concluded that the quagga and plains zebra were very similar, but her result is questionable, because her quagga sample may have come mainly from plains zebra (Thackeray, 1988). Arguably, there are no more than four or five adult quagga skulls with unalloyed pedigrees, and our analysis of three of these below suggests that they differ significantly from their plains-zebra homologues. The rarity of authentic quagga skulls lends special significance to fossil

s::

(1)

:::,

::r.

::t:

()

ll>

::r. 0

:::,

g_

l:T1 ~ ;:: ;::

"'

(J)

~

:= (J)

0 quagga

FIGURE 1. Approximate historic ranges of 'typical' plains zebra, quagga, and mountain zebra in southern Africa. The base map was redrawn after Comrie-Greig (1983); the quagga after Rau (1986); and the 'typical' plains zebra and mountain zebra after Skinner and Smithers (1990). The range distributions are based on information in Comrie-Greig (1983), Groves (1985), Lloyd (1984), and Skinner and Smithers (1990).

(J)

"O (1)

()

if (J)

0 °' .....

602 R.G. Klein, K. Cruz-Uribe or subfossil specimens from within the quagga' s former range. Almost certainly, the most interesting and important specimens are a group of five subfossil adult skulls and two adult mandibles that W. Fowler collected in the early 1900s on the farm Koffiefontein in the eastern Orange Free State and subsequently donated to the McGregor Museum in Kimberley. Cooke (1943, 1948, 1950) believed the Koffiefontein specimens came from quagga and used them to draw cranial and dental distinctions between quagga, plains zebra, and mountain zebra. The immediate purpose of the research we report here was to determine if the Koffiefontein skulls come from quagga as opposed to donkey and/ or domestic horse, both of which occurred widely within the quagga's range by the latter half of the last century. Our more general purpose was to contribute to the methodology for identifying equid skulls from cranial metrics.

Materials and methods The species of Equus are notoriously difficult to identify from bones alone, and authorities frequently disagree on which bony parts are most diagnostic. Bennett (1980), for example, states that the cranium is much more useful than the postcranium, while Groves and Willoughby (1981) (also Groves 1986) found just the reverse. Palaeontologists, who must work mainly with teeth, have often proposed interspecific differences in dental occlusal patterns (see, for example, Cooke [1943] or Churcher and Richardson [1978] for dental differences among African species), but Groves and Willoughby (1981) argue that nearly all supposed dental differences are unreliable or invalid. The difficulty is that interspecific occlusal variation is largely swamped by intraspecific variation, much of which reflects individual differences in age and wear. Our experience with the teeth of African species echoes that of Groves and Willoughby, and we believe that cranial form is far more dependable for species separation. However, analyses by Eisenmann (1980, 1986), Gentry (1975), Groves and Willoughby (1981), Lundholm (1951), Smuts and Penzhorn (1988), Winans (1989), and others all emphasize the high degree of cranial variability among Equus species, particularly when only one feature or dimension is considered at a time. Inevitably then, compelling assignments to species must rely both on multiple criteria and on multivariate statistical methods. Eisenmann (1980, 1986) has shown the power of multiple, comprehensive, clearly defined measurements for distinguishing the adult skulls of various Equus species, and we have employed her suggested measurement scheme here. It entails 33 cranial and 9 mandibular dimensions that can be variously measured by tape, sliding calipers, or digital calipers. Eisenmann found that the cranial measurements provided the greatest discriminating power, and our experience repeats Eisenmann's, although

Identification of Equus skulls to species 603 it is possible that the mandible would prove more useful given a more extensive set of dimensions, including, for example, those employed by Winans (1989) in her multivariate analysis of North American fossil horses. In the meanwhile, we ignore the mandible in what follows. Regarding cranial dimensions, our study differs from Eisenmann's in the museum comparative samples on which it is based, and in two methodological respects. First, we have relied exclusively on ratios between measurements rather than on raw measurements. Second, we have employed a boxplot format based on dispersion around the median for univariate comparisons among species and on discriminant analysis for multivariate comparisons. We emphasize ratios because our samples show clearly that absolute size can vary as much among populations of a single species as it does among different species. This is true not only for horses, donkeys, and onagers where significant intraspecific size variation is well known (Groves, 1986), but also for plains zebras and mountain zebras, which are more commonly assumed to be relatively uniform in size. Figure 2, for example, shows that basal skull length(= Eisenmann's cranial dimension 1) may differ as much between geographically distinct populations within each zebra species as it does between the species. Following Groves and Willoughby (1981), we have used mainly ratios between various skull dimensions and basal length. These are in effect indices that standardize all measurements on basal length. We have also investigated ratios that reflect the shape of particular cranial features, such as the orbit, the choanae, and the muzzle. We tested many possible ratios or standardized measurements for their power to discriminate, particularly among plains zebras, mountain zebras, horses, and donkeys, which are the species that history or geography suggests could have provided the subfossil Koffiefontein skulls. To conserve space, we present results only for those ratios (Table 1) that individually seemed to discriminate best and that not surprisingly, then, allowed nearly perfect discrimination, when used in combination. All of them involve measurements that can be made on each Koffiefontein skull, and they include most of the dimensions that Eisenmann (1980, 1986) found to produce maximum species separation. Unlike the parametric statistics utilized by Eisenmann, which assume symmetry around the mean, the boxplots we use reflect the actual shape of the underlying data distribution. They are thus more useful for detecting outliers or possible sample heterogeneity. As presented in Figures 217, the boxplots were produced by the Macintosh computer program Data Desk in a format described by Velleman (1992). Specifically, in each boxplot, the vertical line near the centre is the median, the vertical lines at the ends mark the range of more or less continuous data, the hachured rectangle is the 95% confidence interval for the median, and the open box

basal skull length (1)

~

""

:;,::i

0

'=~J5:&""

basal skull length (cm) o

37.5

45.o

"

.___,__--'

52.5 -

60.0 ,______

o

median

56

51.45 51.55 5.49 35.0 66.0

31

41.10 41 .96 4.31 36.5 53.9

northern plains zebra

41

45.00 45.07 1.42 41.0 48.5

I southern plains zebra

103

46 .20 46.46 1.85 43.0 56.4

3

45 .60 45.73 1.60 44.2 47.4

horse

___

K'offiefontein i

t--il

■ t--t

O

I o

• I

t---1 a-,

1H

I

o

I

mean

Namibian mountain zebra

29

48.50 48.61

45.0

52.5

min.

max.

1.23 46.5 51.1

Cape mountain zebra

38

46.1 0 45.98 1.50 41.1 49.4

Grevy's zebra

11

54.70 54.90 0.97 52.9 56.5

onager

22

44.85 44.52 2.39 36.6 49.0

I

37.5

s. d.

...........______....._........................"'-""'-------

quagga

0

~

N I

donkey

•

....~

11)

;s

60.0

FIGURE 2. Boxplots and descriptive statistics for basal skull length in Equus samples examined by authors. 1

n 2

2 '"I

S.:

11)

Identification of Equus skulls to species 605 encloses the middle half of the data between the 25th and 75th percentiles. Circles and starbursts indicate extreme values (points that are far removed from the main body of data). In cases where the 95% confidence intervals do not overlap, the associated medians differ significantly in the conventional statistical sense. In contrast to the multivariate methods (mainly correspondence analysis) employed by Eisenmann, the discriminant analysis method on which we rely is not designed to detect likely species groupings from multiple measurements. Instead it assumes that the species assignments in museum comparative collections are valid and then calculates the multivariate mean for each species. Once this calculation is made, it becomes possible to estimate the probability that a particular skull derives from one species or another, based on multiple measurements (or in our case, ratios). This is a particularly desirable feature when the goal is to establish the most likely specific identity of fossil or subfossil skulls like those from Koffiefontein. Since the discriminant results are based on species means, it is perfectly possible for an analysis to misclassify some of the individual specimens on which it is grounded, and the success of an analysis is often judged by how few specimens it misidentifies. The analyses we report here misidentify very few known specimens, but they are not infallible, and we stress that the identifications of unknown specimens and subsequent inferences must be regarded only as probabilities. The analyses were produced with the Macintosh program JMP (SAS Institute, 1989).

Results Using the boxplot format discussed in the last section, Figures 3-17 illustrate interspecific variation in those craniometric ratios that visually exhibit greatest promise for separating horses, donkeys, plains zebras, and mountain zebras. The relevant dimensions are shown in cranial views redrawn after Eisenmann (1986). The numbered position of each dimension in her scheme is presented in parentheses in each view. Table 2 shows that multivariate discriminant analysis of the same ratios allows nearly perfect separation among the known horse, donkey, mountain zebra, and plains zebra skulls included in our samples. With regard to the 305 skulls for which all the ratios could be calculated, the analysis misidentifies only seven (2.3%) - one known plains zebra that it mistakenly assigns to donkey, one known mountain zebra and three known plains zebras that it mistakenly assigns to horse, one known plains zebra that it mistakenly assigns to mountain zebra, and one known horse that it mistakenly assigns to plains zebra. [Text continues on p. 623.]

°' °' 0

::,::,

0 minimum breadth of muzzle/ breadth of muzzle posterior to 13's

0.500

0.625

0.750

0.875

__:'.':L.~.::: ..~ ,.f;;.t-.}l:':;·m!!,

1.000

I horse

IIIU-------1

*

-- -0-;i.1...,.;...;~~,

1----[IJ---l L

r--Cil---1 I H 1

0.500

0

~ 0.625

0.750

0.875

I

'

s. d.

min.

(C

~

n

max.

a

0.689 0.691 0.056 0.563 0.811

-

N I

0.879 0.874 0.059 0.745 1.000

5

0.754 0.764 0.068 0.679 ·0.833

172

0.642 0.648 0.050 0.421 0.821

3

0.576 0.562 0.084 0.472 0.638

mountain zebra 69

0.716 0.709 0.044 0.609 0.810

Grevy's zebra

13

0.712 0.729 0.048 0.672 0.850

onager

21

0.700 0.709 0.051 0.621 0.810

donkey

plains zebra quagga

I

58

median mean

?.

34

--Koffiefontein

. Cl)

N

~

'\. minimum breadth (17b)

·~

'T'~

-

.!!''""""

-

'

{

1.000

FIGURE 3. Boxplots and descriptive statistics for ratio between minimum muzzle breadth and breadth immediately posterior to upper third incisors in Equus samples reported in this paper

C '"1

~

(C

~

diastema/length of the premolar row 1.00

0 .75

diastema (6)

.. ·

1....-...l length of the premolar row (7)

1.50

1.25

N 0

0

1

58

horse donkey

32

_ _ _ _ ,......_..

median mean

s. d.

min.

max.

1.137 1.135 0.176 0.750 1.721 0.909 0.966 0.153 0.725 1.326 ...

-'!I'

5" 1.081 1.013 0.197 0.791 1.247 0

0

0

0

I

~ t----H

I

0.

~ ::r. ::;?) (')

Ill

169

1.150 1.168 0.130 0.920 1.686

3

1.063 0.998 0.150 0.826 1.105

1

mountain zebra 68

1.060 1.071 0.130 0.809 1.427

_£1

1

Grevy's zebra

13

1.111 1.111 0.089 0.914 1.258

onager

20

0.821 0.835 1.127 0.570 1.025

1

plains zebra quagga

0

-

-· g a. :;::: :;:::

{/)

[

--

{ /)

0.75

1.00

1.25

1.50

0

{/)

'"Cl

~

FIGURE 4. Boxplots and descriptive statistics for ratio between length of upper diastema and occlusal length of upper premolar row in Equus samples reported in this paper

1f 0 °' '.:J

0 °' 00

basal skull length (1)

:;,;:,

0 ~ Jr length of the cheek tooth row/basal skull length 0 .300

0 .325

moo!

I I

I

I

■I

I

., j

I

I

0.300

I

I

0.325

!

I

~ 0

N 0

* 0

I

0.350

. __..:::--

14

0.375

-

!

0

0

0 .350

I length of the cheek tooth row (8)

median mean

s. d.

min.

max.

1

horse

56

0.333 0.335 0.020 0.285 0.389

1

donkey

30

0.345 0.344 0.018 0.293 0.397

Koffiefontein

5

0.341 0.336 0.018 0.323 0.371 !:."' ~'f

plains zebra

168

0.322 0.322 0.013 0.283 0.352

3

0.340 0.333 0.020 0.310 0.348

mountain zebra 66

0.326 0.326 0.014 0.284 0.363

Grevy's zebra

11

0.322 0.323 0.007 0.313 0.335

onager

21

0.357 0.355 0.020 0.307 0.394

quagga

0.375

FIGURE 5. Boxplots and descriptive statistics for ratio between length of cheek tooth row and basal skull length in Equus samples reported in this paper

~

n ~

c

~

/length of the premolar row (7)

length of the molar row (7b}'\. ~.◄

..,

length of the premolar row/length of the molar row

*

•1

I

1.4

1.3

1.2

1.1

1.0

- - -

donkey

* Io I

' 1.0

ol

0

: I ■

I

.

I plains zebra

I

I

j

l

!

t I

'

1.1

1.2

170

-- _,.. 1.029 1.344 w

-~

1.183 1.221 - 0.072 1.157 1;328 1.241 1.239 0.048

Ill

;::r•

§

0 .....

I mountain zebra

68

1.181 1.177 0.050 1.014 1.268

Gravy's zebra

13

1.267 1.264 0.047 1.207 1.333

20

1.283 1.288 0.049 1.209 1.397

I

1.3

~

;::r.

::n n

1.194 1.178 0.027 1.147 1.194

onager !

.

......

0..

3

quagga

~ I

max.

min.

1.227 1.223 0.050 1.127 1.323

31

5

s. d.

1.192 1.181 0.056 0.932 1.284

58

I horse

I

median mean

N

1.4

FIGURE 6. Boxplots and descriptive statistics for ratio between occlusal lengths of premolar and molar rows in Equus samples reported in this paper

tT1 ~ ;:: ;:: en

~

==

g

',:j (I)

n

if °' 0 \0

,..... °' 0

basal skull length (1)

!

:;r::1

0 ~ .....

breadth across the occipital condyles (29)

,?

breadth across the occipital condyles/ basal skull length

0.1500

0 .1625

0 .1750

~

0.1875

0 00

N

min.

max.

32 0.1682 0.1661

quagga

~~

- 1 - t - ...

-- ....,_

5 0.1695 0.1680

168 0.1688 0.1691

0.0093 0.1415 0.1791

-0.0089 0.1580 0.0081

0.1771

0.1476 0.1923

3 0.1647 0.1628 0.0073 0.1547 0.1690

mountain zebra 68 0.1603 0.1612 0.0079 0.1437 0.1839

0

0 .1750

s. d.

donkey

plains zebra

0.1625

mean

54 0.1758 0.1745 0.0128 0.1447 0.1994

Koffiefontein

0.1500

median

horse

Grevy's zebra

11 0.1467 0.1465 0.0042 0.1410 0.1521

onager

22 0.1557 0.1571

0.0082 0.1444 0.1719

0 .1875

FIGURE 7. Boxplots and descriptive statistics for ratio between breadth of skull across the occipital condyles and basal skull length in Equus samples reported in this paper

~ N

~ .....

a"

(1)

basal skull length (1)

hormion-to-basion length/basal skull length 0.200

0.225

0.250

0.275

N

horse

0

ii

s. d.

min.

max.

0.249 0.249 0.011 0.210 0.268 0.22860.013 0.211 0.253

..... 0.

~ ::r.

~

t"l

I»

I I •

I

0

1

ol

0

0.225

0 .250

plains zebra

I

1

167

0.239 0.241 0.015 0.199 0.285

3

0.234 0.235 0.008 0.228 0.244

mountain zebra 67

0.233 0.233 0.013 0.198 0.269

Grevy's zebra

11

0.234 0.236 0.010 0.222 0.257

onager

22

0.230 0.232 0.014 0.210 0.258

quagga

I

0.200

55

median mean

0 .275

FIGURE 8. Boxplots and descriptive statistics for ratio between hormion-to-basion length and basal skull length in Equus samples reported in this paper

::r. 0

::,

g_ .£1 ::: :::

-t

s: lb

anterior ocular line/posterior ocular line 1.65

1.80 0

2.10

1.95

N

~

~

.......'

H

n

I.

0

•

.

',

~

1.65

1.80

1.95

min .

max.

57

1.926 1.913 0.077 1.696 2.126

donkey

32

1.670 1.680 0.068 1.557 1.798

.

5

1.720 1.752 0.173 1.623 2.051

Koffiefontein

.,w

I

quagga

--

.....

'

169

1.858 1.862 0.070 1.705 2.016

3

1.876 1.881 0.040 1.844 1.923

mountain zebra 68

1.821 1.816 0.054 1.686 1.930

Grevy's zebra

13

1.974 1.987 0.066 1.875 2.106

onager

22

1.937 1.923 0.058 1.810 2.011

I

~

s. d.

horse

plains zebra

r=-J------1

median mean

2.10

-... p.

~ .....

:::t) I")

I» ::r. 0

:,

.....

0

.£' :;::: :;:::

"' ~

fJ'J

::::

fJ'J

0 fJ'J

] FIGURE 14. Boxplots and descriptive statistics for ratio between anterior ocular line and posterior ocular line in Equus samples reported in this paper

if ..... °' ~

..... °' 00 :;::,

facial breadth (11)

facial breadth/basal skull length 0.30

0.35

0.40

N

horse •I donkey K6ffi;f;~tein '':i15 plains zebra

111 °

I

~ 0.30

0.35

s. d.

min.

max.

56

0.328 0.328 0.015 0.288 0.365

32

0.353 0.355 0.023 0.319 0.444

s o.339 o.~Ef+'o.02~~o.32s o.3a; '

1,

169

0.326 0.327 0.014 0.270 0.360

3

0.328 0.327 0.002 0.325 0.329

mountain zebra 68

0.324 0.324 0.015 0.295 0.354

Grevy's zebra

11

0.303 0.302 0.007 0.290 0.314

onager

22

0.334 0.335 0.014 0.311 0.362

quagga

*

median mean

0.40

FIGURE 15. Boxplots and descriptive statistics for ratio between facial breadth(= breadth across facial crests) and basal skull length in Equus samples reported in this paper

0 ~ ?. ~

n 2N

c::::i.

c1' ro

bizgyomatic breadth (14)

biorbital breadth/bizygomatic breadth 0.975

1.050

1.125

H-0

*

I 0

1 . ·'-" · '

I

~ . ·t· - -

l;

H

N

*

~

min.

max.

58

1.040 1.043 0.030 0.917 1.097

donkey

34

1.053 1.054 0.032 0.949 1.153

'Koffiefontein

5~

, rosa' 1.6°64 "o:fi24 1.04s· 1To2 .~

~

::s

::r. ~

(')

3

1.000 1.006 0.015 0.995 1.022

mountain zebra 68

1.021 1.023 0.029 0.954 1.103

::s 0 ......

DI

::r. 0

t"r1

~

Gravy's zebra

~

onager

t

1.050

0..

It)

0.979 0.982 0.022 0.920 1.044

13

1.023 0.016 0.033 0.962 1.069

!

0.975

-

171

plains zebra

I

s . d.

horse

quagga

I:

median mean

22

1.070 1.075 0.041 1.005 1.156

1.125

.:::

.::: CJ) Cf)

:;,;-'

=-

....0vi Cf)

"Cl

It)

..... (')

It) Cf)

FIGURE 16. Boxplots and descriptive statistics for ratio between biorbital breadth and bizygomatic breadth in Equus sampies reported in this paper

..... °' '°

l(j

0

:;::,

0 ~

15·

braincase breadth/basal skull length 0.20

0.22

0.24

0.26

N

donkey

32

0.241 0.239 0.017 0.199 0.266

VJ:_,;,j:i ."'· ~ !'

0.24

\y

~~C:,'fi~ '/\'.c'.i/ .: i., , _ , '} : -•. ~ ? '

/:""" ,""$

_,}"~'

:,

5 " 0.258 " 0.250J: 0.015 0.234 0.26:4' 169

0.228 0.229 0.010 0.186 0.256

3

0.232 0.2229 0.008 0.219 0.234

mountain zebra 68

0.228 0.229 0.009 0.207 0.249

Gravy's zebra

11

0.202 0.202 0.005 0.1 92 0.211

onager

22

0.236 0.235 0.009 0.218 0.261

quagga

0.22

max.

0.221 0.222 0.016 0.188 0.271

plains zebra

0 .20

min.

56

;:

0

s. d.

horse

Koffieto.olein Y''.:,":i:;: :,~::, ~,.,,;.

00

median mean

0.26

FIGURE 17. Boxplots and descriptive statistics for ratio between breadth of skull across braincase and basal skull length in Equus samples reported in this paper

~

n

~ ~ ....Oro

Identification of Equus skulls to species 621 TABLE l Craniometric ratios found most useful for distinguishing among horse, donkey, plains zebra, and mountain zebra (numbers before ratios refer to relevant figures in this paper; dimensions that underlie the ratios are listed and defined in Eisenmann [1980, 1986); numbered position of each dimension in her list is given in parentheses) 3 4 5 6 7 8 9 IO 11 12 13 14 15 16 17

Minimum breadth of muzzle (l7b)/breadth of muzzle posterior to I3s (17) Length of diastema (6)/length of premolar row (7) Length of cheek tooth row (8)/basal skull length ( l) Length of premolar row (7)/length of molar row (7b) Breadth across occipital condyles (29)/basal skull length ( 1) Hormion-to-basion length (4 )/basal skull length (l) Maximum skull length (l 8)/basal skull length (l) Facial height behind M 3 (27)/basal skull length ( 1) Skull height behind orbits (28)/basal skull length ( 1) Height of orbit (22)/length of orbit (21) Thickness of infraorbital bar ( 19)/basal skull length (l) Anterior ocular line (23)/posterior ocular line (24) Facial breadth (l 1)/basal skull length (l) Biorbital breadth (13)/bizygomatic breadth (14) Braincase breadth (l 5)/basal skull length (I)

TABLE 2 Species assignments predicted by discriminant analysis of donkey, horse, mountain zebra, and plains zebra using 15 cranial ratios listed in Table l and illustrated in Figures 3-17 (predicted species listed across the top of table, known species down left-hand column) horse

Species

donkey

donkey

29

0

horse

0

51

mountain zebra

0

0

plains zebra

0

Total

29

52

mountain zebra

plains zebra

Total 30

0 3

55

0

156

157

63

161

305

63

62

TABLE 3 Probabilities that each Koffiefontein skull and three known quagga skulls derive from donkey, horse, mountain zebra, plains zebra, or a species totally unlike any of these [= prob (0)), according to same discriminant analysis that produced Table 2. The Koffiefontein specimens (MMK numbers) are housed in the McGregor Museum, Kimberley, South Africa. The three known quagga skulls are from the Academy of Natural Sciences of Philadelphia (ANSP), the Peabody Museum of Yale University (YPM), and the Natural History Museum (London) (BMNH). Specimen number

Locality/ taxon

Prob (0)

Prob (donkey)

Prob (horse)

Prob (mountain)

Prob (plains)

Predicted species

MMK 4158

Koffiefontein

0.0000

0.9875

0.0006

0.0119

0.0000

donkey

MMK 4159

Koffiefontein

0.0000

0.0000

1.0000

0.0000

0.0000

horse donkey

MMK 4520

Koffiefontein

0.0000

0.8969

0.0000

0.1031

0.0000

MMK4521

Koffiefontein

0.0000

1.0000

0.0000

0.0000

0.0000

donkey

MMK 4656

Koffiefontein

0.0046

0.5197

0.0055

0.4744

0.0004

donkey

ANSP 6317

quagga

0.0035

0.0000

0.8863

0.0000

0.1137

horse mountain mountain

YPM 1623

quagga

0.0000

0.0000

0.0001

0.8430

0.1569

BMNH 1864723

quagga

0.0000

0.0000

0.0002

0.9996

0.0002

N °' N

:;;:::,

0 ~

l't)

p"

~

n 2N I

C .... .....

O" l't)

Identification of Equus skulls to species 623 Table 3 shows that the same four-species discriminan t analysis suggests that four of the Koffiefontein skulls are most likely to come from donkeys and one from a horse. Since the analysis can assign unknown skulls only to horse, donkey, mountain zebra, or plains zebra, one possible interpretation of this heterogeneo us outcome is that the Koffiefontein specimens actually represent a fifth species, namely quagga. A concomitant implication would be that the close biomolecular similarity between quagga and plains zebra is not reflected in the skull. There is support for this hypothesis in the failure of the analysis to assign the three measured quagga skulls to plains zebra. Instead, Table 3 shows that one of the quaggas is assigned to horse and the remaining two to mountain zebra. However, Figures 3-17 show that many of the Koffiefontein boxplots are highly asymmetric and reflect significant internal variability, which is inconsistent with the idea that the Koffiefontein sample represents a separate, coherent group. Moreover, Table 4 shows that like the four-species discriminan t analysis just described, an additional analysis that considers the Koffiefontein specimens as a possible fifth species separate from horse, donkey, mountain zebra, and plains zebra continues to assign the same Koffiefontein specimen to horse. Equally notable, Table 4 shows that the extended (five-species) analysis continues to assign the known quaggas to horse and mountain zebra and not to 'Koffiefontein,' as would be expected if the Koffiefontein specimens and the quagga skulls represent the same species. In fact, all relevant discriminan t analyses, including ones that can assign skulls to onager, Grevy' s zebra, or both, also consistently place the same Koffiefontein skull in horse and the known quaggas in horse or mountain zebra. Visual assessment of Figures 3-17 suggests that this result might have been anticipated. In an effort to explore further the structure and affinities of the Koffiefontein sample, we undertook a cluster analysis in which the raw data were the means (presented in Figures 3-17) for each Equus species on each of the 15 basic cranial ratios. Such an analysis produces a horizontal tree-like diagram on which samples that are most similar lie on adjoining branches. Figure 18 presents the result calculated by the Macintosh Program Data Desk (Velleman, 1992). Like the four-species discriminan t analysis summarized in Tables 2 and 3, Figure 18 suggests that as a group, the Koffiefontein specimens are more like donkeys than like any other species and manifestly not like any zebra, including quagga. Together, then, the discriminan t and cluster analyses imply that most, if not all, of the Koffiefontein skulls represent domestic donkeys or horse rather than quagga. The same result might occur if one or more of the Koffiefontein specimens represent donkey /horse crosses (mules or hinnies), and we do not rule this out. Finally, like the discriminan t analysis, the cluster result again implies that quagga and mountain zebra are at least as similar as quagga and

TABLE4 Probabilities that the Koffiefontein skulls and three known quagga skulls derive from a hypothetical Koffiefontein species, donkey, horse, mountain zebra, plains zebra, or a species totally unlike any of these [= prob (0)], based on a discriminant analysis involving the same 15 cranial ratios behind Table 3. The Koffiefontein specimens (MMK numbers) are housed in the McGregor Museum, Kimberley, South Africa. The three known quagga skulls are from the Academy of Natural Sciences of Philadelphia (ANSP), the Peabody Museum of Yale University (YPM), and the Natural History Museum (London) (BMNH).

N °'

"'" :;,;:I 0

-. . .

?\ ( t)

.?

Specimen number

Taxon

Prob (o)

Prob (Koffiefontein)

MMK 4158

Koffiefontein

0.0035

0.9841

0.0156

0.0000

0.0002

0.0000

Koffiefontein

MMK 4159

Koffiefontein

0.0000

0.0002

0.0000

0.9998

0.0000

0.0000

horse

MMK 4520

Koffiefontein

0.0002

0.8706

0.1150

0.0000

0.0144

0.0000

Koffiefontein

MMK 4521

Koffiefontein

0.0000

0.6475

0.3524

0.0000

0.0000

0.0000

Koffiefontein

MMK4656

Koffiefontein

0.0412

0.8761

0.0725

0.0015

0.0498

0.0000

Koffiefontein

ANSP 6317

quagga

0.0037

0.0000

0.0000

0.8625

0.0000

0.1375

horse

YPM 1623

quagga

0.0000

0.0000

0.0000

0.0000

0.8821

0.1179

mountain

BMNH 1864723

quagga

0.0000

0.0001

0.0000

0.0003

0.9996

0.0001

mountain

~

Prob (donkey)

Prob (horse)

Prob (mountain)

Prob (plains)

Predicted species

n

c:::

2N I

::!.

O"' (t)

Identification of Equus skulls to species 625 donkey~

Koffiefontein

quagga

-

horse -

mountain zebra ~

plains zebra

onage

Grevy's zebra

FIGURE 18. Single-linkage cluster analysis of Equus species based on species means for the 15 cranial ratios presented in Figures 3-17. The Grevy's zebra and donkey were redrawn after Dorst and Dandelot (1969); the quagga after Rau (1986); the plains zebra and mountain zebra after Skinner and Smithers (1990); and the onager after Clutton-Brock (1981). Complete-linkage analysis produces a very similar result, including the close similarity between donkey skulls and those from Koffiefontein. The clustering need not reflect phylogeny, and it differs significantly from a mitochondrial DNA tree favoured by George and Ryder (1986). Perhaps most important, the mitochondrial tree unites Grevy's zebra, mountain zebra, and plains zebra as a monophyletic group separate from other species.

626 R.G. Klein, K. Cruz-Uribe plains zebra. However, only three quagga specimens are involved, and at this point we argue only that the biomolecular similarity between quagga and plains zebra may not extend to the cranium. To confirm or reject this possibility, we hope to measure additional quagga skulls soon. We also suggest that the biomolecular results linking quagga and plains zebra should be rechecked through analyses of additional quagga tissues.

Summary and conclusion Like previous analyses of cranial variation among Equus species, our analysis of craniometric ratios shows that average values often differ significantly among species, even if the ranges overlap strongly. The most reliable species separation occurs when many ratios are considered at once. Supplemented by cluster analysis, multivariate discriminant analysis based on ratios whose average values tend to differ significantly among species implies that five subfossil skulls from Koffiefontein, Orange Free State, South Africa, derive mainly if not entirely from donkey, horse, or possibly mules and not from the extinct quagga, as was previously proposed. The analyses assign two known quagga skulls to mountain zebra and a third to horse, and we conclude that the systematic affinities of the quagga remain unresolved, despite biomolecular support for a very close relationship to the extant plains zebra.

Acknowledgments We thank P.B. Beaumont (McGregor Museum, Kimberley) for access to the Koffiefontein skulls; the Academy of Natural Sciences (Philadelphia), the American Museum (New York), the Field Museum (Chicago), the National Museum (Bloemfontein), the Natural History Museum (London), the South African Museum (Cape Town), the Transvaal Museum (Pretoria), and the University of Illinois School of Veterinary Medicine (Urbana) for access to comparative Equus specimens; the National Science Foundation for financial support; and J. Clutton-Brock and Q.B. Hendey for helpful reviews of the manuscript.

NOTES

1 For presentation here, the plains zebra sample has been subdivided between specimens from north of the Zambezi River(= 'northern') and south of the Zambezi River(= 'southern'). The tendency for plains zebras from north of the Zambezi to be smaller than their southern counterparts has also been noted by Eisenmann (1980). A similar tendency characterizes some prominent bovid

Identification of Equus skulls to species 627 species (Klein and Cruz-Uribe, 1991), perhaps reflecting the operation of Bergmann's Rule. The mountain zebras have been divided between the two commonly recognized subspecies: (1) Hartmann's (E. zebra hartmannae) from Namibia and southern Angola and (2) Cape (E. zebra zebra) from the Cape Province of South Africa. The tendency for Hartmann's (or Namibian) mountain zebras to be larger has been also been noted by Skinner and Smithers (1990). LITERATURE CITED

Ansell, W.F.H. 1971. Order Perissodactyla. Part 14 in J. Meester and H. Setzer, (eds), The Mammals of Africa: An Identification Manual. Smithsonian Institution Press, Washington, DC. Antonius, 0. 1951. Die Tigerpferde: die Zebras. Monographien der Wildsaugetieren 11:1-148. Bennett, D.K. 1980. Stripes do not a zebra make, Part I: a cladistic analysis of Equus. Systematic Zoology 29:272-287. Churcher, C.S., and M.L. Richardson. 1978. Equidae. Pp. 379-422 in V.J. Maglio and H.B.S. Cooke (eds), Evolution of African Mammals. Harvard University Press, Cambridge, MA. Clutton-Brock, J. 1981. Domesticated Animals from Early Times. British Museum (Natural History), London, 288 pp. - 1992. Horse Power: A Natural History of the Horse and the Donkey in Human Societies. Natural History Museum, London, 192 pp. Comrie-Greig, J. 1983. 1883-1983: centennial of the extinction of the quagga. African Wildlife 37(4):146--154. - 1985. When did the last Quagga die? African Wildlife 39 (3):109-110. Cooke, H.B.S. 1943. Cranial and dental characters of the recent South African Equidae. South African Journal of Science 40:254-257. - 1948. The Fowler collection of fossils from Koffiefontein, O.F.S. South African Science 2:96--98. - 1950. A critical revision of the Quaternary Perissodactyla of Southern Africa. Annals of the South African Museum 31:393-479. Corbet, G.B. 1978. The Mammals of the Palaearctic Region. British Museum (Natural History), London, 314 pp. Dorst, J ., and P. Dandelot. 1969. A Field Guide to the Larger Mammals of Africa. Houghton-Mifflin, Boston, 287 pp. Eisenmann, V. 1980. Les Chevaux (Equus sensu lato) fossiles et actuels: cranes et dents jugales superieures. Cahiers de Paleontologie. Editions du Centre National de la Recherche Scientifique, Paris, 186 pp. - 1986. Comparative osteology of modem and fossil horses, half-asses, and asses. Pp. 67-116 in R.H. Meadow and H.-P. Uerpmann (eds), Equids of the Ancient World. Dr Ludwig Reichert Verlag, Wiesbaden.

628 R.G. Klein, K. Cruz-Uribe Ellerman, J.R., T.C.S. Morrison-Scott, and R.W. Hayman. 1953. Southern African Mammals 1758 to 1951: A Reclassification. British Museum (Natural History), London, 363 pp. Gentry, A.W. 1975. A quagga, Equus quagga (Mammalia, Equidae), at University College, London and a note on a supposed quagga in the City Museum, Bristol. Bulletin of the British Museum of Natural History (Zoology) 28(5):217-226. George, M., and O .A. Ryder. 1986. Mitochondrial DNA evolution in the genus Equus. Molecular Biology and Evolution 3:535-546. Grayson, D.K. 1989. The chronology of North American Late Pleistocene extinctions. Journal of Archaeological Science 16:153-166. Groves, C.P. 1985. Was the quagga a species or subspecies? African Wildlife 39(3): 106-107. - 1986. The taxonomy, distribution, and adaptations of recent equids. Pp. 11-65 in R.H. Meadow and H .-P. Uerpmann (eds), Equids of the Ancient World. Dr. Ludwig Reichert Verlag, Wiesbaden. Groves, C.P., and D.P. Willoughby. 1981. Studies on the taxonomy and phylogeny of the genus Equus. Mammalia 45:321-354. Harley, E. 1988. The retrieval of the Quagga. South African Journal of Science 84:158-159. Harper, F. 1945. Extinct and Vanishing Mammals of the Old World. New York Zoological Park, New York, 850 pp. Harris, W.C. 1840. Portraits of the Game and Wild Animals of Southern Africa (1969 reprint). A.A. Balkema, Cape Town, 195 pp. Higuchi, R., B. Bowman, M. Freiberger, O .A. Ryder, and A.C. Wilson. 1984. DNA sequences from the quagga, an extinct member of the horse family. Nature 312:282-284. Higuchi, R., L.A. Wrischnik, E. Oakes, M. George, B. Tong, and A.C. Wilson. 1987. Mitochondrial DNA of the extinct quagga: relatedness and extent of postmortem change. Journal of Molecular Evolution 25:283-287. Honacki, J.H., K.E. Kinman, and K.W. Koeppl (eds). 1982. Mammal Species of the World. Allen Press, Inc. and the Association of Systematics Collections, Lawrence, KS, 694 pp. Klein, R.G., and K. Cruz-Uribe. 1991. The bovids from Elandsfontein, South Africa, and their implications for the age, palaeoenvironment, and origins of the site. African Archaeological Review 9:21-79. Klingel, H . 1969. The social organisation and population ecology of the plains zebra (Equus quagga) . Zoologica Africana 4:249-263. Kurten, B., and E. Anderson. 1980. Pleistocene Mammals of North America. Columbia University Press, New York, 442 pp. Lloyd, P. 1984. The Cape mountain zebra 1984. African Wildlife 38 (4):144-149. Lowenstein, J.M. 1985. Half-striped quagga was a plains zebra. New Scientist 1465:27. Lowenstein, J.M., and O .A. Ryder. 1985. Immunological systematics of the extinct quagga (Equidae). Experientia 41:1192-1193.

Identification of Equus skulls to species 629 Lundholm, B. 1951. A skull of the true quagga (Equus quagga) in the collection of the Transvaal Museum. South African Journal of Science 47:307-312. MacFadden, B.J. 1992. Fossil Horses: Systematics, Paleobiology, and Evolution of the Family Equidae. Cambridge University Press, Cambridge, 369 pp. Martin, P.S. 1984. Prehistoric overkill: the global model. Pp. 354--403 in P.S. Martin and R.G. Klein (eds), Quaternary Extinctions: A Prehistoric Overview. University of Arizona Press, Tucson. Meester, J.A.J., LL. Rautenbach, N.J. Dippenaar, and C.M. Baker. 1986. Classification of Southern African Mammals. Transvaal Museum Monographs 5:1-359. Mohr, E. 1964. Von Rosshauten und Zebrafellen. Das Pelzgewerbe 15:161-168. Pocock, R.l. 1904. The Cape Colony quaggas. Annals and Magazine of Natural History 14:313-338. Rau, R.E. 1974. Revised list of the preserved material of the extinct Cape Colony Quagga, Equus quagga quagga (Gmelin). Annals of the South African Museum 65:41-87. - 1978. Additions of the revised list of preserved material of the extinct Cape Colony Quagga and notes on the relationship and distribution of southern plains zebras. Annals of the South African Museum 77:27-45. - 1983. The colouration of the extinct Cape Colony quagga. African Wildlife 37(4):136-138. - 1986. The quagga and its kin. Sagittarius 1(2):8-10. Roberts, A. 1951. The Mammals of South Africa. Central News Agency, Cape Town, 700 pp. SAS Institute. 1989. JMP User's Guide. SAS Institute, Inc., Cary, NC, 584 pp. Skead, C.J. 1980. Historical Mammal Incidence in the Cape Province, volume 1. Cape Provincial Department of Nature and Environmental Conservation, Cape Town, 903 pp. Skinner, J.D., and R.H.N. Smithers. 1990. The Mammals of the Southern African Subregion. University of Pretoria, Pretoria, 769 pp. Smuts, M.M.S., and B.L. Penzhorn 1988. Descriptions of anatomical differences between skulls and mandibles of Equus zebra and Equus burchelli from Southern Africa. South African Journal of Zoology 23:328-336. Thackeray, J.F. 1988. Zebras from Wonderwerk Cave, northern Cape Province, South Africa: attempts to distinguish Equus burchelli and E. quagga. South African Journal of Science 84:99-101. Velleman, P.F. 1992. Data Desk 4 Handbook. Ithaca, NY, Data Description Inc., 283pp. Winans, M.C. 1989. A quantitative study of North American fossil species of the genus Equus. Pp. 262-297 in D.R. Prothero and R.M. Schoch (eds), The Evolution of Perissodactyls. Oxford University Press, New York.

Temporal variability in horn-core dimensions of Damaliscus nirofrom Olduvai, Sterkfontein, Cornelia, and Florisbad

J.F. Thackeray, J.S. Brink, I. Plug

Abstract Measurements have been obtained from horn-cores of Damaliscus niro (an extinct alcelaphine) from Olduvai Beds I and II, and Sterkfontein Member 5 (early Pleistocene); Cornelia in South Africa (middle Pleistocene); as well as Florisbad and Maselspoort in South Africa (middle-late Pleistocene). Temporal trends in the relationship between medio-lateral breadth and anterior-posterior length of these horn-cores are identified, suggesting morphological changes within specimens that have been attributed to a single species.

Introduction Damaliscus niro is an extinct alcelaphine that is represented at various Pleistocene localities in both East and South Africa. Hopwood (1936) initially described it as Hippotragus niro, on the basis of a horn-core from Olduvai Bed IV. The curvature of such horns is similar to that of hippotragine antelope such as roan (H. equinus) or sable (H. niger), but Leakey (1965) drew attention to the fact that specimens attributed to Hippotragus niro might be better accommodated within the genus Damaliscus. Gentry and Gentry (1978) reassessed the East African specimens, including horn-cores from Olduvai and Peninj, and agreed that Damaliscus was more appropriate (see also Gentry, 1965). This study focuses on measurements of specimens now attributed to D. niro from various localities in South and East Africa. The objective is to address the question whether horn-core dimensions show any pattern of change within the Pleistocene.

Hom-core dimensions of Damaliscus niro 631

Materials The specimens of D. niro used in this study include Pleistocene horn-cores from several African sites, notably Olduvai Gorge in Tanzania; Sterkfontein in the Transvaal Province of South Africa; Cornelia, Florisbad, and Maselspoort in the Orange Free State, South Africa (Fig. 1). The earliest specimens include those from Olduvai Beds I and II, previously measured by Gentry and Gentry (1978); and horn-cores from Sterkfontein Member 5, based on previously unpublished data obtained from samples from excavations undertaken under the direction of P.V. Tobias, with identifications and measurements by F. Thackeray and I. Plug (this study), supplementing data obtained by Vrba (1976). Middle Pleistocene specimens from Cornelia (Cooke, 1974), and late Pleistocene specimens from Maselspoort and Florisbad have been measured by J.S. Brink (1987; this study).

Measurements The measurements taken in this study include anterior-posterior length (APL) and medio-lateral breadth (MLB) at suitable (undamaged) positions on horn-cores which were rarely if ever complete. Specimens from Olduvai, Cornelia, and Florisbad were generally more complete than those from Sterkfontein. Gentry and Gentry (1978) published dimensions for only the base of horn-cores. For purposes of this study, APL and MLB were measured on fragments of horn-cores from Sterkfontein, as well as more complete specimens from other South African sites. A total of 51 measurements were obtained from specimens from Sterkfontein and Olduvai (combined sample); 22 measurements from the Cornelia assemblage; and 50 measurements for the combined sample from Florisbad and Maselspoort.

Results Relationships between log-transformed medio-lateral breadth and anterior-posterior length dimensions of early Pleistocene horn-core specimens from Olduvai (Beds I and II) and Sterkfontein (Member 5) are shown in Figure 2. The linear regression equation for these data is as follows: y

= 1.047x -

0.202 (r = 0.98)

Equation 1

Data obtained from mid-Pleistocene horn-cores from Cornelia are plotted

632 J.F. Thackeray, J.S. Brink, I. Plug

Equator

OLDUVAI •

STERKFONTEIN I

MASEU,ro::v

FLORISBAD • • CORNELIA

FIGURE 1. Map of localities in Africa from which measurements of horns of D. niro have been obtained

Horn-core dimensions of Damaliscus niro 633

-i

I• a::

~

Ill

5

0 6

Ill

:I

9

1.7 X

1.8 -

◊◊

OLDWAJ STERKFONTEIN

◊

◊

o

.◊◊

0◊ ◊

1.S -

X

X

1AX

1.3 -

X xX X

X

1.2 X

1.1

◊◊ ◊

◊

X

X

X

X XX X

~ X

X

X

X

X XX

X

X

1.3 1.4 1.5 1.8 1.7 LOG ANTERIOR-POSTERIOR LENGTH (nn)

1.8

FIGURE 2. Anterior-posterior length and medio-lateral breadth dimensions of horn-cores attributed to D. niro from Olduvai Beds I and II (Gentry and Gentry, 1978) and Sterkfontein (previously unpublished data obtained from specimens from excavations undertaken under the direction of P.V. Tobias, measurements by J.F. Thackeray and I. Plug [this study))

in Figure 3. The regression equation based on these measurements is given by Equation 2: y

= 1.195x -

0.517 (r =0.97)

Equation 2

Measurements for late Pleistocene horn-cores from Florisbad and Maselspoort are shown in Figure 4. The equation obtained from linear regression analysis is given by Equation 3:

y = 2.105x - 1.935 (r = 0.87)

Equation3

Discussion and conclusions Specimens attributed to D. niro from Olduvai Beds I and II tend to be larger than those from Sterkfontein Member 5 (Figure 2). However, the relationship between log-transformed APL and MLB for penecontemporaneous early Pleistocene specimens from both Olduvai and Sterkfontein is essentially co-linear.

634

-s. E

.i a:

....

s ~

0 6

w :I

9

J.F. Thackeray, J.S. Brink, I. Plug

1.7

•

CORNELIA

• • • •• • ••

1.1 -

. ...

1.5 1.4 1.3 -

•

1.2 1.1

•

• •• • • • •

• 1.3 1.4 1.5 1.1 1.7 LOG ANTERIOR-POSTERIOR LENGTH (111111)

1.8

FIGURE 3. Anterior-posterior length and medio-distal breadth dimensions of horncores attributed to D. niro from Cornelia, Orange Free State, South Africa (measurements by J.S. Brink [this study])

The fact that there is no overlap in the dimensions of specimens from these localities, as plotted in Figure 2, is simply due to the fact that Gentry and Gentry (1978) published only maximum values (at the base of horncores) for specimens from Olduvai. Linear relationships between APL and MLB have also been determined for middle and late Pleistocene samples from elsewhere in Africa, but the slopes and intercepts for Equations 1, 2, and 3 show differences that reflect a trend through the Pleistocene. The slope for Equation 1 (based on early Pleistocene samples) is 1.047 (standard error = 0.03). By contrast, the slope obtained from the analysis of middle Pleistocene specimens is higher (1.195, standard error = 0.07); the slope obtained for late Pleistocene specimens is still higher (slope= 2.105, standard error= 0.17). A similar increasing trend can also be discerned by examining y-intercept values obtained from regression analyses. Notably, the intercept value of -0.202 with a standard error of 0.04 for the early Pleistocene sample (Equation 1) is significantly greater (p = 0.05) than the intercept of -0.516 (associated with standard error of 0.04) obtained for the middle Pleistocene sample of D. niro horn-cores (Equation 2). The intercept of

Hom-core dimensions of Damaliscus niro 635

e'

1.7 -

§

1.8 -

s

a:

1.5 -

~

1A -

Ill

5

g

FLORISBAD

6.

IIASELSPOORT

+++

+ + +

1.3 ++

Ill

1.2

§

1.1 -

::E

+

~

+

+ ++ + 6.

+ +

-t+ "'+ +

+ + ++ ++ + :t + ~+ + + + 6.

+ + + + + + + +

+ I

1.3 1.5 1.4 1.7 1.8 LOG ANTERIOR-POSTERIOR LENGTH {mm)

1.8

FIGURE 4. Anterior-posterior length and media-distal breadth dimensions of late Pleistocene horn-cores attributed to D. niro from Florisbad and Maselspoort, Orange Free State, South Africa (measurements by J.S.Brink [this study])

-1.935 (standard error = 0.07, Equation 3) for the late Pleistocene sample is significantly different (p = 0.05) from the intercept obtained for the middle Pleistocene assemblage. The changes in slope and intercept in these equations reflect a progressive flattening of the horn-cores, resulting from an increase in anteriorposterior length relative to media-lateral breadth. Of particular interest here is that the intercept and slope values show significant differences (p = 0.05) when one compares early and middle Pleistocene samples, and also when one compares middle and late Pleistocene specimens, despite the fact that all of the specimens have been attributed to a single alcelaphine species.

Acknowledgments We thank K. Stewart for the invitation to contribute to this volume in honour of Rufus Churcher. J.F.T. thanks P.V. Tobias and the late Alun Hughes for the opportunity to study bovid specimens from Sterkfontein; Meave Leakey for providing an opportunity to examine bovid specimens

636 J.F. Thackeray, J.S. Brink, I. Plug from East African localities; anonymous referees for helpful comments; and the Foundation for Research Development (South Africa) for financial assistance.

LITERATURE CITED

Brink, J.S. 1987. The archaeozoology of Florisbad, Orange Free State. Memoirs of the National Museum, Bloemfontein 24:1-151. Cooke, H.B.S. 1974. The fossil mammals of Cornelia, O.F.S., South Africa. Pp. 6384 in The Geology, Archaeology and Fossil Mammals of the Cornelia Beds, 0.F.S., Memoirs of the National Museum, Bloemfontein 9. Gentry, A.W. 1965. New evidence on the systematic position of Hippotragus niro Hopwood 1936 (Mammalia). Annals and Magazine of Natural History, ser. 13, 8:335--338.

Gentry, A.W., and A. Gentry. 1978. Bulletin of the British Museum, Natural History (Geology) London 29:289-446. Hopwood, A.T. 1936. New and little-known fossil mammals from the Pleistocene of Kenya Colony and Tanganyika Territory. Annals and Magazine of Natural History, ser. 10, 17:636--641. Leakey, L.S.B. 1965. Olduvai Gorge, 1951-61, Vol. 1. A Preliminary Report on the Geology and Fauna, pp. 1-109. Cambridge University Press, Cambridge. Vrba, E.S. 1976. The Fossil Bovidae of Sterkfontein, Swartkrans and Kromdraai. Memoirs of the Transvaal Museum, 21.

The fossil and living Hyaenidae of Africa: Present status

Lars Werdelin, Alan Turner

Abstract We review the fossil record of the Hyaenidae in Africa. The oldest hyaenids known on this continent are from the middle Miocene Fort Ternan locality and belong to Protictitherium. The evolution of hyaenids in Africa after this time follows the same pattern as that seen in Eurasia, with the dog-like genera Ictitherium and Hyaenictitherium present in the Miocene and the post-Miocene dominated by bone-cracking forms more closely related to the modern taxa. Africa differs from Eurasia, however, in the near absence of bone-cracking forms in the Miocene, and with the common Eurasian species Adcrocuta eximia only known from Sahabi in North Africa. The data are still very incomplete, but suggest two or three dispersal events of hyaenas into Africa during the Miocene. The oldest representatives of all extant hyaena species are from Africa.

Introduction Although consisting today of only four species, the Hyaenidae were a diverse family of carnivores during the Miocene, the Pliocene, and the earlier part of the Pleistocene (Fig. 1; Werdelin and Solounias, 1991). Members of the family are common in the fossil record. It is now close to thirty years since Ewer (1967), offered an overview of the evolution of the African Hyaenidae, and since that time several new and important assemblages have been recorded. The most significant of these have been at Langebaanweg in the Cape Province of South Africa (Hendey, 1974a), at Sahabi in Libya (Howell, 1987), and at Laetoli in Tanzania (Barry, 1987), while studies of new specimens from the Transvaal hominid-bearing sites have been integrated with revisions of older collections (Turner, 1984,

638

L. Werdelin, A. Turner

FIGURE 1. Cladogram showing hyaenid phylogeny (modified after Werdelin and Solounias [1991 ]; *** =exclusively African taxa; * = taxa with part of their range in Africa)

Present status of African Hyaenidae 639 1986, 1987, 1988a, 1993). A critical revision of the extant southern African mammalian fauna has attempted to take account of the fossil record (Meester et al., 1987). Much of the material referred to in these works, together with the results of other discussions, has been summarized in a synthesis confined to events surrounding the appearances of the extant members of the African large carnivore guild during the later Pliocene (Turner, 1990), while a more broadly based treatment dealing with the Hyaenidae as a whole was given by Werdelin and Solounias (1991). Continuing work on both the Mio-Pliocene evolution of the family and the appearance of the extant taxa (Werdelin, in prep.; Werdelin and Barthelrne, in prep.; Werdelin et al., 1994) suggests that the time is now right for a further overview of the Hyaenidae of Africa. We have between us seen a substantial amount of the available material at first hand, as well as considerable quantities of hyaena remains from Eurasia. We therefore offer a complete overview of the fossil and living Hyaenidae of Africa as we currently understand the taxonomy and relationships. The treatment offered here is taxonomically based. It is worth stressing at the outset that the nomenclatural history of many hyaenid taxa has often been immensely complex, and the African material is no exception. We have tried to convey information relevant to an understanding of older references to the same material without becoming enmeshed in minutiae adequately dealt with elsewhere (Werdelin and Solounias, 1991). Synonomies relevant to an understanding of previous literature are nevertheless discussed within the treatment of the senior synonym proposed in the present study. Localities referred to in the text are shown in the map in Figure 2, while the ages of deposits are shown in Figure 3.

Hyaenid taxa The Hyaenidae, as currently viewed, consists of some 69 probable species, although the full nomenclature and allocation of the sometimes sparse and fragmentary material has yet to be completed (Werdelin and Solounias, 1991; in press; this paper). Although representing only a portion of this total diversity (Fig.I), the African fossil record affords an interesting picture of developments, the more so since the earliest representatives of the four extant taxa are recorded from that continent. As discussed in detail by Werdelin and Solounias (1991), there are now substantial reasons for excluding the genera Percrocuta, Dinocrocuta, and Allohyaena from the Hyaenidae and placing them in a separate family, the Percrocutidae. We shall therefore omit consideration of various African specimens discussed by Howell and Petter (1985) that in our view are better referred to this family.

640

L. Werdelin, A. Turner I . 4 Bled ed Douarah

a,

L

Am Brtm

Ver ~ -

llanapot67

y"s.mburue~.:=n I Fort Teman 6 Macod1 t) 6 Olduvai 6 i..,to11

4.::,s.

Lanc

::

Cluster matrix Pairwise comparison matrix

Steatomys pratensis

~

~

s::,_

~

"' .s .9 .l!)

"';;

E

E

.8 U>

iii

8

.8 V)

C

~

e

;;

~

~

-J

U>

I::

::

"'

~

;;

~

::

·i;_

U>

""';;

"';;

·E

:..;

]

8

"'E

;:,.,

~

~

8

8

U> U>

E

,,::

E

8

,,::

...

.8

;:,.,

U>

E C ::

~

::i

;;

"';:

~

0.2332 0.2401 0.1372 0.1756 0.2191 0.1577 0.2238 0.2450

"--....__ __Q.1810 0.2083 0.1238 0.1727 0.2036 0.1772 0.1740 0.2055

Saccostomus isiolae

0.1645

Lophuromys sp.

0.2425 0.1989

Millardia coppensi

0.2596 0.2907 0.3002

Antemus chinjiensis

0.2662 0.1985 0.2287 0.2024

Acomys subspinosus

0.2239 0.2239 0.1860 0.2739 0.2043

Acomys cahirinus

0.2361 0.1712 0.1580 0.2131 0.1238 0.1052

Acomys mabele

0.1663 0.1904 0.1604 0.2871 0.2202 0.0829 0.1163

~

Musauctor

0.2422 0.1768 0.1051 0.2757 0.1977 0.0755 0.1141

0.1199

Uranomys ruddi

~

0.2986 0.1990 0.1701 0.2316 0.1738 0.1515 0.1225 0.1631

0.1055

-....__.....__

._9.1208 0.1256 0.1136 0.1352 0.1545 0.0947 0.1220 -....__.....__

0.1668 0.1401 0.1517 0.1542 0.1298 0.1557

'-

~ 0.0855 0.1249 0.1071 0.1181 0.1541 ~ 0.0686 0.0644 0.0917 0.1166

~ 0.0802 0.0708 0.0885 0.1058 0.1258 0.0616

.....__.....__.

r

► ':-' ~

s·

j"" r'

rs

~ ,..,

g"

Is Acomys a murine? 667 Steatomys pratensis Saccostomus isiolae Lophuromys sp. Millardia coppensi Antemus chinjiensis Acomys cahirinus Acomys subspinosus Acomys mabele Mus auctor Uranomys ruddi 01

0

0.2

Cluster Distance

FIGURE 3. Cluster diagram based on changes in direction and magnitude for seven landmarks among cricetomyines, dendromurines, and murines (data from Table 2, above diagonal line)

2

s'

/

9

'8

FIGURE 4. Graphic representation of landmark positional changes comparing the cricetomyine Saccostomus isiolae (base) with average landmark values determined for all murines listed in Table 1 (vector tips) (numbers indicate cusp positions as designated in Fig. 2)

668 X. Xu, A.J. Winkler, L.L. Jacobs the murines (including Acomys) listed in Table 1. Vectors indicate that the major difference between Saccostomus and murines in the Ml landmarks employed is the posterior displacement of cusps 3, 4, and 6, disrupting the transverse alignment of chevrons. The posterior portion of the tooth becomes narrower by anterolingual displacement of cusp 9. With the addition of the eighth landmark (cusp 1), the pairwise comparison index values for murines (Table 3) decrease in 18 of 28 cases and increase in 10 cases. The range of values increases from 0.0755-0.3002 to 0.0650-0.3793. In nearly two-thirds of the cases the strength of similarity increases with the addition of the eighth landmark. The similarity of the three Acomys species is reflected in the reduction of their pairwise comparison index values. The paired comparison indices for eight landmarks between pairs of Acomys species are the lowest observed for either seven or eight landmarks. Figure 5 is a cluster analysis for murines utilizing eight landmarks. The arrangement of taxa differs from that in Figure 3 in that Millardia coppensi is the most divergent taxon and Uranomys groups with Lophuromys plus Antemus rather than Mus. Acomys species group together again, and then with Mus auctor, as before. Figure 6 shows the landmark positions of modern Acomys species (averaged) and fossil A. mabele to be reasonably uniform. When Mus auctor is compared to Acomys (all species averaged, Fig. 7), vectors indicate a more elongate and narrow tooth, consistent with the qualitative characterization of Mus (Misonne, 1969). Figure 8 shows strong positional shifts in most landmark positions of Millardia coppensi compared to Acomys (average).

Discussion This study is preliminary in that it includes too few taxa, too few characters, and the sample size for each taxon is small. Nevertheless, it has elucidated some interesting patterns. First and foremost, the Murinae as a consistent group distinct from the Cricetomyinae and Dendromurinae is upheld based on Ml morphology. Within the murines, the addition of an eighth landmark decreased the index value of pairwise comparison for some groups (such as the species of Acomys), indicating a stronger morphological similarity when the eighth landmark is taken into account. However, pairwise comparisons also indicate high morphological volatility associated with the cusp 1 position because index values increased with the addition of cusp 1 as the eighth landmark in 10 of 28 cases. Within the constraints imposed by the samples used, all three Acomys species fall within the Murinae and cluster together exclusive of other taxa. Neither Millardia coppensi nor Mus auctor appears to belong in

TABLE3

Value matrices of pairwise comparisons (below diagonal line) and cluster analysis (above diagonal line) using eight landmarks ·;;;

Cluster matrix

ci. .

E C, u

...;:

"'::t

.,

.: :E;

}1

8

E

~

"'

"'>. 8

~ E

E

...;:

...;:

8

;.;

...

.2 u

::t

"'"'::t

:s

""::t...

~ E C,

.:

l:

:::i

0.1525 0.1009 0.1157 0.1558 0.1495 0.1329 0.1082

~~

.1766 0.1674 0.1741

Antemus chinjiensis

0.2027 0.1976

Acomys subspinosus

0.1350

0.2641 0.1516 0.2633 0.1671

0.1764 0.1372 0.1977

~ "0.0983 0.1352 0.1193

0.1278 0.1286

~ t--9.0641 0.0613 0.0964 0.1115

Acomys cahirinus

0.1860

Acomys mabele

0.1720 0.2575 0.1627 0.0650 0.0732

Mus auctor

0.1528

0.2491 0.1661

0.0991 0.0848 0.1148

Uranomys ruddi

~

0.1130

0.3793 0.2102 0.1270 0.1515 0.1411

0.1476

0.0667

~ ....0.0700

~

0.0798 0.1334 0.1097 0.1230 0.1237

~

-

IJJ

::i:.

8

~Ill

g

::t ::s (t)

..,,, 0-,

$

670 X. Xu, A.J. Winkler, L.L. Jacobs Millardia ccppcnsi Uranomys ruddi Lophuromys sp. Antemus chinjiensis Mus auctor Accmys cahirinus

---..i-J-I I

I

Acomys mabele Acomys suospinosus

Ol

0

0.2

Cluster Distance

FIGURE 5. Cluster analysis based on changes in direction and magnitude for eight landmarks among murines (data from Table 3, above diagonal line)

2

3

'

5 6 4

I 8

FIGURE 6. Graphic representation of landmark positional changes comparing fossil Acomys mabele (base) with average landmark values for modern Acomys cahirinus plus A. subspinosus (vector tips)

Is Acomys a murine? 671

1

5

4

9

t

8

FIGURE 7. Graphic representation of landmark positional changes comparing the average landmark values of Acomys cahirinus plus A. subspinosus plus A. mabele (base) to Mus auctor (vector tips)

Acomys, as has been suggested. However, these suggestions were made based on the morphology of the third molars, which in Acomys has a distinct cusp 3. Nevertheless, based on Ml, Mus auctor seems to be clearly outside of Acomys, although the morphologies are similar (Jacobs, 1978; Jacobs et al., 1989), and Millardia coppensi appears to be well removed from Acomys. More extensive evaluations, expansion of the database, and a review of the weighting of characters are called for in deciphering the relationships of Acomys to other murines. Is Acomys a grand example of morphological convergence with murines recognizable only through immunological and DNAxDNA hybridization reactions? We cannot say with certainty, but we would expect some morphological clues to remain even in a case of exceptionally strong convergence. An alternative interpretation could admit that molecular systematic studies are in a preliminary descriptive phase. There is no clear understanding of the meaning of immunological or DNAxDNA hybridization reactions vis-a-vis the whole organism for phe-

672 X. Xu, A.J. Winkler, L.L. Jacobs

y

/1

9

8

FIGURE 8. Graphic representation of landmark positional changes comparing average landmark values of Acomys cahirinus plus A. subspinosus plus A. mabele (base) to Millardia coppensi (vector tips)

notypic traits useful in traditional systematics. How do the discrepancies seen at the molecular level relate to the morphological parameters of Acomys? How do the adaptations for life in arid environments, for instance, that are characteristic of Acomys but uncommon in other murines relate to its molecular constituents? The evolutionary process clearly must involve both the molecules of life and the organisms that put them to use. Discerning how is a continual new frontier.

Acknowledgments Anyone who is familiar with the palaeontological literature of Africa knows a part of the work of C.S. Churcher. Because Africa holds a special place for us, we are especially pleased to offer this contribution in honour of Rufus. We are grateful for the constructive comments and helpful assistance provided by D. Winkler. J. Rensberger kindly reviewed a draft of the manuscript. Ralph Chapman not only reviewed the manuscript, but also patiently answered our many queries.

Is Acomys a murine? 673 LITERATURE CITED

Butler, P.M. 1956. The ontogeny of molar pattern. Biological Review 31: 30-70. - 1980. Functional aspects of the evolution of rodent molars. Pp. 249-262 in M.J. Michaux (ed.), Memoire jubilaire en hommage a Rene Lavocat. Palaeovertebrata. - 1982. Some problems of the ontogeny of tooth patterns. Pp. 45-51 in B. Kurten (ed.), Teeth: Form, Function, and Evolution. Columbia University Press, New York. - 1985. Homologies of cusps and crests, and their bearing on assessments of rodent phylogeny. Pp. 381-401 in W.P. Luckett and J.-L. Hartenberger (eds), Evolutionary Relationships among Rodents. Plenum Press, New York. Catzeflis, F.M., J.P. Aguilar, and J.-J. Jaeger. 1992. Muroid rodents: phylogeny and evolution. Trends in Ecology and Evolution 7:122-126. Chapman, R.E. 1990. Conventional procrustes approaches. Pp. 251-267 in F.J. Rohlf and F. Brookstein (eds), Proceedings of the Michigan Morphometrics Workshop. University of Michigan Museum, Special Publication no. 2, Ann Arbor. Chevret, P., C. Denys, J.-J. Jaeger, J. Michaux, and F.M. Catzeflis. 1993. Molecular evidence that the spiny mouse (Acomys) is more closely related to gerbils (Gerbillinae) than to true mice (Murinae). Proceedings of the National Academy of Science 90:3433-3436. Denys, C. 1978. Rodentia and Lagomorpha. 6.1. Fossil rodents (other than Pedetidae) from Laetoli. Pp. 118-170 in M.D. Leakey and J.M. Harris (eds), Laetoli, a Pliocene Site in Tanzania. Oxford University Press, London. - 1990. The oldest Acomys (Rodentia, Muridae) from the lower Pliocene of South Africa and the problem of its murid affinities. Palaeontographica, Abt. A, 210:79-91. Denys, C., J. Michaux, F. Petter, J.P. Aguilar, and J.-J. Jaeger. 1992. Molar morphology as a clue to the phylogenetic relationship of Acomys to the Murinae. Israel Journal of Zoology 38:253-262. Flynn, L.J., L.L. Jacobs, and E.H. Lindsay. 1985. Problems in muroid phylogeny: relationship to other rodents and origin of major groups. Pp. 589-616 in W. P. Luckett and J.-L. Hartenberger (eds), Evolutionary Relationships among Rodents. Plenum Press, New York. Fraguedakis-Tsolis, S.E., B.P. Chondropoulos, and N.P. Nikoletopoulos. 1993. On the phylogeny of the genus Acomys (Mammalia: Rodentia). Zeitschrift fiir Saugetierkunde 58:240-243. Gaunt, W.A. 1961. The development of the molar pattern of the golden hamster (Mesocricetus auratus W.), together with a re-assessment of the molar pattern of the mouse (Mus musculus). Acta Anatomica 45:219-251. Graaff, G. de. 1961a. A preliminary investigation of the mammalian microfauna in

674 X. Xu, A.J. Winkler, L.L. Jacobs Pleistocene deposits of caves in the Transvaal system. Palaeontologia africana 7(1960):59-118. - 1961b. On the fossil mammalian microfauna collected at Kromdraai by Draper in 1895. South African Journal of Science 56:259-260. Hendey, Q.B. 1981. Paleoecology of the late Tertiary fossil occurrences in 'E' Quarry, Langebaanweg, South Africa, and a re-interpretation of their geological context. Annals of the South African Museum 84:1-104. Hutterer, R., N. Lopez-Martinez, and J. Michaux. 1988. A new rodent from Quaternary deposits of the Canary Islands and its relationships with Neogene and Recent murids of Europe and Africa. Palaeovertebrata 18:241-262. Jacobs, L.L. 1978. Fossil rodents (Rhizomyidae & Muridae) from Neogene Siwalik deposits, Pakistan. Museum of Northern Arizona Press, Bulletin Series 52:1103. Jacobs, L.L. and W.R. Downs. 1994. The evolution of murine rodents in Asia. Pp. 149-156 in Y. Tomida, C.k. Li, and T. Setoguchi (eds), Rodent and Lagomorph Families of Asian Origins and Diversification. National Science Museum Monographs no. 8, Tokyo. Jacobs, L.L., L.J. Flynn, and W.R. Downs. 1989. Neogene rodents of southern Asia. Pp. 157-177 in C.C. Black and M.R. Dawson (eds), Papers on Fossil Rodents, in Honor of Albert Elmer Wood. Natural History Museum of Los Angeles County, Science Series no. 33. Jacobs, L.L., L.J. Flynn, W.R. Downs, and J.C. Barry. 1990. Quo Vadis, Antemus? The Siwalik muroid record. Pp. 573-586 in E.H. Lindsay, V. Falbusch, and P. Mein (eds), European Neogene Mammal Chronology. Plenum Press, New York. Kingdon, J. 1974. East African Mammals, vol. IIB, Hares and Rodents. University of Chicago Press, Chicago, 704 pp. Miller, G.S. 1912. Catalogue of the mammals of western Europe (exclusive of Russia) in the collection of the British Museum. British Museum (Natural History):1-801. Misonne, X. 1969. African and Indo-Australian Muridae, evolutionary trends. Musee Royal de l' Afrique Centrale, Tervuren, Annales, serie 8, 172:1-219. Musser, G.G., and M.D. Carleton. 1993. Family Muridae. Pp. 501-755 in D.E. Wilson and D.M. Reeder (eds), Mammal Species of the World. Smithsonian Institution Press, Washington, DC. Petter, F. 1983. Elements d'une revision des Acomys africains. Un sous-genre nouveau, Peracomys Petter et Roche, 1981 (Rongeurs, Murides). Annales du Musee Royal d' Afrique Centrale, Science and Zoology 237:109-119. Petter, F., and J. Roche. 1981. Remarques preliminaires sur la systematique des Acomys (Rongeurs, Murides). Peracomys, sous-genre nouveau. Mammalia 45:381-383. Pocock, T.N. 1987. Plio-Pleistocene fossil mammalian microfauna of southern Africa: a preliminary report including description of two new fossil muroid genera (Mammalia: Rodentia). Palaeontologia africana 26:69-91.

Is Acomys a murine? 675 Rosevear, D.R. 1969. The Rodents of West Africa. Trustees of the British Museum (Natural History), London, 604 pp. Sabatier, M. 1982. Les rongeurs du site Pliocene a hominides de Hadar (Ethiopie). Palaeovertebrata 12:1-56. Setzer, H.W. 1975. Part 6.5 Genus Acomys. Pp. 1-2 in J. Meester and H.W. Setzer (eds), The Mammals of Africa: An Identification Manual. Smithsonian Institution Press, Washington, DC. Wesselman, H .B. 1984. The Omo micromammals: systematics and paleoecology of early man sites from Ethiopia. Contributions to Vertebrate Evolution 7:1-219.