o ^ ^ Number 498 iJ5 3.x 24 December 2003 Contributions IN Science Geology and Vertebrate Paleontology OE THE Early Pliocene Site oe Kanapoi, Northern Kenya Edited by John M. Harris and Meave G. Leakey JAM 1 6 2004 Natural A ^History Museum ^ of Los Angeles County Serial Publications OL THE Natural History Museum of Los Angeles County The scientific publications of the Natural History Museum of Los Angeles County have been issued at irregular in- tervals in three major series; the issues in each series are numbered individually, and numbers run consecutively, re- gardless of the subject matter. • Contributions in Science, a miscellaneous series of tech- nical papers describing original research in the life and earth sciences. • Science Bulletin, a miscellaneous series of monographs describing original research in the life and earth scienc- es. This series was discontinued in 1978 with the issue of Numbers 29 and 30; monographs are now published by the Museum in Contributions in Science. • Science Series, long articles and collections of papers on natural history topics. Contact the Scholarly Publications Office at 213/763-3330 or visit our website at . A catalog is avail- able on request. Scientific Publications Committee John Lieyning, Deputy Director for Research and Collections John M. Harris, Committee Chairman Brian V. Brown Gordon Hendler Ines Horovitz Joel W. Martin K. Victoria Brown, Managing Editor Natural History Museum OF Los Angeles County 900 Exposition Boulevard Los Angeles, Calieornia 90007 Printed at Allen Press, Inc., Lawrence, Kansas ISSN 0459-8113 Geology and Vertebrate Paleontology oe the Early Pliocene Site oe Kanapoi, Northern Kenya Edited by John M. Harris^ and Meave G. Leakey^ TABLE OF CONTENTS Introduction 1 John M. Harris and Meave G. Leakey Stratigraphy and Depositional Setting of the Pliocene Kanapoi Formation, Lower Kerio Valley, Kenya 9 Craig S. Feibel Fossil Fish Remains from the Pliocene Kanapoi Site, Kenya 21 Kathlyn Stewart Early Pliocene Tetrapod Remains from Kanapoi, Lake Turkana Basin, Kenya 39 John M. Harris, Meave G. Leakey, and Thure E. Cerling with an Appendix by Alisa J. Winkler Carnivora from the Kanapoi Hominid Site, Turkana Basin, Northern Kenya 115 Lars Werdelin 1. George C. Page Museum, 5801 Wilshire Boulevard, Los Angeles, California 90036, USA. 2. National Museums of Kenya, PO Box 40658, Nairobi, Kenya. Contributions in Science, Number 498, pp. 1-132 Natural History Museum of Los Angeles County, 2003 V: 'i Introduction John M. Harris^ and Meave G. Leakey^ The site of Kanapoi lies to the southwest of Lake Turkana in northern Kenya (Fig, 1). Vertebrate fos- sils were recovered from Kanapoi in the 1960s by Harvard University expeditions and in the 1990s by National Museums of Kenya expeditions. The assemblage of vertebrate fossils from Kanapoi is both prolific and diverse and, because of its depo- sitional context of fluviatile and deltaic sediments that accumulated during a major lacustrine phase, exemplifies a time interval that is otherwise not well represented in the Lake Turkana Basin. Kanapoi has yielded one of the few well-dated early Pliocene assemblages from sub-Saharan Africa but hitherto only the hominins, proboscideans, perissodactyls, and suids recovered from this locality have received more than cursory treatment. The four papers pre- sented in this contribution document the geologic context and diversity of the Kanapoi fossil verte- brate biota. HISTORICAL CONTEXT The Lake Turkana Basin (formerly the Lake Rudolf Basin) traverses the western Kenya-Ethiopia border and has been an important source of Neogene ter- restrial vertebrate fossils since the early part of the twentieth century (Coppens and Howell, 1983) (Fig. 1). In 1888, Count Samuel Teleki von Szek and Ludwig Ritter von Hohnel were the first Eu- ropean explorers to reach the lake (Hohnel, 1938), which they named Lake Rudolf after Crown Prince Rudolf of Austria-Hungary (1859-89). The subse- quent French expedition of Bourg de Bozas (1902- 03) recovered vertebrate fossils from Plio-Pleisto-, cene exposures in the lower Omo Valley (Haug, 1912; Joleaud, 1920a, 1920b, 1928, 1930, 1933; Boulenger, 1920). This discovery prompted the Mission Scientifique de I’Omo (1932-33), which further documented the geology and paleontology of the area to the north of the Omo Delta (Aram- bourg, 1935, 1943, 1947). Allied military forces occupied southern Ethiopia during World War II; vertebrate fossils collected during the occupation were forwarded to the Coryndon Museum in Nai- robi (now the National Museums of Kenya) and in 1942 L.S.B. Leakey (honorary curator of the Cor- yndon Museum) sent his Kenyan staff to collect from the southern Ethiopian Omo deposits (Lea- key, 1943). Political unrest in both Kenya and Ethi- opia after the end of the Second World War pre- 1. George C. Page Museum, 5801 Wilshire Boulevard, Los Angeles, California 90036, USA. 2. National Museums of Kenya, PO Box 40658, Nai- robi, Kenya. Contributions in Science, Number 498, pp. 1-7 Natural History Museum of Los Angeles County, 2003 eluded further fieldwork in the area for more than a decade. In the mid-1960s, L.H. Robbins investigated the terminal Pleistocene and Holocene archaeology of the southwestern portion of the Lake Turkana Ba- sin (Robbins, 1967, 1972). Robbins let it be known that the region also contained somewhat older fos- sils and, in 1966, Bryan Patterson initiated a series of Harvard University expeditions to the region be- tween the lower Kerio and Turkwell Rivers. Patter- son’s expeditions focused initially on the Kanapoi region (1966-67) and subsequently on Lothagam (1967-72). Assemblages from the two localities shed much light on the late Miocene-early Pliocene vertebrate biota of sub-Saharan Africa and provid- ed the basis for monographic revisions of elephan- tids (Maglio, 1973), perissodactyls (Hooijer and Patterson, 1972; Hooijer and Maglio, 1974), and suids (Cooke and Ewer, 1972). The Patterson ex- peditions recovered few primate fossils but docu- mented a hominid mandible from Lothagam (Pat- terson et ah, 1970; Leakey and Walker, 2003) and a hominin humerus from Kanapoi (Patterson and Howells, 1967; Ward et ah, 2001). In 1967, a joint French, American, and Kenyan expedition (International Omo Research Expedi- tion) resumed exploration of Plio-Pleistocene ex- posures in the lower Omo Valley. In 1968, the Ken- yan contingent withdrew from the lORE to pros- pect the northeast shore of Lake Rudolf. The East Rudolf Research Project became the Koobi Fora Research Project when the Government of Kenya changed the name of the lake to Lake Turkana in 1975. The International Omo Research Expedi- tions (1967-76) and Koobi Fora Research Project (1968-78) recovered a great wealth of Plio-Pleis- tocene vertebrate fossils, including important new hominin material. Monographic treatment of ma- terial from the Omo Shungura sequence was pub- lished in the Cahiers de Paleontologie series edited by Y. Coppens and F. C. Howell (e.g., Eisenmann, 1985; Gentry, 1985; Eck and Jablonsky, 1987). That from Koobi Fora was published in the KFRP monograph series of Clarendon Press (Leakey and Leakey, 1978; Harris, 1983, 1991; Wood, 1994; Isaac, 1997). During the 1980s, National Museums of Kenya expeditions under the leadership of Richard Leakey explored the sedimentary exposures on the west side of Lake Turkana (Harris et al., 1988a, 1988b). Small but significant Plio-Pleistocene vertebrate as- semblages included the first cranium of Australo- pithecus aethiopicus (Walker et al., 1986) and a rel- atively complete skeleton of Homo ergaster (Brown et al., 1985; Walker and Leakey, 1993). 2 ■ CS 498, Harris and Leakey: Kanapoi Figure 1 Map of late Miocene through Pleistocene fossiliferous localities in the Lake Turkana Basin (after Harris et ah, 1988b) During the 1990s, National Museums of Kenya expeditions, now under the leadership of Meave Leakey, concentrated on the southwest portion of the Lake Turkana Basin, discovering new localities (Ward et ah, 1999) as well as revisiting Lothagam and Kanapoi. Lothagam was reworked from 1989 to 1993 and monographic treatment of the biota has now been published (Leakey and Harris, 2003). The Kanapoi locality was reprospected from 1993 to 1997 (Leakey et ah, 1995, 1998). Hominin ma- terial recovered by the National Museums of Kenya expeditions has been described in detail (Ward et Harris and Leakey: Introduction ■ 3 al., 2001); other recently recovered vertebrate spe- cies and their geologic setting provide the topic of this contribution. GEOLOGICAL CONTEXT The Lake Turkana Basin dates back to the early Pliocene. The present lake is sited in a closed basin that is fed year-round from the north by the Omo River, whose source is in the Ethiopian highlands and seasonally from the southwest by the Turkwel and Kerio Rivers and by other smaller ephemeral rivers. Paleogeographic reconstructions by Brown and Feibel (1991) indicate that, for much of the Pliocene, the Omo River flowed through the basin and directly into the Indian Ocean but occasional tectonic activity disrupted the outflow and resulted in short-lived temporary lakes. After about 1.9 Ma, the history of the region is still not clear. It is pos- sible that the river no longer exited through the southeastern part of the basin, yet mollusks flour- ished until at least 1.7 Ma ago, implying that wa- ters of the lake had not become as alkaline as they are at present. Indeed, mollusk-packed sands are reasonably common until at least 1.3 Ma ago (Har- ris et ah, 1988a), so the basin may have remained open until this time either at the southern end, or alternatively, the lake may have occasionally over- flowed to the northwest through Sanderson’s Gulf into the Nile catchment.The Plio-Pleistocene ter- restrial and lacustrine strata from the northern half of the basin form part of the Omo Group (Brown and Feibel, 1986) and are represented by the Shun- gura, Mursi, and Usno Formations in the lower Omo Valley (de Heinzelin, 1983), the Koobi Fora Formation on the northeast side of the lake (Brown and Feibel, 1991), and the Nachukui Formation on the northwest side of the lake (Harris et ah, 1988a). The Nachukui Formation extends to the southwest of the lake where, at Lothagam, it overlies the late Miocene Nawata Formation (Feibel, 2003a). Figure 2 lists the members of the Koobi Fora and Nachu- kui Formations in stratigraphic order. The oldest paleolake recognized in the basin is referred to as the Lonyumun Lake. It is documented by the lacustrine sediments of the Lonyumun Mem- ber, which was defined as the basal unit of the Koo- bi Fora Formation (Brown and Feibel, 1991) but also forms the basal unit of the Nachukui Forma- tion on the west side of the lake (Harris et ah, 1988a). The Lonyumun Lake is represented in the southwest part of the basin by the upper Apak and Muruongori members of the Nachukui Formation (Feibel, 2003a). The fossiliferous strata from Kan- apoi include a short-lived lacustrine episode that corresponds with the Lonyumun lacustrine interval. Feibel (2003b) interprets the fluvial sediments that enclose the lacustrine phase to have been deposited by the Kerio River and has named the sequence the Kanapoi Formation. The Pliocene strata of Kana- poi thus provide the oldest record of fluvial sedi- ments deposited by the Kerio River and include a deltaic tongue extending into the Lonyumun Lake. They thereby complement the fluvial sediments of the Kaiyumung Member of the Nachukui Forma- tion at the nearby locality of Lothagam that were evidently deposited by the Turkwel River (Feibel, 2003a). PALEONTOLOGICAL CONTEXT As exemplified at the nearby site of Lothagam (Lea- key et ah, 1996; Leakey and Harris, 2003), there was a drastic change in the terrestrial vertebrate biota of sub-Saharan Africa at the end of the Mio- cene due to faunal interchange between Africa and Eurasia, and coincident with the worldwide radia- tion of C4 vegetation (Ceding et ah, 1997). The Kanapoi biota, dated radiometrically between 4.17 and 4.07 Ma (Leakey et ah, 1995, 1998) lacks the large mammalian genera characteristic of the late Miocene at Lothagam — such as the amphicyonid carnivorans, the elephantids Stegotetrabelodon Pet- trochi, 1941 and Primelephas Maglio, 1970, the te- leoceratine rhino Brachypotherium Roger, 1904, the giraffid Palaeotragus Gaudrey, 1861, and bo- selaphin bovids (Leakey and Harris, 2003). Instead, the Kanapoi fauna demonstrates the first post-Mio- cene radiation of endemic African carnivorans (Werdelin, 2003) and a suite of ungulate species that is less progressive than that characteristic of late Pliocene exposures in the Lake Turkana Basin (Harris et ah, 2003). The Kanapoi fish assemblage (Stewart, 2003b) is similar to but less diverse than that from the temporally equivalent strata at Loth- agam (Stewart, 2003a). Partly because of the widespread nature of the Lonyumun Lake, fluvial sediments with vertebrate fossils representing that time interval are rare in the Lake Turkana Basin. Fossils from horizons imme- diately before and after the Lonyumun lacustrine interval at the nearby locality of Lothagam have been described recently (Leakey and Harris, 2003). A hominin-bearing vertebrate assemblage slightly younger than that from Kanapoi has been recov- ered from the Koobi Fora Formation in Allia Bay on the eastern shore of Lake Turkana but thus far only the hominins have been described in detail (Ward et ah, 2001). A few suid teeth from the Mur- si Formation, collected by the Kenyan contingent of the International Omo Research Expedition in 1967, suggests that the oldest formation in the Omo Group (de Heinzelin, 1983) is of broadly sim- ilar age to the Kanapoi Formation. There are sev- eral small assemblages that have been recovered from localities south of the Turkwel River (Eshua Kakurongori, Longarakak, Nakoret, Napudet, etc.) but these have yet to be fully prepared or studied in detail. OVERVIEW The four papers presented in this contribution treat different aspects of the geology and vertebrate pa- leontology of the northern Kenyan locality of Kan- 4 ■ CS 498, Harris and Leakey: Kanapoi Nachukui Fm (WT) Nariokotome Mb Natoo Mb Kaitio Mb Kalachoro Mb Lokalalei Mb Lomekwi Mb Kataboi Mb Lonyuniun Mb Nachukui Fm (LT) Kaitio Mb Kalachoro Mb Kaiyumung Mb Miiruongori Mb and Lothagam Basalt Apak Mb Kanapoi Fm Upper fluvial interval Laeustrine interval Lower fluvial interval Koobi Fora Fm Chari Mb Okote Mb KBS Mb Burgi Mb Tulu Bor Mb Lokochot Mb Moiti Mb Lonyumun Mb Figure 2 Stratigraphic sequence of the formal and informal members of the Kanapoi Formation, the Koobi Fora For- mation and the Nachukui Formation where exposed in West Turkana (WT) and Lothagam (LT); for correlative details, see Harris et al. (1988b: fig. 4) and Feibel (2003a, 2003b) apoi. However, their appearance together in a sin- gle publication will provide a useful source of ref- erence for this interesting site. Feibel describes the stratigraphy and erects a new formation for the Kanapoi succession. The environ- mental setting recorded by the Kanapoi sedimen- tary sequence reflects a progression of fluvial and lacustrine systems that overwhelmed a volcanic landscape. He interprets the vertebrate-bearing flu- vial sediments to have formed part of the Kerio River system as it entered the Lonyumun Lake just over 4 million years ago. The high degree of land- scape heterogeneity and pronounced soil catenas of the Kanapoi setting are indicative of a great mosaic of habitats in the southwestern part of the Turkana Basin during the early Pliocene. Stewart describes the nearly 3,000 fish elements recovered from lacustrine sediments at Kanapoi during the early 1990s. The Kanapoi fish fauna mainly comprises large piscivores and medium to large molluscivores. The paucity of herbivorous fish such as mormyroids, Barbus Cuvier and Cloquet, 1816, Alestes, and distichodids is a little unexpect- ed. While Barbus Muller and Troschel, 1841, and large tilapiine cichlids are scarce in African fossil deposits prior to the Pleistocene (Stewart, 2001), the other groups are represented in the Lothagam succession and one would expect them to be pre- sent in the Pliocene lake. The Kanapoi assemblage has many similarities with that of the Muruongori Member from the Lothagam succession. However, differences in representation of alestid and tetrao- dontid species suggest either that the Kanapoi la- custrine phase correlates temporally more closely with the Apak Member in the Lothagam sequence or that the Kanapoi and Muruongori fish assem- blages sample different habitats. Stewart interprets the Kanapoi lake to be well oxygenated and non- saline; the scarcity of lungfish, bichirs and Heterotis Ruppell, 1829 all of which were well represented in the Nawata Formation at Lothagam, could sig- nify an absence of well-vegetated backwaters or bays. Harris, Leakey, and Cerling document the diver- sity of tetrapods (exclusive of carnivorans) that have been recovered from Kanapoi. The mamma- lian fauna provides a standard for the early Plio- cene in East Africa, with the cercopithecid, ele- phantid, rhinocerotid, suid, giraffid, and bovid spe- cies providing a link between those from upper Miocene levels at Lothagam and those in late Pli- ocene assemblages from elsewhere in the Lake Tur- kana Basin. Even though the microfauna has yet to be studied in detail, the Kanapoi mammalian biota is already larger and more diverse than the prelim- inary report of mammals from the slightly older site of Aramis in Ethiopia or from the Nachukui Eor- mation members at Lothagam. Kanapoi is the type locality for the oldest East African australopithe- cine species yet recognized, Australopithecus ana- mensis (Leakey et ah, 1995), so the Kanapoi biota is of interest for the information it provides about environments in which early bipedal hominins lived. No taphonomic investigation has yet been undertaken at the Kanapoi locality but, as pointed out by Behrensmeyer (1991), broad-scaled paleoen- vironmental reconstructions based on the presence of taxa are likely to be accurate despite the taph- onomic history of the assemblage. The paleosols from the Kanapoi succession sug- gest a suite of habitats similar to those currently found in the vicinity of the modern Omo Delta at the north end of Lake Turkana. On the basis of their modern counterparts, the Kanapoi herbivores suggest a relatively dry climate and a mixture of woodland and open grassland. However, ecological structure analysis (cf. Reed, 1999) suggests closed woodland, and thus is closer to the wooded habitat interpreted for the slightly older hominin Ardipi- thecus ramidus (White et ah, 1994) from Aramis in Ethiopia (WoldeGabriel et ah, 1994). An appendix Harris and Leakey: Introduction ■ 5 by Winkler provides a brief preliminary report on the micromammals. Werdelin describes the carnivoran component of the Kanapoi biota, which is larger and more diverse than those from most Pliocene localities in eastern Africa and provides a substantial addition to our knowledge of early Pliocene African Carnivora. It shares a number of species with the slightly older Langebaanweg (South Africa) and the slightly younger Laetoli (Tanzania), but the overall mixture of species is unique to Kanapoi. The late Miocene Nawata Formation at Lothagam has yielded a number of carnivorans that were evidently mi- grants from Eurasia. The carnivoran assemblage from Langebaanweg also includes a number of rel- ict Miocene forms but that from Kanapoi includes only forms whose immediate forebears are found in Africa. Kanapoi, therefore, provides evidence for the first post-Miocene radiation of endemic African carnivorans. SUMMARY The locality of Kanapoi is significant in that it has yielded an early Pliocene assemblage that includes representatives of the earliest East African species of Australopithecus Dart, 1925, and the vertebrate biota has the potential for providing a detailed pic- ture of the environments exploited by early bipedal hominins. The assemblage is derived from fluvial and lacustrine sediments that are tightly con- strained between tephra dated at 4.17 and 4.07 Ma. Paleosols in the sequence indicate the presence of terrestrial habitats that are today found at the north of Lake Turkana in the vicinity of the Omo Delta. In particular, they indicate the presence of a significant quantity of grass, given that the propor- tion of soil carbonate derived from C4 plants varies from 25% to 40% in the paleosols associated with terrestrial fossils (Wynn, 2000). Much of the terrestrial vertebrate assemblage was collected via surface prospecting and no de- tailed taphonomical investigations have yet been undertaken. Nevertheless, preliminary investigation of the mammalian fossils provides support for the environmental interpretations derived from the pa- leosols. Grazing mammals outnumber browsing forms by nearly two to one in terms of numbers of species and by three to one in terms of numbers of specimens. The microfauna has yet to be studied in detail, but initial investigation of some rodent spe- cies suggests they represent dry and open habitats (see Appendix in Harris et al., 2003). However, ecological structure analysis of the kind advocated by Reed (1997) suggests that the Kanapoi assem- blage may instead be indicative of closed woodland as represented at Lothagam by the Kaiyumung Member of the Nachukui Eormation or in the low- er Omo Valley by Member B of the Shungura Eor- mation. This apparent conflict of interpretation has yet to be resolved but may also be indicative that the habitats present in the region during the initial formation of the Turkana Basin may not be directly comparable with the modern habitats now char- acteristic of eastern Africa. ACKNOWLEDGMENTS This introductory section was compiled at the suggestion of the Scientific Publications Committee of the Natural History Museum of Los Angeles County. We are grateful to John C. Barry, Francis H. Brown, Peter Ditchfield, Peter L. Forey, Nina Jablonski, Alison Murray, Olga Otero, Blaire Van Valkenberg, Xiaoming Wang, Tim White, and an anonymous referee for helpful comments. LITERATURE CITED Arambourg, C. 1935. Mission Scientifique de I’Omo (1932-1933): Geologie-Anthropologie-Paleontolo- gie. Vol. 1, fasc. 1, pp. 1-59. . 1943. Mission Scientifique de I’Omo (1932- 1933): Geologie-Anthropologie. Vol. 1, fasc. 2, pp. 60-230. Memoire du Museum national d’Histoire naturelle (Paris). . 1947. Mission Scientifique de POmo (1932- 1933): Paleontologie. Vol. 1, fasc. 3, pp. 232-562. Memoire du Museum national d’Histoire naturelle (Paris). Behrensmeyer, A. K. 1991. Terrestrial vertebrate accu- mulations. In Taphonomy releasing data locked in the fossil record, eds. P. E. Allison and D. E. G. Briggs, 291-335. New York: Plenum. Boulenger, G. A. 1920. Sur le gavial fossile de I’Omo. Comptes-Rendues de I’Academie des Sciences, Paris 170:913-914. Brown, F. H., and C. S. Feibel. 1986. Revision of litho- stratigraphic nomenclature in the Koobi Fora region, Kenya. Journal of the Geological Society 143:297- 310. . 1991. Stratigraphy, depositional environments and paleogeography of the Koobi Fora Formation. In The fossil ungulates: Geology, fossil artiodactyls and palaeoenvironments, Koobi Fora Research Pro- ject, Vol. 3. ed. J. M. Harris, 1-30. Oxford: Clar- endon Press. Brown, E, J. Harris, R. Leakey, and A. Walker. 1985. Ear- ly Homo erectus skeleton from West Lake Turkana, Kenya. Nature 316:788-792. Cerling, T. E., J. M. Harris, B. J. MacFadden, M. G. Lea- key, J. Quade, V. Eisenmann, and J. R. Ehleringer. 1997. Global vegetation change through the Mio- cene/Pliocene boundary. Nature 389:153-158. Gooke, H. B. S., and R. P. Ewer. 1972. Eossil Suidae from Kanapoi and Lothagam, Kenya. Bulletin of the Mu- seum of Comparative Zoology Harvard 143(3):149- 295. Coppens, Y., and E C. Howell. 1983. Introduction. In The Omo Group: Archives of the International Omo Re- search Expedition; Musee Royal de I’Afrique Cen- tral— Tervuren, Belgique, J. de Heinzelin, ed., An- nales-serie in-8°-sciences geologiques 85:1-5. de Heinzelin, J. 1983. The Omo Group: Archives of the International Omo Research Expedition; Musee Royal de I’Afrique Centrale — Tervuren, Belgique. Annales Serie in-S° -Sciences Geologiques 85:365 pp. Eck, G. G., and N. G. Jablonsky. 1987. The skull of Ther- opithecus brumpti compared with those of other spe- cies of the genus Theropithecus. In Les faunes plio- pleistocenes de la basse vallee de POmo (Ethiopie), 6 ■ CS 498, Harris and Leakey: Kanapoi eds. Y. Coppens and F. C. Howell, 3:18-142. Paris: Cahiers de Paleontologie. Eisenmann, V. 1985. Les equides des gisements de la vallee de POmo en Ethiopie. In Les faunes plio-pleistocenes de la basse vallee de I’Omo (Ethiopie), eds. Y. Cop- pens and E. C. Howell, 1:15-56. Paris: Cahiers de Paleontologie. Eeibel, C. S. 2003a. Stratigraphy and depositional history of the Lothagam sequence. In Lothagam: The dawn of humanity in Africa, eds. M. G. Leakey and J. M. Harris, 17-29. New York: Columbia University Press. . 2003b. Stratigraphy and depositional setting of the Pliocene Kanapoi Formation, Lower Kerio Val- ley, Kenya. In Geology and vertebrate paleontology of the Early Pliocene site of Kanapoi, northern Ken- ya. Edited by John M. Harris and Meave G. Leakey. Contributions in Science 498:9-20. Gentry, A. W. 1985. The Bovidae of the Omo Group de- posits. In Les faunes plio-pleistocenes de la basse val- lee de I’Omo (Ethiopie), eds. Y. Goppens and F. C. Howell, 1:119-191. Cahiers de Paleontologie. Harris, J. M., ed. 1983. The fossil ungulates: Prohoscidea, Perissodactyla and Suidae. Koohi Pora Research Project, Vol. 2. Oxford: Clarendon Press. , ed. 1991. The fossil ungulates: Geology, fossil ar- tiodactyls and palaeoenvironments. Koobi Pora Re- search Project, Vol. 3. Oxford: Clarendon Press. Harris, J. M., F. H. Brown, and M. G. Leakey. 1988a. Geology and palaeontology of Plio-Pleistocene lo- calities west of Lake Turkana, Kenya. Contributions in Science 399:1-128. Harris, J. M., F. H. Brown, M. G. Leakey, A. C. Walker, and R. E. Leakey. 1988b. Pliocene and Pleistocene hominid-bearing sites from west of Lake Turkana, Kenya. Science 239:27-33. Harris, J. M., M. G. Leakey, T. E. Cerling, and A. J. Wink- ler. 2003. Early Pliocene tetrapod remains from Kan- apoi, Lake Turkana Basin, Kenya. In Geology and vertebrate paleontology of the Early Pliocene site of Kanapoi, northern Kenya. Edited by John M. Harris and Meave G. Leakey. Contributions in Science 498: 39-113. Haug, E. 1912. Trade de Geologie. Vol. 2. Les periodes geologiques. Paris: Armand Colin. Hohnel, L. von. 1938. The Lake Rudolf region: Its dis- covery and subsequent exploration 1888-1909. Journal of the Royal African Society of London 37: 16-45, 206-226. Hooijer, D. A., and V. J. Maglio. 1974. The earliest hip- parion south of the Sahara in the late Miocene of Kenya. Proceedings of the Koninklijke Nederlandse Akademie van Wetenschappen Series B 76:311-319. Hooijer, D. A., and B. Patterson. 1972. Rhinoceroses from the Pliocene of northwestern Kenya. Bulletin of the Museum of Comparative Zoology 144(l):l-26. Isaac, G. LL. 1997. Plio-Pleistocene Archaeology. Koobi Pora Research Project, Vol. 5. Oxford: Oxford Uni- versity Press. Joleaud, L. 1920a. Gontribution a Petude des hippopota- mes fossiles. Bulletin de la Societe Geologique de France, ser. 4, 20:13-26. . 1920b. Sur la presence dTin gavialide du genre Tomistoma dans le Pliocene d’eau douce de I’Ethiopie. Comptes-Rendus de la Academie de Sci- ence, Paris 170:816-818. . 1928. Elephants et dinotheriums pliocenes de PEthiopie: Gontribution a Petude paleogeographique des proboscidiens africains. International Geological Congress, Madrid, 14th session, 3:1001-1107. . 1930. Les crocodiliens du Pliocene d’eau douce de POmo (Ethiopie): contribution a Petude paleogeo- graphique des Tomistoma et les crocodiles a museau de gavial. Societe Geologique de France, Pivre Ju- bilaire 1830-1930, 2:411-423. . 1933. Un nouveau genre d’equide quaternaire de POmo (Abyssinie): Pibyhipparion ethiopicum. Bul- letin de la Societe geologique de France, ser. 5, 3:7- 28. Leakey, L. S. B. 1943. New fossil Suidae from Shungura, Omo. Journal of the East African Natural History Society 17:45-61. Leakey, M. G., C. S. Eeibel, R. L. Bernor, J. M. Harris, T. E. Cerling, I. McDougall, K. M. Stewart, A. Walker, L. Werdelin, and A. J. Winkler. 1996. Lothagam: A record of faunal change in the Late Miocene of East Africa. Journal of Vertebrate Paleontology 16(3): 556-570. Leakey, M. G., C. S. Feibel, I. McDougall, and A. Walker. 1995. New four-million-year-old hominid species from Kanapoi and Allia Bay, Kenya. Nature 376: 565-571. Leakey, M. G., C. S. Feibel, I. McDougall, C. Ward, and A. Walker. 1998. New specimens and confirmation of an early age for Australopithecus anamensis. Na- ture 393:62-66. Leakey, M. G., and J. M. Harris, eds. 2003. Lothagam: The dawn of humanity in Africa. New York: Colum- bia University Press. Leakey, M. G., and R. E. Leakey, eds. 1978. The fossil hominids and an introduction to their context, 1968-1974. Koobi Fora Research Project, Vol. 1. Oxford: Clarendon Press. Leakey, M. G., and A. C. Walker. 2003. The Lothagam hominoids. In Lothagam: The dawn of humanity in Africa, eds. M. G. Leakey and J. M. Harris, 249- 257. New York: Columbia University Press. Maglio, V. J. 1973. Origin and evolution of the Elephan- tidae. Transactions of the American Philosophical Society, Philadelphia, n.s. 63:1-149. Patterson, B., A. K. Behrensmeyer, and W. D. Sill. 1970. Geology and fauna of a new Pliocene locality in north-western Kenya. Nature 226:918-921. Patterson, B., and W. W. Howells. 1967. Hominid hu- meral fragment from Early Pleistocene of North- western Kenya. Science 156:64-66. Reed, K. A. 1997. Early hominid evolution and ecological change through the African Plio-Pleristocene. Jour- nal of Human Evolution 32:289-322. Robbins, L. H. 1967. A recent archaeological discovery in the Turkana District of northern Kenya. Azania II: 1-5. . 1972. Archeology in the Turkana District, Kenya. Science 176:359-366. Stewart, K. M. 2001. The freshwater fish of Neogene Af- rica (Miocene-Pleistocene): Systematics and bioge- ography. Fish and Fisheries 2:177-230. . 2003a. Fossil fish remains from the Mio-Pliocene deposits at Lothagam, Kenya. In Lothagam: The dawn of humanity in Africa, eds. M. G. Leakey and J. M. Harris, 75-111. New York: Columbia Univer- sity Press. . 2003b. Fossil fish remains from the Pliocene Kan- apoi Site, Kenya. In Geology and vertebrate pale- ontology of the Early Pliocene site of Kanapoi, northern Kenya. Edited by John M. Harris and Harris and Leakey: Introduction ■ 7 Meave G. Leakey. Contributions in Science 498:21- 38. Walker, A., R. Leakey, J. Harris, and R Brown. 1986. 2.5 myr Australopithecus boisei from west of Lake Tur- kana, Kenya. Nature 322:517-522. Walker, A. C., and R. E. Leakey, eds. 1993. The Narioko- tome Homo erectus skeleton. Cambridge: Harvard University Press. Ward, C. V., M. G. Leakey, B. Brown, R Brown, J. Harris, and A. Walker. 1999. South Turkwel: A new Plio- cene hominid site in Kenya. Journal of Human Evo- lution 36:69-95. Ward, C. V., M. G. Leakey, and A. C. Walker. 2001. Mor- phology of Australopithecus anamensis from Kana- poi and Allia Bay, Kenya. Journal of Human Evo- lution 4:255-368. Werdelin, L. 2003. Carnivora from the Kanapoi Hominid Site, Turkana Basin, Northern Kenya. In Geology and vertebrate paleontology of the Early Pliocene site of Kanapoi, northern Kenya. Edited by John M. Harris and Meave G. Leakey. Contributions in Sci- ence 498:115-132. White, T. D., G. Suwa, and B. Asfaw. 1994. Australopi- thecus ramidus, a new species of early hominid from Aramis, Ethiopia. Nature 371:306-312. WoldeGabriel, G., T. D. White, G. Suwa, P. Renne, J. de Heinzelin, W. K. Hart, and G. Heiken. 1994. Eco- logical and temporal placement of Early Pliocene hominids at Aramis, Ethiopia. Nature 371:330-333. Wood, B. A. 1994. Hominid cranial remains. Koobi Fora Research Project, Vol. 4. Oxford: Clarendon Press. Wynn, J. G. 2000. Paleosols, stable carbon isotopes, and paleoenvironmental interpretation of Kanapoi, northern Kenya. Journal of Human Evolution 39: 411-432. Received 26 December 2002; accepted 23 May 2003. Stratigraphy and Depositional Setting of the Pliocene Kanapoi Formation, Lower Kerio Valley, Kenya Craig S. Feibel^ ABSTRACT. The Pliocene sedimentary sequence at Kanapoi is attributed to the Kanapoi Formation, newly defined here. The formation consists of three sedimentary intervals, a lower fluvial sequence, a lacustrine phase, and an upper fluvial sequence. The entire formation is strongly influenced by paleotopography developed on the underlying Mio-Pliocene basalts, with a landscape of rounded hills and up to 40 m in local relief. The lower fluvial interval is dominated by conglomerates, sandstones, and pedogenically mod- ified mudstones. Two altered pumiceous tephra occur within this interval. A sharp contact marks the transition to fully lacustrine conditions. This interval is characterized by laminated claystones and siltstones, lenticular sand bodies, and abundant ostracods, mollusks, and carbonized plant remains. A single vitric tephra, the Kanapoi Tuff, occurs within this interval. A return to fluvial conditions is recorded first by upward fining cycles reflecting a meandering river system. This is succeeded by deeply incised conglomerates and sands of a braidpiain, capped by the Kalokwanya Basalt. The Kanapoi Formation is richly fossiliferous, and has yielded the type specimen as well as much of the hypodigm of Australopithecus anamensis. Ver- tebrate fossils derive primarily from two depositional settings within the formation: vertic floodplain pa- leosols and deltaic sand bodies. These reflect successional stages in the development of a major tributary system in the Turkana Basin during the early Pliocene. INTRODUCTION The Pliocene sedimentary sequence at Kanapoi pre- sents a complex record of fluvial and lacustrine strata deposited over a landscape of considerable local relief (up to 40 m) on Mio-Pliocene voicanics. Early fluvial systems accumulated predominantly overbank mudstones, with a well-developed soil overprint, associated with lenticular sands and gravels. A lacustrine phase, the Lonyumnn Lake, in the middle of the sequence is marked by laminated claystones and molluskan bioherms, with thick del- taic sand bodies. Following local infilling of the lake, a fluvial regime is again represented. The top of the sedimentary interval is dominated by a thick and deeply incised conglomeratic unit that accu- mulated prior to capping of the entire sequence by the Kalokwanya Basalt. Vertebrate fossils are found throughout the sed- imentary sequence, being particularly abundant in the deltaic sand bodies, but are also found in pa- leosols of the lower and upper fluvial sequences. Fossil invertebrates are common in the lacustrine facies, though the quality of preservation tends to be poor. Lacustrine mudstones preserve abundant plant impressions and carbonized remains at sev- eral levels. Isotopic age determinations on materials from Kanapoi by I. McDougall of the Australia National University (Leakey et ah, 1995, 1998) established a precise chronostratigraphy for the sequence. The 1. Departments of Anthropology and Geological Sci- ences, Rutgers University, 131 George Street, New Bruns- wick, New Jersey 08901, USA. major phase of deposition is constrained to fall be- tween 4.17 and 4.07 Ma, and the capping Kalok- wanya Basalt is placed at 3.4 Ma. The Kanapoi deposits reflect an early stage of accumulation with- in the developing Plio-Pleistocene Turkana Basin. The geological investigations reported here were conducted over eight visits to Kanapoi between 1992 and 1996. Field mapping and 21 stratigraphic sections are the basis for a formal definition of the Kanapoi Formation presented here. Analysis of de- positional environments, postdepositional modifi- cation, and sedimentary architecture are the basis for a reconstruction of the environmental setting for the rich Kanapoi fossil assemblage. BACKGROUND The sedimentary strata around Kanapoi were first described by Patterson (1966). He recognized many of the important features of the local geology. The interval he described was predominantly lacustrine in character and comprises the middle unit in the sequence described here. No formal stratigraphic terminology or subdivision of the strata was pro- posed, although a section measuring 175 ft. (53.3 m) is mentioned. The first report of a hominid fossil from Kanapoi by Patterson and Howells (1967) in- cluded a few additional observations on the geol- ogy of the sequence. Patterson et al. (1970) discussed the Kanapoi fauna, and used the term ‘Kanapoi Formation’ for the sequence, but provided no descriptions, sec- tions, or type locality. The most detailed geological work conducted prior to the 1990s was Powers’ Contributions in Science, Number 498, pp. 9-20 Natural History Museum of Los Angeles County, 2003 10 ■ CS 498, Harris and Leakey: Kanapoi (1980) investigation of strata of the Lower Kerio Valley. He provided sections and descriptions of the sedimentary strata at Kanapoi, as well as an inter- pretation of depositional environments and post- depositional modification. Most of the early dis- cussion of Kanapoi centered around attempts to date the sequence, including isotopic age determi- nations on the overlying Kalokwanya Basalt, as well as biostratigraphic comparisons. The systematic field work undertaken by the Na- tional Museums of Kenya in the early 1990s, under the direction of M. G. Leakey, led to important new fossil discoveries, a reinvestigation of the sedimen- tary sequence, and establishment of detailed chron- ostratigraphic control (Leakey et ah, 1995, 1998). A detailed analysis of the numerous paleosols in the Kanapoi sequence was reported by Wynn (2000). EXPOSURE AND STRUCTURE The entire sedimentary sequence at Kanapoi dips very gently (~1°) to the west. Local depositional dips, however, can be quite high. These are com- monly 12-15° at some distance above the base- ment, and may reach 45° where sediments are draped directly over hills in the volcanic basement. Several small faults (0. 5-1.0 m offset) occur in the study area, and in the southeast, a more significant normal fault (bearing 310°, down to NE) offsets the section by several tens of meters. For the most part, however, the sequence is much more strongly af- fected by deposition over pre-existing topography than by subsequent tectonics. THE KANAPOI EORMATION The Pliocene sedimentary rocks exposed in the Kanapoi region (Fig. 1) are here defined as the Kan- apoi Formation. The type section of the formation, section CSF 95-8 (Fig. 2), is located in the south- eastern part of the exposures and displays most of the major characteristics of the formation. Where exposed, the base of the formation rests uncon- formably on Mio-Pliocene basalts. The Pliocene Kalokwanya Basalt unconformably caps the for- mation. In the type section, the Kanapoi Formation is 37.3 m thick. Some local sections are known to reach nearly 60 m in thickness, and the formation can be seen to pinch out entirely between the ba- salts to the east and north. The new formation designation is justified on both lithostratigraphic and historical grounds. The formation is mappable and lithologically distinc- tive. Unifying characteristics include the dominance of paleotopographic influence in sedimentary ac- cumulation pattern and an early basaltic clast dom- inance later replaced by silicic volcanics. The single tephrostratigraphic marker within the formation that has been geochemically characterized is the Kanapoi Tuff (Leakey et ah, 1998). The formation is related to synchronic deposits of the Turkana ba- sin farther north, but historical usage, complex re- lationships, and lack of correlative marker tephra (boundary stratotypes) preclude assignment to any previously defined stratigraphic units. The lower portion of the formation likely correlates with the upper Apak Member of the Nachukui Formation described from Lothagam (Feibel, 2003). The la- custrine interval in the middle of the Kanapoi For- mation is correlative with the lower Lonyumun Member of the Nachukui and Koobi Fora Forma- tions (Brown and Feibel, 1986; Harris et ah, 1988). The upper sedimentary interval in the Kanapoi For- mation corresponds broadly to lower members of the Omo Group formations to the north (Moiti and Lokochot Members of the Koobi Fora Formation or Kataboi Member of the Nachukui Formation). Three tephra units within the Kanapoi Formation have been isotopically dated. Two devitrified pu- miceous tephra low in the sequence yielded ages of 4.17 ± 0.03 Ma (lower pumiceous tuff) and 4.12 ± 0.02 Ma (upper pumiceous tuff) (Leakey et ah, 1995), while the vitric Kanapoi Tuff was found to contain rare pumices, which were dated to 4.07 ± 0.02 Ma (Leakey et ah, 1998). In addition, the overlying Kalokwanya Basalt has been dated to 3.4 Ma, providing an upper limit on the age of the for- mation. The onset of accumulation is estimated to have begun around 4.3 Ma. Most of the subcon- glomeratic sequence likely accumulated prior to 3.9 Ma, but the sedimentary environments responsible for accumulation of the uppermost Kanapoi strata were likely active up until extrusion of the Kalok- wanya Basalt. The Kanapoi sedimentary sequence was depos- ited on a dissected volcanic landscape with at least 40 m of local relief. This basal topography had a strong influence on the lateral variability of the se- quence and disrupted the sedimentation pattern through nearly the entire stratigraphic thickness. The lowermost stratigraphic units are localized within paleotopographic lows, while the super- posed strata become more and more laterally con- tinuous upwards. The onlapping sequence of sedi- ments is complex, as the strata were deposited more-or-less horizontally, against this topographic surface. The overall stratigraphic sequence of the Kanapoi Formation reflects three intervals, an ini- tial fluvial regime, a subsequent lacustrine phase, and a final return to fluvial conditions. Local basement for the Kanapoi Formation con- sists of Mio-Pliocene basalts. They are typically spheroidally weathered, and present a landscape of conical hills, many of which protrude through the eroding sedimentary sequence today (Fig. 1). There is considerable variation in the nature of the con- tact between the local basement and the Kanapoi Formation, primarily as a function of paleotopo- graphic position. The most common association, seen in paleotopographic lows as well as at other positions, is a scoured relationship, which super- poses basalt cobble- to pebble-conglomerates, or less frequently sandstones, on basalt basement (Fig. 2: sections CSF 95-8, 95-10). Slightly higher paleo- topographic positions sometimes preserve a gravel Feibel: Stratigraphy 111 Figure 1 Geological map of the Kanapoi area showing prominent geographic landmarks and locations of the stratigraphic sections 12 B CS 498, Harris and Leakey: Kanapoi Figure 2 Type section (CSF 95-8) and reference sections of the Kanapoi Formation. Numbered correlations shown are 1, lower pumiceous tuff; 2, upper pumiceous tuff; 3, basal flooding surface of the Lonyumun Lake sequence; and 4, Kanapoi Tuff. See Figure 1 for location of sections. For a key to symbols, see Figure 3 regolith developed on the basalt, along with a blocky structured paleosol developed on silts or clays (Fig. 4; sections CSF 96-10, 95-13). At even higher paleotopographic positions, corresponding to the landscape exposed at the time of inundation by the Lonyumun Lake, molluskan packstones rep- resenting bioherms up to 1 m in thickness are de- veloped on what were then islands of the basalt basement. The highest of the paleotopographic hills preserves a pebbly clay paleosol that has been con- tact baked upon extrusion of the capping Kalok- wanya Basalt (east of CSF 96-8). The initial sedimentary interval of the Kanapoi Formation, informally termed the lower fluvial se- quence, can be constrained between the Mio-Plio- cene basalts below and the lacustrine sequence above. The base of the lacustrine sequence is a sharp boundary in virtually all sections and is easily rec- ognizable by an abundance of ostracods, carbonized plant fragments, and/or mollusks. The lower fluvial sequence is characterized by conglomerates, sand- stones, and claystones with well-developed vertic (paleosol) structure. The conglomerates are generally massive, basalt cobble to pebble units. Sandstones are medium- to fine-grained, quartzofeldspathic or litharenitic, and commonly display well-developed planar crossbedding in 10-20 cm bedsets. Large- scale trough crossbedding is locally seen in coarser sandstones, while the finer grained sands and upper portions of sand bodies are typically massive due to bioturbation. Mudstones are generally quite thick in the lower fluvial sequence (up to 10 m; sections CSF 95-10 in Fig. 2 and 95-5 in Fig. 5), with well-devel- oped paleosols. Wynn (2000) has provided a detailed analysis of paleosols throughout the formation. The most common paleosol is the Aberegaiya pedotype, a thick, often cumulative vertisol with well-devel- oped wedge-shaped peds and slickensided dish frac- tures. This lower sedimentary sequence records a de- positional regime controlled by fluvial systems, and Feibel: Stratigraphy ■ 13 KEY TO SYMBOLS Lithologic Columns Tephra nn CSF 94-7 Section Number 10 Conglomerate Tuff Sand Silt Limestone Clay Basalt c z s g /t V tuff A bentonite © pumice © altered pumice Fossils Sedimentology cobbles/boulders trough cross-strata epsilon cross-strata planar cross-strata gravel lag erosional scour upward-fining unit thinly bedded upward-coarsening unit laminated mammal bone fish M Q; molluscs Etheria reef 0 ostracods w wood plant remains rhizoliths Other Pedogenic Features c vertic structure soil carbonate nodules prismatic/blocky structure flooding surface correlation line Figure 3 Key to symbols for the graphic sections presented in this report there are indications of both braided and meander- ing streams based on internal sequences and primary structures. Several tephra units are intercalated within the lower fluvial sequence. The two most prominent of these display characteristics of airfall tephra that mantled the Kanapoi paleolandscape. The lower pumiceous tephra layer is a thick (up to 3.6 m), poorly sorted unit, with altered angular pumice clasts to 1 cm in diameter scattered throughout. 14 ■ CS 498, Harris and Leakey: Kanapoi CSF 96-10 m CSF 95-13 m c z s g /t Figure 4 Reference sections from the basal contact of the Kanapoi Formation. See Figure 1 for location of sections. For a key to symbols, see Figure 3. The upper pumiceous tephra layer is a thinner unit (ca. 15 cm), and displays laminated basal and up- per subunits with an unsorted pumiceous middle. The vitric component of these tephra has been com- pletely altered to clay and zeolite minerals. Both, however, had a significant pumiceous component. The pumices have been devitrified and slightly flat- tened, but appear as clay pebbles dispersed throughout the units. Devitrification of the pumices has left a residual population of volcanogenic feld- spar crystals, which have been used to control for the age of the strata and associated fossils. Overlying the lower fluvial sequence and locally banked against the higher elements of the eroded volcanic basement is a lacustrine interval. Litho- stratigraphic, chronostratigraphic, and biofacies in- dicators all support correlation of this lacustrine in- terval with the Lonyumun Lake phase well known from the Omo Group deposits of the northern Tur- kana Basin (Brown and Feibel, 1991; Feibel et ah, 1991) as well as from Lothagam (Feibel, 2003) and elsewhere in the lower Kerio Valley (Feibel, unpub- lished). Where the volcanic basement produced lo- cal islands in this lake, they are mantled by a mol- luskan packstone, dominated by the gastropod Bel- lamy a Jousseaume, 1886. Elsewhere, the lacustrine strata begin with a mollusk- and ostracod-rich clay- stone, typically succeeded by a well-laminated clay- stone and siltstone sequence, and continue with an upward coarsening sequence, which is capped by distributary channel sands. The upper portion of the deltaic complex has isolated sand bodies, rep- resenting distributary channels. This deltaic com- plex contains the only vitric tuff preserved at Kan- apoi. This unit, termed the Kanapoi Tuff (Leakey et ah, 1998), is a pale brown, fine-grained tuff with well-preserved climbing-ripple cross-lamination. Upper portions of the tuff commonly show soft- sediment deformation, and in a few localities, the tuff preserves pumice. The composition of this tephra indicates an iron-rich rhyolite (Table 1). Al- though this tephra does not correlate with any of the well-known tephra of the Turkana Basin Omo Group sequence, Namwamba (1993) has suggested that it correlates with his Suteijun Tuff of the Chemeron Formation in the Baringo Basin to the south. Above the Kanapoi Tuff, lacustrine conditions persisted locally for a short period. In an important locality west of Akai-ititi, a distributary channel se- quence is cut into the Kanapoi Tuff (Fig. 6). Here the eroded channel base is draped with a molluskan packstone that includes a well-developed reef of the Nile oyster Etheria Lamarck, 1807. The remainder of the channel is filled with a quartzofeldspathic sand. The record of Etheria in a channel setting documents the perennial nature of the river at this time. The transition from the lacustrine interval to the upper fluvial interval is not sharp, as in the base of the lacustrine sequence, but rather proceeds through an interval of interbedded shallow lacus- trine muds and those with a clear pedogenic over- print indicating exposure. There are also several moderate to well-developed paleosols within the la- custrine sequence, indicative of instability in lake- level as well as local emergence due to delta pro- gradation. Wynn (2000) reports several new pedo- types from this stratigraphic interval due to these particular conditions. In most sections, the overlying sedimentary se- quence again becomes dominated by a fluvial sys- tem, and several coarse gravels with significant ero- sional bases cap the sedimentary deposits. This in- terval is referred to here as the upper fluvial se- quence. Like the lower fluvial sequence, this interval exhibits a high degree of lateral variation (Fig. 7). The influence of the basement topography is consid- erably less, however, and thus the sequence presents more characteristic upward-fining units indicative of a meandering fluvial system. It is noteworthy that, in all but one section, once fully fluvial conditions are re-established, there is no further indication of lacustrine conditions or even of floodplain ponding. The single exception is seen in section CSF 95-10 (Fig. 2). Here a thin interval of ostracod-packed claystones and fissile green claystones clearly indi- cates deposition in a lake or pond. This interval rests on a thin bentonite. It is possible that this sequence represents the Lokochot Lake, a lacustrine phase, which occurred ca. 3.5 Ma in the Turkana Basin. The Lokochot Lake is well documented from Omo Group deposits in the northern Turkana Basin Feibel: Stratigraphy HIS CSF 95-4 m c z s g/t Figure 5 Reference sections from the lower and middle portions of the Kanapoi Formation. Numbered correlations shown are 3, basal flooding surface of the Lonyumun Lake sequence; and 4, Kanapoi Tuff. Unnumbered correlations are lithologic contacts walked out between sections. See Figure 1 for location of sections. For a key to symbols, see Figure 3 (Brown and Feibel, 1991; Feibel et aL, 1991), and has been recognized elsewhere in the lower Kerio Valley (Feibel, unpublished). The uppermost strata of the Kanapoi Formation are a sequence of massive cobble- to pebble-con- glomerates, which incise deeply into the underlying fluvial strata. These conglomerates are dominated by silicic volcanics, with a matrix of litharenite sand. The conglomerates may occur as multiple units and may reach up to 21 m in thickness. They often have thin sand interbeds. Mudstones are rare in this upper part of the section, and by the top of the formation, the depositional setting appears to have developed into a gravel braidplain. These gravels are overlain by the Kalokwanya Basalt (Powers, 1980). The basalt has been dated to 3.4 Ma by McDougall (Leakey et ah, 1995). In some localities, the basalt fills deep channels cut into the conglomerates. The vertical and lateral variations in lithofacies Table 1 Electron microprobe analysis of glass from the Kanapoi Tuff (Leakey et al. 1998).'' Sample SiO, AhOa Fe,03 CaO K^O Na,0 MgO MnO TiO, Cl F Zr Total N KPOl-15-01 70.36 7.86 8.32 0.37 3.93 1.73 0.01 0.25 0.25 0.42 0.03 NA 93.53 6 K92-4846 69.50 7.55 8.31 0.28 0.26 0.16 0.01 0.25 0.23 0.52 0.01 0.29 96.49 13 K92-4847 69.76 7.66 8.42 0.29 0.37 0.24 0.01 0.25 0.24 0.48 0.01 0.28 96.58 18 " Abundances are shown as weight per cent. N, number of shards analyzed; NA, not analyzed 16 ■ CS 498, Harris and Leakey: Kanapoi CSF 96-12 m E then a reef Kanapoi Tuff Figure 6 Reference section from the middle portion of the Kanapoi Formation. The Kanapoi Tuff here is deeply channeled, and the channel-fill includes both an Etheria bioherm and a later channel sand. See Figure 1 for loca- tion of sections. For a key to symbols, see Figure 3 seen at Kanapoi are summarized in Figure 8. This somewhat schematic diagram emphasizes the ge- ometry of the major facies types and their relation- ships to the underlying basement paleotopography. FOSSIL CONTEXT AND PALEOENVIRONMENTS The Kanapoi stratigraphic sequence is summarized in the composite section of Figure 9. This compos- ite forms the basis for a discussion of the context of fossil vertebrate faunas recovered from Kanapoi, as well as for the environmental history recorded in the deposits. There are two major stratigraphic levels produc- ing the bulk of the vertebrate fossil material at Kan- apoi. The lower level is the channel sandstone and overbank mudstone complex associated with the lower and upper pumiceous tephra. Most of the fossils in this interval, including much of the Aus- tralopithecus anamensis Leakey et ah, 1995, hy- podigm, come from vertic paleosois developed on the floodplain through this period. The upper fos- siliferous zone is the distributary channel complex associated with the Kanapoi Tuff. This richly fos- siliferous zone is dominated by aquatic forms (fish and reptiles) but also includes a wide range of ter- restrial mammals. Fossils are also found in the up- per fluvial sequence, where they are associated with both channel and floodplain settings. The environmental setting recorded by the Kan- Figure 7 Reference sections from the upper part of the Kanapoi Formation. Numbered correlations shown are 1, lower pumiceous tuff; 2, upper pumiceous tuff; and 3, basal flooding surface of the Lonyumun Lake sequence. See Figure 1 for location of sections. For a key to symbols, see Figure 3 Feibel: Stratigraphy ■ 17 m Geometry of Major Lithofacies and Marker Beds in the Kanapoi Region Pliocene Kalokwanya basalt conglomerate C 1 Tks sandstone Q I 1 paleosol laminated claystones molluscan limestone Kanapoi Tuff 1 500© 5000 Tkpt altered pumiceous tephra Tmbu Miocene basalts, undifferentiated ^ 10 km ► Figure 8 Schematic drawing of the geometry of major lithofacies and marker beds in the Kanapoi Formation. Note the vertical exaggeration in the diagram. Only major components are depicted, minor facies are shown in white apoi sedimentary sequence reflects a progression of fluvial and lacustrine systems that overwhelmed a volcanic landscape. The fluvial system that domi- nated local environments throughout the Kanapoi record was the ancestral Kerio River. This is sup- ported by evidence from the tectonic heritage of the region, provenance of sedimentary clasts, and the southerly link provided by correlation of the Kan- apoi Tuff into the Baringo Basin. The ancestral Ker- io River was certainly seasonal through this time period. Perennial flow is only demonstrated for the middle of the represented time interval, through the presence of Etheria reefs in a channel setting above the Kanapoi Tuff. At other times, there are indica- tors of strong seasonality in flow, particularly in conglomerates low and high in the section as well as in the prevalence of planar cross-stratification in many of the sands. This may reflect strong season- ality in a perennial stream or ephemeral flow con- ditions. The well-developed upward-fining cycles, particularly in the upper fluvial interval, however, are suggestive of continued perennial flow there. The hills/islands of the volcanic basement provid- ed a considerable degree of local heterogeneity. For the fluvial systems, this would have been manifest not only in the local topographic relief but also in different soil conditions, drainage, and vegetation patterns. This is an element of habitat patchiness which is not typically seen in the Plio-Pleistocene paleoenvironments investigated from elsewhere in the Turkana Basin (e.g., Feibel et ah, 1991). The fluvial systems that encountered this complex land- scape were spatially controlled by paleotopograph- ic lows that restricted both the flow patterns for fluvial channels as well as the extent and connect- edness of the early floodplains. Although the degree of influence this basement topography exerted de- creased through time, it was present throughout the formation. A strong seasonality in precipitation is docu- mented by the prevalence of vertisols in the over- bank deposits. In this sense, the Kanapoi flood- plains are closely comparable with those of the ear- ly Omo Group sequence (e.g., Moiti, Lokochot, Tulu Bor Members) in the Turkana Basin to the north. There does not appear to be a progressive shift in the character of paleosols through time at Kanapoi. Rather, the variability seen in soil types reflects aspects of the soil catena across the Kanapoi paleolandscape. This relates primarily to topo- graphic effects (including drainage and leaching), soil development on different parent materials, and variations in the maturity of soils induced by re- organizations of the landscape. Examples of the lat- ter are the influence tephra fallouts produced in the lower and upper pumiceous tephra. The pervasive 18 ■ CS 498, Harris and Leakey: Kanapoi Kanapoi Composite Stratigraphic Section Kalokwanya Basalt 3.4 Ma V Kanapoi Tuff Lonyumun Lake 0 4.07 Ma 20 upper pumiceous tuff 4.1 2 Ma 10 lower pumiceous tuff 4.1 7 Ma Mio-Pliocene basalts c z s g/t Figure 9 Composite stratigraphic column for the Kanapoi Formation. Note major fossiliferous levels in lower fluvial paleosols and in deltaic sands of the Lonyumun Lake stage. Age control based on work of I. McDougall (Leakey et ah, 1995, 1998) Feibel: Stratigraphy ■ 19 thick profile of the lower pumiceous tephra indi- cates it blanketed the landscape and would have forced a ‘restart’ of a successional regime in soil, vegetation, and ecological communities based on this volcanic parent matter. The thinner upper pu- miceous tephra is only patchily preserved, which implies that it was locally incorporated into the ac- tive soil substrate rather than overwhelming it. The Lonyumun Lake transgression produced the most dramatic reorganization of the Kanapoi land- scape. The sharp basal contact of the lacustrine claystone in this interval demonstrates a rapid drowning of the landscape. The precise chronostra- tigraphic control on the Kanapoi sequence provides the best age control on this event, which can be placed at 4.10 ± 0.02 Ma. The Lonyumun trans- gression affected a major portion of the Turkana Basin (ca. 28,000 km^), most likely due to tectonic or volcanic damming of the basin outlet. The trans- gression was everywhere rapid, and Kanapoi is sit- uated along the drowned paleovalley into which the ancestral Kerio River flowed. The rapid local infilling of the Lonyumun Lake at Kanapoi is to be expected from the minimal ac- commodation space available in this drowned pa- leovalley and the rapid sedimentation induced by proximity of the ancestral Kerio River Delta. The thick accumulation of the Kanapoi Tuff (nearly 11 m) in the central part of the Kanapoi area resulted from the filling of the interdistributary bays of this delta (Powers 1980) following an explosive erup- tion in the rift valley to the south. As the lake re- treated northwards, a progression of minor inun- dations and exposure is reflected in the interbedded fluvial and lacustrine strata that mark the transition from the lacustrine phase to the upper fluvial se- quence. The characteristics of the fluvial strata that suc- ceeded the Lonyumun Lake sequence reflect the considerable infilling that the lake phase produced and the broader floodplains available for a mean- dering river system. In other aspects, however, this river w^as very similar to the system that existed prior to the Lonyumun transgression. This lower portion of the upper fluvial sequence stands in stark contrast, however, to the upper strata of the inter- val, where sands and gravels dominate to the near exclusion of mudstones. This upper portion of the formation reflects two fundamental changes in the system, increased supply of coarse elastics and a gradual drop-off in overall accumulation rates. The elastics are dominated by siliceous volcanic cobbles and pebbles, in contrast to the basaltic suite of con- glomerates at the base of the formation. This likely reflects renewed tectonic activity in the source area to the south. The slowdown in accumulation is im- plied rather than measured, as there are no time markers between the Kanapoi Tuff and the Kalok- wanya Basalt. The dramatic change in sedimentary facies, however, suggests that much of the time be- tween these two chronostratigraphic markers lies within these upper gravels. This upper portion of the sequence would likely have presented the most dramatic deviation in environmental characteris- tics. The substrate of the braidplain would have been well drained, and the coarse siliceous volca- nics would provide a poor medium for growth of vegetation. The starkness of this landscape would be succeeded, however, by the truly inhospitable volcanic landscape produced by eruption of the Ka- lokwanya Basalt. CONCLUSIONS The Pliocene sedimentary sequence of the Kanapoi region, termed here the Kanapoi Formation, was deposited by the ancestral Kerio River in three phases. Initial deposition occurred upon a fluvial floodplain that was broken by numerous hills of the local Mio-Pliocene basaltic basement. These hills strongly influenced patterns of deposition, as the fluvial system mantled the complex topography with channel gravels and sands, while vertic paleo- sols developed on the adjacent floodplains. Two pu- miceous airfall tephra accumulated on this land- scape (lower pumiceous tuff, 4.17 Ma; upper pu- miceous tuff, 4.12 Ma), allowing precise chrono- stratigraphic control on this phase of deposition. The Lonyumun Lake transgression replaced the flu- vial system with a lacustrine setting and the rapidly prograding Kerio River Delta. The vitric Kanapoi Tuff (4.07 Ma) was deposited primarily in interdis- tributary floodbasins at this stage. The prograda- tion locally replaced the Lonyumun Lake with a second floodplain system, somewhat less con- strained by basement topography. A shift in this system from a meandering sand/mud fluvial system to a gravel braidplain reflects tectonic activity in the source area to the south and a lowering of accu- mulation rates. Eruption of the Kalokwanya Basalt effectively ended significant sediment accumulation at Kanapoi. The rich vertebrate fossil assemblages of Kanapoi are found in floodplain paleosols of the lower and upper fluvial intervals, as well as in the distributary sands of the Kerio River Delta during the Lonyu- mun Lake phase. The high degree of landscape het- erogeneity and pronounced soil catenas of the Kan- apoi setting provided some of the greatest habitat patchiness recorded from the Turkana Basin Plio- Pleistocene. ACKNOWLEDGMENTS Support for this work was provided by the Leakey Foun- dation, the National Geographic Society, the National Sci- ence Foundation (U.S.A.), and the National Museums of Kenya. Special thanks to M.G. Leakey for her enthusiasm and support. I would like to thank Tomas Muthoka for assistance in the field, I. McDougall and J.G. Wynn for discussions of Kanapoi geology, and S. Hagemann for help in the laboratory. Much of the analysis and writing of this report was made possible by a fellowship from the Insti- tute for Advanced Studies of the Hebrew University of Jerusalem. 20 ■ CS 498, Harris and Leakey: Kanapoi LITERATURE CITED Brown, F. H., and C. S. Feibel. 1986. Revision of litho- stratigraphic nomenclature in the Koobi Fora region, Kenya. Journal of the Geological Society, London 143:297-310. . 1991. Stratigraphy, depositional environments and paleogeography of the Koobi Fora Formation. In Koobi Fora Research Project, Vol. 3. Stratigraphy, artiodactyls and paleoenvironments, ed. J. M. Har- ris, 1-30. Oxford: Clarendon Press. Feibel, C. S. 2003. Stratigraphy and depositional history of the Lothagam sequence. In Lothagam: The dawn of humanity in eastern Africa, eds. M. G. Leakey and J. M. Harris, 17-29. New York: Columbia Univer- sity Press. Feibel, C. S., J. M. Harris, and F. H. Brown. 1991. Pa- laeoenvironmental context for the late Neogene of the Turkana Basin. In Koobi Fora Research Project, Vol. 3. Stratigraphy, artiodactyls and paleoenviron- ments, ed. J. M. Harris, 321-370. Oxford: Claren- don Press. Harris, J. M., F. H. Brown, and M. G. Leakey. 1988. Ge- ology and paleontology of Plio-Pleistocene localities west of Lake Turkana, Kenya. Contributions in Sci- ence 399:1-128. Leakey, M. G., C. S. Feibel, I. McDougall, and A. Walker. 1995. New four-million-year-old hominid species from Kanapoi and Allia Bay, Kenya. Nature 376: 565-571. Leakey, M. G., C. S. Feibel, L McDougall, C. Ward, and A. Walker. 1998. New specimens and confirmation of an early age for Australopithecus anamensis. Na- ture 393:62-66. Namwamba, F. 1993. Tephrostratigraphy of the Cheme- ron Formation, Baringo Basin, Kenya. Unpublished M.S. Thesis. University of Utah, Sait Lake City. 78 pp. Patterson, B. 1966. A new locality for early Pleistocene fossils in northwestern Kenya. Nature 212:577-578. Patterson, B., A. K. Behrensmeyer, and W. D. Sill. 1970. Geology and fauna of a new Pliocene locality in northwestern Kenya. Nature 226:918-921. Patterson, B., and W. W. Howells. 1967. Hominid hu- meral fragment from early Pleistocene of northwest- ern Kenya. Science 156:64-66. Powers, D. W. 1980. Geology of Mio-Pliocene sediments of the lower Kerio River Valley. Ph.D. dissertation, Princeton University. 182 pp. Wynn, J. G. 2000. Paleosols, stable carbon isotopes, and paleoenvironmental interpretation of Kanapoi, northern Kenya. Journal of Human Evolution 39: 411-432. Received 26 December 2002; accepted 23 May 2003. Fossil Fish Remains from the Pliocene Kanapoi Site, Kenya Kathlyn Stewart* ABSTRACT. Over 2,800 fossil fish elements were collected in the 1990s from the Pliocene site of Kanapoi, located in the Turkana Basin, northern Kenya. The Kanapoi fish fauna is dominated by large piscivores and medium to large molluscivores, whereas herbivorous fish are rare. The genera Labeo, Hydrocynus, and Sindacharax are abundant in the deposits, as are large percoids and catfish. While the Kanapoi fauna has many similarities with both the near-contemporaneous fauna recovered from the Muruongori Member at nearby Lothagam and the site of Ekora, including the extinct genera Sindacharax and Semlikiichthys, it differs significantly in two features. The Kanapoi fauna is dominated by a Sindacharax species that is absent at Muruongori and it lacks two other Sindacharax and two Tetraodon species which are common in the Muruongori deposits and at Ekora. The Kanapoi fauna is similar to that from the Apak Member at Lothagam, in particular by the domination of Sindacharax mutetii. INTRODUCTION The presence of fossil fishes at Kanapoi had been reported by Behrensmeyer (1976) among others, but no systematic recovery was initiated until 1993. Over 2,800 fossil fish elements were recovered from Kanapoi deposits in the 1993 and 1995 field sea- sons (see map in Introduction, page 2). Most col- lecting was undertaken by the author and Sam N. Muteti, of the National Museums of Kenya, with some additional collecting by the National Muse- ums of Kenya fossil team. As discussed by Feibel (2003b), the major phase of deposition of the Kan- apoi deposits date from 4.17 Ma to about 4.07 Ma with three sedimentary intervals: a lower fluvial se- quence, a lacustrine phase, and an upper fluvial se- quence. The fish fossils were collected from six sites located in the lacustrine phase of the formation, and from one site probably deposited during the upper fluvial sequence and hence slightly younger than 4.07 Ma. Fieldwork at Kanapoi followed three years of in- tensive collection of vertebrate and invertebrate fos- sils from the nearby site of Lothagam (Leakey et ah, 1976; Leakey and Harris, 2003), with fossilif- erous deposits ranging in age from late Miocene to Holocene, as well as from the western Turkana Ba- sin Pliocene sites of Ekora, South Turkwel, North Napudet and Eshoa Kakurongori. Reference will be made in this report to the detailed description of over 7000 fish fossils collected at the nearby site of Lothagam (Stewart, 2003). Collection offish fossils at Lothagam was extensive, in order to obtain in- formation on systematics, environment and bioge- ography, previously poorly known from this peri- od. Most fish elements collected from Lothagam 1. Canadian Museum of Nature, PO Box 3443, Station D, Ottawa, Ontario KIP 6P4, Canada. Contributions in Science, Number 498, pp. 21-38 Natural History Museum of Los Angeles County, 2003 derived from the Lower and Upper Nawata Mem- bers of the Nawata Eormation, and the Apak Mem- ber of the Nachukui Eormation, ranging in age from 7.44 Ma to about 4.2 Ma (McDougall and Feibel, 1999). Fish bones were also collected from the Muruongori Member, and the Kaiyumung Member of the Nachukui Formation, which date to about 4.0 Ma, and approximately 4.0 to 2.0 Ma respectively (C. Feibel, F. Brown, personal com- munication). More detailed information about the stratigraphy and geochronology at Lothagam is provided by Feibel (2003a) and McDougall and Feibel (1999). Fish collecting at Kanapoi was less extensive than at Lothagam, as only elements with potential taxonomic and systematic information were col- lected. The Kanapoi fish elements derive from sed- iments which date close to 4.07 Ma, and, like the Muruongori Member sediments at Lothagam, were probably deposited during the Lonyumun Lake transgression (Feibel, 2003b). Reference will also be made to the fish collected from the Ekora site, located about 50 km southeast of Lothagam and about 25 km north of Kanapoi, near the mod- ern Kerio River. The Ekora fauna is of Pliocene age and probably also derives from Lonyumun Lake deposits (Eeibel, personal communication). In the descriptions and discussions below, eco- logical and zoogeographical information on mod- ern fish was referenced from the Checklist of the Freshwater Fishes of Africa volumes (Daget et ah, 1984, 1986) and from Hopson and Hopson (1982). The Kanapoi fishes have not yet been acces- sioned into the collections of the National Muse- ums of Kenya. In the systematic description, the specimens are listed by the field number for their site of origin. 22 ■ CS 498, Harris and Leakey: Kanapoi Figure 1 Hyperopisus sp., SEM of isolated tooth, ventral view, from Kanapoi Figure 2 Hyperopisus sp., SEM of isolated tooth and base, dorsal view, from Kanapoi SYSTEMATIC DESCRIPTION Order Polypteriformes Family Polypteridae Polypterus Geoffrey Saint-Hilaire, 1802 Polypterus sp. KANAPOI MATERIAL. 3156, scale. Polypterus material is extremely rare in Kanapoi deposits, with only one element identified. As Po- lypterus scales, spines, and cranial fragments are robust and preserve well, this poor record suggests a minimal Pliocene presence at Kanapoi. The family Polypteridae is today represented by two genera: Polypterus and Calamoichthys Smith, 1866 (rather than Erpetoichthys Smith, 1865; see discussion in Stewart, 2001), both restricted to Af- rica. Most fossil elements comprise scales, verte- brae, and spines, and have been referred to the larg- er and today much more widely distributed genus Polypterus or only to the family Polypteridae. Polypterus is a long, slender fish with a distinc- tive, long dorsal fin that is divided by spines into portions resembling sails; they have a lung-like or- gan to breathe air. Polypteridae have several prim- itive features with similarities to Paleozoic paleon- iscoids (Carroll, 1988). Their earliest fossil record from Africa is from Upper Cretaceous deposits in Egypt, Morocco, Niger, and Sudan (Stromer, 1916; Dutheil, 1999). Their Cenozoic record includes fos- sils from Eocene deposits in Libya (Lavocat, 1955); Miocene deposits in Rusinga, Loperot, and Lotha- gam, Kenya (Greenwood, 1951; Van Couvering, 1977; Stewart, 2003), and Bled ed Douarah, Tu- nisia (Greenwood, 1973); Pliocene deposits at Wadi Natrun, Egypt (Greenwood, 1972); Pliocene depos- its at Lothagam, Kenya (Stewart, 2003); and Plio- Pleistocene deposits at Koobi Fora (Schwartz, 1983). Polypterus has never been recovered from the Western Rift sites. Two extant species are known from Lake Turkana — P. senegalus Cuvier, 1829, and F. bichir Geoffroy Saint Hilaire, 1802. Polypterus is widespread from Senegal to the Nile Basin up to Lake Albert, as well as the Congo Basin and Lake Tanganyika. Order Mormyriformes Family Mormyridae Hyperopisus Gill, 1862 Hyperopisus sp. Figures 1, 2 KANAPOI MATERIAL. 3156, 1 tooth; 3845, 96 teeth; 3847, 7 teeth; 3848, 3 teeth; 3849, 2 teeth. Hyperopisus teeth appear as truncated cylinders with smooth, relatively flat tops and bases (Figs. 1, 2), and attach to the parasphenoid and basihyal bones. The average diameter of the Kanapoi teeth (1-4 mm) is within the range of large extant Hy- peropisus individuals (up to 90 cm total length). Hyperopisus teeth are relatively common throughout the Kanapoi deposits. While absent from the Nawata Formation deposits at Lothagam, the teeth are common in the Nachukui Formation deposits and at the Pliocene South Turkwell site (personal observation). Modern Hyperopisus (and other mormyroids) generate weak electromagnetic fields in order to sense their environment. They are therefore absent from modern Lake Turkana and other bodies of water with high salinity values, which apparently impede this sensory ability (Bea- dle, 1981). Fossil Hyperopisus teeth (see summary in Stew- art, 2001) are known from Pliocene deposits of Wadi Natrun, Egypt (Greenwood, 1972), from Plio-Pleistocene deposits in the Lakes Albert and Edward Basins (Greenwood and Howes, 1975; Stewart: Fish ■ 23 Stewart, 1990), Mio-Pleistocene Lakes Albert and Edward Basins deposits (Van Neer, 1994), from Pli- ocene deposits at Lothagam {Stewart, 2003) and from Plio-Pleistocene deposits at Koobi Fora (Schwartz, 1983). Modern H. bebe Lacepede, 1803, is known from the Omo River Delta of Lake Turkana, and from the Senegal, Volta, Niger, Chad, and Nile Basins. Large teeth referred to }Hyperopisus have been reported in Pliocene Lake Edward Basin deposits and Pliocene Wadi Natrun deposits (Greenwood, 1972; Stewart, 1990, 2001). These teeth, although identical to those of modern Hyperopisus, far ex- ceed the size range of modern teeth, and as no iden- tified bone has been recovered with the teeth, their affiliation is problematic. These were not recovered in the Turkana Basin deposits, and to date have a restricted Nile River and Western Rift presence. Family Gymnarchidae Gymnarchus Cuvier, 1829 Gymnarchus sp. KANAPOI MATERIAL. 3156, 18 teeth; 3845, 3 teeth; 3847, 3 teeth; 3848, 3 teeth; 3849, 7 teeth. Gymnarchus teeth line the premaxilla and den- tary. They are common throughout the Kanapoi de- posits. The Kanapoi teeth average 3-4 mm in width, which is within the size range of large mod- ern individuals (60-100 cm total length). Gymnarchus is piscivorous, although mollusks and insects are also eaten. As in Hyperopisus, these fish use an electromagnetic field to sense the envi- ronment and are therefore intolerant of highly sa- line waters. Gymnarchus teeth are common throughout the Kanapoi deposits, as they are throughout the Lothagam deposits. Fossil elements are reported from Miocene-Pleistocene deposits in Lakes Albert and Edward Basins (Van Neer, 1994), Pliocene deposits in the Lakes Albert and Edward Basins (Stewart, 1990; Van Neer, 1992), late Mio- cene and Pliocene deposits at Lothagam, Kenya (Stewart, 2003), and Plio-Pleistocene deposits at Koobi Fora (Schwartz, 1983). Modern G. niloticus Cuvier, 1829, is known from the Omo River Delta in Lake Turkana, and in the Gambia, Senegal, Ni- ger, Volta, Chad, and Nile Basins. Order Cypriniformes Family Cyprinidae Laheo Cuvier, 1817 Laheo sp. (Figures 3, 4) KANAPOI MATERIAL. 3156, 12 teeth; 3845, 24 teeth; 3846, 4 teeth; 3847, 37 teeth; 3848, 6 teeth; 3849, 26 teeth teeth, 1 trunk vertebra. Laheo is essentially represented by its pharyngeal teeth, which were not identifiable to species (Figs. 3, 4). One vertebra was also recovered, and while Figure 3 Laheo sp., SEM of pharyngeal tooth, side view, from Kanapoi similar to Barbus Cuvier and Cloquet, 1816, ver- tebrae, Laheo vertebrae can be distinguished by tra- becular morphology. These elements represent in- dividuals up to 90 cm in total length, which is with- in the modern size range of the Turkana species. Laheo teeth are surprisingly common throughout the Kanapoi deposits. Its teeth are rare in the Na- wata Formation sites at Lothagam, but more com- 24 ■ CS 498, Harris and Leakey: Kanapoi Figure 4 Labeo sp., SEM of pharyngeal tooth, ventral view, from Kanapoi mon in the Nachukui Formation sites. Labeo is an inshore bottom fish, eating algae and organic de- tritus. The fossil record is scanty (Stewart, 2001), but reported from late Miocene deposits at Lotha- gam, Kenya (Stewart, 2003); Pliocene deposits in Wadi Natrun, Egypt (Greenwood, 1972), Koobi Fora, Kenya (Schwartz, 1983), the Lakes Albert and Edward Basins (Stewart, 1990); and Pleisto- cene deposits in the Western Rift (Van Neer, 1994). A reported Miocene occurrence from western Uganda may be in error; the author states that cer- tain Mio~Pliocene sites had Pleistocene-aged fossils mixed in (Van Neer, 1994:90). Labeo-like teeth are also reported from the mid-Miocene of Loperot but are not confirmed (Van Couvering, 1977). In Lake Turkana, extant Labeo is represented by one spe- cies— L. horie Heckel, 1846. Elsewhere, the genus is widespread throughout the continent, including the Nile Basin, West Africa, eastern Africa, and the Congo and Zambezi Basins. Barbus Cuvier and Cloquet, 1816 Barbus sp. (Eigures 5, 6) KANAPOI MATERIAL. 3156, 4 teeth; 3845, 2 teeth; 3846, 3 teeth; 3849, 6 teeth. Barbus is exclusively represented by its pharyn- geal teeth (Figs. 5, 6), which represent small indi- viduals, probably under 30 cm total length. These teeth do not resemble those of B. bynni Boulenger, 1911, the only similar sized Barbus now inhabiting Lake Turkana, but do resemble those of B. altian- alis Boulenger, 1900; no other comparison with modern Barbus species was made. The teeth do not have the rows of small cusps observed on some Barbus} teeth recovered from Miocene deposits in Saudi Arabia (Otero and Gayet, 2001). Like Labeo, Barbus is an inshore demersal (bot- tom-dwelling) fish, with a varied diet of ostracods, mollusks, insects, aquatic vegetation, and occasion- ally fishes. Barbus teeth are not common at Kana- poi, nor are they common at nearby Lothagam Figure 5 Barbus sp., SEM of pharyngeal tooth, ventral view (left) and side view (right), from Kanapoi Stewart: Fish ■ 25 Figure 6 Barbus sp., SEM of pharyngeal tooth (different from Fig. 5), side view, from Kanapoi (Stewart, 2003). Their fossil record in Africa is vir- tually nonexistent prior to the Pliocene (Stewart, 2001), with the earliest reported finds being from Pliocene deposits at Lothagam (Stewart, 2003), Pli- ocene deposits in the Western Rift, Congo (Stewart, 1990), Plio-Pleistocene deposits from Koobi Fora (Schwartz, 1983), and Pleistocene deposits from the Western Rift (Greenwood, 1959; Van Neer, 1994). Van Couvering (1977) notes that “Brtr^ws-like” teeth are known from mid-Miocene deposits in Kenya. The report of a probable Barbus in Miocene deposits in Saudi Arabia (Otero, 2001; Otero and Gayet, 2001) indicates these fishes could have en- tered Africa from the Arabian region during land connections in the Burdigalian (early Miocene) (see discussion in Otero, 2001). At present, Barbus is represented by three species in Lake Turkana, with only B. bynni attaining a length of at least 30 cm in Lake Turkana. Order Characiformes Family Distichodontidae Distichodus Muller and Troschel, 1845 Distichodus sp. (Figure 7) KANAPOI MATERIAL. 3156, 1 tooth; 3845, 2 teeth; 3847, 2 teeth; 3849, 3 teeth. Distichodus teeth are oral, lining the premaxilla and dentary (Fig. 7). The average height of the Kan- apoi teeth was 5 mm long, which is within the size range of modern individuals. Distichodus remains are not common at Kanapoi nor at Lothagam, but this may reflect their small size and probable poor preservation. The fossil re- cord is poor (Stewart, 2001) but is known from Mio-Pliocene deposits in the Lakes Albert and Ed- ward Basins (Van Neer, 1994), Pliocene deposits in the Lakes Albert and Edward Basins (Stewart, 1990), late Miocene deposits at Lothagam, Kenya (Stewart, 2003), and Pleistocene deposits at Koobi Fora (Schwartz, 1983). Extant D. niloticus (Lin- naeus, 1762) is known from Lake Turkana and from the Nile Basin up to Lake Albert. Figure 7 Distichodus sp., SEM of oral tooth, side view, from Kanapoi Family Alestidae Hydrocynus Cuvier, 1817 Hydrocynus sp. (Figure 8) KANAPOI MATERIAL. 3156, 23 teeth, 1 den- tary fragment with tooth; 3845, 33 teeth, 1 dentary fragment with tooth; 3846, 3 teeth; 3847, 13 teeth, 1 dentary fragment with tooth, 1 premaxilla frag- ment; 3848, 32 teeth, 4 dentary fragments, 4 den- tary fragments with teeth, 1 premaxilla fragment; 3849, 2 teeth. Hydrocynus teeth are long and conical in shape (Fig. 8), with considerable size range. At Kanapoi, both teeth and jaw elements were recovered, often with the teeth in situ, usually in replacement sock- ets within the jaw element. Teeth and jaw elements Figure 8 Hydrocynus sp., SEM of oral tooth and base, side view, from Kanapoi 26 ■ CS 498, Harris and Leakey: Kanapoi Figure 9 Brycinus macrolepidotus, SEM of second inner premaxillary tooth, occlusal view, from Kanapoi; labial side at bottom, lingual at top represent individuals of up to 1 m in total length, although modern individuals in Lake Turkana do not exceed 65 cm in total length (Hopson and Hop- son, 1982). Several small teeth with a broader base and more triangular shape than most teeth were determined, from modern Hydrocynus jaws, to be teeth which were just erupting. The fossil record of Hydrocynus has been pri- marily based on teeth (Stewart, 2001); therefore, the recovery of jaw elements potentially provides new information about the fossil genus. Hydrocy- nus are pelagic and are voracious piscivores. Fossil Hydrocynus teeth are known from Mio-Pleistocene deposits in the Western Rift, Uganda (Van Neer, 1994), Miocene deposits of Sinda, Congo (Van Neer, 1992), late Mio-Pliocene deposits at Lotha- gam, Kenya (Stewart, 2003), Pliocene deposits in Wadi Natrun, Egypt (Greenwood, 1972), and the Western Rift, Congo (Stewart, 1990), and Plio- Pleistocene deposits in the Omo Valley (Aram- bourg, 1947) and at Koobi Fora (Schwartz, 1983). Hydrocynus is represented by one species, H. for- skalii Cuvier, 1819, in Lake Turkana, but a second species, H. vittatus Castelnau, 1861, is present in the Omo River. Hydrocynus is widespread from Se- negal to the Nile, including the Volta, Niger, and Chad Basins. Brycinus Myers, 1929 B. macrolepidotus (Cuvier and Valenciennes, 1849) (Figures 9-11) KANAPOI MATERIAL. 3156, 1 second inner premaxillary tooth. Brycinus is represented only by a second inner premaxillary tooth (Fig. 9) which is identical to the Figure 10 Brycinus macrolepidotus, SEM of first, second, third, and fourth inner premaxillary teeth (from left to right) and some outer teeth, occlusal views; lingual side at top right, labial at bottom left; modern specimen Stewart: Fish ■ 27 Figure 11 Brycinus macrolepidotus, SEM of second inner premaxillary tooth, occlusal view; labial side at bottom, lingual at top; modern specimen same tooth in modern B. macrolepidotus specimens (Figs. 10, 11). Brycinus macrolepidotus has distinc- tive inner premaxillary teeth, which are not found in other alestid specimens. The tooth represents an individual of about 30 cm in total length. A recent study has transferred Alestes macrole- pidotus and twelve other Alestes Muller and Tro- schel 1844 species to the genus Brycinus, leaving five species in the genus Alestes, and five in a poly- phyletic grouping referred to as “Brycinus” (Mur- ray and Stewart, 2002). Both Alestes and Brycinus are genera within the Alestidae, which possess sim- ilar multicusped molariform teeth. In the following discussions, I will use “alestid” to refer to Alestes and/or Brycinus, but not to Hydrocynus, which is also alestid but with very different, conical teeth (see Murray and Stewart, 2002, for discussion of the terms Alestidae, “Flydrocyninae,” and “Alesti- nae”). Use of fine-meshed screens at several sites result- ed in recovery of many small cusped teeth, which in the field were thought to belong to Alestes or Brycinus. Closer inspection with a microscope re- vealed that these teeth actually belonged to Sinda- charax, based on similarity to larger specimens (see discussion under Sindacharax Greenwood and Howes, 1975). The recovery of only one Brycinus tooth, particularly when fine-meshed screens (1- mm mesh) were used at several sites indicates the scarcity of this taxon at Kanapoi. Brycinus and Alestes were slightly more common at Lothagam. The fossil record of Brycinus sp. (and Alestes sp.) is poor (Stewart, 2001), with remains known from Mio-Pliocene deposits at Lothagam, Kenya (Stew- art, 2003), Plio-Pleistocene deposits in the Lakes Albert and Edward Basins (Stewart, 1990), Mio- Pleistocene deposits in the Western Rift (Van Neer, 1994), and Manonga, Tanzania (Stewart, 1997). Miocene teeth with alestid affinities are reported from Loperot and Mpesida, Kenya (Van Couvering, 1977). Modern alestids are represented by six spe- cies in Lake Turkana, including A. baremoze de Joannis, 1835; A. dentex Linnaeus, 1758; B. nurse Riippell, 1832; B. macrolepidotus; B. ferox Hop- son and Hopson, 1982; and B. minutus Hopson and Hopson, 1982. Modern alestids are found in the Volta, Niger, and Chad Basins to the Nile River, and in the Congo, Zambezi, and Limpopo Basins. Modern alestid species span a range of trophic ad- aptations and habitats. In modern Lake Turkana, they are generally pelagic and omnivorous. Sindacharax Greenwood and Howes, 1975 A total of 2,272 teeth from Kanapoi are attributed to Sindacharax. This preponderance of Sindachar- ax teeth compared to numbers of elements of other fish reported here does not reflect actual abun- dance, but a selective collection policy. Very few Sindacharax dentaries and/or premax- illae are known with in situ teeth, and none from Kanapoi, so identification of isolated teeth was ac- complished by comparison with the complete den- tary and premaxilla of S. greenwoodi Stewart, 1997, found at Lothagam (Stewart, 1997). How- ever, because there is considerable individual vari- ation in cusp patterns of teeth in known Sinda- charax jaws, placement of these isolated teeth is tentative. Analysis of the in situ teeth and the iso- lated teeth at Lothagam indicated that outer pre- maxillary and dentary teeth were so similar among Sindacharax species, as were third and fourth inner premaxillary teeth, that no species designations for these teeth are made. As is the common convention, in this article, teeth are numbered sequentially, starting from the midline of the jaw (#1 left or right) and moving laterally. As mentioned above in the discussion on modern alestids, many very small teeth (<2 mm) were re- covered, which were initially thought to belong to Brycinus or Alestes, until closer examination indi- cated they were very small S. mutetii Stewart, 2003, and S. lothagamensis Stewart, 2003, teeth. The sim- ilarity in shape and cusping between small Sinda- charax and modern alestid teeth leads to specula- tion on the development of the characteristic cusped ridges in Sindacharax teeth for which the genus is named (Greenwood and Howes, 1975; Greenwood, 1976). Examination of a range of Sm- dacharax inner premaxillary teeth indicates that, while the smaller teeth (ca. <2 mm) are cusped, in larger teeth, the cusps morph to form ridges. Once the ridges are formed, the tooth pattern remains consistent. On the other hand, in several of the modern ales- tid specimens observed by the author (including B. macrolepidotus, A. dentex, A. baremoze), the inner teeth remained cusped in both large and small spec- imens, with one exception. Small A. stuhlmanni 28 ■ CS 498, Harris and Leakey: Kanapoi Figure 12 Sindacharax lothagamensis, SEM of second in- ner premaxillary tooth, occlusal view, from Kanapoi; la- bial side at top, lingual at bottom Pfeffer, 1896, individuals had cusped teeth, but the larger specimens (ca. 24 cm in total length) had teeth with ridges (personal observation). Of partic- ular interest therefore is whether teeth of other modern alestid species develop ridges after achiev- ing a certain length and whether these teeth can be distinguished from Sindacharax teeth. This leads to some taxonomic difficulties, as the genus Sinda- charax was erected based on its supposedly unique ridged teeth. While the analysis of existing Sinda- charax jaw elements demonstrates enough differ- ences with modern alestid elements to keep Sinda- charax as a separate genus, the diagnosis of the Sin- dacharax genus needs to be re-examined in order to define it more accurately. Flowever, more Sin- dacharax cranial and postcranial elements must be recovered for such revision. Sindacharax lothagamensis Stewart, 2003 (Figures 12, 13) KANAPOI MATERIAL. 3156, 2 second inner premaxillary teeth; 3848, 1 first inner premaxillary tooth; 3849, 17 first inner premaxillary teeth, 57 second inner premaxillary teeth; 29287, 7 first in- ner premaxillary teeth. Teeth of Sindacharax lothagamensis are smaller on average than those of other Sindacharax and are relatively common at Kanapoi. The Kanapoi sec- ond inner premaxillary teeth are identical to both the holotype and the Isolated teeth found at Loth- agam (e.g., Stewart, 2003: fig. 3.5) (Fig. 12). The size range of teeth differs slightly from that at Loth- agam; at Lothagam, second inner premaxillary teeth ranged up to 5.5 mm in length with most un- der 3 mm, whereas the Kanapoi teeth ranged to 5 mm, with most under 2 mm. Numerous first inner teeth were found associated with the second inner Figure 13 Sindacharax lothagamensis, SEM of first inner premaxillary tooth, occlusal view, from Kanapoi; labial side at top, lingual at bottom teeth at Kanapoi, particularly at site 3849, and they showed a slightly different cusp pattern than that described for the Lothagam teeth (Stewart, 2003). These Kanapoi first teeth are long and narrow, with the dominant cusp at the lingual end of the tooth. A smaller cusp, not two, veers in a diagonal line towards the presumed bucco-labial side, and anoth- er cusp is positioned anterior to the dominant cusp. Anterior to this are one or more ridges traversing the width of the tooth (Fig. 13). Sindacharax lothagamensis teeth were the second most numerous of Sindacharax teeth at Kanapoi. This abundance of S. lothagamensis is a surprise, as they were primarily recovered in the late Mio- cene Lower Nawata deposits at Lothagam and only occasionally in later Pliocene deposits. Their abun- dance at Kanapoi suggests that absence in later Lothagam deposits may reflect a collection bias or different environmental conditions. Collection bias seems unlikely, as intensive collecting occurred at Pliocene deposits in Lothagam. Flowever, different environmental conditions from the late Miocene to Pliocene deposits at Lothagam is certainly possible, with the latter providing unfavorable habitats for Stewart: Fish ■ 29 Figure 14 Sindacharax mutetii, SEM of first inner pre- maxillary tooth, occlusal view, from Kanapoi; labial side at top, lingual at bottom S. lothagamensis, and Kanapoi may have provided a more favourable environment for them, Sindacharax mutetii Stewart, 2003 (Figures 14-17) EMENDED DIAGNOSIS. Second inner premax- illary tooth distinguished from Sindacharax leper- sonnei Greenwood and Flowes, 1975 and S. loth- agamensis by cusps forming ridges rather than dis- crete cusps as in S. lepersonnei and S. lothagamen- sis. Distinguished from S. deserti Greenwood and Howes, 1975, by absence of raised circular ridge radiating from the dominant lingual cusp; distin- guished from S. greenwoodi Stewart, 1997, by lack of the ridged arc surrounding dominant lingual cusp, and distinguished from all other Sindacharax by broad oval shape. HOLOTYPE. A second inner premaxillary tooth, collected from Lothagam by Sam N. Muteti and Peter Kiptalam in 1993 from Site 1944 in the Apak Member of the Nachukui Formation, and now housed in the collections of the National Mu- seums of Kenya, Nairobi, with the accession num- ber KNM-LT 38265. Figure 15 Sindacharax mutetii, SEM of second inner pre- maxillary tooth, occlusal view, from Kanapoi; labial side at top, lingual at bottom Figure 16 Sindacharax mutetii, SEM of second inner pre- maxillary tooth, occlusal view, from Kanapoi; labial side at top, lingual at bottom 30 ■ CS 498, Harris and Leakey: Kanapoi Figure 17 Sindacharax miitetii, SEM of in situ second and third inner premaxillary teeth, occlusal view, from Kanapoi; labial side at top, lingual at bottom, medial to the right, lateral to the left KANAPOI MATERIAL. 3156, 3 second inner premaxillary teeth; 3845, 21 first inner premaxil- lary teeth, 26 second inner premaxillary teeth; 3846, 14 first inner premaxillary teeth, 39 second inner premaxillary teeth; 3847, 51 first inner pre- maxillary teeth, 75 second inner premaxillary teeth; 3848, 43 first inner premaxillary teeth, 67 second inner premaxillary teeth; 3849, 50 first inner pre- maxillary teeth, 73 second inner premaxillary teeth; ngui site, 28 first inner premaxillary teeth, 65 sec- ond inner premaxillary teeth. Most first and second inner premaxillary teeth recovered were identical to those recovered from the Apak and Kaiyumung Members at Lothagam (Stewart, 2003) (Figs. 14, 15); however, a few sec- ond inner premaxillary teeth showed a slight devi- ation in cusp pattern (Fig. 16). Instead of the first ridge, which is anterior to the dominant cusp, tra- versing the whole width of the tooth, in some teeth it was shortened and often bracketed by one or both ends of the ridge anterior to it. The large number of S. mutetii teeth recovered at Kanapoi reflect a range of individual variations in their cusp patterns. Several of the second and third inner premaxillary teeth recovered have similar pat- terns to their counterparts in situ on the premaxilla recovered from the Apak Member at Lothagam, which was ascribed to cf. 5. mutetii (Stewart, 2003). Therefore, the Lothagam premaxilla is now included in S. mutetii (Fig. 17). This premaxilla re- mains the only jaw element recovered which is as- cribed to S. mutetii. A total of 555 teeth ascribed to S. mutetii were recovered at Kanapoi, making it the most abundant of the Sindacharax species at that site. Sindacharax mutetii teeth were also the most common teeth re- covered from the Apak Member deposits at Loth- agam, although this species was not recovered from the Murongori Member. Stewart (2003) stated that S. mutetii was the only Sindacharax found at Kanapoi. Llowever, further study of the Kanapoi specimens showed that, while S. mutetii is by far the most common species re- covered, teeth of both S. lothagamensis and S. how- esi Stewart, 2003, are also present. Sindacharax howesi Stewart, 2003 KANAPOI MATERIAL. 3845, 3 first inner pre- maxillary teeth; 3846, 1 first inner premaxillary tooth; 3847, 8 first inner premaxillary teeth; 3848, 7 first inner premaxillary teeth, 2 second inner pre- maxillary teeth; 29287, 1 second inner premaxil- lary tooth. Sindacharax howesi teeth were not common at Kanapoi. Mainly first inner premaxillary teeth were recovered, and these were identical to those found in the northern Kaiyumung deposits at Lothagam. Sindacharax howesi teeth were exclusively found in the north Kaiyumung deposits at Lothagam, where they are numerous. Their appearance in the Stewart: Fish ■ 31 Figure 18 Sindacharax sp., SEM of in situ third inner pre- maxillary tooth, from Kanapoi; labial side at top, lingual at bottom Kanapoi deposits indicates a slightly earlier pres- ence (ca. 4.0 Ma) in the Turkana Basin than pre- viously thought. Sindacharax sp. (Figures 18, 19) As discussed above, outer premaxillary and dentary teeth are indistinguishable between the species, and therefore are referred as Sindacharax sp. Similarly, third and fourth inner premaxillary teeth are simi- lar among the species, and again were referred to Sindacharax sp. Third and Fourth Inner Premaxillary Teeth KANAPOI MATERIAL. 3845, 16 third or fourth inner premaxillary teeth; 3846, 11 third or fourth inner premaxillary teeth; 3847, 43 third in- ner premaxillary teeth, 15 fourth inner premaxil- lary teeth; 3848, 25 third or fourth inner premax- illary teeth; 3849, 32 third or fourth inner premax- illary teeth; 29287, 12 third or fourth inner pre- maxillary teeth. No third or fourth inner premaxillary teeth could be assigned to species, as they were very similar throughout the deposits. Often the third and fourth teeth could not be distinguished from each other, as the only conhrmed fourth premaxillary tooth (preserved in situ on the S. greenwoodi type speci- Figure 19 Sindacharax sp., SEM of in situ fourth inner premaxillary tooth, from Kanapoi; labial side probably to the right, lingual probably to the left men) is very worn and the cusp pattern almost in- distinguishable. However, based on comparison with the S. greenwoodi premaxilla and modern Alestes and Brycinus premaxillae, I have tentatively assigned some teeth as third inner teeth (Fig. 18) and fourth inner teeth (Fig. 19). Outer Teeth While outer teeth are difficult to distinguish be- tween species, there are several distinct types. Sim- ilar to the Lothagam outer teeth (Stewart, 2003a), I have classified the Kanapoi teeth into types, with an indication to which species they are most con- sisistently associated. Outer Premaxillary Teeth, Type A KANAPOI MATERIAL. 3156, 4 outer premax- illary teeth; 3845, 26 outer premaxillary teeth; 3846, 10 outer premaxillary teeth; 3847, 86 pre- maxillary outer teeth; 3848, 48 premaxillary outer teeth; 3849, 22 premaxillary outer teeth; 29287, 19 premaxillary outer teeth. Type A teeth consist of one dominant and two much smaller flanking cusps, which slope into a short, uncusped platform on one side but have a steep shelf on the other side (see figs. 3.26 and 3.27 32 ■ CS 498, Harris and Leakey: Kanapoi in Stewart, 2003). They have a round or oval at- tachment base. At Kanapoi, Type A teeth are as- sociated with both S. lothagamensis and S. mutetii teeth; when associated with S. mutetii, they often have elongated attachment bases. Outer Premaxillary Teeth, Type B KANAPOI MATERIAL. 3845, 5 outer premax- illary teeth; 3846, 2 outer premaxillary teeth; 3847, 11 outer premaxillary teeth; 3848, 11 outer pre- maxillary teeth; 3849, 3 outer premaxillary teeth; 29287, 1 outer premaxillary tooth. Type B teeth are similar to Type A but have one or more discrete cusps at the base of the platform (Stewart, 2003: figs. 3.26, 3.27). Their attachment base is round or a roundish oval. These teeth were most common in sites where 5. howesi was also found. Outer Premaxillary Teeth, Type C KANAPOI MATERIAL. 3848, 5 outer premax- illary teeth. Type C teeth have a dominant central cusp, flanked by concentric or semiconcentric rows of small cusps (Stewart, 1997: figs. 2, 3). Their at- tachment base is an elongated oval. These teeth were rare at Kanapoi, and no particular affiliation can be ascertained. Outer Dentary Teeth The outer dentary teeth recovered were mainly first, second, and third teeth; fourth teeth are much smaller and fewer have been recovered. The first tooth is usually truncated posteriorly, to accom- modate the inner tooth. There is considerable wear visible on most dentary teeth, and it is often diffi- cult to describe any morphology on the teeth. As with premaxillary outer teeth, outer dentary teeth can be divided into types, although only Type A was recovered at Kanapoi. Outer Dentary Teeth, Type A KANAPOI MATERIAL. 3156, 5 outer dentary teeth; 3845, 85 outer dentary teeth; 3846, 61 outer dentary teeth; 3847, 205 outer dentary teeth; 3848, 134 outer dentary teeth; 3849, 295 outer dentary teeth; 29287, 122 outer dentary teeth. All dentary teeth recovered at Kanapoi belonged to Type A, although many were too worn to ascer- tain type. These teeth have a dominant, pointed cusp and are flanked by two smaller cusps, which form a shelf on one side, more elongated and less steep than that of the premaxillary teeth (illustrated in Stewart, 2003). On the other side, the cusps slope into a broad platform, which is usually un- cusped but may be weakly cusped. The attachment base is much more elongated than in most premax- illary teeth. Outer dentary teeth were by far the most abundant of all Sindacharax teeth, probably because of their size and robust attachment bases. Inner Dentary Teeth KANAPOI MATERIAL. 3845, 8 inner dentary teeth; 3846, 4 inner dentary teeth; 3847, 22 inner dentary teeth; 3848, 7 inner dentary teeth; 3849, 5 inner dentary teeth; 29287, 5 inner dentary teeth. These teeth are very similar in both Alestes and Sindacharax. There is only one inner tooth on each dentary in living individuals, and it is positioned posterior to a notch in the first outer dentary tooth. Inner dentary teeth are small and round in shape, with a single elongated centrally placed cusp. Worn and/or Eragmented Teeth, Unassigned to Position KANAPOI MATERIAL. 3156, 1 tooth; 3845, 6 teeth; 3846, 48 teeth; 3847, 85 teeth; 3848, 50 teeth; 3849, 4 teeth; 29287, 50 teeth. Order Siluriformes Family Bagridae or Family Claroteidae KANAPOI MATERIAL. 3156, 1 pectoral spine fragment; 3847, 1 cranial spine. These catfish elements are referred to the family level both because of their incomplete nature and the similarity of these elements between some bag- rid and claroteid species. They represent small in- dividuals, probably no longer than 50 cm in total length. Bagrid and/or claroteid catfish elements were common in the field at Kanapoi, often representing large individuals (approximately 1 m in length). Most of these elements were not collected. Many of these appeared to belong to Clarotes Kner, 1855, a large catfish which today often inhabits deltaic regions. Bagrids and claroteids, particularly Bagrus Bose, 1816, and Clarotes, are known from the Mio-Pliocene deposits at Lothagam (Stewart, 2003) and at Koobi Fora (Schwartz, 1983), as well as other Cenozoic deposits in Africa. They do not appear to have radiated in the Turkana Basin as they did in the Western Rift (Stewart, 2001), al- though they are more common in the Plio-Pleisto- cene deposits at eastern Turkana. Family Clariidae Clarias Scopoli, 1777 Heterobranchus Geoffroy Saint-Hilaire, 1809 Clarias sp. or Heterobranchus sp. KANAPOI MATERIAL. 3847, 2 caudal verte- brae; 3849, 2 trunk vertebrae, 1 caudal vertebrae. These vertebrae are referred to Clarias or Het- erobranchus because of great similarity between the elements. These vertebrae derive from small indi- viduals (<50 cm total length). Clariid elements were abundant at Kanapoi, but not collected. As with the bagrid catfish, many el- ements appeared to come from large individuals, up to 2 m in length. Clarias is a bottom-dwelling, in- Stewart: Fish ■ 33 Figure 20 Synodontis sp., SEM of dentary tooth, from Kanapoi; side view shore fish, which can tolerate highly deoxygenated waters. Clariid remains were common throughout the Lothagam deposits (Stewart, 2003) and in late Cenozoic deposits of Africa (Stewart, 2001). Ten- tative identifications are reported in the mid-Mio- cene from Bled ed Douarah, Tunisia (Greenwood, 1973), and Ngorora, Kenya (Schwartz, 1983). Def- inite clariid remains are known from Miocene de- posits in Sinda, Congo (Van Neer, 1992), and Chal- ouf, Egypt (Priem, 1914); Mio-Pliocene deposits in Manonga, Tanzania (Stewart, 1997); Mio-Pleisto- cene deposits in the Western Rift (Van Neer, 1994); Pliocene deposits in Wadi Natrun, Egypt (Green- wood, 1972); and Plio-Pleistocene deposits at Koo- bi Fora (Schwartz, 1983). Extant Clarias is repre- sented by C. lazera Cuvier and Valenciennes, 1840, in Eake Turkana. Clarias is widespread throughout Africa, including the Nile, Congo, and Zambezi Ba- sins. Heterohranchus has a similar appearance and size to Clarias, but may be more sensitive to high salinity values. Modern H. longifilis Valenciennes, 1840, is present in Lake Turkana, but is rare. Like Clarias, Heterohranchus is widespread throughout the major river basins of Africa. It was identified in late Pleistocene Lake Edward Basin (Congo) depos- its (Greenwood, 1959). Family Mochokidae Synodontis Cuvier, 1817 Synodontis sp. (Eigures 20, 21) KANAPOI MATERIAL. 3845, 5 teeth; 3847, 1 cranial spine base. Synodontis teeth were not common in the Kan- apoi sites sampled, suggesting Synodontis was not a dominant presence at Kanapoi. The teeth are lo- cated on the dentary and are curved (Eig. 20). They averaged about 1 mm in width, similar to modern Synodontis in Lake Turkana, suggesting the fossil fish reached 30-35 cm in total length (and much larger than the other Lake Turkana mochokid, Mochocus de Joannis, 1935, which reaches only 6.5 cm in length). The cranial spine base recovered is fragmentary (Fig. 21) but very similar to that of modern Synodontis. In life, it is positioned anterior to the dorsal cranial spine and resembles a trun- cation of the spine. Synodontis was probably not common at Kana- poi, as its remains normally preserve well. It was also not common at Lothagam, although consis- tently present through the deposits. Synodontis in- habits all zones of lakes and rivers, and is omniv- orous, eating insects, small fish, mollusks, and zoo- plankton. Eossil Synodontis is also known from Miocene deposits at Rusinga and Chianda, Kenya (Greenwood, 1951; Van Couvering, 1977), Mog- hara and Chalouf, Egypt (Priem, 1920), and Bled ed Douarah, Tunisia (Greenwood, 1973); Mio- Pleistocene deposits in the Western Rift (Green- wood and Howes, 1975; Van Neer, 1992, 1994); Pliocene deposits in the Western Rift (Stewart, 1990) and Wadi Natrun (Greenwood, 1972); and Plio-Pleistocene deposits at Koobi Fora (Schwartz, 1983). Two species of Synodontis inhabit modern Lake Turkana — S. schall Bloch and Schneider, 1801, and S. frontosus Vaillant, 1895. Synodontis is also widespread in systems throughout the Afri- can continent. Order Perciformes Suborder Percoidei Family Latidae hates Cuvier, in Cuvier and Valenciennes, 1828 hates niloticus (Linnaeus, 1758) KANAPOI MATERIAL. 3845, 1 hyomandibu- lar, 1 premaxilla, 1 dentary, 1 first trunk vertebra, 2 trunk vertebrae, 1 caudal vertebra; 3847, 2 pre- maxillae, 2 posttemporal, 1 quadrate, 1 articular, 1 basioccipital fragment, 1 vomer, 6 first trunk ver- tebra, 1 trunk vertebra; 3848, 1 basioccipital, 1 premaxilla, 1 vomer. These elements are identical to those in modern hates niloticus and represent fish of a diverse size range. Several large fossil elements were compared with modern L. niloticus elements recovered from the lake margin, and these indicated that many of the Kanapoi fish had an estimated total length of over 2 m. Many elements of hates niloticus were observed in the field at Kanapoi, but only those listed above were collected, for their diagnostic value. Many of 34 ■ CS 498, Harris and Leakey: Kanapoi Figure 21 Synodontis sp., SEM of cranial spine base, from Kanapoi, ventral view the bones represented large individuals, estimated to be approximately 2 m in length. Clearly, L. nil- oticus was a common component of the Kanapoi fish fauna, and with many large individuals must have been one of the most voracious consumers of fish in the aquatic food chain. Modern hates in- habits most zones of lakes and rivers, although it only tolerates well-oxygenated waters. It is highly piscivorous. Elements of fossil hates spp. are common in Af- rican deposits (Stewart, 2001) and are known from Miocene deposits from Rusinga, Kenya (Green- wood, 1951), Gebel Zelten and Cyrenaica, Libya, (Arambourg and Magnier, 1961), Moghara and Chalouf, Egypt (Priem, 1920), Bled ed Douarah, Tunisia (Greenwood, 1973); Mio-Pliocene deposits at Eothagam, Kenya (Stewart, 2003); Plio-Pleisto- cene deposits from Eakes Albert and Edward Basins (together with Semlikiichthys rhachirhinchus) (Greenwood, 1959; Greenwood and Howes, 1975; Stewart, 1990; Van Neer, 1994); an unpublished report from Marsabit Road, Kenya (in Schwartz, 1983), the lower Omo Valley (Arambourg, 1947), and Koobi Fora (Schwartz, 1983); and Pliocene de- posits from Manonga, Tanzania (Stewart, 1997), and Wadi Natrun, Egypt (Greenwood, 1972). hates niloticus has also been reported from Messinian (late Miocene) deposits in Italy, the only confirmed report of this species in Europe (Otero and Sorbini, 1999). Elements formerly identified as hates from Eocene deposits in Fayum, Egypt (Weiler, 1929), were referred to Weilerichthys fajumensis (Otero and Gayet, 1999b). Modern hates is known from Lake Turkana (L. niloticus and L. longispinis Wor- thington, 1932) and is widespread throughout northern, eastern, and western Africa from Senegal to and including the Nile and Congo River Basins. Percoidei incertae sedis Semlikiichthys Otero and Gayet, 1999 Semlikiichthys rhachirhinchus (Greenwood and Howes, 1975) Semlikiichthys cf. S. rhachirhinchus KANAPOI MATERIAL. 3845, 4 first trunk ver- tebrae, 16 trunk vertebrae, 2 caudal vertebrae; 3846, 1 dentary, 1 first trunk vertebra, 5 trunk ver- tebrae, 4 caudal vertebrae; 3847, 4 dentaries, 1 first trunk vertebra, 9 trunk vertebrae, 3 caudal verte- brae; 3848, 1 vomer, 2 basioccipital, 4 first trunk vertebrae. All material collected at Kanapoi is identical to drawings of Semlikiichthys rhachirhinchus (former- ly hates rhachirhinchus [Greenwood and Howes 1975] but renamed S. rhachirhynchus [Otero and Gayet, 1999a]; this author adheres to the original spelling [Greenwood and Howes, 1975] for “rhach- irhinchus”), and is also identical to material col- lected at Lothagam, figured and described as Sem- likiichthys cf. 5. rhachirhinchus (Stewart, 2003). Full descriptions and photos of the extensive Sem- likiichthys cf. 5. rhachirhinchus material from Lothagam are found in the Lothagam volume (Stewart, 2003), where it is compared with the orig- inal drawings of L. rhachirhinchus (Greenwood and Howes, 1975). In particular, the vomer recovered from Kanapoi is fully described in the Lothagam volume (Stewart, 2003), as it is the only vomer recovered from either Kanapoi or Lothagam which is almost identical to the type S. rhachirhinchus vomer (Greenwood and Howes, 1975), and is important in the naming of these fossils {rhachirhinchus means loosely “snout with spine”). The Kanapoi elements establish a strong presence of Semlikiichthys cf. S. rhachirhinchus at Kanapoi. As at Lothagam, Lates niloticus and Semliki- ichthys cf. S. rhachirhinchus appear to have coex- isted at Kanapoi. Both groups of fish seem to have attained large size, up to 2 m in length. This author has previously suggested (Stewart, 2001, 2003) that the presence of Semlikiichthys in the Turkana Basin resulted from exchange of faunas with the Lakes Albert and Edward Basins, the only other basins in which Semlikiichthys is known. Its reasonably com- mon presence in Kanapoi further supports the sug- gestion of exchange. The identification of a palatine in Wadi Natrun Pliocene deposits probably refer- able to Semlikiichthys (Greenwood, 1972; Green- wood and Howes, 1975; discussed in Stewart 2001, 2003) also supports a more widespread faunal ex- Stewart: Fish ■ 35 change within the Nile-linked systems, extending to the Egyptian Nile area. hates or Semlikiichthys KANAPOI MATERIAL. 3156, 3 first trunk ver- tebrae, 1 caudal vertebra; 3845, 1 dorsal spine; 3846, 1 trunk vertebra; 3847, 1 maxilla fragment, 1 basioccipital, 1 pelvic spine; 3848, 1 basioccipital fragment, 4 vertebra fragments. Perciformes Indeterminate KANAPOI MATERIAL. 3156, 5 pelvic spines. Pelvic spines are often difficult to distinguish be- tween cichlids and hates! Semlikiichthys. One pelvic spine appears to be more similar to those of cich- lids, but not enough for positive identification. If so, it would be the only cichlid fossil recovered at Kanapoi. Cichlids are similarly rare at Lothagam. PALEOECOLOGY The Kanapoi fish fauna is characterized by two tro- phic components: large, piscivorous fish, in partic- ular the polypterids, Gymnarchus, Hydrocynus, hates (and probably Semlikiichthys by analogy), and the bagrid and clariid catfish; and medium to large molluscivores, including Hyperopisus, Gym- narchus (which is also a piscivore), Sindacharax, and possibly habeo. Together with the numerous elements of the crocodile Euthecodon Fourteau, 1920, identified in the Kanapoi deposits, there was clearly much piscivory in this region of the Lon- yumun lake (see, e.g., Tchernov, 1986, for details of Euthecodon). While Euthecodon's diet probably consisted of the plethora of fish in the lake, the diet of the piscivorous fish must have included the nu- merous Sindacharax, Hyperopisus, and Eabeo in- dividuals, as well as smaller fish whose elements were not preserved. The near-absence of herbivo- rous fish, including Barbus, Alestes, and the large tilapiine cichlids is surprising. Most of these groups are common in modern Lake Turkana and the Nile system, and would be expected in the Pliocene lake. Certain absences may be explained by unfavorable environmental conditions (see DISCUSSION AND SUMMARY section). Barbus and the large tilapiine cichlids are generally scarce in African fossil de- posits prior to the Pleistocene (Stewart, 2001). The diversity and composition of taxa represent- ed at Kanapoi is reminiscent of the modern Omo River Delta in northern Lake Turkana, which is in- habited, among other fish, by mormyriforms, char- acoids, bagrids, claroteids, and percoids. Gymnar- chus and Clarotes in particular prefer delta regions (Lowe-McConnell, 1987). Many of the fish in the modern Omo River Delta region are intolerant of saline waters and therefore inhabit the delta and the lower reaches of the Omo River because they can- not tolerate the more saline Lake Turkana waters. Modern hates is intolerant of deoxygenated waters, and the modern mormyriforms are intolerant of sa- line waters. By analogy with the modern Omo Riv- er Delta, Kanapoi waters may also therefore have been well-oxygenated and fresh. The scarcity of fish such as Protopterus, Polyp- terus Saint-Hilaire, 1802, and Heterotis Ruppell, 1829, which were relatively common in the Na- wata Formation at Lothagam, may signify an ab- sence of vegetated, shallow backwaters or bays, as these are the type of habitats frequented by modern members of these taxa. The nearby Pliocene site of Eshoa Kakurongori contained numerous Protopte- rus toothplates, indicating a very different environ- ment from Kanapoi. DISCUSSION AND SUMMARY Because the Kanapoi fish fauna comes primarily from one phase in the Kanapoi Formation — the la- custrine phase — there are no evolutionary or envi- ronmental transitions documented as was apparent through the Nawata and Nachukui Formations at the nearby Lothagam site. Nevertheless, the fauna alters some of the evolutionary and biogeographic interpretations made from the Lothagam fauna, as discussed below. Most surprising is the comparison of the taxo- nomic composition from the Lothagam Muruon- gori deposits, Kanapoi deposits, and what the au- thor has observed from the Ekora site deposits, all of which are presumed to derive from the Lonyu- mun Lake. The Muruongori deposits contain sim- ilar taxa to that at Kanapoi (Table 1), but also in- clude two Sindacharax taxa — S. deserti and S. greenwoodi — and two Tetraodon Linnaeus, 1758, taxa — T. fahaka Hasselquist, 1757, and Tetraodon sp. nov., Stewart, 2003 — which are common at Muruongori and at Ekora but which are completely absent from Kanapoi. Further, the most common Sindacharax species at Kanapoi — S. mutetii — is ab- sent in the Muruongori Member and apparently in the Ekora fauna, although common in the Apak Member of Lothagam. There are several possible explanations for this disparity in taxa between Kanapoi on one hand and Muruongori and Ekora on the other, which derive from the same lake. First, the different sites may represent different time intervals in the lake’s his- tory: the Ekora tetrapod fauna is said to be younger than that at Kanapoi (Maglio in Behrensmeyer, 1976), and the Muruongori Member may also be slightly younger than at Kanapoi. The “new” Sin- dacharax and Tetraodon taxa from Muruongori and Ekora may represent immigrants from a new inflow which was not present during Kanapoi de- position. The abundance of S. mutetii at both Kan- apoi and in the Apak Member of Lothagam, which dates earlier than Muruongori and Ekora, may sug- gest an earlier deposition of the Kanapoi deposits, and a faunistic change between the Kanapoi, and Muruongori and Ekora waters. Sindacharax mu- tetii is completely absent from Muruongori and Ekora. Alternatively, the Kanapoi and Muruongori and 36 ■ CS 498, Harris and Leakey: Kanapoi Table 1 Fish taxa found in Mio-Pliocene deposits in Nawata, Apak, Muruongori, and Kaiyumung Members, Lothagam (Stewart, 2003) and Kanapoi (this report); see Feibel (2003a) and McDougall and Feibel (1999) for detailed information about the geochronology, geological formations, and members at Lothagam Nawata Apak Muruongori Kaiyumung Kanapoi Protoptems sp. + + + Polypterus sp. + + + + + Heterotis sp. + + + Hyperopisus sp. + + + + Gyrnnarchns sp. + + + + -h Labeo sp. ? + + + + Barbus sp. + + + Distichodus sp. + + + Hydrocynus sp. + + + + + Brycimts macrolepidotus + Alestidae sp. + + Sindacharax lothagamensis + + + S. mutetii + + + S. howesi + + S. deserti + + S. greenwoodi + + Sindacharax sp. + -f + + + Bagrus sp. -t- Aff. Bagrus sp. + + + Clarotes sp. + -r + Bagridae/Claroteidae + Schilbe sp. + Clarias or Heterobranchus + -r + -r Synodontis sp. + + + + + Lates niloticus + + + + Lates sp. + + + + -f Semlikiichthys cf. S. rhachirhinchus + + + + + Cichlidae + -1- Tetraodon sp. nov. + Tetraodon sp. + + Ekora deposits may represent different ecological zones in the Lonyumun Lake, with the “new” taxa restricted ecologically to local habitats and/or ba- sins. Again the analogue of the modern Omo River Delta is appropriate here, with many taxa restricted to the delta region and not occurring in Lake Tur- kana proper, A third alternative is that the field sampling was not extensive enough, and elements of the “new” taxa were not recovered at Kanapoi. This alterna- tive seems less likely, as extensive screening was un- dertaken at all areas, and teeth of the “new” taxa should at least be somewhat represented at Kana- poi. Further, the Tetraodon toothplates are very ro- bust and distinctive as fossils, and extensive survey- ing at Kanapoi should have recovered at least some toothplates, if this taxon had been present. A third “new” taxon — cf. Semlikiichthys rhach- irhinchns — was rare in earlier Lothagam deposits, but was common in the Muruongori Member at Lothagam, at Kanapoi, and at Ekora. This percoid apparently coexisted with Lates niloticus, which was also recovered in large numbers. Stewart (2001) has reported that S. rhachirhinchus elements were also found in the Western Rift and probably at Wadi Natrun, suggesting interchange between the three systems. Otero and Sorbini (1999) have suggested that the genus Lates diversified in the fresh waters of Europe and Africa in the Miocene, from a Mediterranean origin. Further work is need- ed to clarify the relationships of S. rhachirhinchus. In sum, the Kanapoi fauna is of considerable in- terest for several reasons. It has an unusual com- position of mainly large piscivorous fish and me- dium to large molluscivores, and a scarcity of her- bivorous fish. This composition is considerably dif- ferent from that of modern Lake Turkana, which is much more evenly balanced between herbivores and piscivores. The scarcity of herbivores such as Barbus and the large tilapiine cichlids, common in the modern lake and the Nile River system, is par- ticularly enigmatic. While the Kanapoi fauna shares many taxa with the similarly aged Muruongori Member fauna from Lothagam and also from Ekora, it also shows some differences. The dominance of Sindacharax mutetii Stewart: Fish ■ 37 at Kanapoi and in the Apak Member at Lothagam contrasts with its absence in the Muruongori de- posits and at Ekora, as does the dominance of S. deserti and Tetraodon at Muruongori and Ekora, and their absence from Kanapoi and the Apak Member. Whether these disparities reflect ecologi- cal variants or chronological differences needs to be further studied. ACKNOWLEDGMENTS I express my gratitude to Dr. Meave Leakey, who invited me to study the fossil fish from Kanapoi and provided support in the field. My gratitude also to Sam N. Muted, who worked extremely hard with me in the field and pro- vided additional help at the National Museums of Kenya. Thanks also to the members of the National Museums of Kenya fossil team for their help in collecting fossils. Spe- cial thanks to Kamoya Kimeu for leadership and support in the field, and to the Paleontology staff at the National Museum in Nairobi for assistance in the lab. Special thanks to Donna Naughton for scanning electron micro- scopic photography. My thanks to the Canadian Museum of Nature Research Advisory Council for transport and field assistance. Many thanks as well to John Harris for editorial assistance with the manuscript, to Frank Brown for discussions in Kenya on the fish and geology, to Craig Feibel for information on the sedimentary sequence, and to O. Otero, A. Murray, and P. Forey for helpful com- ments on the article. LITERATURE CITED Arambourg, C. 1947. Contribution a V etude geologique et paleontologique du Basin du lac Rodolphe et de la Basse Vallee de I’Omo. Deuxieme partie. Paleon- tologie. In Mission scientifique de I’Omo, 1932- 1933. Tome I. Geologie-Anthropologie. Paris: Mu- seum National d’Histoire Naturelle, Fasc. 3, 229- 562. Arambourg, C., and P. Magnier. 1961. Gisements de ver- tebres dans le bassin tertiare de Syrte (Libye). Comp- tes-Rendus de I’Academie des Sciences, Paris, 151\ 1181-1183. Beadle, L. C. 1981. The inland waters of tropical Africa. London: Longman, 475 pp. Behrensmeyer, A. K. 1976. Lothagam, Kanapoi, and Eko- ra: A general summary of stratigraphy and fauna. In Earliest man and environments in the Lake Rudolf Basin, eds. Y. Coppens, F. C. Howell, G.Ll. Isaac, and R. E. F. Leakey, 163-173. Chicago: University of Chicago Press. Carroll, R. L. 1988. Vertebrate paleontology and evolu- tion. New York: W. H. Freeman and Co., 698 pp. Daget, ]., J. P. Gosse, and D. F. E. Thys van den Auden- aerde. 1984. Checklist of the freshwater fishes of Af- rica. Vol. 1. Paris: Orstom, 410 pp. . 1986. Checklist of the freshwater fishes of Africa. Vol. 2. Paris: Orstom, 520 pp. Dutheil, D. B. 1999. The first articulated fossil cladistian: Seremoichthys kemkemensis, gen. et sp. nov., from the Cretaceous of Morocco. Journal of Vertebrate Paleontology 19(2):243-246. Feibel, C. S. 2003a. Stratigraphy and depositional history of the Lothagam Sequence. In Lothagam — The dawn of humanity in Eastern Africa, ed. M. G. Lea- key and J. M. Harris, 1 7-3 1 . New York: Columbia University Press. . 2003b. Stratigraphy and depositional setting of the Pliocene Kanapoi Formation, lower Kerio Valley, Kenya. In Geology and vertebrate paleontology of the Early Pliocene site of Kanapoi, northern Kenya. Edited by John M. Harris and Meave G. Leakey. Contributions in Science 498:9-20. Greenwood, P. H. 1951. Pish remains from Miocene de- posits of Rusinga Island and Kavirondo Province, Kenya. Annals and Magazine of Natural History 12(4):1192-1201. . 1959. Quaternary fish-fossils. Exploration du Parc National Albert, Mission J. de Heinzelin de Beaucourt (1950). Fasc. 4:1-80. . 1972. New fish fossils from the Pliocene of Wadi Natrun, Egypt. Journal of Zoology, London 168: 503-519. . 1973. Fish fossils from the late Miocene of Tu- nisia. Notes du Service Geologique de Tunisie 37: 41-72. . 1976. Notes on Sindacharax Greenwood and Howes, 1975, a genus of fossil African characid fish- es. Revue de Zoologie africaine 90:1-13. Greenwood, P. H., and G. J. Howes. 1975. Neogene fossil fishes from the Lake Albert-Lake Edward rift (Con- go). Bulletin of the British Museum (Natural Histo- ry) (Geology) 26(3):69-127. Hopson, A. J., and J. Hopson. 1982. The fishes of Lake Turkana. In Lake Turkana: A report on the findings of the Lake Turkana Project (1972-1975), ed. A. J. Hopson, 283-347. London: Overseas Development Administration. Lavocat, R. 1955. Decouverte de Dipneustes du genre Protopterus dans le Tertiare ancien de Tomaguilelt (Soudan frangais). Comptes-Rendus de I’Academie des Sciences 240:1915-1917. Leakey, M. G., C. S. Feibel, R. L. Bernor, J. M. Harris, T. E. Cerling, K. M. Stewart, G. W. Storrs, A. Walker, L. Werdelin, and A. J. Winkler. 1976. Lothagam: A record of faunal change in the Late Miocene of East Africa. Journal of Vertebrate Paleontology 16(3): 556-570. Leakey, M. G., and J. M. Harris, eds. 2003. Lothagam: Dawn of humanity in Eastern Africa. New York: Columbia University Press, 678 pp. Lowe-McConnell, R. H. 1987. Ecological studies in trop- ical fish communities. Cambridge: Cambridge Uni- versity Press, 381 pp. McDougall, L, and C. S. Feibel. 1999. Numerical age con- trol for the Miocene-Pliocene succession at Lotha- gam, a hominoid-bearing sequence in the northern Kenya Rift. Journal of the Geological Society, Lon- don 156:731-745. Murray, A. M., and K. M. Stewart. 2002. Phylogenetic relationships of the African genera Alestes and Bry- cinus (Teleostei, Characiformes, Alestidae). Canadi- an Journal of Zoology 80:1887-1899. Otero, O. 2001. The oldest-known cyprinid fish of the Afro-Arabic plate, and its paleobiogeographical im- plications. Journal of Vertebrate Paleontology 21(2): 386-388. Otero, O. and M. Gayet. 1999a. Semlikiichthys (Percifor- mes Incertae sedis), genre nouveau, et position sys- tematique nouvelle pour Lates rhachirhynchus Greenwood and Howes, 1975, du Plio-Pleistocene africain. Cybium 23:13-27. . 1999b. Weilerichthys fajumensis (Percoidei incer- tae sedis), new name and systematic position for La- tes fajumensis Weiler, 1929, from the Eocene of the 38 ■ CS 498, Harris and Leakey: Kanapoi Fayum (Egypt). Neues Jahrbuch fiir Geologie und Paldontologie, Mh. (1999)2:81-94. . 2001. Palaeoichthyofaunas from the Lower Oli- gocene and Miocene of the Arabian Plate: Palaeoe- cological and palaeobiogeographical implications. Palaeogeography, Palaeoclimatology, Palaeoecology 165:141-169. Otero, O., and L. Sorbini. 1999. Etude systematique et anatomique du Lates (Perdformes, Latidae) du Mes- sinien continental de Monte Castellaro, Italic. Con- siderations paleobiogeographiques. Studi e recherche sui giacimenti terziari di Bolca 8:29-42. Priem, R. 1914. Sur les poissons fossiles et en particulier des Silurides du Tertiaire superieur et des couches recentes d’Afrique. Mhnoires de la Societe geolo- gique de France 21:1-13. . 1920. Poissons fossiles du Miocdie de I’Egypte. \n Contribution a I’Etude des Vertebres miocenes de I’Egypte, ed. R. Eourtau, 12-15. Cairo: Government Press. Schwartz, H. L. 1983. Paleoecology of late Cenozoic fish from the Turkana basin, northern Kenya. Ph.D. Dis- sertation, University of California, Santa Cruz. Stewart, K. M. 1990. Eossil fish from the Upper Semliki. In Evolution of environments and Hominidae in the African Western Rift Valley, ed. N. T. Boaz, 141- 162. Martinsville: Virginia Museum of Natural His- tory Memoir 1. . 1997. A new species of Sindacharax (Teleostei: Characidae) from Lothagam, Kenya, and some im- plications for the genus. Journal of Vertebrate Pale- ontology 17(l):34-38. . 2001. The freshwater fish of Neogene Africa (Miocene-Pleistocene): Systematics and biogeogra- phy. Fish and Fisheries 2:177-230. . 2003. Fossil fish remains from Mio-Pliocene de- posits at Lothagam, Kenya. In Eothagam: Dawn of humanity in Eastern Africa, eds. M. G. Leakey and J. M. Harris, 75-115. New York: Columbia Univer- sity Press. Stromer, E. 1916. Die Entdeckung und die Bedeutung der Land- und Siisswasser bewohnenden Wirbeltiere in Tertiar und der Kreide Agypte. Zeitschrift der Deutschen Geologischen Gesellschaft 68:397-425. Tchernov, E. 1986. Evolution of the crocodiles in East and North Africa. Paris: Editions du Centre National de la Recherche Scientifique, 65 pp. Van Couvering, J. A. H. 1977. Early records of fresh-wa- ter fishes in Africa. Copeia 1:163-166. Van Neer, W. 1992. New late Tertiary fish fossils from the Sinda region, eastern Congo. African Study Mono- graphs Supplement 17:27-47. . 1994. Cenozoic fish fossils from the Albertine Rift Valley in Uganda. In Geology and palaeobiology of the Albertine Rift Valley, Uganda-Congo. Vol. II: Palaeobiology, eds. M. Pickford and B. Senut, 89- 127. Orleans: CIEEG Occasional Publications. Weiler, W. 1929. Die mittel- und obereocane Fischfauna Agyptens mit besonderer Beriicksichtigung der Te- leostomi. Abbandlungen der Bayerische Akademie der Wissenscbaften 1:1-57. Received 26 December 2002; accepted 23 May 2003. Early Pliocene Tetrapod Remains erom Kanapoi, Lake Turkana Basin, Kenya John M. Harris,' Meave G. Leakey, ^ and Thure E. Cerling^ Appendix by Alisa J. Winkler^ ABSTRACT. Kanapoi, located in the southwestern portion of the Lake Turkana Basin, is the type locality of the oldest East African australopithecine species, Australopithecus anamensis, and has yielded a diverse tetrapod assemblage that includes more than 50 species. The terrestrial fossils derive from fluviatile sands that date between 4.17 and 4.07 Ma and which sandwich a lacustrine interval that represents the extensive Lonyumun Lake — a predecessor of Lake Turkana. The tetrapod assemblage is broadly similar to compa- rably aged assemblages from the nearby locality of Lothagam and the slightly older site of Aramis in Ethiopia. Soil horizons from the succession suggest a mixture of habitats similar to those in the vicinity of the Omo Delta at the north end of the modern Lake Turkana. The Kanapoi biota, however, appears to represent a mixture of open and closed xeric habitats. INTRODUCTION The early Pliocene locality of Kanapoi is located to the southwest of Lake Turkana in northern Kenya at 36°3'51.1"E, 2°18'32.2"N. The Kanapoi sites were first collected in the mid-1960s (1965-67) by expeditions from Harvard University under the leadership of Bryan Patterson. Significant among the 400 specimens collected by the Harvard Uni- versity parties are the holotypes of the elephantid Loxodonta adaurora Maglio, 1970, and rhinocer- otid Ceratotherium praecox Hooijer and Patterson, 1972, two of the most characteristic components of the East African Pliocene biota, together with a hominid humerus that was tentatively identified as Australopithecus sp. (Patterson et ah, 1970). Some elements of the fauna were described in detail, in- cluding proboscideans (Maglio, 1970, 1973), suids (Cooke and Ewer, 1972), and rhinos (Hooijer and Patterson, 1972), but the remainder of the material was largely overlooked. Kanapoi was subsequently recollected in the mid- 1990s by National Museums of Kenya expeditions under the leadership of Meave Leakey. The verte- brate fauna was considerably augmented and the hominid hypodigm was expanded to include cra- nial, dental, and postcranial material of Australo- pithecus anamensis Leakey et ah, 1995, currently the oldest recognized East African australopithecine 1. George C. Page Museum, 5801 Wilshire Boulevard, Los Angeles, California 90036, USA. 2. National Museums of Kenya, PO Box 40658, Nai- robi, Kenya. 3. Departments of Biology and Geology & Geophysics, University of Utah, Salt Lake City, Utah 84112, USA. 4. Shuler Museum of Paleontology, Southern Method- ist University, Dallas, Texas 75275, USA. Contributions in Science, Number 498, pp. 39-113 Natural History Museum of Los Angeles County, 2003 (Ward et ah, 2001). The age of the Kanapoi biota is tightly constrained to between 4.17 and 4.07 Ma (Leakey et ah, 1998) and, as such, it documents a perilacustrine assemblage from an interval of time that is not well represented elsewhere in the Lake Turkana Basin. We provide here a description of much of the fossil vertebrate material recovered by the Harvard University and National Museums of Kenya expe- ditions. The fish and carnivorans are described sep- arately by Kathy Stewart and Lars Werdelin, re- spectively, in adjacent parts of this Contributions in Science issue, as is an assessment of the geologic context by Craig Feibel. The micromammals will be investigated by Frederick Kyalo of the National Museums of Kenya for his Ph.D. dissertation, but a preliminary assessment of this material by Alisa Winkler is appended to this contribution. GEOLOGIC SETTING The stratigraphic sequence at Kanapoi reflects de- position in fluviatile and lacustrine environments and has been studied by Denis Powers (1980) and Craig Feibel (2003b). A basal fluvial complex of channel sandstones and floodplain paleosols lies on the dissected surface of Miocene volcanic rocks. Two devitrified tuffs in this lower fluviatile unit have yielded dates of 4.17 ± 0.03 Ma (lower pu- miceous tuff) and 4.12 ± 0.02 Ma (upper pumi- ceous tuff) (Leakey et ah, 1995). In this unit, ver- tebrate fossils occur primarily in the vertic flood- plain paleosols. The basal complex is overlain by claystones with ostracods and mollusks; these were deposited within the early Pliocene Lonyumun Lake that extended throughout much of the Lake Turkana Basin. The vitric Kanapoi Tuff from the upper portion of the lower lacustrine sequence con- 40 ■ CS 498, Harris and Leakey: Kanapoi tained pumices that were dated to 4.07 ± 0.02 Ma (Leakey et ah, 1998). This tuff cannot be matched with any tephra exposed in the northern part of the Lake Turkana Basin but shows affinity with tephra from the Lake Baringo Basin. The lacustrine strata are overlain by extensive deltaic sandstones that are rich in vertebrate remains, and these are in turn overlain by a second fluvial interval that has exten- sive conglomeratic channels near its top. The se- quence is capped by the Kalokwanya Basalt that has been dated to 3.4 Ma, providing an upper limit on the age of the formation (Feibel, 2003b). CONVENTIONS AND ABBREVIATIONS The full accession number for Kanapoi specimens housed at the National Museums of Kenya in Nai- robi begins with the prefix KNM-KP (e.g., KNM- KP 435). This prefix is abbreviated to KP in the descriptive portions of the text and omitted alto- gether in the list of Kanapoi material at the begin- ning of each systematic section. The following ab- breviations appear in the lists of specimens and in the tables: ant = anterior ap anteroposterior BL = buccolingual dist = distal EK = Ekora frag = fragment h/c = horn core juv juvenile KP = Kanapoi LL = labiolingual LT Lothagam Lt = left It = length MD mesiodistal mand = mandible max = maxilla met = width at metaloph(id) m/p metapodial m/t = metatarsal PL = plastron length post = posterior prot = width at protoloph(id) prox = proximal p/c = postcranial Rt = right tr = transverse SYSTEMATIC DESCRIPTION Order Chelonia Family Pelomedusidae Turkanemys Wood, 2003 Turkanemys pattersoni Wood, 2003 KANAPOI MATERIAL. 435, carapace; 436, carapace; 437, carapace frags; 451, carapace frags; 562, partial carapace and plastron; 30174, cara- pace and plastron; 30438, carapace and plastron; 30470, carapace and plastron; 30595, carapace; 30597, carapace and plastron; 30618, carapace and plastron. Turkanemys pattersoni was recently described from Lothagam (Wood, 2003). Its shell differs from all other African pelomedusid species by virtue of having trapezoidally shaped first vertebral scute and in the tendency of the nuchal bone to be pro- portionally broader than in other species. The shell differs from all South American species of Podoc- nemis (to which many African fossil pelomedusids have been referred in the past as a matter of con- venience) in having six rather than seven neural bones and a triangular rather than pentagonal in- tergular with gulars meeting in the midline behind it. Turkanemys pattersoni also differs from all other pelomedusid species in the structure of the cervical series, with articular surfaces being intermediate in shape between saddle joints of typical podocnemi- nes and procoelous condition of pelomedusines and Erymnochelys madagascariensis Grandier, 1867. Turkanemys pattersoni is the most common che- lonian from Kanapoi. Kenyemys Wood, 1983 Kenyemys sp. indet. KANAPOI MATERIAL. 30419, carapace frags; 30468, juvenile carapace and plastron. Kenyemys williamsi Wood, 1983, differs from all other known pelomedusids by the following com- bination of characters: (a) a series of elongate tu- berosities forming an interrupted keel extending along the midline rearward from the dorsal surface of the second neural bone; (b) six neural bones forming a continuous series, the anterior end of the first abutting directly against the rear margin of the nuchal bone and the sixth one being heptagonal; (c) outer corners of nuchal bone extending beyond lateral margins of first vertebral scute; (d) pentag- onal shape of first vertebral scute; (e) only eighth and posterior part of seventh pair of pleural bones meet at midline of carapace; (f) anterior plastral lobe truncated; (g) triangular intergular scute not overlapping anterior end of entoplastron and only partially separating the gular scutes along the mid- line axis of the plastron (Wood, 1983). The occur- rence of two pelomedusid specimens with a midline keel on the carapace indicates the presence of an as yet undetermined species of Kenyemys. Family Trionychidae Tribe Cyclanorbini Cyclanorbis Grey, 1854 Cyclanorbis turkanensis Meylan et ah, 1990 KANAPOI MATERIAL. 17196, carapace miss- ing the eight pair of costals and the lateral portions of all the left costals (holotype). Only one specimen (the holotype) of this taxon Harris, Leakey, Ceding, and Winkler: Tetrapods ■ 41 Figure 1 KNM-KP 10552; Geochelone crassa shell, dorsolateral view; scale = 25 cm 42 ■ CS 498, Harris and Leakey: Kanapoi is known from Kanapoi. The distinctly concave lat- eral margins of the carapace posterior to the fourth costal together with the hypertrophied and distinct- ly V-shaped cross-section of the vertebrae centra distinguish this species from other species of Cy- clonorbis. Cyclanorbini gen. and sp. indet. KANAPOI MATERIAL. 17202, carapace. A relatively complete cyclanorbin carapace un- fortunately lacks the diagnostic scutes that would permit identification beyond the tribal level. Trionychidae gen. and sp. indet. KANAPOI MATERIAL. 30605, carapace and plastron frags. This specimen comprises fragments of carapace and plastron of a single individual that have the sculpted morphology characteristic of trionychid turtles but which cannot be identified beyond the family level. Family Testudinidae Geochelone Fitzinger, 1835 Geochelone crassa Andrews, 1914 (Figure 1) KANAPOI MATERIAL. 10552, complete shell; 30199, carapace and plastron. Geochelone is an almost cosmopolitan tortoise genus in which the carapace and plastron are never hinged and which often achieved large size. Geo- chelone crassa is a large (>700 mm PL) African form in which the pectorals are widest at the mid- line but narrow considerably on either side of the midline; gulars and pectorals fall short of the en- toplastron. This species is based on a partial plas- tron from the early Miocene locality of Karungu (Meylan and Auffenberg, 1986:281) but evidently survived into the Pliocene. Meylan and Auffenberg (1986) identified KP 10552 — a nearly complete but crushed, meter-long shell of a large land tortoise — as G. crassa (Fig. 1) but erroneously documented the accession number as 10052. They noted that most of the shell was badly broken but sulci and bone sutures are visible on the ventral surface of the anterior lobe of the plastron. The pectorals are widest at the midline (110 mm) but narrow later- ally (to 30 mm). The gular scutes appear to reach the entoplastron but the pectorals do not. cf. Geochelone sp. indet. KANAPOI MATERIAL. 30439, carapace and plastron; 30614, half carapace and plastron. Several specimens comparable in size to, or smaller than, the extant leopard tortoise Geoche- lone pardalis Bell, 1828 were recovered from Kan- apoi by National Museums of Kenya expeditions. Order Crocodylia Family Crocodylidae Crocodylus Laurenti, 1768 Crocodylus niloticus Laurenti, 1768 (Figure 2) KANAPOI MATERIAL. 18333, Lt. mand frag; 18334, Lt. mand frag; 18336, incomplete skull; 18337, young skull; 18338, adult skull; 30196, ar- ticulated skull and mand; 30437, skull frags; 30492, cranial frags, scutes, and p/c frags; 30594, partial skull and mandible; 30604, skull. The extant Nile crocodile is a moderate- to large- sized crocodylid with generalized rostrum of mod- erate proportion, median nasal promontorium, typ- ically 14 maxillary and 15 mandibular teeth and with the anterior nuchal osteoderms well developed (Storrs, 2003). Ten relatively complete Crocodylus niloticus crania and/or mandibles were collected by National Museums of Kenya expeditions (Fig. 2). Tchernov (1976, 1986) had reported that the broad-snouted Rimasuchus lloydi (Fourteau, 1920) was the common crocodilian in the Lake Turkana Basin and that C. niloticus had a very sparse record in the Plio-Pleistocene. However, Storrs (2003) documented the presence of five crocodilian taxa, including both R. lloydi and C. niloticus at Loth- agam. Even so, it was somewhat surprising to find ten C. niloticus cranial specimens in the Kanapoi biota but only one of R. lloydi. Nevertheless, the apparent dominance of the extant Nile crocodile in the Kanapoi biota must be viewed with caution be- cause only the more complete crocodilian speci- mens were collected. } Crocodylus sp. indet. (Figure 3) KANAPOI MATERIAL. 30451, rostrum and symphysis frags. Associated rostrum and symphysis fragments document the presence of a second species of broad-nosed crocodile. This species can be distin- guished from broad-nosed crocodilians previously documented from the Pliocene of East Africa by its unusually long mandibular symphysis. R. lloydi and the extant species C. niloticus and C. cata- phractus Cuvier, 1824, all have short mandibular symphyses whereas that of KP 30451 is more than 10 cm long. This undetermined species is also char- acterized by the broad and straight anterior margin of the snout. No other fossil or extant crocodilian species is known to possess this combination of fea- tures (G. Storrs, personal communication). Rimasuchus Storrs, 2003 Rimasuchus lloydi (Fourteau, 1920) (Figure 4) KANAPOI MATERIAL. 30619, anterior ros- trum and mandible frags. Harris, Leakey, Ceding, and Winkler: Tetrapods ■ 43 Figure 2 KNM-KP 18338; Crocodylus niloticus cranium, dorsal view; scale = 5 cm 44 ■ CS 498, Harris and Leakey: Kanapoi Figure 3 KNM-KP 30451; }Crocodylus sp., cranium and mandible fragments; A = occlusal views of anterior rostrum and mandibular symphysis; B = dorsal view of anterior rostrum, ventral view of mandibular symphysis; scale = 5 cm Harris, Leakey, Cerling, and Winkler: Tetrapods ■ 45 Figure 4 KNM-KP 30619; Rimasuchus Uoydi, anterior rostrum and mandible fragments, A = dorsal view; B = ventral view; C = lateral view; scales = 10 cm 46 ■ CS 498, Harris and Leakey: Kanapoi The genus Rimasuchus was proposed by Storrs (2003) for a very large, brevirostrine crocodylid that had been previously referred to C. Iloydi. The species is characterized by premaxillae that are broader than long and that had a relatively straight premaxillae/maxillae palatal suture, a deep “ca- nine” occlusal notch, a slight dorsal maxillary boss, closely spaced anterior dentary teeth, broadly di- verging mandibular rami, and prominent dentary festoons. It is readily distinguished from C. niloti- cus by the lack of a nasal promontorium. An anterior portion of a very large rostrum with associated mandible was recovered from just below the Kanapoi Tuff. Measuring 326 mm across the widest preserved portion of the anterior rostrum, the specimen represents the largest crocodylian in- dividual yet recovered from the Lake Turkana Ba- sin. The most distinctive features of R. Iloydi, other than the absence of a nasal promontorium, are its relatively short but broad premaxillae and its short and deep mandibular symphysis. The conformation of the Kanapoi specimen leaves no doubt that it represents R. Iloydi. Euthecodon Fourtau, 1920 Euthecodon brumpti (Joleaud, 1920) KANAPOI MATERIAL. 18330, complete skull and mand; 18331, rostrum frag; 18332, articulated skull and mand frag; 30653, mand symphysis; 30176, edentulous mand; 30266, mand frags; 30407, mand. Euthecodon brumpti is a very large eusuchian in which the extremely elongate and narrow rostrum has deeply scalloped dental margins. The premax- illae and nasals are attenuated but the narial ridge is prominent. A moderate premaxillary/maxillary diastema is located behind four premaxillary teeth. The skull table is small and nearly square, the oc- ciput vertical, the mandibular symphysis long, and the teeth isodont and slender (Storrs, 2003). Half a dozen specimens of slender-snouted croc- odilians were recovered by National Museums of Kenya expeditions. These clearly belong to E. brumpti, a species distinguished by its slender snout and long slender curved teeth, rather than to the gavial reported by Storrs (2003) from the lower Nawata Formation at the nearby site of Lothagam. Order Struthioniformes Family Struthionidae Struthio Linnaeus Struthio sp. indet. KANAPOI MATERIAL. 29300, eggshell frags; 30221, eggshell frags; 30223, eggshell frags; 30262, eggshell frags; 30490, eggshell frags; 32522, eggshell frags; 36599, eggshell frags. Several occurrences of fossil ostrich eggshell frag- ments were noted on the surface of the fossiliferous strata at Kanapoi and several voucher specimens were collected by National Museums of Kenya ex- peditions. These shell fragments all had the stru- thious pore pattern characteristic of living ostriches (Sauer, 1972) and the pore basins are of similar size and shape to those reported from the upper Na- wata at Lothagam (Harris and Leakey, 2003a). The content of the fossil eggshell is more negative than those of extant ostriches from Kanapoi and the is more positive. It would appear that there was a greater proportion of C3 vegetation in the Pliocene ostrich diet compared with that of modern ostriches on the west side of Lake Turkana. Order Pelecaniformes Family Anhingidae Anhinga Brisson, 1760 Anhinga sp. indet. KANAPOI MATERIAL. 39325, Lt. humerus. This family is represented by the left humerus of a darter of comparable size to the extant African darter. Order Ciconiiformes Family Ciconiidae Mycteria Linnaeus Mycteria sp. indet. KANAPOI MATERIAL. 30231, Lt. dist tibia, prox Rt. tibiotarsus, Lt. radiale, and long bone frags. Storks were represented at Kanapoi by a single association of long bone fragments. Order Charadriiformes Family Anatidae Gen. and sp. indet. KANAPOI MATERIAL. 39326, proximal Lt. humerus. The Anatidae was represented at Kanapoi by a goose-sized left humerus fragment. Order Primates Family Galagidae Galago Geoffroy, 1796 The modern species range in size from the largest Galago crassicaudatus Geoffroy, 1812 (1,122- 1,750 g) to the smallest. Galago demidovii (Fisher, 1806) (69-81 g), and they inhabit a variety of hab- itats including primary and secondary lowland and montane forests, gallery forests. Acacia woodlands, savannahs, thorn scrub, and forest edge. cf. Galago sp. indet. (Figure 5) KANAPOI MATERIAL. 30260 Rt. mand frag. (M,). Harris, Leakey, Ceding, and Winkler: Tetrapods ■ 47 Figure 5 KNM-KP 30260; cf. Galago sp. indet., right mandible fragment (M2), occlusal view; scale = 5 mm Mandibular fragment, KP 30260, is approxi- mately the size of Galago demidovii, the smallest living galago. The specimen is too fragmentary to allow precise identification. Due to their small size, galagos are rare in the fossil record and little is known of their distribution or evolution in the Pli- ocene. Family Cercopithecidae Subfamily Colobinae Although rare elements in fossil assemblages, the Colobinae were a diverse group in the Pliocene, at which time they appear to have undergone a major radiation. Several species of large colobines are known from Pliocene and early Pleistocene deposits in the Turkana Basin, the Baringo Basin, and from Olduvai (Leakey, 1982, 1987). The postcranial morphology indicates differing degrees of arboreal- ity but all appear to have been more terrestrial than extant east African colobines. Fossil Colobinae in- clude at least three genera — Paracolobus Leakey, 1969, Rhinocolobus Leakey, 1982, and Cercopithe- coides Mollet, 1947 — and are known from several relatively complete skeletons. Smaller colobines were also diverse but are less well represented by fossils. The genus Kuseracolobus Frost, 2001 has been described from early Pliocene deposits at Ar- amis in the Awash Valley, Ethiopia, where colo- bines constitute an unusually high proportion (56%) of the cercopithecid assemblage. At Kana- poi, at least three species are represented, two mod- erately large but of indeterminate affinities, and the other an indeterminate species of Cercopithecoides. Such diversity at Kanapoi indicates that the colo- bine radiation began prior to 4.1 million years ago. Cercopithecoides Mollet, 1947 Cercopithecoides is one of the best-represented co- lobine genera in the East African Pliocene and early Pleistocene. Until recently, only two East African species were recognized, C. williamsi Mollet, 1947, and C. kimeui Leakey M. G., 1982, but additional taxa are now described from Lothagam (Leakey et ah, 2003) and at Leadu in Ethiopia (Erost and Del- son, 2002). Eeatures of the postcranial anatomy in- dicate that Cercopithecoides was probably the most terrestrial of the Pliocene colobines. cf. Cercopithecoides sp. indet. (Eigure 6; Tables 1, 2, 4) KANAPOI MATERIAL. 29255, mandible (Rt. P3-M3, Lt. P.-Mi); 31741, broken M' or M’; 32870, Lt. P^; 36967, Lt. M^; 37382, Rt. /C; 43120, P3 talonid. A handful of specimens represent a small species of colobine. The well-preserved mandible permits tentative taxonomic assignment to Cercopithecoi- des. The mandible probably represents a female: al- though the canines are lost, their alveoli are not large and the honing facet of the P. is short. The symphysis is broken toward the alveolar margin, the right body extends behind the M3, and the left body terminates behind Mj. The MjS are quite heavily worn, the M, and the premolars less so, and the M3 has relatively light wear. Slight damage to the body beneath the P3S on both sides may well be from carnivore chewing. The mandibular morphology is similar to that of Cercopithecoides with a relatively shallow and moderately thick mandibular corpus. However, the anterior face of the symphysis lacks a median men- tal foramen and is more sloping and less flattened than in other known species. Below the deep genio- glossal pit, the inferior torus is well developed. Al- though the canines are missing and the alveoli dam- aged, the dental arcade appears to be narrower be- tween the canine alveoli. The inferior margin of the mandibular body is clearly defined anteriorly where there is a distinct and deep digastric fossa that merges posteriorly with the submandibular fossa. The mylohyoid line is prominent, providing consid- erable strength to the body and defining the upper extent of the submandibular fossa. With the excep- tion of the less worn M3, the teeth are heavily worn, as is so often the case with Cercopithecoides. Colobinae gen. et sp. indet. A (Eigure 7; Tables 1-3) KANAPOI MATERIAL. 31736, skull frag and male associated lower teeth (I, and Lt. P4, partial Lt. and Rt. C, Rt. P3, and molar, and molar frag); 29307, Rt. juv mand (dP4, I,-P4 in crypts); 30408, Lt. mand (erupting P3 and P4, M^_2); 30496, Lt. /C; 32803, Rt. M,o^.2; 32525 male partial Lt. C/; 32821, Lt. M,; 36830, Lt. mandible (M^, crypt for M2). The specimens assigned to this taxon are frag- mentary, largely subadult, and mostly from the lower jaw. They represent a fairly large colobine that has relatively large high-crowned premolars (compared with the size of the molars). The un- erupted P3 of KP 30408 has a large protoconid and distinct metaconid as well as a large distal fovea. The remaining specimens are tentatively referred to this taxon but could equally well belong to the fol- lowing taxon. 48 ■ CS 498, Harris and Leakey: Kanapoi 3 cm Figure 6 KNM-KP 29255; cf. Cercopithecoides sp. indet., mandible (Lt. P3-M1, Rt. P3-M3), A = occlusal view; B = inferior view; C = right lateral view; D = anterior view; scale = 3 cm Colobinae gen. et sp. indet. B (Figure 8; Table 2) KANAPOI MATERIAL. 29308, male Lt. mand (erupting P3, roots C, P4-M]). This is a juvenile mandible with the P3 erupting. Although of similar size, the P3 differs significantly from that of Colobinae gen. et sp. indet. A. It has a large single cusp, the protoconid, and a small dis- tal fovea in contrast with the two cusps and large distal fovea of KP 30408. Table 1 Kanapoi Colobinae upper dentition measure- ments cf. Cercopithecoides sp. indet. M3 MD LL 36967 8 8.2 a Colobinae gen. et sp. indet. A MD LL 32525 7.3 12.9 Subfamily Cercopithecinae Tribe Papionini The Papionini are the dominant cercopithecids in most Pliocene and Pleistocene sites in East Africa. Farapapio Jones, 1937, was the common genus in the early Pliocene, but was later replaced by Ther- opithecus Geoffroy, 1843. At Aramis, in the Middle Awash, a new 4.4 million year old papionin, P//o- papio alemui Frost, 2001, has been recovered (Frost, 2001). Pliopapio Frost, 2001, is distin- guished from Farapapio by the presence of an an- teorbital drop, a distinct ophryonic groove, and a shorter and more rounded symphysial profile. The absence of an anteorbital drop is the sole diagnostic character separating Farapapio from Papio Muller, 1773. Genus Farapapio Jones, 1937 Farapapio is recorded from almost all East African Pliocene sites, and is generally believed to have giv- en rise to Papio, from which it can be differentiated only by its facial profile. A number of species have been described and differ largely on the basis of size — a criterion that is not easy to apply when Table 2 Kanapoi Colobinae lower dentition measurements Harris, Leakey, Cerling, and Winkler: Tetrapods ■ 49 od K VO oo OV ON CQ oo o\ lyd iri oo r\ VO ^ C3V ro iri VO vd O ro K VO A _DC '5 X oo >o VO vd V vd vd U ov K VO d d oj S rt PtH VO VO O 0x1 O 03 C OO o VO VO ro O VO VO rx- oo 0x1 ON ro O Ox] ro OJ (N oo ro d o o|- OO OO OO ON ov T 1 0x1 rx. ro o m O 0x1 0x1 VO (N 0x1 ro ro ro o U ro ro ro ro ro G . F Lt. P3; 30233, Lt. male mand (M3, roots /C-M2), Rt. mand frag (root /C); 30398, worn and broken Rt. M2 and mand frag; 30399, Rt. M3; 30434, Rt. dP; 30483, Lt. M3; 30531, squashed anterior female mand (broken Lt. and Rt. F-P4); 30532, Lt. mandible (broken P4-M2); 30535, Rt. M3; 29311, Lt. P; 30538, female max, mand, and skull frags: Rt. mandible (broken and Table 4 cf. Cercopithecoides sp. indet mandibular mea- surements 29255 Depth below P4/Mj junction 20.9 Depth below M2/M3 junction3 21.4 Max thickness below Mj 10.4 Max thickness below M3 10.9 Length tooth row P3-M3 39.8 2 cm Figure 7 KNM-KP 30408; Colobinae gen. et sp. indet. A, left mandible, occlusal view; scale = 2 cm 2 cm Figure 8 KNM-KP 293080; Colobinae gen. et sp. indet. B, left mandible, occlusal view; scale = 2 cm Harris, Leakey, Ceding, and Winkler: Tetrapods ■ 51 Figure 9 KNM-KP 29306; Parapapio ado, left male mandible (P3-M3), A = lateral view; B = medial view; C = occlusal view; scale = 3 cm cracked I1-M3), Lt. mandible Lt. Ij, M3, /M; Lt. maxilla (P-P^) broken Rt. C/, broken Lt. P\ M>, M3; 30539, Rt. M^; 32512, Lt. M3; 32520, female edentulous mand (roots Lt. Ij-Mj and Rt. I,), Rt. I2; 32527, Lt. dP^; 32534, Rt. M' or M2; 32535, Rt. M^; 32554, Rt. male C/; 32572, Lt. dP4; 32804, mesial frag Lt. M3; 32805, partial Rt. Mb 32806, Rt. M3; 32811, broken Rt. M3; 32816, Rt. I,; 32817, worn Rt. M3; 32819, Rt. max (P3-4), frag edentulous Lt. premax; 32869, Rt. M3, and Lt. /C; 32878, Rt. dP3; 32884, Lt. d/C; 36914, Lt. M2, partial /M; 36969, Rt. female /C, M frag, Lt. P"^; 37374, tooth frags — molar and premolar up- per and lower, dist phalanx; 37378, broken M, tooth frag; 37379, Lt. /I, talonid Lt. M3; 37380, broken Lt. I2; 43121, Rt. female /C; 43122 dP4. The Kanapoi Parapapio material is rather frag- mentary and consists mainly of isolated teeth. Two mandibular specimens are well preserved in addi- tion to the mandible KP 286 that was recovered by the earlier expeditions and initially attributed to Parapapio jonesi (Patterson, 1968). KP 29306, a male left mandibular body, includes the entire sym- physis extending to the right canine alveolus. The symphysial morphology is close to that of KP 286, although it is shallower and in profile less steeply inclined with a slight curvature. The body is not deep and is quite slender with well-preserved P3- M3. KP 30147 is a mandibular symphysis that has suffered some postmortem expansion, making it appear larger than KP 286. If allowance is made for the distortion, the symphysial morphology is similar to the other two specimens. All three man- dibles have a distinct mental foramen and may be distinguished from Pliopapio alemui by the steeply sloping symphysis and the angle of the incisor roots, which shows the incisors to have been pro- cumbent rather than vertically set. Two specimens, KP 30149 and KP 30538, have well-preserved up- per and lower juvenile dentitions. The M3 has an unusually large talonid and tal- onid basin, such that the hypoconulid is of almost equal size or in some cases (KP 30233) larger than the distal lingual cusp (entoconid). This feature, which occurs with a relatively high frequency at Kanapoi but is unusual in other species of Para- papio, may possibly be of taxonomic significance. Theropithecus Geoffroy, 1843 Theropithecus first appears in the Turkana Basin about 3.5 million years ago, although one older tooth from Allia Bay may possibly represent this genus and a partial tooth from Kanapoi is here as- signed to Theropithecus. At Pliocene localities younger than 3.5 million years, T. brumpti (Ar- ambourg, 1947) became the dominant cercopithe- coid until it was replaced by T. oswaldi (Andrews, 1916) at about 2.5 million years (Leakey, 1993). cf. Theropithecus sp. indet. KANAPOI MATERIAL. 32879, broken Rt. Miorl- Table 5 Kanapoi Parapapio ado upper teeth measurements 52 ■ CS 498, Harris and Leakey: Kanapoi ^ oo od \d A O oo A od od ro 0-) cd od cd P3 U rn o^ A A O oo ^ MO \d I K K MO od 0^ kJ o^ooa^o^oooo^M'^^>oa^■^o^ rOrOT-l'^^Loo-)ioiOOOOOOoaN O^O^OOOOO^^^^^Ir'^^Mr^l^ (Nc^lrorrjrOrorOirnrDrorrjr^ro This tooth is larger than any of the Parapapio first and second molars (MD length 11.6; BL distal breadth 8.0). It is incomplete, lacking most of the lateromesial cusp both mesial and lateral to the pre- served cusp tip. The deep valleys and high cusps indicate affinities with Theropithecus. As such, this is one of the earliest occurrences of this genus. Cercopithecidae gen. et sp. indet. (Figure 10) KANAPOI MATERIAL. 458, prox Lt. ulna, dist femur; 29292, partial Lt. femur shaft; 29303, dist Rt. humerus; 29309, Lt. astragalus; 30424, prox Rt. femur; 30431, prox Rt. radius; 30460, prox Lt. femur; 30529, prox Rt. femur; 32516, Lt. C/; 32562, phalanges, phalanx shaft, prox m/p, patella, rib frags; 32818, frags of calcaneum, femur head, long bone frags; 32880, caudal vertebra; 36601, proximal phalanx; 36831, prox Lt. ulna frag; 36834, distal Rt. humerus; 36837, Rt. astragalus. A complete, slender, high-crowned canine (KP 32516) represents a small monkey. Crown mea- surements are MD length, 3.9; LL width, 6.3; max crown height, 13.0. The height of the crown sug- gests it must be from a male. The size is too small for it to belong to the Kanapoi Parapapio. It may possibly be referable to cf. Cercopithecoides sp. in- det., the small colobine described above, but the slender nature of the canine is unlike other colobi- nes. For now it is left as Cercopithecidea gen. et sp. indet. Two fragments of distal humerus are morpholog- ically similar but very different in size. The medio- lateral width of the smaller, KP 29303, is 22.5 mm, in contrast with the larger KP 36834, which mea- sures 30.4 mm. Both are reasonably well preserved although, on both, there is some weathering of the medial epicondyle and along the lateral edge of the capitulum. The trochlear flange is relatively lightly developed and the olecranon fossa elongated later- omedially, quite deep but not perforated. Com- pared with the Parapapio lothagamensis humeri de- scribed from Lothagam (Leakey et ah, 2003), the medial epicondyle is less posteriorly directed, the trochlear flange less strongly developed, and the olecranon fossa more elongated lateromedially without any proximal extension to accommodate the ulna olecranon process in extension. The distal humerus of Cercopithecoides is closer in all these characters to those from Lothagam than to the two specimens from Kanapoi. Two partial ulnae were recovered, KP 458 and KP 36831. The former is well preserved, small, lightly built, and most probably colobine. Its short olecranon process is slightly deflected posterome- dially. The radial notch appears to show some in- dication of a double articular facet. KP 36831 is slightly larger, less well preserved, and squashed, but is morphologically similar to KP 458. The shaft of the proximal radius, KP 30431, ex- tends 30 mm below the radial tuberosity. The ar- Table 6 Parapapio ado lower dentition measurements Harris, Leakey, Cerling, and Winkler: Tetrapods >53 03 U T-H o 0\ ON m oo o K od ON ON oo K od od ON to rs| oo ON to r-H (N oi ro O ^ (N ^ od ON od od od ON oo ro o o ^ O T-H On od o ^ ON oo r4 m o ro rd (N to ON fO od T-H od rd ON ro O r4 ^ ro CO O od od o o ^ to NO ON A (N oo rd od fO rn ^ "A NO NO A to rn to to A to ^ NO NO Oj to to NO ON rd rd (N rn lo to to to to ^ rd fO (O NO to to oS OJ NONOf0'^i0t00to oooooooo rororotOtOrOroro rt oj 5.2 6.8 4.5 7.1 5.9 30149 R 3.6 4.3 6.9 4.4 7 6 32527 L 6.8 5.5 32884 L 2.9 4.3 43122 6.5 5.2 ticular surface of the head is almost circular, with a distinct depression for articulation with the hu- merus capitulum. It is tilted with the posteriomedial edge higher than the opposite side; the anterolateral portion of the collar surrounding this depression is widest. The head measures 15.5 mediolaterally by 14.5 anteroposteriorly. The radial tuberosity is well developed with a raised and rounded anterior bor- der and a less prominent crest marking the poste- rior margin. The interosseous crest commences about 10 mm below the most distal extent of the radial tuberosity. Three fragments of proximal femora, KP 30460, 30424, and 30529, and a proximal femur shaft, KP 29292, have been recovered. 30460 has lost the margins of the greater tuberosity, 30424 lacks the head, and 30529 has lost both the greater and less- er tuberosities. KP 458 is a distal epiphysis with approximately 25 mm of shaft. The neck of KP 30424 is relatively long compared with those of the Lothagam femora, and the articular surface of the slightly larger head projects onto the femoral neck. The greater tuberosity of KP 30460 is relatively large and would have extended proximal to the head. KP 30529 is significantly smaller than the other fragments. The proximal shaft KP 29292 has Table 8 Kanapoi Parapapio ado mandibular measure- ments 29306 Depth below P4/M1 junction 27.1 Depth below M2/M3 junction 24.7 Max thickness below M, 9 Max thickness below M3 9.3 Length tooth row P3-M3 35.4 a pronounced crest on the posterior face proximal to the commencement of the linea aspera. Two complete tali were recovered. The maxi- mum length of the smaller, KP 29309, is 23 mm; that of the larger, KP 36837, is 27 mm. KP 29309 has a carnivore tooth mark on its trochlear surface. Both specimens have relatively long necks. The affinities of the postcranial elements are un- clear. In general, they appear to be from similar sized individuals, but the size discrepancy between the two morphologically similar distal humeri sug- gests these elements may represent one sexually di- morphic species. DISCUSSION Sites close in age to Kanapoi that have yielded early cercopithecids include Aramis (4.4 Ma), Allia Bay (3.9 Ma), Laetoli (3.6 Ma), and the Sidi Hakoma (3.4-3.22 Ma), Denen Dora (3.22-3.18 Ma), and lower Kada Hadar (3.18-2.92 Ma) members of the Hadar Formation in the Awash Valley. Aramis is unusual in having a larger proportion of colobines to cercopithecines (56% colobines). Only two cer- copithecoid species are recognized from Aramis— Kuseracolobus aramisi Frost, 2001, and Pliopapio alemni. Neither is represented at Kanapoi. The Al- lia Bay cercopithecids are largely represented by isolated teeth that are difficult to assign taxonom- ically. Flowever, papionins far outnumber colobi- nes. At Laetoli, the common cercopithecid is Par- apapio ado, although cf. Paracolobus sp. is also rel- atively common. Unfortunately there are no well preserved P3S in the Laetoli assemblage that can be compared with the two differing large colobine P3S from Kanapoi. It is possible that one of the Kana- poi colobines is the Laetoli species. Present but un- common at Laetoli was a small indeterminate co- Harris, Leakey, Ceding, and Winkler: Tetrapods ■ 55 3 cm Figure 10 Cercopithecidae gen. et sp. indet., postcranial elements; distal right humerus: A,C = KNM-KP 29303, B, D = KNM-KP 36834; A, B = posterior view, C, D = anterior view; E = KNM-KP 458, proximal left ulna, F = KNM- KP 36831, proximal left ulna; G = KNM-KP 30431, proximal right radius, E, F, G = lateral view;. H, I = KNM-KP 30424, proximal right femur, H = anterior view, I = posterior view lobine and a large papionin cf. Papio sp. (Leakey and Delson, 1987). The cercopithecids from Fiadar include Theropithecus darti (Broom and Jenson, 1946) and Parapapio cf. P. jonesi, as well as the colobines Rhinocolobus turkanaensis Leakey M. Table 9 Kanapoi Deinotherium bozasi tooth measure- ments Accession No. 401 30152 Lt. M" ap 92.1 prot 92.4 met 90.8 Lt M3 ap 96.5 prot 99.0 met 90.1 Lt. M3 ap 87.5 prot 79.9 met 67.9 G., 1982, and a new species of Cercopithecoides (Frost and Delson, 2002). Family FJominidae Australopithecus Dart, 1925 Australopithecus anamensis Leakey et ah, 1995 KANAPOI MATERIAL. 271, distal Lt. humerus; 29281, holotype mandible and Lt. temporal; 29282, Lt. 29283, maxilla; 29284, Rt. P3 and Rt. /C germs; 29285, Rt. tibia lacking midshaft re- gion; 292H6, mandible fragments and associated lower teeth (Rt. I,, Lt. and Rt. L-M3); 29287, man- dible with teeth; 30498, Lt. and Rt. maxilla frag- ments and associated dentition; 30500, mandibular fragments and associated dentition (Rt. I2-P4, ^2-3); 30502, Rt. M3, partial Lt. /M, molar frags; 30503, proximal manual middle phalanx; 30505, broken molar germ; 30942, five molar frags; 31712, associated juvenile mandibular and dental fragments; 31713, Rt. mandible with tooth frag- 56 ■ CS 498, Harris and Leakey: Kanapoi ments; 31714, Lt. CIP4; 31715, Lt. M1/2 and two associated tooth frags; 31716, fragment and C/ frags; 31717, Lt. M^, Rt. M3, and Lt. M2 frags; 31718, Rt. mandible frag (M2_3); 31719, L; 31720, maxillary M fragment; 31721, Rt. M^ and M^ par- tial crowns; 31723, Rt. M^; 31724, Lt. capitate; 31726, Rt. P4; 31727, Rt. /C; 31728, Lt. M^; 31729, Rt. dP2; 31730, Lt. M2, Rt. P3; 31732, tooth frags; 34725, associated juvenile dentition and skull frags; 35838, Lt. M3; 35839, Lt. L, Rt. C/, and Lt. P^; 35840, Lt. M^ and upper tooth frags; 35841, M crown; 35842, Rt. 35844, M frag; 35845, M frag; 35847, Lt. M2; 35850, M/ frag; 35851, Lt. Mi^" frag; 35852, Lt. C/; 37522, Lt. /M; 37523, M frag; 37524, tooth frags. Australopithecus anamensis is distinguished from all other australopithecine species by the following features: a small external acoustic meatus present- ing a narrow ellipse in outline; long axes of man- dibular bodies and tooth rows nearly parallel and close together; mental region of mandible not strongly convex; long axis of symphysis slopes markedly posterioinferiorly; canines with very long robust roots, trigons of upper molars much wider than talons, distal humerus with thick cortex en- closing a small medullary cavity. It can be distin- guished from A. afarensis Johanson et ah, 1978, by the following: upper canine root and associated fa- cial skeleton less posteriorly inclined; lower molars tend to have more sloping buccal sides and upper molars more sloping lingual sides; tympanic plate horizontal without defining grooves. It can be dis- tinguished from Ardipithecus White et ah, 1994, by the following features: absolutely and relatively thicker enamel; upper canine buccal enamel thick- ened apically; molars more buccolingually expand- ed; first and second molars not markedly different in size; tympanic tube extends only to the medial end of the postglenoid process, rather than to the lateral edge or beyond it; lateral trochlear ridge of humerus weak (Leakey et ah, 1995). Australopithecus anamensis is the oldest reliably dated Australopithecus species. It was a habitual biped and is readily distinguishable from A. afar- ensis to which it may have been ancestral. A de- tailed account of the Kanapoi material, and of slightly younger material of A. anamensis from Al- lia Bay, was provided by Ward et al. (2001). Order Proboscidea Family Deinotheriidae Deinotheres are Neogene proboscideans that dif- fered from gomphotheriids and elephantids by re- taining only the lower tusk and by never developing horizontal tooth replacement for their low-crowned lophodont teeth. Adapted for browsing on forest vegetation, they were common elements of the early Miocene African assemblages but are encountered only in small numbers in Pliocene and Pleistocene assemblages. Deinotherium Kaup, 1829 Deinotherium bozasi Arambourg, 1934 (Table 9) KANAPOI MATERIAL. 388, enamel frags and postcranial elements; 393, atlas; 401, Lt. M^ and M3; 30152, Lt. M3; 30557, partial M. Deinotherium bozasi was a large African dein- othere with teeth of similar size to the European species Deinotherium giganteum Kaup, 1829. The skull rostrum was steeply downturned anteriorly (like that of Prodeinotherium hobleyi [Andrews, 1911]) but the external nares and rostral trough were narrower than in D. giganteum. As in P. hob- leyi but in contrast with D. giganteum, the preor- bital swelling is reduced and situated just in front of P3, the occiput is steeply inclined, and the nasal bones have a slight anterior median projection (Harris, 1978). Deinotheres are represented at Kanapoi by a cou- ple of isolated teeth and, perhaps, by ribs and ver- tebrae of a juvenile proboscidean that were asso- ciated with unerupted deinothere enamel frag- ments. The upper second and third molars are dis- tinctly smaller than the equivalent teeth on the somewhat younger D. bozasi skull from Koobi Fora (Harris, 1976a) and the lower third molar is smaller than the equivalent tooth in early Pleisto- cene deinotheres. Family Gomphotheriidae Gomphotheriids were the common proboscideans in African Miocene assemblages but only Anancus lingered into the early Pliocene. Anancus Aymard in Dhorlac, 1855 Anancus kenyensis (MacInnes, 1942) (Table 10) KANAPOI MATERIAL. 384, Lt. M2; 410, Lt. M2, Lt. and Rt. M3. Anancus kenyensis has been diagnosed by Cop- pens et al. (1978) as “a progressive species of An- ancus with four and one half to five or more cone pairs on M2 and five and one half or more on M3; the crown is complicated by the presence of enamel columns partially fused into the faces of the main cones.” Anancus kenyensis is the last gomphotheriid spe- cies to survive in sub-Saharan Africa. Three An- ancus teeth were recovered from Kanapoi by the Patterson expeditions of 1965 and 1966. Tassy (2003) recognized two different populations of A. kenyensis — a primitive or kenyensis morph with te- tralophodont second molars based on the holotype of A. kenyensis, from Kanam (MacInnes, 1942) and a derived or A. k. petrocchi morph with pen- talophodont second molars based on the sample from Sahabi (Libya) described by Petrocchi (1954). The A. kenyensis morph was recognized in samples collected from Kanam, Lukeino, Mpesida, and Harris, Leakey, Ceding, and Winkler: Tetrapods ■ 57 Table 10 Kanapoi Anancus kenyensis tooth measurements Accession No. 384 410 410 410 Tooth LM2 LM2 RM3 LM, No. plates 5 5 6 No. plates in wear 5 5 3 Length 156.25 146.6 220 Length wear surface 91.6 Width (greatest) 75.89 68.64 81 79 Height (middle plate) 67.54 60.9 64.09 Enamel thickness 5.04 Lothagam, whereas the A. k. petrocchi morph is represented at Sahabi (Libya) and at Aterir, and in the Chemeron Formation in the Baringo Basin, as well as at Lothagam. The second molars recovered from Kanapoi display the derived traits—accessory conules, fifth lophid, and strong anancoidy— -that are characteristic of the A. k. pettrochi morph (Tas- sy, 2003). Family Elephantidae The Family Elephantidae originated in the late Miocene of Africa. Stegotetrabelodon Pettrochi, 1941, and Frimelephas Maglio, 1970, are the dom- inant elephantids at Lothagam (Tassy, 2003). Ele- phas ekorensis Maglio, 1970, first appears in the Apak Member at Lothagam together with an early Loxodonta F. Cuvier, 1925, species. Elephas eko- rensis, and Eoxodonta adaurora appear to be the characteristic elephantids of the African early Pli- ocene. Elephas Linnaeus Elephas ekorensis Maglio, 1970 (Figure 11; Tables 11, 12) KANAPOI MATERIAL. 382, Rt. P^ frag; 392, P2; 395, M3 frag; 400, mand (P2); 409, Rt. and Lt. mand frags (M2); 411, Lt. M3 frag; 412, M^; 452, Rt. P3; 28442, Lt. and Rt. P3 frags; 30189, Lt. M^; 30197, skull (RP^-C Lt. P2-3); 30236, M^; 30404, Rt. and Lt. P^"^, skull frags; 30625, Mj frag and unerupted M2 frags; 30639, mand (Rt. M2, Lt. Mj_2); 32575, Lt. mandible (P3_4); 30198, Lt. mand (M,_2); 387, RM3; 26956, Rt. mand (MJ; 30169, Lt. juv mand (P2_3); 30170, Lt. juv mand (P3) and symphysis; 30173, Lt. mand frag (P3), frags un- erupted P4; 30175, mand (Lt. P3, d?4, Rt. P4); 30270, Lt. and Rt, and tusk frag eaten by hy- ena; 30471, Lt. mand (M2 and M3) and symphysis; 38975, Rt M3. Elephas ekorensis is an early species of Elephas that was diagnosed by Maglio (1973) as having molars with crown height 10-15% greater than width, M3 broad anteriorly, becoming very narrow posteriorly; greatest width at base of crown; enamel loops prominent; enamel only very weakly folded near the base, 3-4 mm thick. The enamel plates are well separated with lamellar frequency of 3. 8-4. 8. The plate formulae are M3 11/12, M2 ?/?, Ml 7/8 (Maglio, 1973). Elephas ekorensis and Eoxodonta adaurora are the two common proboscideans from Kanapoi. The teeth of E. ekorensis may be generally distinguished on the basis of their narrower width, greater num- ber of plates, greater lamellar frequency, and thin- ner and more convoluted enamel (Fig. 11). Unworn tooth plates have more apical digitations than those of Eoxodonta teeth. Mandibles of E. ekorensis are proportionately narrow; the ventral surface is hor- izontal but curves down in front of the symphysis to terminate in a beak, which curves abruptly an- teroventrally and is smaller and more gracile than that of L. adaurora. The upper second premolar sometimes has two plates (KP 30197) but the lower has three (KP 30150). The third premolars have six plates and are wider posteriorly. The fourth pre- molar and first molar have eight plates. Elephas ekorensis is one of the oldest recognized species of Elephas. It has been recovered from the Apak Member of the Nachukui Formation at Loth- agam, where it is preceded in the Upper Nawata Member by Elephas nawataensis Tassy, 2003, which Tassy (2003) interprets as intermediate between the earlier Frimelephas gomph other oides Maglio, 1970, and E. ekorensis. Loxodonta F. Cuvier, 1825 Loxodonta adaurora Maglio, 1970 (Figure 12; Tables 13, 14) KANAPOI MATERIAL. 381, Lt. and Rt. mand (MJ; 383, Lt. and Rt. M^; 383, Lt. M3 and frags; 385, holotype mand and skeleton; 389, M^ frags; 390, Rt. and Lt. M^ frags; 391, Mj frag and pc; 394, radius and ulna; 396, Rt M2; 403, mand (Lt. M2_3), pelvis, and femur; 406, Rt. mand frags (Mi_2); 407, RM3; 548, Lt. M3 frags; 28441, M^ frag; 30150, P2; M2„r3 talonid; 30188, Rt. maxilla (worn MA; 30191, max frag (Lt. M^); 30193, Lt. and Rt. mand (M3) and frags; 30204, skull (Rt. and Lt. M^), damaged posteriorly; 30269, mand (Rt. and Lt. M2_3); 30436, P2; 30596, mand Mp3 (needs preparation); 30616, mandible (Lt. and Rt. 58 ■ CS 498, Harris and Leakey: Kanapoi Figure 11 KNM-KP 30175; Elephas ekorensis mandible with Lt. P3, P4, Rt. P4, occlusal view; scale = 5 cm P3, P4); 30654, Rt. M3; 32542, mand (Rt. and Lt. M,_3). Loxodonta adaurora was diagnosed by Maglio (1970) as a Loxodonta species with low-crowned molars, height equal to or less than width; enamel not folded, 3-5 mm in thickness; very large ante- rior and posterior columns, partially fused into plates but free at their apices, forming prominent median loops with wear; plates thick and well sep- arated, lamellar frequency from 2.6 to 4.4. The plate formulae: M3 8-lOX/lO-l lx, M2 7-8x/6- 8x, Ml 7/6-7 dm4 5/5-6, dm3 5/5-6, dm2 3/3 (Maglio 1973). The holotype of Loxodonta adaurora, a skull and partial skeleton, was collected from Kanapoi in 1965 and described by Maglio in 1970 but un- fortunately was badly damaged during its subse- quent transportation back to Kenya. Some less complete specimens were collected by National Museums of Kenya expeditions during the 1990s. In general, L. adaurora teeth may be distinguished from those of E. ekorensis because they are wider, have fewer plates, and have thicker and less con- voluted enamel; unworn tooth plates have a pos- terior median pillar and fewer apical digitations. The horizontal ramus of the mandible tends to be bulbous because of the greater width of the L. adaurora teeth. The ventral surface of the man- dible curves downward anteriorly at the rear of the symphysis to terminate in a robust beak. The cusps of the second premolars tend to remain iso- lated rather than fused into transverse plates. Fourth premolars and first molars have seven plates. Loxodonta exoptata (Dietrich, 1941) (Figure 13) Kanapoi Material. 30611, Rt. mandible (M2). Loxodonta exoptata was originally recognized from the 3.5 Ma Laetolil Beds of Laetoli in Tan- zania (Beden 1987). The teeth are more hypsodont than those of L. adaurora and have more plates but thinner enamel. In early wear, the plates form a lox- odont pattern strongly reminiscent of the extant L. africana (Blumenbach, 1797). Molar fragments at- tributed to Loxodonta aff. L. exoptata have been recovered from the Apak Member at Lothagam (Tassy, 2003). A partial mandible with lower second molar from high in the succession is the sole representa- tive of Loxodonta exoptata from Kanapoi. It has eight plates, each with a posterior median pillar. Harris, Leakey, Cerling, and Winkler: Tetrapods ■ 59 Table 1 1 Kanapoi Elephas ekorensis upper teeth measurements Accession No. 30197 30197 382 30197 Tooth LP" RP" LP^ LP^ No. plates 5 + 6 No. plates in wear 2 6 Length 24.75 23.57 68.7 Length wear surface 16.75 62.3 Width (widest plate) 23.12 21.29 40.24 Height (middle plate) 37.74 Enamel thickness 1.7 Laminar frequency 9E 10 Accession No. 30197 30404 30404 30404 30404 Tooth RP^ LP' RP^ LP‘» RP^ No. plates 6 6 6 5 + 5 + No. plates in wear 6 6 6 1 1 Length 70.54 60.4 62.11 Length wear surface 57.76 59.45 59.33 8.42 9.59 Width (widest plate) 43.44 40.68 39.63 51.74 52.43 Height (middle plate) 42.58 42.67 Enamel thickness 1.73 1.84 2.14 2.39 Laminar frequency 10 9 9 6 6 Accession No. 412 387 411 30189 Tooth LM^ M?3 RM^ LM3 No. plates 7 + No. plates in wear 4 Length Length wear surface 122.87 Width (widest plate) 77.95 80.18E 77.69 85.42 Height (middle plate) 99.05 Enamel thickness 2.9 3.34 5.42 4.24 Laminar frequency 5E4.5 4 + 5.5 5 Accession No. EK 424 EK424 Tooth RM3 LM^ No. plates 10 + 12 No. plates in wear p 4 Length 269 + 280 Length wear surface 128.11 Width (widest plate) 91 95.46 Height (middle plate) 112 120.06 Enamel thickness 5.15 Laminar frequency 5 5 and thick enamel. The rear of tooth is wider than the front. Order Perissodactyla Family Rhinocerotidae The extant genera Ceratotherium Gray, 1867, and Diceros Gray, 1821, first appear in the late Mio- cene but at Lothagam are accompanied by the te- leoceratine rhino Brachypotherium Roger, 1904 (Harris and Leakey, 2003b). Thus far, only the ex- tant genera have been recovered from Kanapoi. Ceratotherium Gray, 1867 Ceratotherium praecox Hooijer and Patterson, 1972 (Figure 14; Table 15) KANAPOI MATERIAL. 30, Rt. nasal boss, oc- ciput frag, and skull frags; 32, incomplete rami (Lt. and Rt. P3-M3); 33, Lt. mand frag (M2); 36, skull (Lt. M^“3, Rt. P^-M^) holotype; 38, Rt. P“^ frag, tooth frags; 538, dist Lt m/t III; 30187, partial skull (P^“^); 30195, Lt. humerus; 30217, Lt. mand (P3); 30554, Lt. P,; 32556, Lt. dP/ frag; 32868, Lt. P4. Table 12 Kanapoi Elephas ekorensis lower teeth measurements Accession No. 400 30169 411 30169 30170 30173 30175 32575 60 ■ CS 498, Harris and Leakey: Kanapoi rn ^ o Ph (U ^ c ^ O c VO CL, VO VO o\ c -o (N o wd O c ^ c OV lo _2 _S _C ex CL, ' § 6 6 Z Z tJD bX) c c :2 (U OJ 173 CL t/, — c IS "O u CO G *c:> 5 (U jP d: tS + VO (N + T ro o K vd (N VO C'l Lp lyp r\ OV ro ^ pq o ov o VO >o o\ ^ rd vd VO (^1 Lo r\ + w OO ro VO Lp O u Z 2 o IT u d: b "Sh ^ S 4-' IH W IS 2 3 0 5 -5 oo ^ ^ c "2 •- OJ tC 2h ^ ^ ^ X X ov ov h- ro ^ w VO lo rd vd + ro ^ vd rd ON VO T P-1 ro f2 ^ ^ ^ X ■S) g pj Table 12 Continued Harris, Leakey, Ceding, and Winkler: Tetrapods ■ 61 ON OO ^ + ecj + O OO 22 ap 31 tr 25 p4 ap 22 tr 23.5 Ml ap 34 tr 33 M" ap 39.7 tr 41.1 ap 42.6 tr whose distribution was limited to the Western Rift (Harris and Leakey, 2003c). The teeth of Ny. pat- tersoni may be distinguished from those of Ny. syr- ticus and Ny. devauxi (Arambourg, 1968) by their larger size, more complex third molar talon(id)s, and proportionately smaller premolars. The third molars are shorter and have less complex talon(id)s than those formerly assigned to ''Nyanzachoerus'' jaegeri Coppens, 1971. The teeth display consider- able size variation, polarizing between larger male and smaller female specimens. The larger male teeth are only slightly smaller than those of Ny. aus- tralis Cooke and Hendey, 1992, from Langebaan- weg, but we think the variation seen in the Kanapoi sample reflects sexual dimorphism rather than the presence of two dentally similar species. The mandible of Nyanzachoerus pattersoni is ro- bustly constructed. The symphysis is relatively nar- row across the canine alveoli and moderately con- cave. The anterior border, which bears three pairs of well-developed and closely proximate incisors, is arched and projects in front of the canines. The concave superior symphysial surface is deeply ex- cavated in front of the posterior border. The rela- tively small but stout canines are set at an oblique rather vertical angle. The inferior surface is smooth and only slightly flattened toward the incisive al- veoli. The mandible constricts to its minimum width midway along the relatively short postcanine diastema. The posteroventral border merges smoothly with the inferior body of the corpus, which is short, robust, and becomes gradually deeper posteriorly. At Lothagam, Nyanzachoerus syrticus and the much smaller Ny. devauxi predominate in the Na- wata Formation. The more progressive Ny. patter- soni has not been recovered from strata earlier than the Kaiyumung Member although the dentally sim- 72 ■ CS 498, Harris and Leakey: Kanapoi Figure 17 KNM-KP 30186, Nyanzachoerus pattersoni, cranium; A = left lateral view; B = dorsal view; C = anterior view. Scale = 5 cm ilar but somewhat larger Ny. australis (Cooke and Hendey, 1992) occurs in the Upper Nawata and Apak Members. Notochoerus Broom, 1925 Notochoerus species are tetraconodontines that have wide and flat mandibular symphyses, small premolars, elongate molars, and very long and hyp- sodont third molars. The M3 is the longest of all tetraconodontines (van der Made, 1999). Notochoerus jaegeri (Coppens, 1971) (Figures 19, 20; Tables 24, 27, 28) KANAPOI MATERIAL. 203, C/ frag; 225, bro- ken Rt. M^ and partial Lt. M^; 209, partial M3; 210, frag Rt. mand (M3 broken); 211, Lt. M^; 226, damaged mand (broken Rt. M3 and P3); 234, Lt. M^, frag Rt. M^; 235, mand (Lt. M3, broken Rt. P4 and /C); 241, damaged mandible (Rt. P3); 242, male skull frags (zygomatic knob); 245, partial M3; 251, partial skull and mand; 252, upper female ca- nines; 253, Lt. M^; 254, Rt. M3 and tooth frags; 257, partial skull (Lt. and Rt. P^-M^); 265, M^ Harris, Leakey, Cerling, and Winkler: Tetrapods ■ 73 Figure 17 Continued frags; 267, mand; 269, juv mand (dP3^, MJ; 270, Lt. M3; 26944, Rt. JVI^ and tooth frags; 30178, large mand (Lt. /C, Rt. Ij.g, Lt. and Rt. P3-M3); 30180, Lt. mand (P3-M3); 30182, Lt. M^, Rt. M^; 30185, tusk frags; 30402, mand (M2_3); 30452, mand (Rt. and Lt. /C-M3); 30484, talon; 30550, mand (M2_3); 30617, male skull (Rt. P^- M3); 32528, M3_3; 32801, Lt. mand. frag (M3). Notochoerus jaegeri is a large progressive tetra- conodontine possessing three pairs of premolars, of which the third and fourth are proportionately smaller than Ny. syrticus or Ny. pattersoni. The third molar is longer and taller and has more pillars than those of either of the latter taxa. There is a tendency for the molar enamel to be folded. The cranium and mandible are larger and more elongate than in Nyanzachoerus species; strong sexual di- morphism is evident, with the zygomatic swellings more localized but more protuberant in males (after Harris and White, 1979). Cooke and Ewer (1972) assigned cranial and dental material of a large suid from Kanapoi and Lothagam to their new species Nyanzachoerus pli- catus, but this taxon proved to be a junior synonym of the species Nyanzachoerus jaegeri erected by Coppens in 1971 for suid dentitions from Chad. Known mainly from its teeth, this species was in- terpreted by Harris and White (1979) to be mor- phologically and phylogeneticaly intermediate be- tween, on one hand, nyanzachoeres represented by 74 ■ CS 498, Harris and Leakey: Kanapoi i! !■ Figure 18 KNM-KP 239, Nyanzachoerus pattersoni, mandible; A = right lateral view; B = occlusal view; scales — 5 cm Harris, Leakey, Ceding, and Winkler: Tetrapods ■ 75 Figure 19 KNM-KP 30617, Notochoerus jaegeri, cranium; A = dorsal view; B = occlusal view; scales = 5 cm Ny. kanamensis and Ny. pattersoni and, on the oth- er, the somewhat younger notochoeres represented by Notochoerus capensis Broom, 1925, from South Africa and Not. euilus (Hopwood, 1926), from eastern Africa. New mandibles recovered from Kanapoi by National Museums of Kenya expedi- tions have augmented our understanding of this species and clarified its phylogenetic relationships. In contrast with the relatively narrow mandibu- lar symphyses of Ny. pattersoni and Ny. syrticus, that of Notochoerus jaegeri resembles the mandib- ular symphysis of Notochoerus euilus in its breadth and flatness. The corpus of the Not. jaegeri man- dible is longer and less robustly constructed than that of either Ny. pattersoni or Ny. syrticus. The symphysis is distinctly spatulate and flattened at the widest point across the canine alveoli. The anterior border, which carries three pairs of rather small but well-spaced incisors is almost straight and thick- ened ventrally at the alveolus. It projects only slightly in front of the anterior edges of the canine alveoli. Behind the canines, the superior surface is less deeply excavated. The large heavy canines ex- tend laterally and anteriorly from their alveoli at a more horizontal angle, curving posteriorly in their distal portions. The inferior surface thins anteriorly 76 ■ CS 498, Harris and Leakey: Kanapoi Figure 20 Notochoerus jaegeri, mandibles; A = KNM-KP 30178, occlusal view; B = KNM-KP 30452, occlusal view scales = 5 cm Harris, Leakey, Cerling, and Winkler: Tetrapods ■ 77 toward the incisor alveoli and between this and the canines is shallowly excavated either side of the midline. The posteroventral border projects below the inferior surface of the corpus. Posterior to the canines, the mandible constricts to its minimum width midway along the rather long postcanine di- astema. The corpus is long, narrow, and relatively lightly built. The inferior surface is elevated be- tween the posterior edge of the symphysis and the angle of the ramus. The teeth of Not. jaegeri resemble a progressive version of Ny. kanamensis and Ny. australis teeth: the premolars have similar morphology but are proportionately smaller, the molars have similar morphology but the talon(id) of the third molar is more complex. However, the similarity of the ori- entation of the lower canines and the mandibular morphology to that of Not. euilus argues for in- corporation of this species in Notochoerus rather than Nyanzachoerus. Harris and White (1979) noted tetraconodontine upper third molars from Laetoli that they inter- preted as primitive for Not. euilus on account of their larger size and more massive appearance. Re- comparison of the Laetoli suids with those from Kanapoi suggests that the massive upper molars from Laetoli are better attributed to Not. jaegeri than to Not. euilus. Notochoerus jaegeri is also pre- sent in the Apak Member at Lothagam (Harris and Leakey, 2003c). Notochoerus euilus (Hopwood, 1926) Notochoerus euilus is a species of Notochoerus with a deep and narrow cranium bearing sexually dimorphic zygomatic knobs. The has two strong lingual talon pillars. The M3 typically has three pairs of talonid pillars plus single midline ter- minal pillar or series of pillars. Individual major lateral pillars of molars are moderately tall, widely separated from adjacent lateral pillars, and taper sharply from their bases (Harris and White, 1979). Table 23 Kanapoi Suidae cranial measurements Nyanzachoerus pattersoni 30186 M Notochoerus jaegeri 30617 Width canine flanges 136 110.3 Width canine alveoli 160 Nasal boss width 70 Bizygomatic width 400 Width postorbital processes 140 Max width occiput 136 Nuchal crest-foramen magnum 185 133 + Bicondylar width 76 98.5 Table 24 Kanapoi Suidae mandible measurements Nyanzachoerus mandibles N. pattersoni M N. pattersoni N. pattersoni N. pattersoni N. pattersoni 220 239 30161 30410 38978 Length symphysis 144 + 118 124 126 140+ Width at canines 75 77.7 110 + Width at I3 76 61 62 78 + Length P3-M3 153 138 149.6 157 145 Length P3_4 48.3 44.6 51.8 48.6 44 Min width diastema 73 55 52.8 60.2 63 Notochoerus mandibles N. euilus M N. euilus F N. jaegeri M N. jaegeri M N. jaegeri M ER3540 KP30184 KP226 KP30178 KP30452 Length symphysis 172 131 189 195 185 + Width at canines 164 103 + 178 150.5 148 Width at I3 85 120.4 88.6 82 Length P3-M3 159 173 180 163 Length P3_4 36 44 46.7 45.2 Min width diastema 63 68 78 85 74 78 ■ CS 498, Harris and Leakey: Kanapoi Table 25 Kanapoi Nyanzachoerus pattersoni upper tooth measurements 201R 205R 205L 222R 223L ap tr p3 ap 23.85 tr 19.13 p4 ap 19.97 tr 22.31 Ml ap tr 16.99 16.7 M2 ap 28.69 30.93 tr 24.27 M^ ap 49.71 49.99 tr 29.62 33.69 dP^ ap tr 12.78 12.34 dP^ ap 19.62 15.82 tr 14.47 18.01 239L 239R 244L 244R 260 F ap 18.8 tr 11.6 P ap 17.8 tr 8.4 P md a ap 25.8 22.2 tr 16.4 17 p' ap tr P2 ap 12 7.3 12.36 tr 7.1 7 7.65 p3 ap 20.5 21.9 tr 19.4 18.9 p4 ap 18.8 17.7 19.31 20.24 tr 21.6 20.6 24.49 25.93 Ml ap 18e 20e 20 22.52 tr 20.9 20.84 M2 ap 26.7 24 32.01 30.64 27.9 tr 25.3 25 24.68 25.6 20.4 M" ap 46.6 47.6 tr 34 33.6 264L 18566L 30159R 30162R P2 ap 10.03 tr 6.5 p3 ap 21.89 tr 19.17 18.48 p4 ap 20.22 18.3 17.47 tr 23.45 22.5 21.7 Ml ap 17.98 18.3e tr 19.5e M2 ap 28.51 tr 28.51 M3 ap 50.72 tr 31.24e Table 25 Continued 30177 30177L 30177R 30268L 30268L a ap 26.5 tr 15.4 pi ap tr p^ ap tr p3 ap 22.27 22.54 tr 21.53 22.56 p4 ap 18.51 18.91 tr 21.96 23.2 M> ap 21.08 tr 17.62 M2 ap tr M3 ap 51.89 51.5e tr 30268R 30413 30433L 30433R 30453R P2 ap 12.1 tr 7.9 p3 ap 22 22.3 23.7 tr 21.5 20.7 p4 ap 19 18.4 18.7 tr 23.7 24.3 23.2 M> ap 21.8 15.7e 15.7e tr 19.9 20.7 18.4e M2 ap 26e 27.4 tr 24 27.7 M3 ap 47.69 tr 31.4 28.9 30474 30615 32515 30474R p3 ap 22.1 19.9 tr 16.3 p4 ap tr 22.7 M' ap tr M2 ap 24.9 25.43 tr 24.8 24.91 M3 ap 48.4 48.2 49.07 tr 31.4 Notochoerus cf. Not. euilus (Figure 21; Tables 24, 29) KANAPOI MATERIAL. 30184, female mand with symphysis (Rt. P"^”M3, roots Lt. and Rt. I^, /C). One mandible, female from the size of its ca- nines, has a third molar that terminates in a com- plex of pillars rather than in a fourth pair of pillars, thereby resembling the third molar of Not. euilus. The incisors and premolars are small but not as reduced as those of Not. euilus from the Lokochot Member at Koobi Fora (Flarris, 1983b). At Loth- agam, Notochoerus euilis is the predominant suid Harris, Leakey, Ceding, and Winkler: Tetrapods ■ 79 A B Figure 21 KNM-KP 30184, Notochoerus cf. N. eiiilus mandible; A = right lateral view; B = occlusal view; scales = 5 cm in the Kaiyumung Member but there is a canine from the Apak Member that may also belong to that species. Family Giraffidae Giraffoids are rare elements of middle Miocene fau- nas in East Africa, being best represented by C/i- macoceras africanus Macinnes, 1936, at Maboko, Climacoceras gentry i Hamilton, 1978, at Fort Ter- nan, and Palaeotragus primaevus Churcher, 1970, from Fort Ternan and Ngorora (Hamilton, 1978). Palaeotragus germaini, first described from late Miocene localities in North Africa (Arambourg, 1959), has been reported from Fothagam (Church- er, 1979), as has a second smaller Palaeotragus, species (Harris, 2003a). Palaeotragines are not found at African Pliocene localities, their place be- ing taken by more derived sivatheriines {Sivather- ium Falconer and Cautley, 1832) and giraffines {Giraffa Brunnich, 1771), both probably being de- rived from Eurasia immigrant stock (Harris, 2003a). Giraffa Brunnich, 1771 Giraffa stillei (Dietrich, 1942) (Table 30) KANAPOI MATERIAL. 30151, Rt. M"-^. 30428, Rt. P-' and Mg 30635, Ft. Pg 30636, Rt. 30637, upper tooth frags. Giraffa stillei is a small species of Giraffa with teeth that are often smaller than those of G. Ca- melopardalis (Finnaeus, 1798) and G. jumae Lea- key, 1965, but always larger than those of G. pyg- 80 ■ CS 498, Harris and Leakey: Kanapoi Table 26 Kanapoi Nyanzachoerus pattersoni lower teeth measurements 202R 213L 215 219R 220R P2 ap 10.87 tr 7.55 P3 ap 24.56 28.31 tr 19.47 22.28 P4 ap 21.78 22.11 24.48 tr 19.5 22.75 M, ap tr M, ap 27.79 28.68 28.85 tr 21.69 22.78 M3 ap 52.88 49.77 56.61 tr 24.04 25.19 25.21 28.61 221R 228 228L 236L 239L I. ap 00 tr 12 I2 ap 12.3 tr 14.6 I3 ap 10.2 tr 10 /C ap 19.9 tr 13.2 Pi ap tr P2 ap 10.9 tr 6.6 P3 ap 27.52 tr 24.46 16.8e P4 ap 26.38 tr 20.45 Ml ap 21.79 19 tr 13.62 13.9 M, ap 32.16 31.41 23.7 tr 25.99 M3 ap 52.8 tr 23 239R 240L 240R 255L 256R 258L b ap 8.7 tr 11.7 L ap 10.9 tr 13.4 I3 ap 10.6 tr 8.7 /C ap 21.5 tr 13.4 Pi ap tr P2 ap 11 tr 6.7 P3 ap 22.8 22.74 tr 17 19.88 P4 ap tr Ml ap tr M, ap 27.27 26.46 26.73 27.41 tr 19.81 20.45 20.8 21.11 M3 ap 53.3 54.62 53.87 60.17 tr 22.9 24.05 23.78 + 25.32 Harris, Leakey, Cerling, and Winkler: Tetrapods ■ 81 Table 26 Continued 259L 260R 260R 262L 263L 264L P2 ap 11.26 tr 7.78 P3 ap 23.26 26.62 tr 19.78 20.69 P4 ap 20.61 23.71 tr 19.4 21.58 M, ap 19.87 19.21 19.5 tr 13.64 14.5 13.18 M, ap 25.98 33.73 26.19 27.64 tr 20.14 25.8 18.06 M3 ap 54.21 48.79 56.32 tr 22.53 28.86 266L 533R 534L 30160 30161 Ii ap 13.4 tr 15.6 I2 ap 12.4 tr 17.4 I3 ap tr /C ap 22.8 tr 15.3 Pi ap tr P2 ap 8.2 tr 7.5 P3 ap 27.3 24.7 tr 20.9 17.5 P4 ap 24.4 tr 21.3 Ml ap tr M, ap 24.64 26.4 tr 19.39 ll.lt M3 ap 56.48 50.6 49.2 tr 25.6 26.72 26.4 25.6 30161R 30168R 30177L 30177R 30177R 30179L b ap 13.4e tr 10.9 I2 ap 14.7 tr 13.8 I3 ap 8.6 tr 12.5 /C ap tr Pi ap tr P2 ap 9.08 9.85 tr 8.25 7.78 P3 ap 24.4 25.23 25.71 tr 17.5 21.36 19.51 P4 ap 20.7 tr 16.3 M, ap tr M, ap 24.3 tr 19.7 M3 ap 52.9e 48.43 tr 24.1 25.65 75.67e 82 ■ CS 498, Harris and Leakey: Kanapoi Table 26 Continued 30183L 30205R 30267L 30267R 30267L I, ap 9.1 tr 12e h ap 12e tr 13.6e P2 ap 9.97 8.7 tr 7.81 7.42 P3 ap 25.61 27.45 24.84 23.56 25.3 tr 19.75 21.03 18.27 18.31 18.2 P4 ap 22.47 22.5 21.78 21.22 22 tr 20.34 21.33 18.51 20.06 20 Ml ap 17.83 18 tr M, ap 26.09 28.1 28.5 tr 23.52E 21.31 21.5 M3 ap 54.34 52.16 54.06 54.8 tr 24.23 27.64 25.26 24.9 30267R 30403 30409 30410L 30410R 30456 /C ap 32.3 tr 22.9 P. ap tr P2 ap 10 tr 7.1 P3 ap 25.3 25.0 25.5 tr 18.5 17.5 18.7 P4 ap 22.2 23 23.5 22.0 tr 19.5 19 20.9e 20.5 Ml ap 19.2 20.8e tr 15.5 M, ap 28.5 26 26 tr 22 19.2 17.1 M3 ap 54.5 50.7e 50e 55.6e tr 24e 23.3 22.6 30458L 30458 30462 30475 30475L 30545 I3 ap 10.1 tr 8.5 /C ap tr Pi ap tr P2 ap 10.32 tr 7.12 P3 ap 22.91 26.6 tr 18.29 24.4 P4 ap 21.45 26.2e tr 19.64 Ml ap 18.06 tr 14.88 M, ap 27.4 28.9 29 33 30.24 tr 21.26 20.8 21.5e 26 19.87 M3 ap 54.43 tr 23.94 Harris, Leakey, Cerling, and Winkler: Tetrapods ■ 83 Table 26 Continued 30620 32539 32553 I3 ap tr /c ap tr Pi ap tr P2 ap tr P3 ap tr P4 ap 23.2 21.2 tr 20.3 19.1 Ml ap 19.5 18.1 tr 15.6 14.6e M2 ap 27.4 25 tr 21.8 19.7 M3 ap 58.7e 50 tr 24.3 maea Harris, 1976. The ossicones are uprightly in- serted like those of G. Camelopardalis but appre- ciably smaller than male specimens of the extant species and often lack well-developed terminal knobs; they are larger than specimens assigned to G. pygmaea but are not backwardly raked like those of G. jumae. Five isolated teeth attributed to G. stillei are smaller than those of extant giraffes except for one anterior lower premolar (KP 30365) that is about the same size as that of the extant species. Giraffa jumae Leakey, 1965 (Table 31) KANAPOI MATERIAL. 30450, Lt. mandible (P2-M3). Giraffa jumae is a large species of giraffe of sim- ilar size to the extant G. Camelopardalis. The sur- face of the frontal bone between the external rims ! of the orbits is nearly flat. The width of the skull roof between the orbits is greater than in G. ca- ! melopardalis. The longitudinal median section from the posterior edge of the nasals to the lateral ossicone is flat or slightly concave. The lateral os- sicones originate immediately above the orbit and project more posteriorly than in G. Camelopardalis. No secondary bone apposition is known to occur on the lateral ossicones. The median ossicone is poorly developed. The basilar process of the occip- ital is longer than the external width of the palatal area at M^. The ascending ramus of the mandible is wide and stout. The corpus is deep and long, the anterior portion of the corpus being inclined up- ward from the premolars to the symphysis and then downward in the incisive region. Only one mandible of G. jumae has been recov- ered from Kanapoi. The lower teeth are somewhat larger than those of extant giraffes and the anterior premolar {P2) is significantly larger. Giraffa sp. indet. (Table 32) KANAPOI MATERIAL. 98, distal metacarpal; 480, prox phalanx; 472, Lt. astragalus; 473, Rt. astragalus frag; 30446, prox metacarpal; 42091, Lt. radioulna; 42092, Lt. radioulna, Lt. metacarpal, Lt. cuneiform, Lt. semilunar; 42093, Lt. tibia and Rt. humerus. None of the postcranials were found in associa- tion with teeth or ossicones. All are somewhat smaller than those of extant giraffes and thus are more likely to represent Giraffa stillei rather than G. jumae. Sivatherium Falconer and Cautley, 1832 Sivatherium hendeyi Harris, 1976 Sivatherium hendeyi is of similar size and dental morphology as the Asian S. giganteum Falconer and Cautley, 1832, and the later African species S. maurusium Pomel, 1892, but the posterior ossico- nes are short, extending laterally and backward from the cranium, and are unornamented by knobs or flanges or palmate digitations. The metacarpals longer than in S. giganteum or in Pleistocene spec- imens of S. maurusium (after Harris, 1976b). Sivatherium cf. S. hendeyi (Table 33) KANAPOI MATERIAL. 135, Lt. fibula; 30227, upper and lower molar frags; 30449, prox Rt. os- sicone frag; 32551, Rt. M/ and Rt. M3 frags. The proximal right ossicone fragment is trian- 84 ■ CS 498, Harris and Leakey: Kanapoi Table 27 Kanapoi Notochoerus jaegeri upper teeth 211L 225L 225R 234L p2 ap 11.48 11.37 tr 7.15 7.42 p3 ap 23.01 22.49 21.12 tr 20.03E 19.07 19.43 p4 ap 15.86 19.05 18.38 tr 19.21 21.83 21.89 ap tr M" ap 25.56 23.95 tr 26.39 24.36 ap 47.22 49.37 66.89e tr 33.45 33.68 32.59e dP^ ap tr dP" ap tr 253R 257L 257R 26944 26944L P^ ap tr P3 ap 21.83 24.15 tr 19.42 18.39 P4 ap 18.76 16.39 tr 20.98 21.37 M' ap tr ap 30.86 29.79 tr 25.14 25.71 M" ap 71.78 63.82 63.77 71.2 72.31 tr 33.23 35.46 36 36.8 36.99 30617L 30617R P" ap tr P^ ap 23.0 tr 21.0 p4 ap 22.3e tr 20.8 Ml ap tr M2 ap 35.8 tr 25.9 27e M2 ap 74.0 74.2 tr 35e 36.9 gular in transverse section with convex anterome- dial and anterolateral surfaces and a flattened pos- terior surface. The ossicone has a basal sinus and is massive proximally but tapers above the base. The ossicone is somewhat bovine in appearance and appears to have extended outward and back- ward at the base but curves medially. The anterior surfaces are marked with deep longitudinal grooves. Compared with ossicones of S. mauru- sium, this specimen is smaller, more gracile, and lacks the lateral bosses. Compared with the horn cores of the bovin Simatherium Dietrich, 1941, the ossicone lacks the strong dorso-ventral compres- sion of the bovin horn core and tapers less gradu- ally distally. The giraffid tooth fragments assigned to Sivath- erium are much larger than the teeth in the G. ju- mae mandible. Harris (1976b) proposed the name S. hendeyi for the Langebaanweg sivatheres, which were charac- terized by ossicones that were shorter and simpler than those of S. maurusium from younger sites. Churcher (1978) did not use this name but thought the Langebaanweg sivatheres may represent an an- cestral African sivathere population. We now know that Pliocene sivathere metapodials associated with S. maurusium ossicones were originally of similar length to those of 5. hendeyi but underwent a dra- matic reduction in length when S. maurusium adopted a grazing diet at the end of the Pliocene (Harris and Cerling, 1998; Cerling et ah, in press). The Kanapoi ossicone is referred to S. hendeyi on the basis of its size and morphology. Family Bovidae The family Bovidae is sparsely represented at early Miocene localities in East and North Africa (Ham- ilton, 1973; Gentry, 1978) but bovids were the most numerous terrestrial mammals at the mid- Miocene site of Fort Ternan (Gentry, 1970) and they dominate younger vertebrate fossil assemblag- es from eastern Africa. Bovids occur in both Eur- asia and Africa during the early Miocene, and thereafter, migrations occurred between them re- peatedly (Vrba, 1985, 1995) but Gentry (1990) ar- gues for Africa as the origin of this family. Because the sole apomorphic character characterizing the family Bovidae is the presence of horn cores (Janis and Scott, 1987), it would be difficult to substan- tiate a hornless bovid ancestor with any degree of certainty. The Antilopini and Caprini, both of Eurasian or- igin, are first represented in Africa about 14 million years ago; the endemic Cephalophini and Neotra- gini make their appearance shortly thereafter. To- ward the end of the Miocene, Ovibovini and Bovini migrated into Africa from Eurasia while the endem- ic Tragelaphini, Hippotragini, Alcelaphini, and Ae- pycerotini are documented for the first time. The Reduncini, whose continent of origin is uncertain, also appear in the late Miocene whereas the Bose- laphini become extinct in Africa near the Mio-Pli- ocene boundary (Vrba, 1985). The Kanapoi bovid assemblage is dominated by tragelaphins, alcelaphins, and aepycerotins — sug- gesting a mixture of open and closed mesic to xeric habitats. Boselaphins, common elements in the Na- wata Formation at Lothagam and persisting into the Apak Member (Harris, 2003b), have yet to be recovered from Kanapoi. Harris, Leakey, Ceding, and Winkler: Tetrapods ■ 85 Table 28 Kanapoi Notochoerus jaegeri lower teeth mea- surements 210R 226 235R 235L P2 ap tr P3 ap tr 13.4 P4 ap 21 tr M, ap tr ap tr M3 ap 80.56 67.66 tr 27.89 27.06 241R 267R 251L 251R 252R Pa ap tr P3 ap 23.86 22.55 23.97 22.48 tr 16.71 13.74 16.84 16.78 P4 ap 18.1 19.89 tr 15.32 17.18 Ml ap 20.44 tr 13.81 M, ap tr M3 ap 64.55 61.18 tr 25 25.07 267L 269R 30178L 30178R Pa ap tr P3 ap 25.85 25.55 tr 17.92 17 P4 ap 20.88 tr 18.48 17.8e Ml ap 26.6 tr 14.05 M, ap 34.06 35.79 tr 23.55 24.65e M3 ap 81.97 77.28 80.87 tr 28.37 30.34 30.37 30180L 30182L 30402 30550 32258 Pa ap tr P3 ap 23.6 tr 16.4 P4 ap 19.7 tr 17.1 Ml ap tr 16 M, ap 31 32 35.6 tr 22e 20.2 + 22.3 23.6 M3 ap 70.2 76.6 82 78.9 tr 24.9 30.3 29 + 29.6 Table 28 Continued 32801 30180L 30182R 30452R 30452L Pa ap tr P3 ap 23.62 26.61 26.48 tr 16.49 17.64 P4 ap 19.47 20.68 20.46 tr 17.11 M. ap 19.4 tr 15 M, ap 31.78 35.51 29.18 tr 19.95 21.22 M3 ap 76e 71.99 71.65 71.51 tr 26e 25.04 23.8 24.7e Tribe Tragelaphini Tragelaphus Blainville, 1816 Tragelaphus species are medium to large tragela- phins with spiraled horn cores inserted close to- gether and having an anterior keel and sometimes a stronger posterolateral one; small- to medium- sized supraorbital pits, which are frequently long and narrow, and an occipital surface tending to have a flat top edge and straight sides (Gentry, 1985). The genus Tragelaphus is diverse, including the kudus, the nyalas, the sitatunga, and the more ubiquitous bushbuck. Greater and lesser kudus oc- cur today in the northern half of the Turkana Basin. Tragelaphus kyaloae Harris, 1991 (Table 34) KANAPOI MATERIAL. 68, occiput; 77, Lt. h/c frags; 78, Lt. h/c frag; 79, Rt. and Lt. h/c frags; 80, Lt. h^ frag; 81, Rt. h/c frags; 82, Rt. h/c frags; 84, dist Rt. h/c frag; 86, Lt. h/c frags; 87, dist Lt. h/c; 88, dist. Rt. h/c frag; 89, dist. Rt. h/c frag; 90, prox Table 29 Kanapoi Notochoerus euilus lower tooth mea- surements 30184R la md 11.7 bl 13.6 Pa ap tr P3 ap tr P4 ap 20.5 tr 16.7 Ml ap 21.0 tr 14.9 M, ap 32.4 tr 21.4 M3 ap 73.5 tr 25.8 86 ■ CS 498, Harris and Leakey: Kanapoi Table 30 Kanapoi Giraffa stillei tooth measurements KP 30635 KP 30636 KP 30637A KP 3063 7B KP 30428 30151 P/ ap 18.51 tr 16.43 M/ ap 25.37 21.22 + 23.16 prot 25.41 24.45 met 21.65 P2 ap 15.25 tr 10.61 P3 ap tr P4 ap tr Ml ap 23.74 prot 16.47 met 16.64 M, ap prot met M3 ap prot met Rt. h/c frag; 91, prox Lt. h/c frag; 92, frontlet and prox Rt. h/c; 29258, prox Rt. h/c frag; 29260, prox Rt. h/c; 29261, prox Rt. h/c; 29268, prox Rt. h/c; 29269, dist Rt. h/c; 29272, prox Lt. h/c; 29278, dist Rt. h/c; 30156, h/cs and frontlet; 30158, cal- varia and h/cs; 30445, prox Rt. h/c; 30448, h/c frags; 30486, dist. Rt. h/c; 30628, dist Lt. h/c frag; 30629, Rt. h/c frag; 30634, prox Lt. h/c; 32519, prox Rt. h/c frag; 32559, dist. Rt. h/c frag; 29266, dist Lt. h/c. Tragelaphus kyaloae is a medium-sized tragela- phin with horn cores inserted close together, at a low inclination, that diverge rapidly from their base but converge distally, and that spiral 180° anti- Table 3 1 Kanapoi Giraffa jumae tooth measurements KP 30450 Pz ap 20.62 tr 13.62 + P3 ap 25.03 tr 19.83 P4 ap 26.15 tr 22.17 Ml ap 28.81 prot met M2 ap 29.76 prot 22.52 met 22.62 M3 ap 46.42 prot 23.89 met 21.56 clockwise in the right horn core. Proximally, there is a strong posterolateral keel and a fainter antero- lateral keel; distally, these become anterolateral and posteromedial, respectively. The shape of the horn cores is reminiscent of the extant sitatunga (T. spek- ei Sclater, 1863); they are less helically coiled than in T. strepsiceros (Pallas, 1766), T. imberbis Blyth, 1869, or T. gaudreyi (Thomas, 1884) but converge closer distally than in T. nakuae Arambourg, 1941, or the kudu-like species. The cranial vault bears a faint but distinct transverse bar immediately in front of the nuchal crest. The paroccipital processes are short but stout, and the posterior tuberosities of the basioccipital are much wider than the ante- rior ones. This is by far the most abundant bovid in the Kanapoi biota. The holotype of T. kyaloae is from the lower Lokochot Member at Kosia (Harris, 1991b) and it is the common tragelaphin in the Lo- kochot and Moiti Members of the Koobi Fora For- mation (Harris, 2003b). The oldest documented oc- currence of this taxon is a single specimen from, the Upper Nawata at Lothagam but it is also repre- sented by a dozen horn cores from the Apak and Kaiyumung Members of the Nachukui Formation at Lothagam. The horn cores are similar in cross- section to those of T. nakuae, the common Pliocene tragelaphin of the Lake Turkana Basin, but have more pronounced torsion. Tragelaphini gen. indet. (Table 35) KANAPOI MATERIAL. 32514, Lt. maxilla (P^- M^); 66, Rt. mandible (M2_3); 67, Lt. mandible frag (Ml); 76, Lt. Mb 109, Lt. Mb 29273, Lt. and Rt. Harris, Leakey, Cerling, and Winkler: Tetrapods ■ 87 Table 32 Kanapoi Giraffa sp. indet. postcranial measurements Specimen Length Prox ap Prox tr Dist ap Dist tr Epic tr 42093 Humerus 119.5 110.42 125.58 42092 Radius 70.43 115.14 66.48 106.77 42092 Prox ulna 63.31 98 Metacarpal 61.8 92.15 30446 Metacarpal 37.43 65.01 42092 Metacarpal 66.7 97.7 57.63 86.53 42093 Tibia 76.59 102.95 480 Proximal phalanx 66.98 33.2 33.15 22.48 27.86 Max ap Max tr Max dv 42092 Cuneiform 69.5 35.0 60.5 42092 Semilunar 61.4 43.5 48.0 Mi-2; 30395, Lt. Mi-2; 30396, Lt. M^; 30421, Rt. M„ Lt. M„ Rt. M3; 30441, Lt. M'; 32545, Lt. mandible frags (P2-M1); 32570, Lt. mandible frags, P3-M1; 32573, M/, Lt. Mi, and tooth frags; 32574, Lt. Mj; 32829, upper and lower molar frags; 32881, Rt. P3; 36861, Rt. mandible (P3_4) and bone frags. A number of isolated tragelaphin dentitions were recovered. Most are of similar size and presumably represent Tragelaphus kyaloae but the smaller KP 32514 indicates that at least two tragelaphin spe- cies were present at Kanapoi. Tribe Bovini Simatherium Dietrich, 1941 Only one extant bovin species, Syncerus caffer (Sparrman, 1779), is endemic to sub-Saharan Af- rica but three other genev2i~Ugandax Cooke & Corydon, 1970, Simatherium, and Pelorovis Reck, 1928 — are represented in the fossil record. Simath- erium differs from Ugandax by being larger and having more widely separated horn cores that are more divergent basally and more curved anteriorly. The more derived Pelorovis, is unknown before the late Pliocene. Simatherium kohllarseni Dietrich 1942 was originally described from Laetoli in Tan- zania whereas the somewhat older Simatherium de- missum Gentry, 1980, was first described from Lan- gebaanweg in South Africa. Simatherium demissum Gentry, 1980 Simatherium cf. S. demissum (Figures 22, 23; Tables 36, 37) KANAPOI MATERIAL. 96, Rt. mandible (P^- M3) and assoc postcranial; 29265, skull frags, Lt. Table 33 Kanapoi Sivatherium fibula measurements Length Prox tr Dv 135 55.79 31.42 33.85 maxilla (P^~M0, Rt. maxilla (P'^, M^"^); 30612, broken skull with Rt. and Lt. h/cs; 32560, Lt. M' frag. This Kanapoi bovin is represented by a very bat- tered calvaria with portions of both horn cores and by a partial upper dentition, both collected by Na- tional Museums of Kenya parties, and by a man- dible and associated postcranial elements that were collected by the Harvard University expedition. The horn cores are strongly dorso-ventrally com- pressed. They extend outward and backward from their bases but ascend upward toward their tips. The horn cores taper gently in their proximal por- tion but more rapidly toward their tip. The surface of the horn cores is not well enough preserved to confirm the presence or absence of a lateral keel but are otherwise ornamented by strong longitudi- nal ridges and furrows. Although the bone is very incompletely preserved, the braincase evidently ex- tended for some distance behind the horn cores and was bordered by well-developed temporal fossae. There was a strong interfrontal crest. Tribe Hippotragini There are three extant genera of hippotragins — Hippotragus Sundevall, 1846 (roan and sable an- telopes), Oryx Blainville, 1816, and Addax Rafin- esque, 1815, and several extinct genera. Vrba (1987) suggested that, early in their history, the hip- potragins diverged into two major subclades — one exemplified by Hippotragus, with uprightly insert- ed and mediolaterally compressed horn cores, and the other by the less water-dependent Oryx and Ad- dax, with horn cores that are less mediolaterally compressed but bent more strongly backwards. Hippotragini gen. indet. (Figure 24; Table 38) KANAPOI MATERIAL. 483, B-D, Lt. Lt. M^, Rt. M-; 29274, Rt. mandible (M,_3); 30631, proximal Rt. h/c; 32526, Lt. ML This taxon is represented by the proximal por- tion of a large right horn core that is mediolaterally 88 ■ CS 498, Harris and Leakey: Kanapoi i Figure 22 KNM-KP 30612, Simatherium cf. S. demissum, calvaria, dorsal view; scale = 5 cm Harris, Leakey, Cerling, and Winkler: Tetrapods ■ 89 C Figure 23 Simatherium cf. S. demissum, dentitions; A = KNM-KP 29265, left maxilla, dorsal view; B = KNM-KP 96, right mandible (P3-M3), lateral view; C == KNM-KP 96, occlusal view; scales = 5 cm compressed with a flattened lateral surface. The horn core curves gently backward from its base with a slight lyrate flexure in anterior view. There is faint transverse annulation on the anterior sur- face. At its base, the horn core measures 46.1 mm anteroposteriorly; the transverse measurement is 33.9 mm. A few hippotragin teeth were recovered although none were associated with the sole recognized hip- potragin horn core. Tribe Reduncini Reduncins are small to large antelopes with hyp- sodont teeth for grazing and today characteristical- ly occur in grasslands near permanent water. There are two extant genera — Kobus Smith, 1840 (water- 90 ■ CS 498, Harris and Leakey: Kanapoi Figure 24 KNM-KP 30361, Hippotragini gen. indet., proximal right horn core; A = lateral view; B = medial view; scales = 5 cm Harris, Leakey, Ceding, and Winkler: Tetrapods ■ 91 Table 34 Kanapoi Tragelaphus kyaloae horn core measurements prox ap (Rt) prox tr (Rt) It (Rt) prox ap (Lt) prox tr (Lt) It (Lt) 77 38 53.12 + 78 38 50 79 41 50 80 33 44 81 31 + 86 33 47 90 37 45 91 43 56 92 38 46 29258 38 49 29260 40 53 29261 37 48 29268 39 51 29272 38 46 30156 38 50 38 49 30158 42 54 41 55 30445 38 53 30634 34 52 bucks, kobs, etc.) and Redunca Smith, 1827 (reed- were originally described from Langebaanweg bucks). The earliest recognized species of Kobus are (Gentry, 1980) and the genus was subsequently rec- from Lukeino and Lothaga m. ognized at Lothagam (Harris, 2003b) and else- where. A cladistic analysis of fossil and living Al- Kobus Smith, 1840 celaphini by Vrba (1997) suggested that two alce- nhij < Qn iTTnpf laphin subtribes diverged during or before the Mio- l\L/LyirlO olJ. Pliocene transition — the Alcelaphina {Damalacra KANAPOI MATERIAL. 461, Rt. h/c frags; neanica Gentry, 1980, and Beatragus) and the large 29267, proximal Rt. h/c. clade Damalascina (which includes the Damaliscus KP 29267 is a small but stout horn core that curves backward and slightly outward from its base. The horn core tapers gently upward from its base and the lateral surface is slightly flattened. At its base, KP 29267 measures 35.5 mm anteropos- teriorly and 28.2 mm transversely. Corresponding measurements for KP 461 are 34.9 and 28.8 mm. The two horn cores are smaller than those of K. presigmoidalis Harris, 2003b, from Lothagam. Reduncini gen. indet. (Table 39) KANAPOI MATERIAL. 75, Rt. P3; 463, Lt. mandible (MJ; 29275, Rt. maxilla (P^'-M^); 29276, Rt. mandible (P2-M1); 29279, Rt. mandible (P2, M, M, 30626, Lt. mandible (P4-M1); 27450, Lt. A few reduncin teeth were recovered but none were associated with the Kobus horn core. Tribe Alcelaphini Alcelaphins are represented by four extant genera: Damaliscus Sclater and Thomas, 1894, Beatragus Heller, 1912, Alcelaphus de Blainville, 1816, and Connochaetes Lichtenstein, 1814, plus a number of extinct genera. All the extant genera are specialist bulk grazers that are capable of going without wa- ter for varying intervals of time. Damalacra species and Parmularius Hopwood, 1934 lineages). Both subtribes appear to be represented at Kanapoi. Damalacra Gentry, 1980 Damalacra neanica Gentry, 1980 Horn cores of Damalacra neanica are without com- pression or are slightly compressed anteroposteri- orly, have no flattened lateral surface, taper fairly sharply from base to tip, show much increased di- vergence distally, have either slight forward or slight backward curvature in profile, and are in- serted behind or above the back of the orbits. The boundary between the pedicel top and the base of the horn core is higher on the medial than on the lateral side of the horn cores. Female horn cores are smaller than those of males, as in extant alce- laphins (after Gentry, 1980). Damalacra cf. D. neanica (Figure 25) KANAPOI MATERIAL. 71, frontlet with Rt. and Lt. h/cs. The sole specimen comprises a frontlet with proximal horn cores that have long pedicels hol- lowed out by very large basal sinuses. The horn cores diverge outward and taper rapidly upward from their bases. They are ornamented by strong longitudinal ridges and furrows. The face was evi- 92 ■ CS 498, Harris and Leakey: Kanapoi Figure 25 KNM-KP 71, Damalacra cf. D. neanica frontlet with proximal horn cores; A = anterior view lateral view; scales = 5 cm Harris, Leakey, Ceding, and Winkler: Tetrapods ■ 93 Table 35 Kanapoi Tragelaphini dentition measurements 109 29273L 29273R 30395 32514 p4 ap 10.3 tr 14.4 Ml ap 23.0 21.2 21.5 18.9 13.8 tr 16.9 15.9 15.4 23.0 14.2 ap 21.8 22.8 27.8 18.1 tr 16.6 17.0 24.0 17.8 ap 21.7 tr 16.2 30396 30441 M‘ ap 19.7 tr 16.5 ap tr ap 26.1 tr 14.4 66 67 32574 32881 32570 P3 ap 11.9 12.9 tr 7.4 8.0 P4 ap 14.4 tr 8.9 M, ap 17.6 15.9 tr 10.2 10.8 M, ap 19.0 18.9 tr 13.0 M3 ap 29.9 tr 12.7 32545 32573 30421 36861 P2 ap 10.2 tr 5.7 P3 ap 10.6 tr 6.0 P4 ap 16.5 11.0 tr 8.3 5.8 M, ap 16.0 tr 10.4 M, ap 20.9 18.9 20.0 tr 12.5 12.3 12.2 M3 ap 26.1 tr 12.0 dently strongly angled on the braincase. From the width of the preserved portion of the braincase at the base of the horn cores, this was evidently a much larger animal than Damalacra species A or B from Lothagam (Harris, 2003b). The base of the right horn core measures 36 mm (anteroposterior- ly) by 33 mm (transversely); the left horn core mea- sures 24 X 33 mm. This specimen is closer to D. neanica than to D. acalla Gentry, 1980, by virtue of the uncompressed and widely divergent horn cores, but the horn cores are smaller than male representatives of this species from Langebaanweg. Table 36 Kanapoi Simatherium cf. S. demissum horn core measurements 30612 Lt h/c ap 86 tr 65 Rt. h/c ap tr It 290 + Width between h/cs 160 Damalacra sp. A (Figure 26; Table 40) KANAPOI MATERIAL. 64, proximal Lt. h/c; 26560, proximal Lt. h/c; 29270, Rt. h/c; 30630, distal h/c frags; 30447, proximal Lt. h/c. This species had a long, slender horn core very reminiscent of specimens attributed to Damalacra sp. A from Lothagam (Harris, 2003b). The horn core was mediolaterally compressed with a flat- tened lateral surface. It tapers gently upward from the base. It is teardrop-shaped in transverse section with a carinate posterior border. The horn cores diverge outward and slightly backward from the Table 37 Kanapoi measurements Simatherium cf. S. demissum teeth 29265L 29265R pi ap 19.18 tr 17.47 ap 18.52 tr 22.37 p4 ap 16.72 16.88 tr 22.3 M‘ ap 23.1 tr 24.73 M2 ap 27.86 27.03 tr 27.43 M3 ap 29.98 30.36 tr 15.77 29.43 96 Pi ap 14.64 tr 9.31 P3 ap 20.19 tr 11.61 P4 ap 21.66 tr 13.24 M, ap 21.75 tr 17.18 M, ap 27.97 tr 17.71 M3 ap 37.98 tr 17.09 94 ■ CS 498, Harris and Leakey: Kanapoi Figure 26 KNM-KP 29270, Damalacra sp. A, right horn core; A = lateral view; B = medial view; scale - 5 cm Harris, Leakey, Ceding, and Winkler: Tetrapods ■ 95 Table 38 Kaeapoi Hippotragini teeth measurements 32526 483 B-D p4 ap 11.5 tr 11.2 Ml ap 21.1 19 tr 23.8 13.7 M" ap 20.2 tr 15.5 ap 25.1 tr 15.9 29274 M, ap 24.91 tr 14.41 M3 ap 30.93 tr 14.1 base but recurve to upright in their distal portion. There are faint transverse annulations on the an- terior surface and strong longitudinal ridges and furrows on the lateral and medial surfaces. Damalacra acalla Gentry, 1980 The cranium, of Damalacra acalla is of similar size and proportions to that of D. neanica. The horn Table 39 Kaeapoi Reduecini tooth measurements 29275 p2 ap tr ap tr p4 ap tr Ml ap 13.8 tr 17.6 M2 ap 18.1 tr 21.8 M2 ap 22.1 tr 75 463 29275 29276 29279 30626 P2 ap 8.0 7.7 tr 6.9 4.6 P3 ap 10.4 11.7 tr 6.9 9.0 P4 ap 14.9 14.1 tr 9.2 7.8 Ml ap 16.1 15.8 16.0 17.3 tr 9.6 10.7 11.8 13.0 M, ap 19.6 tr 14.3 M3 ap 31.1 tr 14.8 Table 40 Kanapoi Damalacra sp. A horn core measure- ments 26560 64A 29270 30447 Rt h/c ap 23.9 tr 17.6 It 160 + Lt h/c ap 23.8 22.3 21.1 tr 18.3 17.4 18.1 It cores are mediolaterally compressed with localized (usually medial) swelling at their bases, sometimes with flattened lateral surface along part of their length, with more definite backward curvature than in D. neanica and less marked distal divergence, inserted close behind orbits, and with a more nearly horizontal boundary between the top of the pedicel and the horn core proper. Other differences from D. neanica are that the braincase roof is less steeply inclined, more curved in profile, and with a parietal hump that is as well developed as in living Dam- aliscus; the mastoid exposure is large but less ex- panded especially medioventrally, and the auditory bullae are slightly larger and much more inflated. Female horn cores are smaller than those of the males. The teeth are of similar morphology to D. neanica (after Gentry, 1980). Damalacra cf. D. acalla (Figure 27; Table 41) KANAPOI MATERIAL. 30157, calvaria with Rt. and Lt. h/cs; 32557, proximal Rt. h/c frag; 32876, distal h/c frag. This alcelaphin is best represented by KP 30157, a calvaria with proximal horn cores that differs from KP 71 by its smaller size and less strongly divergent but more strongly compressed horn cores. The horn cores are inserted close together above the orbits on long pedicels. At their base, they are oval in transverse section. They rise upward but diverge gently outward about halfway up and become more strongly mediolaterally compressed in their distal portion. There is slight anticlockwise torsion in the right horn core when viewed from above. The face is strongly angled on the braincase. There is a large Table 41 Kanapoi Damalacra cf. D. acalla horn core measurements 30157 32557 Rt h/c ap 33.1 33.9 tr 25.7 26 It Lt h/c ap 34.2 tr 24.7 It 96 ■ CS 498, Harris and Leakey; Kanapoi Figure 27 KNM-KP 30157, Damalacra cf. D. acalla, calvaria with proximal horn cores; A = anterior view; B - left lateral view; C = dorsal view; D = right lateral view; scales = 5 cm Harris, Leakey, Cerling, and Winkler: Tetrapods ■ 97 Table 42 Kanapoi Alcelaphini tooth measurements 493 110 111 83 73 31739 M' ap 19.8 21.5 18.9 tr 11.9 15.6 12.5 M" ap 26.3 tr 22.1 ap 30.7 26.5 tr 17.1 17.4 dP4 ap 22.9 tr 8.2 387 462 44128 ap 18.5 tr 13.3 M2 ap 24.8 tr 13.5 ap 27.3 27.7 24 tr 19.3 17 16.8 74 108 no 30406 31733 31033 462 P2 ap 5.9 tr 4.5 P3 ap 11.1 tr 6.6 P4 ap 14.9 tr 8.2 M. ap 20.6 17.4 tr 9.8 9.5 M, ap 25.7 tr 10.7 9.8 M3 ap 26.3 27.1 tr 11.9 9.9 9.5 10.8 9.7 10.2 ''Parmularius-\ike'' bulge on the top of the brain- case, midway between the horn core bases and the nuchal crest. The shape of the calvaria and horn cores are close to those of Damalacra acalla from Lange- baanweg but the horn cores would group in size with the smallest D. acalla specimens and the cal- varia has a much more pronounced parietal boss. Alcelaphini gen. indet. (Table 42) KANAPOI MATERIAL. 73, A = Lt. M', B = Rt. M^; 74, Lt. M3 frag; 83, Rt. M^; 108, Rt. M3; no, Rt. M3 frag and Rt. M/; 111, Lt. M^; 287, Lt. Mh 462, Rt. M^, Lt. M^, Rt. M3 frag, tooth and bone frags; 493, Rt. dP4; 30406, Lt. mandible frags (M,_3); 31733, Lt. mandible (P2-M3); 31739, Lt. M>; 311033, Rt. Mg 44128, Lt. M^ and tooth frags. A few alcelaphin teeth were recovered but none were associated with alcelaphin horn cores. Tribe Aepycerotini This tribe is today represented by the single but variable species Aepyceros melampus (Lichtenstein, 1812). Aepycerotins may be a sister group of the alcelaphins (Vrba, 1984) but the two groups have different habitat preferences and dental morpholo- gy. Lyrate horned impalas were the dominant bovid in the Nawata Formation of Lothagam {Aepyceros premelampus Flarris, 2003b). A smaller and less ly- rate-horned species predominated in the middle Pli- ocene strata of the Shungura, Nachukui, and Koobi Fora Formations but the extant species had become dominant by the late Pliocene. Aepyceros Sundevall, 1847 Aepyceros sp. indet. (Figures 28 and 29; Tables 43 and 44) KANAPOI MATERIAL. 70, prox Rt. h/c; 95, Lt. M^; 99, prox Lt. h/c; 106, Rt. M^; 29259, Rt. man- dible (M, 2), Lt. Rt. Mi-2; 29277, calvaria and prox Rt. h/c; 30394, Rt. M; 30417, Lt. M"; 30418, prox Lt. h/c; 30627, Lt. h/c; 30633, h/c frags; 32543, Lt. h/c; 32544, Rt. mandible (P4-M,); 32546, Lt. maxilla (P^-M^); 32561, Rt. M^; 32812, Lt. maxilla frag (M^-^); 32823, Rt. M^; 36836, Lt. ML KP 29277 is a rather battered calvaria with the proximal portion of the right horn core but it is sufficiently well preserved to confirm the generic identification. The horn cores are mediolaterally compressed at their base with slight lyrate curva- ture in anterior view and a gentle sigmoid curvature in lateral view. In this regard, the Kanapoi impala more strongly resembles A. shungurae Gentry, 1985, from the northern portion of the Lake Tur- kana Basin (Gentry, 1985; Flarris, 1991b) by the lack of lyration than A. premelampus Flarris, 2003, from the nearby but somewhat older site of Loth- agam (Harris, 2003b). The majority of specimens Table 43 Kanapoi Aepyceros sp. indet. horn core measurements prox ap (Rt) prox tr (Rt) It (Rt) prox ap (Lt) prox tr (Lt) it (Lt) 70 42 41 99 35 30 29277 37 32 30418 36 30 30627 29 23 32543 33 28 98 ■ CS 498, Harris and Leakey: Kanapoi Figure 28 KNM-KP 29277, Aepyceros sp. indet., calvaria with proximal right horn core; A = anterior view; B - right lateral view; scales = 5 cm Harris, Leakey, Cerling, and Winkler: Tetrapods ■ 99 f Figure 29 KNM-KP 32543, Aepyceros sp. indet. left horn core; A = anterior view; B — lateral view; scales — 5 cm 100 ■ CS 498, Harris and Leakey: Kanapoi Table 44 Kanapoi Aepyceros sp. indet. teeth measure- ments 95 106 29259L 29259R 30417 Ml ap 13.5 tr 11.1 M2 ap 16.6 16.4 16.3 15.3 tr 13 12.8 11.4 M^ ap 19.2 18.5 tr 11.9 12.3 32546 32561 32812 32823 P2 ap 7.8 tr 7.2 P^ ap 7.1 tr 8.1 p4 ap 9.1 tr 9.7 Ml ap 11.4 tr 12.2 M2 ap 15.4 16.5 14.5 17.1 tr 12.7 11.6 13.2 12.5 M^ ap tr 29259R 32544 P4 ap 00 bo tr 6.1 M, ap 12.1 12 tr 7.4 8.9 M, ap 15.1 tr 7.9 M3 ap tr show slight to moderate transverse annulation; this is most strongly developed in KP 70 — the largest specimen whose attribution to Aepyceros is slightly less certain than that of the other specimens. Tribe Antilopini The Antilopini are small- to medium-sized hypso- dont antelopes that appear well adapted to arid ar- eas. The tribe has a long geological history and in- determinate species of gazelles have been reported from the middle Miocene of North (Thomas, 1979) and East Africa (Gentry, 1970). Gazella de Blainville, 1816 Gazella sp. indet. (Table 45) KANAPOI MATERIAL. 29264, Rt. h/c frags; 32513, Lt. and Rt. h/c frags. Two specimens appear to represent the small and slightly medio-laterally compressed proximal horn cores of an indeterminate species of gazelle. Table 45 Kanapoi Gazella sp. indet. horn core measure- ments 29264 32513 Rt h/c ap 21.5 20.1 tr 19.3 17.1 It Lt h/c ap 19.5 tr 17.7 It Tribe Caprini Caprins, which include domesticated sheep and goats, are documented as common elements of the middle Miocene assemblage from Tort Ternan (Gentry, 1970) but thereafter only appear sporadi- cally in sub-Saharan Africa as rare elements of late Neogene biotas. Caprini gen. et sp. indet. (Eigure 30) KANAPOI MATERIAL. 36604, Lt. h/c. A single, large left horn core with a very pro- nounced basal sinus may represent an indetermi- nate species of goat. The horn core is strongly me- diolaterally compressed with a flattened lateral sur- face. It evidently curved backward and outward from the base with a strong lyrate curvature from the anterior view. In some ways, it resembles a large but very flattened impala horn core. At its base, it measures 45.4 mm anteroposteriorly and 31.4 mm transversely. Tribe Neotragini Neotragins include about a dozen species of small antelopes that represent the root stock from which the Aegodontia (alcelaphins, antilopins, caprins) emerged (Kingdon, 1982). The tribe is today re- stricted to Africa. Neotragins representing species of steinbuck and dikdik have been reported from late Miocene assemblages at Lothagam (Harris, 2003b) but their small size tends to make them in- frequent components of Pliocene assemblages, al- though they are known from Langebaanweg (Gen- || try, 1980) and elsewhere. || Raphiceras H. Smith, 1827 The steenboks (or steinbucks) are moderate-sized neotragins that are distributed discontinuously ) through open bushland or light woodland in east- ern and southern Africa. The horn cores are short | to moderately long with little mediolateral com- pression, inserted widely apart above the back of jj the orbits, are parallel to one another and, in pro- file, have a slightly concave anterior edge. Postcorn- ual fossae are present and the supraorbital pits are wide apart (Gentry, 1980). Harris, Leakey, Ceding, and Winkler: Tetrapods ■ 101 I Figure 30 KNM-KP 36604, Caprini gen. and sp. indet. left horn core; A = anterior view; B = medial view; scales = 5 cm 102 ■ CS 498, Harris and Leakey: Kanapoi Figure 31 KNM-KP 29263, Raphiceras sp. indet. frontlet with horn cores; A = anterior view; B = right lateral view; scale = 5 cm Table 46 Kanapoi Raphiceras sp. indet. measurements 29263 Rt h/c ap 14.3 tr 12.7 It 71.6 Lt. h/c ap 13.7 tr 14 It 66.6 Width between h/cs 25.2 93 30443 36833 30273 P, ap 4.3 tr 2.8 P3 ap 8.5 tr 4.5 P4 ap 7.5 tr 4.8 Mj ap 10.9 9.5 tr 6.3 6.2 6.3 M, ap 12.0 10.3 10.4 tr 6.6 6.7 7.3 M3 ap 17.0 16.3 tr 5.8 6.4 Raphiceras sp. indet. (Figure 31; Table 46) KANAPOI MATERIAL. 93, Rt. mandible (P2, Mi_3); 29263, frontlet with Lt. and Rt. h/cs; 30273, Lt. Mi_2; 30443, Lt. mandible (P4-M3); 36833, Rt. mandible frag (P3_4). A frontlet with right and left horn cores (KP 29263) is a little larger than the equivalent portion of an extant steinbuck cranium but is evidently con- generic. A number of neotragine dentitions that are larger than those of dikdiks are interpreted to rep- resent this genus. Madoqua Ogilby, 1837 The dikdiks are small, shy antelope that occur sin- gly or in pairs in dry bush country in eastern and southwestern Africa. The horn cores are short, of- ten keeled, compressed anterolaterally to postero- medially, and with some flattening of the postero- medial surface. They are inserted wide apart above the back of the orbits, parallel to one another, and not very upright in lateral view. Postcornual fossa are small and shallow but the supraorbital pits are small and wide apart (Gentry, 1987). Harris, Leakey, Ceding, and Winkler: Tetrapods ■ 103 Table 47 Kanapoi Madoqua sp. indet. tooth measure- ments 103 30207 30416 30427 30537 32547 ap tr M' ap 9.3 8.2 tr 8.6 7.4 P2 ap tr P3 ap 4.8 tr 3.3 P4 ap 6 6.5 tr 4.1 3.7 Ml ap 7.3 tr 4.3 M2 ap 7.5 7.2 9.5 tr 5.1 4.5 5.1 M3 ap 10.5 11.1 tr 4.7 5.1 5.1 36832 36835 36840 30206A 30206B M2 ap 6.9 tr 8.2 M^ ap 8 tr 7.6 P2 ap tr P3 ap tr P4 ap 8.1 7.5 tr 3.9 4.2 Ml ap 7.6 7.5 7.3 7.2 tr 4.6 4.5 4.6 4.7 4.8 M2 ap 8.9 8.1 8.1 tr 5.2 5 5.1 5.2 M3 ap tr Madoqua sp. indet. (Table 47) KANAPOI MATERIAL. 103, Rt. M^; 30206, Rt. mandible (Mi_2), Rt. mandible (Mi_2), Lt. M^, Rt. M^, humerus frags; 30207, Lt. mandible frag (P3_4), prox scapula, dist Lt. and Rt. humerus; 30416, Rt. mandible frags (P4, M2_3); 30427, Rt. maxilla frag (M"-3); 30537, Lt. mandible (Mi_3); 32547, Rt. mandible (M2_3), seven foot bones and distal radius; 36832, Lt. mandible (P4-M3); 36835, Rt. mandible (Mi_2); 36840, Rt. mandible (P3-M2). Small partial dentitions of comparable size to ex- tant dikdiks represent one of the earliest known species of this genus. PALEOENVIRONMENTS Chronologically and stratigraphically, the Kanapoi Formation is equivalent to the lower part of the Nachukui Formation sequence as exposed at Loth- agam. The lacustrine interval documented in the middle of the Kanapoi sequence represents a south- erly extension of the Lonyumun Lake, which at the nearby locality of Lothagam is represented by the Muruongori Member. The terrestrial sediments at Kanapoi may thus be correlated to an upper por- tion of the Apak Member and the lower portion of the Kaiyumung Member in the Lothagam succes- sion (Feibel, 2003a). More than fifty mammalian species have been recovered from the Kanapoi For- mation (Table 48). The percentage distribution of the larger herbivore families in the Kanapoi biota is broadly comparable with that seen in the Apak Member although the represented taxa suggest a slightly younger age for the Kanapoi assemblage and at Kanapoi large suids seem to have replaced part of the bovid population that dominated at Apak (Table 49). The Nawata Formation sequence at Lothagam documents the interval of time in which C4 plants (and particularly grasses) underwent a major ex- pansion in Africa, Asia, and North and South America (Cerling et ah, 1997). As documented by Cerling et al. (2003a:fig. 12.5), the composi- tion of tooth enamel of mammals with a pure C4 diet ranges from 1-3 permil whereas that of mam- mals with a pure C3 diet in open habitats ranges from —12 to —16 permil. Preliminary results of iso- topic analysis of mammalian herbivores from Kan- apoi are mostly consistent with the younger radio- metric age determinations for the Kanapoi Forma- tion (Table 50). As expected, the elephantoid pro- boscideans [Anancus kenyensis, Loxodonta adaurora, Elephas ekorensis) all had a C4 grazing diet, as did the suids Nyanzachoerus pattersoni and Notochoerus jaegeri and the equid Eurygnathohip- pus sp. All these species had assumed a grazing diet before or during the accumulation of the Apak Member at Lothagam. In contrast, C3 browse was a major component of the diet of deinotheres {Deinotherium bozasi), sivatheres {Sivatherium cf. S. hendeyi), impalas {Aepyceros sp.), and ostriches {Struthio sp.). The sole rhino tooth analyzed also indicated a browsing diet. The 8’^C content of the fossil ostrich eggshell is more negative than those of extant ostriches from Kanapoi and the is more positive; it would appear that there was a greater proportion of C3 vegetation in the Pliocene ostrich diet compared with that of modern ostrich- es on the west side of Lake Turkana. Hence, the Lake Turkana Basin in the vicinity of Kanapoi ev- idently supported a variety of vegetation types and habitats during the early Pliocene. The very positive 8^^0 values for the analyzed rhino and sivathere teeth (Table 50) suggests these animals were ob- taining most of their water from the plants that they ate. Wynn (2000) recognized seven types of paleosols at Kanapoi, all of which provide information about the local ecosystem during the process of soil for- mation. 104 ■ CS 498, Harris and Leakey: Kanapoi Table 48 Kanapoi mammalian faunal list Taxon Sample size Body weight Locomotion Diet Chiroptera Hipposideros spp. A Ae I Insectivora Myosorex sp. A SA I Macroscelidae Elephantiihis sp. A SA I Lagomorpha Leporidae gen. and sp. indet. C T Hg Rodentia Tatera sp. A SA Hb Murini gen. indet. A SA Hb Xerus sp. B T HF cf. Steatomys sp. A T Hb Carnivora Enhydriodon ekecaman 1 D Aq C cf. Torolutra sp. 1 C Aq C Earahyaena bowel li 31 E T c Dinofelis petteri 3 D T-S c Homotheriimi sp. 4 E T-S c Felis sp. 1 C T-S c Helogale sp. 2 B T c Genetta sp. nov. 3 C T c Primates Australopithecus anamensis 59 D/E TA o cf. Galago sp. indet. 1 B A o cf. Ceropithecoides sp. indet. 6 C TA HF Colobinae sp. A 8 D TA HF Colobinae sp. B 1 D TA HF Parapapio ado 50 D TA HF cf. Theropithecus sp. 1 D TA HF Proboscidea Deinotherium bozasi 5 H T Hb Anancus kenyensis 2 H T Hg Elephas ekorensis 25 H T Hg Loxodonta adaurora 26 H T Hg Loxodonta exoptata 1 H T Hg Perissodactyla Ceratotheriiim praecox 11 H T Hg Diceros bicornis 4 H T Hb Eurygnathobippus sp. indet. 34 F T Hg Artiodactyla Hexaprotodon protamphibius 60 H Aq BG Hexaprotodon sp. 1 H Aq BG Nyanzacboerus pattersoni 77 F T Hg Notochoerus jaegeri 29 F T Hg Notochoerus cf. N. euilus 1 F T Hg Giraffa stillei 5 G T Hb Giraffa jumae 1 H T Hb Giraffa sp. indet. 9 G T Hb Siuatherium cf. S. bendeyi 3 G T Hb Tragelaphus kyaloae 33 F T BG Harris, Leakey, Cerling, and Winkler: Tetrapods ■ 105 Table 48 Continued Taxon Sample size Body weight Locomotion Diet Tragelaphini gen. indet. 17 F T BG Simatherium cf. S. demissum 4 H T Hg Hippotragini gen. indet. 4 G T BG Kobus sp. 2 F T Hg Reduncini gen. indet. 6 F T Hg Damalacra cf. D. neanica 1 F T Hg Damalacra sp. A 5 E T Hg Damalacra cf. D. acalla 3 F T Hg Alcelaphini gen. indet. 14 F T Hg Aepyceros sp. 18 F T Hb Gazella sp. 2 D T Hg Caprini gen. and sp. indet. 1 E T Hg Raphiceras sp. 5 D T Hb Madoqua sp. 10 C T Hb Ecovariabie categories Code Body weight Code Locomotor patterns Code Feeding preferences A 0-100 g T Terrestrial I Insectivore B 100-1,000 g TA Terrestrial-arboreal F Frugivore C 1-10 kg SA Semiarboreal HF Herbivore-frugivore D 10-45 kg A Arboreal Hb Browser E 45-90 kg S Scansorial Hg Grazer F 90-180 kg TS Terrestrial-scansorial BG Browser-grazer G 180-360 kg Aq Aquatic C Carnivore H 360+ kg Ae Aerial Cl Carnivore-insectivore F Fossorial O Omnivore TF Terrestrial-fossorial 1. Modern analogues of the Aberegaiya paleosols are found throughout semiarid regions of East Africa on floodplains flanking large river sys- tems and support edaphic savanna grasslands within low tree-shrub savanna mosaics. 2. The Lorenyan paleosol is found in delta deposits where it represents brief emergence of the delta flats and desiccation of the clayey parent mate- rial; modern analogues occur on the Omo River Delta at the north end of Lake Turkana where they support sparsely vegetated grasslands with variable admixtures of ephemeral forbs. 3. Dite paleosols, from which most of the hominid fossils were derived, were better drained but formed in broadly similar conditions to the Aberegaiya and Lorenyang paleosols. Modern analogues to the Dite paleosols are found in the lower Omo Valley, where well-drained soils sup- port low tree-shrub savanna vegetation in semi- arid to arid climatic regimes. 4. Abilat paleosols are calcic xerosols. Their char- acteristics are intermediate between Dite and Aberegaiya paleosols and they are believed to represent transitional habitats between these two pedotypes. Vegetation growing on Abilat soils would thus have been transitional between the forb-dominated edaphic grasslands of the Aberegaiya pedotype and the low tree-shrub sa- vanna of the Dite pedotype. 5. The Nasua paleosol is found on delta sediments. Like the Lorenyang, the Nasua paleosol repre- sents brief emergence of the delta flats but sup- ported denser vegetation; similar modern clayey soils are found on the Omo Delta, where they support shrub thickets of aquatically adapted Acacia and Cadaba species. 6. Kabisa paleosols with vertical rhizoliths are found in channel sandstones in the upper fluvial series; modern counterparts in the lower Tur- kana Basin occur in the sandy soils of ephemeral channels, where they support gallery woodland and thicket vegetation — often dominated by Acacia tortilis. 7. Akai-Ititi paleosols have horizontal rhizoliths representing root mats in nearshore environ- ments where groundwater is shallow; modern analogues fringe the Omo Delta and support lakeside and streamside grasslands with root mats of Phragmites, Typha, and Cyperus species and Loudetia phragmatoides or Sporobolis spi- catus. Wynn’s (2000) overall interpretation of the pre- vailing habitats at Kanapoi was that they strongly resembled habitats that presently occur in the re- 106 ■ CS 498, Harris and Leakey: Kanapoi Table 49 Percentage distribution of large herbivores at Lothagam and Kanapoi Lr Naw^ata Ur Naw^ata Apak Kanapoi Kaiyumung Elephantoidea 4.73 1.97 11.26 13.95 1.52 Rhinocerotidae 4.73 5.90 5.30 4.26 6.06 Equidae 10.11 11.52 8.61 8.04 13.64 Hippopotamidae 20.65 18.54 12.58 14.42 0.00 Suidae 24.52 18.82 13.25 25.06 36.36 Giraffidae 2.80 2.53 5.96 4.26 4.55 Bovidae 32.47 40.73 43.05 30.02 37.88 100.00 100.00 100.00 100.01 100.00 Table 50 Results of stable isotope analysis of Kanapoi herbivores Struthio sp. WT 3597 -8.66 10.89 Anancus kenyensis KP 30442 -0.13 -1.11 Elephas ekorensis KP 30173 -1.74 -1.32 WT 3570 -1.48 2.37 Loxodonta adaurora KP 390 -2.73 0.87 30196 -2.5 1.05 30196 -2.34 1.09 30596 -1.08 2.2 KP 383 -2.99 -0.36 Deinotherium bozasi WT 3617 -12.43 -1.75 Rhinocerotidae gen. indet WT 3223 -9.98 4.71 Eurygnathohippus sp. 30233 1.24 2.09 Nyanzachoerus pattersoni KP 205 -6.5 1.19 WT 3222 (M3) -1.94 1.14 KP-XIO -1.19 -0.22 Notochoerus jaegeri KNM-3348 B -1.74 1.43 KP 241 -2.01 1.07 KP 265 -5.47 -2.19 WT 3227 -1.74 1.78 Sivatherium cf. 5. hendeyi KP 30227 -9 8.6 Aepyceros sp. WT 3240-Aepy -10.02 1.36 WT 3240 -9.77 1.46 gion of the modern Omo River Delta at the north end of Lake Turkana. Most of the hominids derive from Dite paleosols that he interpreted to have sup- ported low tree-shrub savanna vegetation in a semiarid climate with an annual rainfall that ranged from 350 to 600 mm. Wynn (2000) sug- gested that Kanapoi may represent the earliest known occurrence of hominids venturing into rel- atively open habitats. However, Ward et al. (1999) caution that many of the hominid specimens show evidence of carnivore damage. Thus, it is possible that the hominid remains were transported by car- nivores to the locations where they accumulated. Other paleosols ranged from poorly drained Verti- sols with forb-dominated edaphic grassland in local shallow depressions to well-drained alluvial soils supporting gallery woodland adjacent to stream courses. The proportion of soil carbonate contrib- uted by C4 plants (i.e., grasses) ranges from 25% in the Kabisa paleosol to 40% in the Dite and Abi- lat paleosols (Wynn 2000). Confirmation of the nature of the terrestrial j habitats during the interval that the early Pliocene assemblage accumulated at Kanapoi is provided | by the dietary adaptations of the terrestrial ver- ! tebrate fossils (Table 48). Proboscideans amount i to about 14% of the large vertebrate biota: Lox- odonta adaurora and Elephas ekorensis predomi- t nate, whereas Anancus and Deinotherium are only sparsely represented. Elephantids and gomphoth- eres had acquired a grazing diet by the beginning of the Pliocene and their presence at Kanapoi ap- j pears to signify open habitat in the near vicinity of the Kanapoi Delta, although the presence of Deinotherium confirms an ample supply of C3 vegetation. Rhinos were represented by the pre- i decessors of the two extant African species, both j of which had acquired their different (grazing ver- ' sus browsing) dietary specializations by the begin- ning of the Pliocene. Equids are represented by i Eurygnathohippus, a grazing hipparionine. Earge suids are represented by Nyanzachoerus patter- soni and Notochoerus jaegeri; both were inter- preted as closed habitat species by Bishop (1994) on the basis of their postcranial anatomy but both | species were evidently specialized grazers on the basis of the isotopic composition of their tooth Harris, Leakey, Cerling, and Winkler: Tetrapods ■ 107 Table 5 1 Stratigraphic distribution of bovid tribes at Lothagam and Kanapoi Lr Nawata Ur Nawata Apak Kanapoi Kaiyumung Tragelaphini 3.33 2.08 20.63 39.37 16.00 Bovini 1.33 2.08 9.52 3.15 28.00 Boselaphini 24.67 7.64 3.17 Hippotragini 5.33 5.56 4.76 3.94 Reduncini 11.33 21.53 9.52 7.09 8.00 Alcelaphini 4.67 17.36 14.29 18.11 12.00 Aepycerotini 48.00 40.28 31.75 14.17 36.00 Antilopini 0.67 1.39 4.76 2.36 Neotragini 0.67 2.08 1.59 11.81 100.00 100.00 100.00 100.00 100.00 enamel (Harris and Cerling, 2002; Cerling et al., 2003b). Giraffids were relatively uncommon con- stituents of the Kanapoi assemblage but both siv- atheres and giraffines were C3 browsers at that point in time. Bovids, as usual, are the most com- monly encountered terrestrial herbivores. Trage- laphins, aepycerotins, and neotragins — now all closed habitat or forest-edge forms — make up two thirds of the Kanapoi bovid assemblage, but the presence of bovins, reduncins, hippotragins, and alcelaphins is indicative of nearby sources of grass. In contrast with the bovid assemblages from the Apak and Kaiyumung Members, the Kanapoi bovid assemblage has more tragelaphins and al- celaphins but fewer bovins and aepycerotins — which may be interpreted to represent drier and less wooded habitats in the vicinity of Kanapoi (Table 51). Overall, grazing species outnumber browsing species by a ratio of nearly two to one; in terms of numbers of individuals, browsers out- Frugivores versus Non-arboreal 30 n 25 -I W (U E E 20 (0 E V) 3 15 - Oi □ 10 o o X O Closed woodland □ Closed wd/bshland ABushland O Open woodland -l-Shrubland XPIains • Kanapoi ♦ Lothagam ■ Shungura AKoobi Fora XAramis 0 ^ 1 , 80 90 100 % Terrestrial mammals Figure 32 Graph depicting ecological structural analysis of frugivorous mammals and nonarboreal mammals from mod- ern African habitats and from Pliocene assemblages from the Lake Turkana Basin. Data for extant habitats and assem- blages from the Shungura and Koobi Fora Formations are from Reed (1997); those for Lothagam assemblages are from Leakey and Harris (2003); those for Aramis are from WoldeGabriel at al. (1994). Of the two data points representing Kanapoi, that on the right is for the assemblage from below the lacustrine interval, that on the left is for the assemblage from above the lacustrine interval 108 ■ CS 498, Harris and Leakey: Kanapoi Fresh Grass Grazers versus Terrestrial O Closed woodland □ Closed wd/bshind A Bushland O Open woodland + Shrubland • Kanapoi ♦ Lothagam ■ Shungura AKoobi Fora X Aramis Figure 33 Graph depicting ecological structural analysis of fresh grass grazing mammals and terrestrial mammals from modern African habitats and from Pliocene assemblages from the Lake Turkana Basin. Data points as for Figure 32 number grazers by three to one. In confirmation, the Kanapoi small mammals suggest a relatively dry climate and open habitat (Appendix). The Kanapoi large-mammal assemblage shares many elements with that from the 4.4 Ma site of Aramis, Ethiopia (WoldeGabriel et ah, 1994), al- though, with the exception of the primates, most of the Aramis fossil material has yet to be described in detail. Taxa shared by Kanapoi and Aramis in- clude elephantids (unidentified at Aramis), Anan- cus, Deinotherium, Ceratotherium cf. C. praecox, “Hipparion” sp., Hexaprotodon sp., Nyanzachoe- nis pattersoni, Notochoerus jaegeri, a large and small species of Giraffa, and tragelaphins, bovins, and neotragins. However, with the exception of an unidentified kudu {Tragelaphus sp.), large verte- brates are rare at Aramis, as are aquatic verte- brates. WoldeGabriel et al. (1994) interpret the dominance of colobines and kudus at Aramis to be a strong indication of a closed, wooded environ- ment. However, the presence at an early Pliocene site of elephantids, Anancus, Ceratotherium, an equid, a hippo, Nyanzachoerus, Notochoerus, and bovins is indicative of a diverse open-country graz- ing component in the biota. Moreover, the larger genera of early Pliocene colobines were more ter- restrial than their extant counterparts (Leakey et ah, 2003) and hence the dominance of colobines may not necessarily mandate the presence of closed woodland. As at Aramis, tragelaphins were the most abundant bovids recovered from Kanapoi. However, in contrast with the situation at Aramis, papionin primates far outnumbered the colobines. On balance, Kanapoi would appear to represent habitats that were similar to but more open than those of Aramis. Andrews and Humphrey (1999) stated that the fossil assemblages from both Aramis and Kanapoi are underrepresented by small mam- mals that often provide a more precise assessment of the prevailing paleoenvironments than do large migratory species. Large quantities of micromam- mals that were retrieved by the National Museums of Kenya expeditions have yet to be studied in de- tail (see Appendix) and the bulk of the nonprimate component of the Aramis fauna has also yet to be studied in depth. Ecological diversity provides a different means from taxonomic uniformitarianism for analyzing the community structure of fossil mammalian fau- nas in order to obtain information about the hab- itats represented by faunal assemblages (Andrews et ah, 1979). Using this methodology, Andrews and Humphrey (1999) interpret the Kanapoi material collected by American expeditions in the 1960s, as “. . . dominated by medium- to large-sized terres- trial species and both browsers and grazers are well represented. The high terrestrial component in the fauna places it with the drier end of the present- day woodland and bushland ecosystems suggesting open woodland with abundant grass.” Harris, Leakey, Ceding, and Winkler: Tetrapods >109 Ecological structural analysis as formulated by Reed (1997) provides a different interpretation. Reed established that habitats were predicted by lo- comotory adaptations and characterized by trophic ecovariables and found that it was possible to dif- ferentiate between different terrestrial habitats by plotting the percentage of frugivorous mammals against the percentage of nonarboreal mammals (Reed, 1997:fig. 5 A). Similar results were obtained by plotting the percentage of fresh grass grazers against percentage of terrestrial mammals and this also established the presence/absence of edaphic grassland (Reed, 1997:fig. 5B). The estimated body weights, locomotor adaptations, and feeding pref- erences for the Kanapoi mammals are listed in Ta- ble 48. When Pliocene assemblages from the Lake Turkana Basin are superimposed on the modern habitat plots for frugivores versus nonarboreal mammals (Fig. 32), the two Kanapoi assemblages (above and below the Lonyumun Lake incursion) plot with the closed woodland assemblages. A sim- ilar grouping occurs when the Kanapoi assemblages are superimposed on the graph of fresh grass graz- ers versus terrestrial mammals (Fig. 33). In both cases, the Kanapoi assemblages plot close to those from the Kaiyumung Member at Lothagam and Shungura Member B in the lower Omo Valley. The reason why the Kanapoi assemblages appear to in- dicate closed woodland is because the primates have been interpreted as at least partly arboreal (re- quiring shelter in trees at night). Attribution of ter- restrial adaptation to one or more of the primate species would skew the sample toward an interpre- tation of more open habitat. Given that the homi- nin was bipedal and that early Pliocene cercopith- ecoids were more terrestrial than later forms, such an interpretation would be more in keeping with the dietary adaptations of the fossil species and the ecological adaptations of their modern counter- parts. SUMMARY The early Pliocene site of Kanapoi, southwest of Lake Turkana in northern Kenya, has yielded the oldest australopithecines from eastern Africa al- though a similar age has recently been claimed for a new australopithecine locality in southern Africa (Partridge et ah, 2003). Associated with the homi- nin remains from Kanapoi is a diverse vertebrate fauna that has been recovered from fluvial and la- custrine sediments dating between 4.17 and 4.07 Ma. The Kanapoi tetrapod fauna is much more prolific than that from any other similarly aged lo- cality elsewhere in the Lake Turkana Basin (Nachu- kui and Koobi Fora Formations). The assemblage includes 24 species of fish, 7 species of chelonians, 4 species each of crocodylians and birds, and over 50 species of mammals. Few of the recognized spe- cies are new but micromammals have not yet been thoroughly investigated. Relatively complete suid specimens mandate the transfer of the species Nyanzachoerus jaegeri to the genus Notochoerus. Paleosols associated with the fauna remains suggest a mixture of open and closed habitats similar to those currently found in the vicinity of the Omo Delta at the north end of Lake Turkana. Such an interpretation is supported by the dietary adapta- tions of the herbivorous fossil mammals and the habitats exploited by their living representatives. ACKNOWLEDGMENTS We thank the Government of Kenya and the Trustees of the National Museums of Kenya for permission to study the Kanapoi fossils and the directors of the National Mu- seums of Kenya and Natural History Museum of Los An- geles County for logistical support that helped expedite this manuscript. Fieldwork at Kanapoi was funded by the National Geographic Society and National Science Foun- dation (SBR 9601025) and made possible by donations of fuel by Caltex (Kenya) and the loan of light aircraft be- longing to Jonathan and Richard Leakey. We thank also the preparation and curatorial staff of the Division of Pa- laeontology at the National Museums of Kenya, Nairobi, and the many people who participated in the 1994-97 field expeditions and whose names are acknowledged in Ward et al. (2001). We gratefully acknowledge helpful dis- cussions with Craig Feibel, Nina Jablonski, Kathy Stew- art, Lars Werdelin, and Alisa Winkler, and we thank Hank Wesselman for his observations on the Kanapoi micro- mammals, Diana Mattheissen for her notes on the Kana- poi birds, and Anthony Macharia for his help with the GIS database. Isotopic analyses were undertaken in the Geochemistry and SIRFER Laboratories of the University of Utah and were underwritten by support from NSF and the Packard Foundation. Drs. Nina Jablonski and Tim White, an anonymous referee, and members of the Sci- entific Publications Committee for the Natural History Museum of Los Angeles County, who kindly offered sug- gestions that helped improve the manuscript. LITERATURE CITED Andrews, P. J., and L. Humphrey. 1999. African Miocene environments and the transition to early hominines. In African biogeography, climate change, and human evolution, eds. T. Bromage and F. Schrenk, 282-300. Oxford: Oxford University Press. Andrews, P. J., J. M. Lord, and E. M. Nesbit Evans. 1979. Patterns of ecological diversity in fossil and modern mammalian faunas. Biological Journal of the Lin- nean Society 11:177-205. Arambourg, C. 1934. Le Dinotherium des Gisements de I’Omo. Comptes Rendues de la Societe geolologique de France 1934:86-87. . 1944. Les hippopotames fossiles d’Afrique. Compte Rendu de I’Academie des Sciences 218:602- 604. . 1959. Vertebres continentaux du Miocene stiper- ieure de I’Afrique du Nord. Publ. Service de la Carte. Geologique d’ Algerie (Alger), Nouveau Series, Pa- leontologie Memoires 4:1-161. Beden, M. 1987. Fossil Elephantidae from Laetoli. In Lae- toli: A Pliocene site in northern Tanzania, eds. M. D. Leakey and J. M. Harris, 259-293. New York: Oxford University Press. Bernor, R. L., and J. M. Harris. 2003. Systematics and evolutionary biology of the Late Miocene and Early Pliocene Hipparionine equids from Lothagam, Ken- 110 ■ CS 498, Harris and Leakey: Kanapoi ya. In Lothagam: The dawn of humanity in eastern Africa, eds. M. G. Leakey and J. M. Harris, 387- 438. New York: Columbia LIniversity Press. Bishop, L. C. 1994. Pigs and the ancestors: Hominids, suids, and environments during the Plio-Pleistocene of East Africa. Ph.D. dissertation, Yale University. Boisserie, J-R. 2002. Nouveaux Hippopotamidae du Tchad et Ethiopie: Implications phylogenetiques et paleoenvironmentales. Ph.D. dissertation, Universite de Poitiers. Broom, R. 1925. On evidence of a giant pig from the Late Tertiaries of South Africa. Records of the Albany Museum 3:307-308. Cerling, T. E., J. M. Harris, and M. G. Leakey. 2003a. Isotope ecology of the Lothagam sequence. In Loth- agam: The dawn of humanity in Africa, eds. M. G. Leakey and J. M. Harris, 605-624. New York: Co- lumbia University Press. . In press. Environmentally driven dietary adapta- tions in African mammals. In History of atmospheric CO2 and the impacts on plants, animals, and eco- systems, eds. J. R. Ehleringer, M. D. Dearing, and T. E. Cerling. New York: Springer- Verlag. Cerling, T. E., J. M. Harris, M. G. Leakey, and N. Mu- dida. 2003b. Isotopic ecology of the Lake Turkana Basin. In Lothagam: the dawn of humanity in Africa, eds. M. G. Leakey and J. M. Harris, 583-603. New York: Columbia University Press. Cerling, T. E., J. M. Harris, B. J. MacFadden, M. G. Lea- key, J. Quade, V. Eisenmann, and J. R. Ehleringer. 1997. Global; vegetation change through the Mio- cene/Pliocene boundary. Nature 389:153-158. Churcher, C. S. 1978. Giraffidae. In Evolution of African mammals, eds. V. J. Maglio and H. B. S. Cooke, 509-535. Cambridge: Harvard University Press. . 1979. The large paleotragine giraffid {Palaeotra- gus germaini) from the Late Miocene of Lothagam Hill. Breviora 453:1-8. Coe, M. 1972. Ecological studies of the small mammals of South Turkana. Geographical Journal 138:316- 338. Cooke, H. B. S., and R. E. Ewer. 1972. Fossil Suidae from Kanapoi and Lothagam, Kenya. Bulletin of the Mu- seum of Comparative Zoology Harvard 1 43 (3): 149- 295. Cooke, H. B. S., and Q. B. Hendey. 1992. Nyanzachoerus (Mammalia: Suidae: Tetraconodontinae) from Lan- gebaanweg. South Africa. Durban Museum Novita- tes 17:1-20. Coppens, Y. 1971. Une nouvelle espece de Suide du Vil- lafranchien du Tunisie, Nyanzachoerus jaegeri nov. sp. Comptes Rendues Hebdomedaires des Sceances, Academie des Sciences (Paris) 272:3264-3267. Coppens, Y., V. J. Maglio, C. T. Madden, and M. Beden. 1978. Proboscidea. In Evolution of African mam- mals, eds. V. J. Maglio and H. B. S. Cooke, 333- 367. Cambridge: Harvard University Press. Coryndon, S. C. 1977. The taxonomy and nomenclature of the Hippopotamidae (Mammalia, Artiodactyla) and a description of two new fossil species. Proceed- ings of the Koninklijke Nederlandse Akademie van Wetenschappen B80(2):61-88. Dietrich, W. O. (1942). Altestquartare Saugetiere aus den siidlichen Serengeti, Deutsch Ost-afrika. Paleonto- graphica 94:43-133. Feibel, C. S. 2003a. A: Stratigraphy and depositional his- tory of the Lothagam sequence. In Lothagam: The dawn of humanity in eastern Africa, eds. M. G. Lea- key and J. M. Harris, 17-30. New York: Columbia University Press. . 2003b. Stratigraphy and depositional setting of the Pliocene Kanapoi Formation, Lower Kerio Val- ley, Kenya. In Geology and vertebrate paleontology of the Early Pliocene site of Kanapoi, northern Ken- ya. Edited by J. M. Harris and M. G. Leakey. Con- tributions in Science 498:9-20. Fourtau, R. 1920. Contribution a V etude de vertebres Miocenes de LEgypt. Cairo: Geological Survey of Egypt, Ministry of Finance. 121 pp. Frost, S. R. 2001. New early Pliocene Cercopithecidae (Mammalia: Primates) from Aramis, Middle Awash Valley, Ethiopia. American Museum Novitates 3350: 1-36. Frost, S. R., and E. Delson. 2002. Fossil Cercopithecidae from the Hadar Formation and surrounding areas of the Afar Depression Ethiopia. Journal of Human Evolution 43:687-748. Gentry, A. W. 1970. The Bovidae of the Fort Ternan fossil fauna. Fossil Vertebrates of Africa 2:243-323. . 1978. Bovidae. In Evolution of African Mammals, eds. V. J. Maglio and H. B. S. Cooke, 540-572. Cambridge: Harvard University Press. . 1980. Fossil Bovidae (Mammalia) from Lange- baanweg. South Africa. Annals of the South African Museum 79(8):2134-337. . 1985. The Bovidae of the Omo Group deposits. In Ees faunes Plio-Pleistocenes de la basse vallee de I’Omo (Ethiopie), eds. Y. Coppens and F. C. Howell, 1:119-191. Paris: Cahiers de Paleontologie. . 1987. Fossil Bovidae from Laetoli. In Eaetoli: A Pliocene site in northern Tanzania, eds. M. D. Lea- key and J. M. Harris, 378-408. Oxford: Clarendon Press. . 1990. Evolution and dispersal of African Bovidae. In Horns, pronghorns, and antlers, eds. G. A. Bub- enik and A. B. Bubenik, 195-227. New York: Springer- Verlag. Geze, R. 1980. Les Hippopotamidae (Mammalia, Artio- dactyla) du Plio-Pleistocene de I’Ethiopie (Afrique orientale). These, Universite de Paris VI. . 1985. Repartition paleoecologique et relations phylogenetiques des Hippopotamidae (Mammalia, Artiodactyla) du neogene d’ Afrique Orientale. In E’environnement des hominides an Plio-Pleistocene, eds. Y. Coppens and M. Beden, 81-100. Paris: Fon- dation Singer-Polignac, Masson. Hamilton, W. R. 1973. The lower Miocene ruminants of Gebel Zelten, Libya. Bulletin of the British Museum (Natural History), Geology 21:76-150. . 1978. Fossil giraffes from the Miocene of Africa and a revision of the phylogeny of the Giraffoidea. Philosophical Transactions of the Royal Society, se- nes B 283:165-229. Harris, J. M. 1976a. Cranial and dental remains of Dein- otherium bozasi (Mammalia, Proboscidea) from East Rudolf, Kenya. Journal of Zoology 178:57-75. . 1976b. Pliocene Giraffidae (Mammalia, Artiodac- tyla) from Cape Province. Annals of the South Af- rican Museum 69:325-353. . 1978. Deinotheriidae and Barytheriidae. In Mam- malian evolution in Africa, eds. V. J. Maglio and H. B. S. Cooke, 315-332. Cambridge: Harvard Univer- sity Press. . 1983a. Family Deinotheriidae. In Koobi Fora Re- search Project, Vol. 2 — The fossil ungulates: Probos- Harris, Leakey, Cerling, and Winkler: Tetrapods Hill cidea, Perissodactyla, and Suidae, ed. J. M. Harris, 22-39. Oxford: Clarendon Press. . 1983b. Family Suidae. In Koobi Fora Research Project, Vol. 2 — The fossil ungulates: Proboscidea, Perissodactyla, and Suidae, ed. J. M. Harris, 215- 302. Oxford: Clarendon Press. . 1991a. Family Hippopotamidae. In Koobi Fora Research Project, Vol. 3 — The fossil ungulates: Ge- ology, fossil artiodactyls, and palaeoenvironments, ed. J. M. Harris, 31-85. Oxford: Clarendon Press. . 1991b. Family Bovidae. In Koobi Fora Research Project, Vol. 3 — The fossil ungulates: Geology, fossil artiodactyls, and palaeoenvironments, ed. J. M. Har- ris, 139-320. Oxford: Clarendon Press. . 2003a. Lothagam Giraffidae. In Lothagam: The dawn of humanity in Africa, eds. M. G. Leakey and J. M. Harris, 523-530. New York: Columbia Uni- versity Press. . 2003b. Lothagam Bovidae. In Lothagam: The dawn of humanity in Africa, eds. M. G. Leakey and J. M. Flarris, 531-579. New York: Columbia Uni- versity Press. Harris, J. M., F. H. Brown, and M. G. Leakey. 1988. Stra- tigraphy and palaeontology of Pliocene and Pleisto- cene localities west of Lake Turkana, Kenya. Con- tributions in Science 399:1-128. Harris, J. M., and T. E. Cerling. 1998. Isotopic changes in the diet of giraffids. Journal of Vertebrate Pale- ontology 18(suppl.):49A. . 2002. Dietary adaptations of extant and Neogene African suids. Journal of Zoology 256:45-54. Harris, J. M., and M. G. Leakey. 2003a. Lothagam birds. In Lothagam: The dawn of humanity in Africa, eds. M. G. Leakey and J. M. Harris, 161-166. New York: Columbia University Press. . 2003b. Lothagam Rhinocerotidae. In Lothagam: The dawn of humanity in eastern Africa, eds. M. G. Leakey and J. M. Harris, 371-385. New York: Co- lumbia University Press. . 2003c. Lothagam Suidae. In Lothagam: The dawn of humanity in Africa, eds. M. G. Leakey and J. M. Flarris, 485-519. New York: Columbia Uni- versity Press. Harris, J. M., and T. D. White. 1979. Evolution of the Plio-Pleistocene African Suidae. Transactions of the American Philosophical Society 69:1-128. Hendey, Q. B. 1981. Palaeoecology of the Eate Tertiary fossil occurrences in “E” Quarry, Eangebaanweg, South Africa, and a reinterpretation of their geolog- ical context. Annals of the South African Museum 84:104. Hooijer, D. A. 1972. A late Pliocene rhinoceros from Lan- gebaanweg. Cape Province. Annals of the South Af- rican Museum 29(9):151-191. Hooijer, D. A., and V. J. Maglio. 1974. Hipparions from the late Miocene and Pliocene of northwestern Ken- ya. Zoologische Verhandelingen Eeiden 134:3-34. Hooijer, D. A., and B. Patterson. 1972. Rhinoceroses from the Pliocene of northwestern Kenya. Bulletin of the Museum of Comparative Zoology 144(l):l-26. Hopwood, A. T. 1936. New and little-known fossil mam- mals from the Pliocene of Kenya Colony and Tan- ganyikas Territory. Annals and Magazine of Natural History series 10, 17:636-638. Janis, C. M., and K. M. Scott. 1987. The inter-relation- ships of higher ruminant families with special em- phasis on the Cervoidea. American Museum Novi- tates 2893:1-85. Joleaud, E. 1920. Sur la presence d’un Gavialide du genre Tomistoma dans le Pliocene d’eau douce de I’Ethiopie. Comptes Rendus de V Academic des Sci- ences, Paris 70:816-818. Kingdon, J. 1974. East African mammals, Vol. IIB. Hares and rodents. Chicago: University of Chicago Press. 704 pp. . 1982. East African mammals, Vol. 3 (part C). Bovids. London: Academic Press. . 1958. Some East African Fossil Suidae. Fossil Mammals of Africa 14:1-133. Leakey, M. G. 1982. Extinct large colobines from the Plio-Pleistocene of Africa. American Journal of Physical Anthropology 58:153-172. Leakey, L. S. B. 1965. Olduvai Gorge 1951-1961: Fauna and background. Cambridge: Cambridge University Press. . 1987. Colobinae (Mammalia, Primates) from the Omo Valley, Ethiopia. In Les faunas Plio-Pleisto- cenes de la Basse Vallee de VOmo (Ethiopie), Tome 3, Cercopithecidae de la Formation Shungura, 148- 169. Cahiers de Paleontologie, Travaux de Paleon- tologie Est-Africaine, Paris: Editions du CNRS. . 1993. Evolution of Theropithecus in the Turkana Basin. In Theropithecus: The rise and fall of a pri- mate genus, ed. N. G. Jablonski, 85-123. Cam- bridge: Cambridge University Press. Leakey, M. G., and E. Delson. 1987. Eossil Cercopithe- cidae from the Laetolil Beds, Tanzania. In Eaetoli: A Pliocene site of northern Tanzania, eds. M. D. Lea- key and J. M. Harris, 91-107. New York: Oxford University Press. Leakey, M. G., C. S. Eeibel, I. McDougall, and A. Walker. 1995. New four-million-year-old hominid species from Kanapoi and Allia Bay, Kenya. Nature 376: 565-571. Leakey, M. G., C. S. Feibel, I. McDougall, C. Ward, and A. Walker. 1998. New specimens and confirmation of an early age for Australopithecus anamensis. Na- ture 393:62-66. Leakey, M. G., and J. M. Harris. 2003. Lothagam: Its significance and contributions. In Eothagam: The dawn of humanity in east Africa, eds. M. G. Leakey and J. M. Harris, 625-660. New York: Columbia University Press. Leakey, M. G., M. E Teaford, and C. V. Ward. 2003. Cercopithecidae from Eothagam. In Lothagam: The daivn of humanity in East Africa, eds. M. G. Leakey and J. M. Harris, 201-248. New York: Columbia University Press. Macinnes, D. G. 1942. Miocene and post-Miocene Pro- boscidea from East Africa. Transactions of the Zoo- logical Society of Eondon 25:33-106. Maglio, V. J. 1970. Eour new species of Elephantidae from the Plio-Pleistocene of northwestern Kenya. Bre- viora 341:1-43. . 1973. Origin and evolution of the Elephantidae. Transactions of the American Philosophical Society 63(3):149 pp. Meylan, P. A., and W. Auffenberg. 1986. New land tor- toises (Testudines: Testudinidae) from the Miocene of Africa. Zoological Journal of the Einnaean Soci- ety 86:279-307. Meylan, P. A., B. S. Weig, and R. C. Wood. 1990. Eossil soft-shelled turtles (Family Trionychidae) of the Lake Turkana basin, Africa. Copeia 1990(2):508-528. Partridge, T. C., D. E. Granger, M. W. Caffee, and R. J. 112 ■ CS 498, Harris and Leakey: Kanapoi Clarke. 2003. Lower Pliocene hominid remains from Sterkfontein. Science 300:607-612. Patterson, B. 1968. The extinct baboon, Parapapio jonesi, in the early Pleistocene of northwestern Kenya. Bre- viora 282:1-4. Patterson, B., A. K. Behrensmeyer, and W. D. Sill. 1970. Geology of a new Pliocene locality in northwestern Kenya. Nature 256:279-284. Petrocchi, C. 1954. I proboscidati di Sahabi. Rendicont delPAccademia Nazionale dei XL 4(1953-1954):l-66. Pickford, M. 1983. On the origins of Hippopotamidae together with descriptions of two new species, a new genus and a new sub-family from the Miocene of Kenya. Geobios 16:193-217. Powers, D. W. 1980. Geology of Mio-Pliocene sediments of the lower Kerio River Valley. Ph.D. dissertation, Princeton University. 182 pp. Reed, K. A. 1997. Early hominid evolution and ecological change through the African Plio-Pleistocene. Journal of Human Evolution 32:289-322. Sauer, E. G. E. 1972. Ratite eggshells and phylogenetic questions. Bonner Zoologische Beitrdge 23:3-48. Storrs, G. W. 2003. Late Miocene-Early Pliocene croco- dilian fauna of Lothagam, southwest Turkana Basin, Kenya. In Lothagam: The dawn of humanity in east- ern Africa, eds. M. G. Leakey and J. M. Harris, 137- 159. New York: Columbia University Press. Tassy, P. 2003. Elephantoidea from Lothagam. In Lotha- gam: The dawn of humanity in eastern Africa, eds. M. G. Leakey and J. M. Harris, 331-358. New York: Columbia University Press. Tchernov, E. 1976. Crocodylidae from the Pliocene/Pleis- tocene formations of the Rudolf Basin. In Earliest man and environments in the Lake Rudolf Basin, eds. Y. Coppens, E. C. Howell, G. Isaac, and R. E. E. Leakey, 370-378. Chicago: University of Chicago Press. . 1986. Evolution of the crocodiles in East and North Africa. Cahiers de Paleontologie. Paris: Cen- tre National de la Recherche Scientifique. Thomas, H. 1979. Les Bovides du Miocene superieur des couches de Mpesida et de la formation de Lukeino (District de Baringo, Kenya). Proceedings of the 8th Pan-African Congress of Prehistory and Quaternary Studies, Nairobi, 1977, 82-91. van der Made, J. 1999. Biometrical trends in the Tetra- conodontinae, a subfamily of pigs. Transactions of the Royal Society of Edinburgh: Earth Sciences 89: 199-225. Vrba, E. S. 1984. Evolutionary pattern and process in the sister group Alcelaphini-Aepycerotini. In Living fos- sils, eds. H. Eldredge and S. M. Stanley, 62-79. New York: Springer- Verlag. . 1985. African Bovidae: Evolutionary events since the Miocene. South African Journal of Science 81: 263-266. . 1987. New species and a new genus of Hippotra- gini (Bovidae) from Makapansgat Limeworks. Pa- laeontologia Africana 26(5):47-58. . 1995. The fossil record of African antelopes (Mammalia, Bovidae) in relation to human evolution and climate. In Paleoclimate and evolution with em- phasis on human origins, eds. E. S. Vrba, G. H. Den- ton, T. C. Partridge, and L. H. Burkle, 385-424. New Haven: Yale University Press. . 1997. New fossils of Alcelaphini and Caprinae (Bovidae, Mammalia) from Awash, Ethiopia, and phylogenetic analysis of Alcelaphini. Palaeontologia Africana 34:127-198. Ward, C. V, M. G. Leakey, and A. C. Walker. 1999. The new hominid species Australopithecus anamensis. Evolutionary Anthropology 7:197-205. . 2001. Morphology oi Australopithecus anamensis from Kanapoi and Allia Bay, Kenya. Journal of Hu- man Evolution 4:255-368. Wesselman, H. B. 1984. The Omo micromammals: Sys- tematics and paleoecology of early man sites from Ethiopia. Contributions in Vertebrate Evolution 7: 1-219. Weston, E. M. 1997. A biometrical analysis of evolution- ary change within the Hippopotamidae. Ph.D. dis- sertation, University of Cambridge. . 2003. Eossil Hippopotamidae from Lothagam. In Lothagam: The dawn of humanity in eastern Africa, eds. M. G. Leakey and J. M. Harris, 441-483. New York: Columbia University Press. Winkler, A. J. 1998. New small mammal discoveries from the early Pliocene at Kanapoi, West Turkana, Kenya. Journal of Vertebrate Paleontology 18(suppl.)3:87A. WoldeGabriel, G., T. D. White, G. Suwa, P. Renne, J. de Heinzelin, W. K. Hart, and G. Heiken. 1994. Eco- logical and temporal placement of Early Pliocene hominids at Aramis, Ethiopia. Nature 371:330-333. Wood, R. C. 1983. Kenyemys williamsi, a fossil pelomedusid turtle from the Pliocene of Kenya. In Advances in her- petology and evolutionary biology: Essays in honor of Ernest E. Williams, eds. A. J. G. Rhodin and K. Miyata, 74-85. Cambridge, Massachusetts: Museum of Com- parative Zoology, Harvard University. . 2003. Eossil turtles from Lothagam. In Lothagam: The dawn of humanity in eastern Africa, eds. M. G. Leakey and J. M. Harris, 115-136. New York: Co- lumbia University Press. Wynn, J. G. 2000. Paleosols, stable carbon isotopes, and paleoenvironmental interpretation of Kanapoi, northern Kenya. Journal of Human Evolution 39: 411-432. APPENDIX PRELIMINARY ASSESSMENT OF THE KANAPOI MICROMAMMALS Alisa J. Winkler Order Chiroptera Family Hipposideridae Hipposideros Gray, 1831 Hipposideros spp. Winkler (1998) reported a large and small species of Hip- posideros (Old World leaf-nosed bats), represented mainly by postcrania and isolated teeth, from the “Bat Site” lo- cality (WT 3409). In Africa, Hipposideros is currently known from seven superspecies or species groups (King- don, 1974). Hipposideros may roost in a variety of shel- ters, including caves, holes in the ground, hollow trees, and in undergrowth. Several species are known to roost with other species of bats. The size of roosting aggrega- tions varies from several to over 1,000 individuals. At the “Bat Site” locality, the large concentration of bats, relative Harris, Leakey, Cerling, and Winkler: Tetrapods ■ 113 scarcity of other taxa, and little evidence of alteration or transport of the bones and teeth suggest that this assem- blage represents an attritional accumulation under a roost site. Order Insectivora Family Soricidae Subfamily Crocidurinae Crocidurinae gen. and sp. indet. Two incomplete mandibles of shrews are currently known from Kanapoi. At least one of these may be referable to Myosorex Gray, 1838. This genus is known from Africa at present and is also reported from the African Plio-Pleis- tocene (Hendey, 1981; Wesselman, 1984). Order Macroscelidea Family Macroscelididae Elephantulus Thomas and Schwann, 1906 Elephantulus sp. Three specimens of Elephantulus have been recovered so far from Nzube’s Mandible Site (WT 3227). Kanapoi in- cludes one of the earliest known records of the genus, the other being Langebaanweg (Hendey, 1981). The present range of E. rufescens (Peters, 1878) (spectacled or long- eared elephant shrew) includes thickets in South Turkana (Coe, 1972). Order Lagomorpha Family Leporidae Leporidae gen. and sp. indet. Dental and postcranial remains of lagomorphs are known from Nzube’s Mandible Site and several other sublocalities at Kanapoi. Lagomorphs are well represented in modern and fossil African faunas. Order Rodentia Winkler (1998) reported a rich microvertebrate fauna as- sociated primarily with the locality that yielded the holo- type mandible of Australopithecus anamensis (Nzube’s Mandible Site). Recovered materials included cranial and postcranial remains of fish, amphibians, reptiles, birds, and mammals. The microvertebrates from Nzube’s Man- dible Site likely represented a concentration of owl pellets, with some attritionally incorporated taxa. The mamma- lian specimens were dominated by a small species of the gerbil Tatera Lataste, 1882, and by murine rodents. There is also a new species of ground squirrel [Xerus) Hemprich and Ehrenberg, 1832, and a new dendromurine genus that appears to be the sister taxon to Steatomys Peters, 1846, (fat mice). Although the Kanapoi rodents have yet to be studied in detail, the fauna appears most similar to that from the lower Shungura Formation of the lower Omo Valley (Wesselman, 1984). At the generic level, the Kan- apoi rodents are strikingly similar to those present today in the South Turkana region (Coe, 1972). Based on a com- parison with modern analogues, the Kanapoi small mam- mals suggested a relatively dry climate and open habitat. Received 26 December 2002; accepted 23 May 2003. Ri4rmll9lli r I 'k-cp .*r>j V-'-i ' .<*r#.'-.. .4!'- .>'i I .n, v;' ; Vt»>- -yr 1 0'y “a m W,! ■ » ’* > ' <, k:-:iL S^ i. ■' ' ■ ' 'c • \ ■ 1: '-:' ''*^V • “‘-" ; "k ' ." . r*" •*' ■;Y.’-' ' ; '■ , ■ A ■ '...' - u''"^ ’ i w$ - ,r>-. ^ 1,^,. 'v^V ■*t .4 L* : a ■■ v-^-'. ’ ■: - ‘r f A'^ ^tA T- '^'- ®U' ‘ Ml h m>W: 0'-'.r -Vj -m> W’,-; Carnivora from the Kanapoi Hominid Site, Turkana Basin, Northern Kenya Lars Werdelini ABSTRACT. Kanapoi is the earliest Pliocene site yet described in eastern Africa with a substantial carni- voran record. It includes eight species in as many genera, representing five families. The material is dom- inated by the hyaenid Parahyaena howelli n. sp., but also includes a new Enhydriodon species, E. ekeca- man, the lutrine cf. Torolutra sp., the felids Dinofelis petteri, Homotherium sp., and Felis sp., the herpestid Helogale sp., and the viverrid Genetta n. sp. The Kanapoi Carnivora includes the remains of the first post- Miocene radiation of endemic African Carnivora. INTRODUCTION The earliest Pliocene (here taken as ca. 5. 2-4.0 Ma) has provided relatively few eastern African locali- ties containing mammalian fossils compared with the million years that followed. Most of the local- ities that do exist from this time interval have either few carnivores associated with them or are as yet undescribed. Thus, the Apak Member at Lothagam includes only four carnivore taxa (Werdelin, 2003), the Lonyumun Member of the Koobi Fora Forma- tion three (Werdelin and Lewis, unpublished), the Kataboi Member of the Nachukui Formation and Kosia (also West Turkana) also three (personal ob- servations). Outside of Kenya, localities of this time period from Uganda have also yielded few carni- vores (Fetter et ah, 1991). A richer site is Aramis, in Ethiopia, although the carnivores there have yet to be described (Howell, personal communication). Kanapoi, with its somewhat larger sample of car- nivores, thus adds significantly to our knowledge of carnivoran evolution in the earliest Pliocene of east- ern Africa. The bulk of the material discussed herein was obtained by the National Museums of Kenya ex- peditions to Kanapoi in the early 1990s. However, a few carnivore specimens were recovered by the earlier American expeditions. These have been mentioned a few times in the literature under var- ious guises (Behrensmeyer, 1976; Savage, 1978; Howell and Fetter, 1980; Turner et ah, 1999). In particular, Behrensmeyer (1976) lists Enhydriodon n. sp.. Hyaena sp., and Machairodontinae indet. as present at Kanapoi. Of these, the first is here still referred to Enhydriodon Falconer, 1863, the second to Parahyaena Hendey, 1974, while the third is here listed as Carnivora indet., as it cannot be de- termined if the specimen belongs to Dinofelis Zdansky, 1924, or Homotherium Fabrini, 1890. The remaining taxa described below are new to the Kanapoi fauna. 1. Department of Palaeozoology, Swedish Museum of Natural History, Box 50007, S-104 05, Stockholm, Swe- den. Contributions in Science, Number 498, pp. 115-132 Natural History Museum of Los Angeles County, 2003 CONVENTIONS AND ABBREVIATIONS Kanapoi fossils in the collections of the National Museums of Kenya are formally accorded the ac- ronym KNM-KP before the accession number; to save repetition, this acronym is omitted from the lists of Kanapoi specimens and abbreviated to KP in the body of the text. Tooth abbreviations in upper case (I, C, P, M) indicate upper teeth; those in lower case (i, c, p, m) indicate lower teeth. Hence Ml is the upper first molar, p4 is the lower fourth premolar. The follow- ing abbreviations are used in the text and tables: a-p Dist E LpP, MC MT max min Prox Sust transv W WaP" WblP^ anteroposterior distal anatomical length (long bones); mesio- distal length (teeth) length of main cusp of P4 metacarpal metatarsal maximum minimum proximal sustentaculum transverse buccolingual width anterior width (including protocone) of P4 minimum blade width of P^ SYSTEMATIC DESCRIPTION Order Carnivora Family Mustelidae Generally speaking, mustelids are rare in the fossil record of eastern Africa. For the most part, this may reflect a preservational bias against smaller species of Carnivora. However, those localities that are particularly rich in smaller Carnivora also differ among themselves in the richness and diversity of their mustelids. Thus, Hadar is rich in small mus- telids (personal observations), while Olduvai has fewer taxa, despite the smaller Carnivora being well represented at the latter site (Fetter, 1973). 116 ■ CS 498, Harris and Leakey: Kanapoi Figure 1 Enhydriodon n. sp., A, KP 10034 A, right P4 in occlusal view; B, KP 10034B, right ml in (top) occlusal and (bottom) lingual view; C, KP 10034C, right Ml in occlusal view. In A and C, anterior is to the right; in B, anterior is to the left / This variation may reflect real differences in paleo- ecology between the sites. Nevertheless, the major- ity of mustelids found in the eastern African fossil record are larger species of the genera Mellivora Storr, 1780, Enhydriodon and Torolutra Fetter, Pickford, and Howell, 1991. Kanapoi is no excep- tion to this pattern, with both of the latter genera represented. Enhydriodon Falconer, 1863 The genus Enhydriodon encompasses a number of species of Enhydrini (sea otters) of large to very large size. The genus is known from the Siwaliks of Pakistan and India, where it was first described (Lydekker, 1884; Willemsen, 1992). However, the main diversity within Enhydriodon is found in Af- rica, with several species of varying size known from a number of localities dating between ca. 6.0 Ma (Lukeino Formation; Pickford, 1978) and ca. 2.5 Ma (Omo, Shungura Formation Members E+F; Howell and Fetter, 1976). The Kanapoi ma- terial represents one of the earlier members of this lineage. Enhydriodon ekecaman sp. nov. (Figure 1) Enhydriodon n. sp. Behrensmeyer, 1976 Enhydriodon pattersoni Savage, 1978 (nomen nu- dum) Enhydriodon pattersoni Turner, Bishop, Denys, and McKee, 1999 (nomen nudum) DIAGNOSIS. Differs from Siwalik Enhydriodon in its smaller size. Differs from E. africanus Strom- er, 1931 (the other described African Enhydriodon) by having a broader ml with a more open talonid basin and relatively smaller paraconid, more robust and larger Ml hypocone(s) and more stoutly built i P4. HOLOTYPE. KNM-KP 10034, A = right P4 (Fig. lA), B = right ml (Fig. IB), C = right Ml i (Fig. 1C), D = right C). Written documentation re- i' garding the association in the field of these four specimens was provided by Dr. J. C. Barry (in cor- i respondence). Given this information, it is reason- i able to assume that all four belonged to a single j individual and therefore the entire hypodigm is chosen as the holotype of this new species. ETYMOLOGY. After the Turkana “ekecaman,” | meaning fisherman. The diet of this animal was i composed at least in part of fish. KANAPOI MATERIAL. 10034; associated teeth , (holotype). MEASUREMENTS. KP 10034A, buccal length = 16.5; KP 10034B, total length = 21.2, trigonid length = 11.5, talonid length = 9.3, trigonid width = 13.3, talonid width = 13.5; KP 10034C, buccal length = 12.1, lingual length = 15.8, anterior width = 19.8, posterior width = 19.0. Measure- ments defined as in Willemsen (1992). The right upper carnassial, KP 10034A (Fig. lA), is missing the protocone and most of the hypocone and posterior shelf. The parastyle is short and low but has a distinct anterior cusp. The paracone is high and much the largest buccal cusp. There are blunt crests leading anteriorly to the parastyle, an- terolingually toward the protocone and posteriorly toward the metacone. The latter crest is interrupted by a shallow valley before it meets the metacone. The latter cusp is set far posteriorly and is much lower and smaller than the paracone. There was an extensive basin formed posterobuccally to the hy- pocone and the posterobuccal corner and posterior margin of the tooth lie at nearly right angles to the anteroposterior axis of the tooth. The cingulum ex- i tends along the entire preserved part of the tooth except for the posterobuccal corner. The right lower carnassial, KP 10034B (Fig. IB) is low, broad, and stoutly built. The three trigonid cusps are all low and pyramidal with their major axes directed either anteroposteriorly (protoconid, metaconid) or transversely (paraconid). In occlusal view, the paraconid is the smallest of the three cusps but all three are worn down to about the same height. The apex of the paraconid is set slight- ly lingual to the middle of the cusp and it is also slightly broader lingually than buccally. The apex of the protoconid is set slightly anterior to the mid- dle of the cusp. There is a blunt crest that runs down the lingual side of this cusp toward the tal- onid, making the cusp almost diamond-shaped, though with a gently curved buccal side. Anteriorly, the protoconid is separated from the paraconid by a shallow valley. Posteriorly, the accessory proto- conid cusp is very weakly developed, merely form- ing a low bulge on the posterior face of the cusp. The metaconid is set slightly posterior to the pro- toconid. Its apex is at the middle of the cusp. The Werdelin: Carnivorans ■ 117 cusp is triangular in occlusal view, with the apex of the triangle directed toward the buccal side. Ante- riorly, the metaconid is separated from the para- conid by a deep valley. Between the trigonid cusps is formed a shallow, flat basin that is about 3 mm wide and 2 mm long. The talonid basin is low and wide. The entoconid is well developed but low. In occlusal view, it is about the size of the paraconid. None of the other talonid cusps is well defined. In- stead, they form a low, broad, gently undulating ridge surrounding the central basin. The buccal cin- gulum is strong and extends from the anteriormost part of the tooth to the posterior end of the hypo- conid. The right Ml, KP 10034C (Fig. 1C), is broad, low, and robust, with a very broad lingual basin. The paracone is small, and buccal to it, there is a large parastyle, which is in fact larger than the paracone itself. The paracone is low and worn and connected to the metacone by a narrow valley. The metacone is considerably higher and larger than the paracone and has a pyramidal base. It is set at the posterior margin of the tooth. Lingual to the meta- cone and separated from it by a narrow valley there is a metaconule that is almost as large as the meta- cone itself. This cusp is also set at the posterior margin of the tooth. Lingual to the metaconule there is a wear facet for occlusion with m2. This wear facet is confined to the posterior margin of the tooth. At the posterolingual corner of the tooth there is a large swelling of the cingulum, forming a low cusp. The protocone is double, with one cusp set slightly anterobuccal to the other. These two cusps are of about equal size. The cingulum runs around the entire tooth, except anteromedially, where it is worn down by the wear facet for P4, and posterolingually, where the aforementioned m2 wear facet is located. The cingulum is otherwise weakest around the metacone. The right C, KP 10034D, is short and straight, with a crown that is only slightly longer than it is wide. The tip is worn flat. There is a strong medial cingulum but no lateral one. DISCUSSION. The material of Enhydriodon from Kanapoi is limited, but can nevertheless be distinguished from other African Enhydriodon of similar age. The lower carnassial, for example, dif- fers from that of E. africanus from Langebaanweg among other features in being shorter and relatively wider, in the somewhat more widely spaced trigo- nid cusps, and in the flatter and more open talonid basin (cf. Hendey, 1978b). The development of the posterolingual cusplets is more pronounced in the Kanapoi form. The Ml differs from a partial Ml from Kosia (West Turkana, ca. 4.0 Ma) in that the anterolingual corner of the Kanapoi Ml forms a nearly right angle, while the homologous angle of the Kosia tooth is closer to 135 degrees (personal observations). The Kanapoi Enhydriodon further differs from younger Enhydriodon from formations such as Hadar, Koobi Fora, and Shungura in its much smaller size. Hence, we can infer that the Kanapoi Enhydriodon belongs to a hitherto un- known species of that genus, a species that is not at present known from any other site. All four En- hydriodon teeth from Kanapoi originate from the American expeditions (1966-72) and were collect- ed in 1967. Torolutra Petter, Pickford, and Fiowell, 1991 Torolutra is a genus of otters similar in size to the living European otter Eutra (Linnaeus, 1758), or slightly larger. Only a single species, T. ougan- densis, has been described (Petter et ah, 1991). This material is from Nyaburogo and Nkondo in Ugan- da, while the species has also been tentatively iden- tified from Ethiopia in the Usno Formation of the Omo Group. These localities bracket Kanapoi in age. cf. Torolutra sp. KANAPOI MATERIAL. 30155, right 13, left and right P4 fragments, premolar fragment, ml talonid fragment, proximal left radius fragment, proximal right tibia fragment, humerus shaft fragment, par- tial cervical, thoracic and caudal vertebrae, assort- ed indeterminate fragments. The 13 is strongly recurved and has a short crown. The enamel reaches farther down on the lateral than on the medial side. There is a promi- nent cingulum surrounding this tooth. This cingu- lum is best developed on the medial side. The root of 13 is relatively straight. The P4 preserves the me- tastyle, a partial paracone and a part of the lingual protocone shelf. There is no carnassial notch. The protocone shelf is not as long as in Enhydriodon, being instead more similar to that of Eutra Brisson, 1762. The ml preserves the posterior part of the talonid, with a prominent hypoconid and low bumps indicating entoconulid and entoconid. The cingulum is prominent around the posterior end of the tooth. The proximal radius fragment is small and broken. The proximal tibia fragment shows only parts of the proximal articular surface. All the vertebrae are relatively robust, the proximal caudal vertebra extremely so, while the cervical vertebrae are relatively much smaller. DISCUSSION. These specimens compare well, as far as comparisons can be made, with specimens of Torolutra ougandensis described by Petter et al. (1991). They are a little larger than the specimens from Uganda, but match the Omo specimens fig- ured by those authors quite well. Family Hyaenidae Hyenas are common elements in eastern African Mio-Pliocene faunas. As in Eurasia, there is an ex- tinction event at the end of the Miocene that elim- inates the dominant ‘dog-like’ hyenas that are still present at localities such as Lothagam (Werdelin, 2003). At Pliocene localities, hyenas are mainly rep- resented by relatives of the living hyenas. These Pli- ocene forms had adaptations to a scavenging life- 118 ■ CS 498, Harris and Leakey: Kanapoi style similar to those of living hyenas, but less ac- centuated. At Laetoli, hyenas are abundant, and in- clude fossil relatives of three of the four living hyenas (Werdelin, unpublished). One of these spe- cies is also present at Kanapoi. Parahyaena Hendey, 1974 There has been some debate regarding the validity of Parahyaena as a genus distinct from Hyaena (Werdelin and Solounias, 1991; Jenks and Werde- lin, 1998). On the one hand, today these are two monospecific sister taxa, and from this perspective, generic distinction may be deemed unnecessary. On the other hand, the split between the two, as in- ferred from both molecular and paleontological ev- idence, extends down into the Miocene, and most generic splits among carnivores are of about this age or even younger. From this perspective, generic separation is valid. Here I follow the latter path, as I believe that geological age is the only criterion with which to resolve ranking issues when these become critical. Until now, Parahyaena was known from the single extant species F. brunnea (Thun- berg, 1820), which has a limited distribution and geological age (Jenks and Werdelin, 1998). A single record extends the range of F. brunnea into eastern Africa in the middle Pleistocene (Werdelin and Bar- thelme, 1997). This makes the presence of an an- cestral species of Parahyaena at Kanapoi highly sig- nificant. Parahyaena hoivelli sp. nov. (Figures 2-5) Hyaena sp. Behrensmeyer, 1976 Pachycrocuta sp. Howell and Fetter, 1980 DIAGNOSIS. Hyaenid of large size (larger than Hyaena hyaena Linnaeus, 1758). Mandibular ra- mus robust, premolars moderately developed for bone cracking (weaker than in Pliocrocuta Kretzoi, 1938, Pachycrocuta Kretzoi, 1938 and Crocuta Kaup, 1828). Masseteric fossa clearly subdivided by a ridge into a ventral and dorsal part (unlike H. hyaena). Metastyle of P4 clearly longer than para- cone (unlike Ikelohyaena abronia (Hendey, 1974) and H. hyaena). Metacarpals short and robust (un- like all modern hyenas). HOLOTYPE. KNM-KP 30235, associated par- tial skeleton (Figs. 2, 3). ETYMOLOGY. After Dr. F. Clark Howell, lead- ing scholar of African fossil carnivores. KANAPOI MATERIAL. 10033, complete right mandibular ramus with c-ml (Fig. 2A, Hyaena sp. in Behrensmeyer 1976; Pachycrocuta sp. in Howell and Fetter 1980); 29249, right mandibular ramus fragment with p4; 29280, proximal fragment of left radius; 29290, right mandibular ramus fragment with p4, distal left radius fragment, right radius shaft fragment; 29293, left distal radius fragment; 29294, right mandibular ramus fragment with p2; 29296, right mandibular ramus fragment with par- tial alveolus for c, alveolus for p2, roots of p3, an- terior root of p4; 29297, left mandibular ramus fragment with roots of p4 and anterior root of ml; 29299, proximal, shaft, and distal fragments of right femur; 29301, left mandibular ramus frag- ment with p4 and anterior root of ml, separate p3; 29302, left P4; 30229, right femur lacking distal end, proximal right tibia; 30234, associated partial left forelimb with left ulna lacking olecranon, left radius, left humerus lacking proximal articulation, left scapholunar, left magnum, left pisiform, left un- ciform; 30235, associated partial skeleton including right mandibular ramus with i2 and c-ml (Fig. 2B), left distal humerus fragment, right proximal and distal humerus fragments, right radius (Figs. 3A-B), right calcaneum, right tibia lacking distal articula- tion, right ulna lacking distal articulation, frag- ments of the right and left scapulae, left cuboid, right and left navicular, right scapholunar, left lat- eral cuneiform, damaged right lateral cuneiform, right unciform, Pright patella, left MC II lacking distal articulation, right MC II lacking proximal ar- ticulation, proximal part of right MC III, left and right MC V lacking proximal articulations, right MC I, distal part of right MT II, right MT IV lack- ing proximal articulation, left and right MT V lack- ing proximal articulations, proximal, middle and distal phalanges including proximal phalanx of dig- it 1 of the manus, pisiform, sternebrae, and cervi- cal, thoracic, lumbar, and caudal vertebral frag- ments; 30272, associated partial skeleton with right mandible fragment with c, damaged p3, roots of p2 and p4, left mandible fragment with roots of p2~ p3, right maxilla fragment with C root, PI alveolus, nearly complete P2, roots of P3, anterior root of P4, left maxilla fragment with alveolus for 13, dam- aged C, roots of P1-P2, anterior root of P3, right zygomatic, proximal end of right and left scapulae, fragment of distal right humerus shaft, right tibia, proximal, shaft and distal fragments of left tibia, pelvic fragment, left navicular, proximal fragment of left MT V, tuber fragment of left calcaneum, par- tial distal right femur, proximal and distal frag- ments of left femur, fragment of left pisiform, prox- imal end of right MC III, pathological left MT II, distal end of left MT III, right MT III, proximal fragment of right and left MT IV, proximal end of right MT V, proximal phalanx of left manus digit 3, proximal phalanx of right manus digit 4, prox- imal phalanx of right manus digit 5, middle pha- lanx of right manus digit 4, middle phalanx of right and left pes digit 4, middle phalanx of right and left pes digit 5, sternebrae, fragments of cervical, thoracic, lumbar, and caudal vertebrae; 30306, left distal femur fragment; 30463, right mandibular ra- mus fragment with roots of p2-p4; 30482, associ- ated complete left MC III-V (Fig. 4); 30487, prox- imal part of left MT III; 30495, proximal fragment of MT II; 30534, proximal right MT III; 30536, left p3; 30540, right 13; 30541, right lower canine; 30544, left mandibular ramus fragment with p3 and ml, p2 and p4 crowns separate; 31734, prox- Werdelin: Carnivorans ■ 119 imal right ulna; 31735, proximal left ulna lacking olecranon; 32538, right ml; 32550, left P3; 32552, mandible fragments with associated left c and p3- ml; 32813, proximal right ulna fragment; 32822, left lower canine; 32865, postcranial fragments in- cluding a fragment of a proximal MC II, distal me- tapodials, a distal humerus fragment, and vertebral fragments. MEASUREMENTS. See Tables 1 and 2 The following is a composite description of the material. The craniodental material shows limited variability, except in size, but where there is varia- tion, this is noted. There is limited duplication be- tween postcranial elements, which allows for a very limited grasp of variation in the taxon, but does mean that a significant proportion of the skeleton of Parahyaena hoivelli is actually known (Fig. 5). Some comparisons with representative morpholo- gies of extant Hyaena hyaena are made. SKULL AND UPPER DENTITION. The skull is represented only by some very small fragments from KP 30272, which unfortunately are too small and damaged to provide any information about the morphology of the species. This specimen has some very damaged teeth and tooth roots that indicate that the upper canine was slightly larger than the lower, that the PI was small and single rooted, and that P2 was considerably smaller than P3. All these features are normal in hyenas and the only fact of interest is the presence of PI. The 13 is represented by specimen KP 30540, which is worn, but shows the distinctive derived hyaenid subcaniniform mor- phology of this tooth. The P3 is also represented by a damaged specimen, KP 32550. This tooth probably had a small anterolingual accessory cusp, though damage in this area makes this somewhat uncertain. The main cusp is high and conical, as is typical of derived hyenas, and the posterior acces- sory cusp is substantial, though precisely how large cannot be determined due to specimen damage. The tooth is similar to P3 of modern Hyaena, but the main cusp is larger and stouter. The upper carnas- sial is represented by the isolated tooth KP 29302. It is typically hyaenid in morphology, with substan- tial parastyle and protocone, a large and relatively narrow paracone and a long metastyle. The par- astyle is robust with a distinct anterior ridge leading down to an anterior cingulum with a hint of a pre- parastyle. The protocone is large but low and set in line with the anterior margin of the parastyle. The paracone is tall and trenchant and the metas- tyle long. There is a lingual cingulum that reaches from the posterior root of the protocone to the pos- terior quarter of the metastyle. MANDIBLE AND LOWER DENTITION. The mandible and associated material are known from sixteen specimens, most of which are fragmentary and damaged, but include two nearly complete rami, KP 10033 and KP 30235A (Figs. 2A-B). The mandible increases gradually in depth from anteri- or to posterior and is deepest just posterior to ml. Posterior to this point, the ventral margin of the mandible rises in a shallow S shape to the angular process. The mandibular condyle is relatively slen- der in comparison with modern Hyaena and con- sists of two semidistinct areas, a lateral, higher one and a medial, lower one. The latter does not taper medially as is the case in Hyaena. The coronoid process is tall and slender compared with that of modern Hyaena, and has a distinct backward tilt. The anterior margin of the coronoid process is steeply S shaped. The masseteric fossa is deep and flat. The ventral part (insertion area for the M. mas- seter intermedins) is delimited by a strong ridge and is set distinctly lateral to the more dorsal parts of the masseteric fossa (insertion area for the M. mas- seter profundis). This is in contrast with the situa- tion in Hyaena, where the ridge separating these two insertion areas is lower and less distinct and the two areas are located in the same vertical plane. There is a single, large mental foramen situated be- neath p2. In KP 10033, the symphysis is broken and all the incisors lost. The lower canine is broken and chipped, but can be seen to have been a rela- tively robust tooth whose anteroposterior axis is angled relative to the main axis of the ramus. The diastema is of about the same length as in Hyaena. The cheek tooth row curves gently to buccal from p2 to p3, then curves gently back from p4 to ml, much as in modern Hyaena. Unlike the latter, how- ever, the p2 of KP 10033 is set at an angle to the main axis of the ramus. The i2 is present only in KP 30235A. It is heavily worn, but is clearly longer (anteroposteriorly) than wide (mesiodistally). The tip is worn flat, though the wear facet is angled from distal (higher) to me- sial. On the mesial side, the wear has reached the enamel-dentine juncture. The lingual face is also worn and there is an angle of about 70° between the apical and buccal wear facets. The p2 is the most variable tooth in both size and shape. There is no anterior accessory cusp, but a small swelling at the anterior end of the tooth is present. The main cusp is narrow and conical, with slightly convex anterior and posterior margins. The posterior ac- cessory cusp is low and narrow. The posterior shelf of this tooth is variable in width. In KP 10033, it is narrow in comparison with modern Hyaena, which has a small shelf that is not present in this Kanapoi specimen. In KP 30235A and KP 30544, p2 is much broader posteriorly and more similar to the condition in Hyaena. In KP 29296, the p2, as judged from the alveolus, must have been notice- ably shorter than in the other specimens that pre- serve traces of this tooth. The p3 has no anterior accessory cusp. Instead, the anterior margin of the main cusp is formed into a low crest that reaches the anterior end of the tooth. The anterior and pos- terior margins of the robust, conical main cusp are slightly convex. The posterior accessory cusp is low and round and set centrally in a posterior cingulum shelf. The tooth is variable in size and shape, though not to the extent seen in p2. KP 29301 has a p3 that is nearly identical to that of KP 10033, 120 ■ CS 498, Harris and Leakey: Kanapoi Figure 2 A, Parahyaena howelli, KP 10033, right mandibular ramus in (top to bottom) buccal, lingual, and occlusal view; B, Parahyaena howelli, KP 30235A, right mandibular ramus in (top to bottom) lingual, buccal, and occlusal view whereas in KP 30235A and KP 30544, the p3 is narrower and has a more distinct waist. KP 30536 and KP 32552 are intermediate in morphology. The tooth is broadly similar to p3 in Hyaena except for the absence of an anterior accessory cusp. The p4 has a small, round anterior accessory cusp ap- pressed to a narrow, conical main cusp with more or less straight anterior and posterior margins. The posterior accessory cusp is relatively high and tren- chant. The posterolingual part of p4 has been dam- Werdelin: Carnivorans ■ 121 Figure 2 Continued aged and it is not possible to determine the width of the posterior part of the tooth. The lower car- nassial is long and relatively low. The paraconid is slightly longer and wider than the protoconid. The metaconid is very small but is distinctly developed and the talonid has two cusps, presumably the hy- poconid and entoconid. Compared with modern Hy- aena, the tooth is relatively much longer, the meta- conid smaller, and the talonid relatively shorter. FORELIMB. The humerus is known from KP 30235 (proximal and distal pieces) and KP 30234 (shaft and distal articulation). The proximal frag- ment is too worn for detailed comparisons with Hyaena, but is larger and appears relatively nar- rower. The distal articulation is transversely broad- er than in modern Hyaena, but is relatively more slender anteroposteriorly. There is a large supra- trochlear foramen present in KP 30235. The radius is known from KP 30235 (Figs. 3A-B) and KP 29280. It is in general very similar to that of Hyaena, but is shorter for the same robusticity. The grooves for the extensor digitis communis, ex- tensor carpi radialis, and abductor pollicis longus are all more deeply incised than in Hyaena. The ulna is known from several fragmentary specimens, KP 30234, KP 30235, KP 31734, KP 31735, and KP 32813. The first two are the most complete and indicate that the ulna of this taxon was slightly shorter but more robust than that of Hyaena. The shaft is more rounded than in Hyae- na, the triceps groove is wider, the ridge on the cra- nial surface of the olecranon is narrower, the at- tachment area for the flexor carpi ulnaris is less distinct, and the pit beneath the radial notch is shal- lower. In addition, the rugosity for the articulation with the radius begins more proximally on the 122 ■ CS 498, Harris and Leakey: Kanapoi A Figure 3 Parahyaena howelU, KP 30234, right radius in A. shaft, while the groove for the abductor pollicis longus is much more distinct than in Hyaena. Several of the carpals are known from different specimens. The scapholunar is known in specimens KP 30234 and KP 30235AC. It is slightly larger and more slender anteroposteriorly than that of modern H. hyaena. Specimen KP 30235AC is bro- ken on the medial side, but in KP 30234, it can be seen that the articular face for the radius extends down onto the medial rim, while in modern H. hy- aena, there is a much more distinct ridge limiting the radial articulation to the dorsal side of the bone. This may indicate greater mobility of this ar- ticulation in the fossil form. In addition, the sulcus B anterior and B, posterior view for the flexor carpi radialis tendon is deeper and bounded medioventrally by a more prominent ridge than in modern H. hyaena. The magnum is known in KP 30234. It is shorter and wider than that of modern H. hyaena but is morphologically very sim- ilar in all other respects. The unciform is also known from KP 30234. It is more robust than that of modern H. hyaena and has a more open articular face for MC III. The pisiform is known from KP 30234. It is larger than that of modern H. hyaena, but aside from size is practically indistinguishable from it. All the metacarpals except MC II are known from complete specimens. MC I, known from KP Werdelin: Carnivorans ■ 123 Figure 4 Parahyaena hoivelli, KP 30482, associated left MC III-V in dorsal view. Scale = 50 mm 30235P, is a substantial element associated with at least one phalanx. This is corroborated by KP 39235V, which is tentatively identified as the prox- imal phalanx of the same digit. This phalanx bears a large articular surface for an ungual phalanx, which has not been identified in the material. It is far larger than that of any extant hyaenid, though relatively smaller than its counterpart in Ikelohyae- na ahronia Fiendey, 1974, from Langebaanweg, a species that is otherwise smaller than the Kanapoi form. The second metacarpal is known from a dis- ! tal and a proximal fragment, KP 30235AF4 and AO, respectively. This metacarpal is more robust than MC II in modern H. hyaena. Metacarpals III to V are associated in specimen KP 30482 (Fig. 4) and MC III is in addition known in KP 30235AG and KP 30235BH and m KP 30272. The third metacarpal has larger proximal and distal articular surfaces than MC III in modern H. hyaena, while the shaft is more robust but markedly shorter than in the extant species. The plantar side of the prox- imal articular surface is narrower relative to the dorsal side than in the modern species. The fourth metacarpal is, like the third, shorter and more ro- bust than that of modern H. hyaena. The two spe- cies are similar in their MC IV morphology, but the proximal articulation in KP 30482 is somewhat more triangular in shape, with a broader dorsal and narrower plantar side than in the modern form. The fifth metacarpal is also more robust and short- er than that of modern H. hyaena, the difference being more accentuated in this element. The prox- imal articular surface with MC IV is set less obliquely and more directly anteroposteriorly than in modern H. hyaena and is also very wide com- pared with the condition in the modern species. F4INDLIMB. The femur is represented by vari- ous fragments from specimen KP 30272, including a partial distal femur KP 30272N, proximal frag- ments of specimen KP 30229, distal fragments, specimen KP 30306, and proximal, shaft and distal fragments, KP 29299. These fragments suggest a femur that is somewhat larger and more robust than the corresponding element in modern H. hy- aena, but otherwise do not show any distinguishing features of note. The tibia is known from fragments from the par- tial skeletons KP 30235D and KP 30272P, Q, as well as from KP 30229. In general, the tibia is not a very diagnostic bone in Fiyaenidae and this is true in the present case as well, the fossil specimens only being distinguished from modern H. hyaena by their greater size and by the slightly greater devel- opment of the medial malleolus of KP 30272Q. The navicular is known from KP 30235AE and AD, and KP 30272F. This bone is generally similar to that of modern H. hyaena, but is slightly larger. It differs in that the plantar process is wider than high, the reverse of the condition in H. hyaena. The process for the separation between the articulation with the cuboid and the plantar side of the bone is less prominent than in H. hyaena. The lateral cuneiform is known only from KP 30235BJ and BP. It is larger than the corresponding bone in modern H. hyaena, apparently relatively more so than other tarsals and carpals reported here. The proximal articular surface of the fossil is more deeply indented laterally and medially than in H. hyaena and, in addition, the distal articular sur- face (for MT III) is concave in the fossil rather than slightly convex as in H. hyaena. The cuboid is known from KP 30235W. It is the tarsal that differs most from the corresponding el- ement in modern H. hyaena. The medial side of the proximal articular surface of the fossil has a medial extension that probably buttressed the medial part of the sulcus for the M. peritoneus longus tendon. This sulcus is shallow and nondescript in modern H. hyaena, deep and well developed in KP 30235W. The cuboid of Crocuta crocuta (Erxleben, 1777), on the other hand, is short and square with a deep, narrow sulcus. The second metatarsal is known from KP 30495, 30235AM, and 30272AB. The two former are proximal ends that are more robust than MT II in modern H. hyaena but otherwise too worn for 124 ■ CS 498, Harris and Leakey: Kanapoi Figure 5 Farahyaena howelli, known skeletal parts of P. howelli (gray) superimposed on a skeleton of a dog; skeleton adapted from Evans (1993) meaningful comparisons to be made. The third specimen, KP 30272, is pathological in that the bone appears to have been broken in life and sub- sequently healed. The distal end of this bone is composed of an amorphous mass of secondary bone suggestive of healing. The third metatarsal is known from specimens KP 30272U and KP 30272AO, KP 30534, and KP 30487. These specimens are more robust than the corresponding element in modern H. hyaena, but are otherwise similar except for the proximodorsal articular surface being concave rather than flat to convex. The fourth metatarsal is known from KP 30235AF and KP 30272T and AC. It differs from that of H. hyaena only in its greater size and in the less expanded proximopalmar eminence. The fifth metatarsal is known from KP 30235AK and AL and KP 30272AA and KP 30272AP. In this case, the greater size is the only clear difference from extant striped hyaena. DISCUSSION. KP 10033 was recovered by the American expeditions. It was referred to Hyaena sp. by Behrensmeyer (1976) and to Pachycrocuta sp. by Flowell and Petter (1980). The referral of this species to Farahyaena rests chiefly on the length of the metastyle of P4. This is only known from a single specimen, KP 29302, but can also be inferred from the length of the ml trigonid in re- lation to p4 and ml talonid length. The length of the P4 metastyle is one feature that clearly distin- guishes all extant hyena species. Hyaena hyaena has a short P4 metastyle, of about the length of the paracone or slightly shorter. In C. crocuta, the me- tastyle of P4 is exceptionally long and straight. In Farahyaena brunnea, the metastyle of P4 is longer than that of H. hyaena but shorter than that of C. crocuta. In the present case, the P4 metastyle has the relative length of that of F. brunnea. This con- trasts with the condition in Ikelohyaena abronia, a possible ancestor of H. hyaena (Fiendey, 1978a; Werdelin and Solounias, 1991), in which the P4 metastyle is short, as in its putative descendant. No other features contradict assignment of the Kanapoi hyena to Farahyaena, and rather than posit the ex- istence of a previously unknown hyaenid lineage, I prefer to suggest a link to the living brown hyena. This represents the first direct indication of the an- cestry of the brown hyena, as all other known fossil Farahyaena fit comfortably within the extant spe- cies (e.g., Hendey, 1974). Family Felidae Felids are common elements in the fossil faunas of eastern Africa. Both Machairodontinae and Felinae are present throughout the Plio-Pleistocene, but up to about 1.5 Ma, the former are by far the more common in the fossil record. Table 1 Dental measurements of Par a hyaena howelli n. sp.; measurement parameters as in Werdelin and Solounias (1991) Werdelin: Carnivorans ■ 125 oo a^ o VD OO rn ^ (N O h- ^ On O ^ r-H (N o o O o r\ OO O T-H ,-1 rN| T-i r\| ON Tf o^ rl O r/N K ^ rl ^ ^ (N ^ O ro UT) ON O O ON rsj ro CX, C'J CX ro CX ^ CX I— 1 CX 1-1 cx ^ c cx S ^ § 2 ^ E: _ £ kJ hJ kJ -o kJ a £ cxi-ii-i CK-iH-iK-irt-OK-iH-i 126 ■ CS 498, Harris and Leakey: Kanapoi Table 2 Postcranial measurements of Parahyaena howelli n. sp. KP30235 KP30234 KP29280 KP30482 KP30534 KP29293 KP32865 Humerus DistW 46.6 46.1 Radius L 215.5 Radius ProxW max 22.3 a21.4 Radius ProxW min 15.7 16.0 Radius DistW transv 31.9 32.4 Radius DistW a-p 20.9 20.7 MC II ProxW transv 10.6 11.8 MC II ProxW a-p 14.8 14.5 MC III L 91.5 MC III ProxW transv 13.5 12.5 MC III ProxW a-p 14.8 15.2 MC III DistW transv al4.3 MC III distW a-p 11.5 MC IV L 89.3 MC IV ProxW transv 12.2 MC IV ProxW a-p 14.6 MC IV DistW transv 13.6 MC IV DistW a-p 12.5 MC V L 75.9 MC V ProxW transv 18.6 (14.0) MC V ProxW a-p 15.6 (13.9) MC V DistW transv 15.8 (12.9) 13.1 MC V DistW a-p 11.9 12.3 KP30272 KP29299 KP30229 KP30235 Femur HeadW 24.3 24.7 Femur ProxW 51.4 a53.3 Femur DistW 41.4 a42.4 Tibia ProxW 44.0 43.7 Tibia DistW 23.9 Calcaneus Head W transv 14.4 Calcaneus HeadW a-p 20.5 Calcaneus tuber Wmax 20.4 Calcaneus tuber Wmin 11.0 Calcaneus SustW 24.3 KP30495 KP30487 KP30272 MT II ProxW transv a9.9 MT II ProxW a-p al2.5 MT III ProxW transv 11.5 10.8 MT III ProxW a-p 17.6 18.3 MT IV ProxW transv 9.8 MT IV ProxW a-p 16.5 MT V ProxW transv 8.2 MT V ProxW a-p 13.1 Dinofelis Zdansky, 1924 Species of the genus Dinofelis are among the most common Felidae in the fossil record of eastern Af- rica. The earliest record there and possibly the ear- liest anywhere is from Lothagam, where material referred to the genus is known from all members (Werdelin, 2003). The genus is subsequently pre- sent at most Plio-Pleistocene sites in eastern Africa until its last occurrence at Kanam East (ca. 1.0 Ma; Werdelin and Lewis, 2001). Dinofelis petteri Werdelin and Lewis, 2001 (Figure 6) KANAPOI MATERIAL. 30397, complete right mandibular ramus with c~ml (Fig. 6); 30542, distal left ulna; P30429, P4 metastyle. MEASUREMENTS. KP 30397, Lc = 13.9, Wc = 9.7, Lp3 = 13.1, Wp3 = 7.3, Lp4 = 20.6, Wp4 = 9.7, Lml = 23.2, Wml = 10.6. Measurement parameters defined in Werdelin and Solounias Werdelin: Carnivorans ■ 127 100 mm Figure 6 Dmofelis petteri, KP 30397, right mandibular ramus in (top to bottom) buccal, lingual, and occlusal view (1991). The horizontal ramus is low, but broad, with a noticeable thickening of the corpus. The depth is about the same throughout. The symphysis is deep and short and nearly vertically oriented, producing a small anteromedial chin. There are two mental foramina, one below the diastema between the canine and p3 and one beneath the anterior root of p3. Both are set low on the ramus. The masseteric fossa is deep and the masseteric foramen large, while the coronoid process is relatively short anteroposteriorly. The condyle is thickest medially and tapers gradually toward the lateral end. The 128 ■ CS 498, Harris and Leakey: Kanapoi angular process is robust and angled ventrally rel- ative to the horizontal ramus. The space for the incisors is very narrow, sug- gesting that they were either staggered or very small. The lower canine is short and robust and angled outward with respect to both the antero- posterior axis of the ramus and the sagittal plane. The diastema is long. The p3 has a small anterior accessory cusp, a low, conical main cusp, and a posterior basin that forms the widest part of the tooth but lacks a posterior accessory cusp. The p4 is long and slender. The anterior accessory cusp is well developed and set far anteriorly, well away from the main cusp. The main cusp is triangular with straight anterior and posterior margins. The posterior accessory cusp is similar in size to the an- terior but set closer to the main cusp. There is a small posterior cingulum cusp and a low lingual cingulum crest, making the posterior basin the wid- est part of the tooth. The lower carnassial is typi- cally felid, with a broad paraconid and narrower and somewhat longer protoconid. There is a mi- nute, posteriorly located talonid. The ml is set in a groove at the posterior end of the horizontal ra- mus. This groove is bounded laterally by the mas- seteric fossa wall and medially by the root of the ascending ramus. The tip of the anconeal process of the ulna is broken and the specimen is somewhat eroded. It is a relatively small, gracile bone compared with later, better known Dinofelis (see Werdelin and Lewis, 2001 for a discussion). DISCUSSION. This and other Dinofelis material from Africa and other regions is described and ex- tensively discussed elsewhere (Werdelin and Lewis, 2001). The Kanapoi material is referred to the spe- cies D. petteri, which is also known from a number of other sites (Allia Bay, Laetoli, Hadar Sidi Hak- oma, and Denen Dora Members, Omo Shungura Members B-F, Koobi Fora Tulu Bor Member, West Turkana Lomekwi Member) in eastern Africa. This gives the species a temporal range of ca. 4.2 Ma (Kanapoi) to 2.3 Ma (Shungura Member E/F). Homotherium Fabrini, 1890 Material that can be referred to Homotherium is relatively ubiquitous at eastern African Plio-Pleis- tocene sites. Unfortunately, much of this material is fragmentary or undescribed. Therefore, the taxon- omy of eastern African Homotherium is confused. Petter and Howell (1988) described a skull from Hadar as Homotherium hadarensis, noting its dif- ferences from Eurasian Homotherium. On the oth- er hand, Harris et al. (1988) described a skull from West Turkana, tentatively affiliating it with Hom- otherium problematicum (Codings et ah, 1976). The latter comparison cannot be maintained, but neither does the West Turkana skull seem to belong to H. hadarensis. African Homotherium requires renewed investigation for the resolution of these problems. 10 mm Figure 7 Homotherium sp., KP 30420, right ml in (top) buccal and (bottom) occlusal view Homotherium sp. (Figure 7) j KANAPOI MATERIAL. 30420, right ml (Eig. j 7); 32558, complete right MC IV; 32820, complete | proximal phalanx; 32882, proximal metatarsal fragment. The lower carnassial is very long and slender. The paraconid is slightly broader than the proto- conid, but the latter is the longer of the two cusps. There is no metaconid and no talonid. The shaft of MC IV is relatively straight and quite rounded in cross-section, widest just below proximal articula- tion and gradually tapering distally. The distal ar- ticulation is about as tall as it is wide. The proximal phalanx is large and robust. The shaft is gently arched. The proximal articulation is broad and low, while the distal articulation is more nearly equal in I height and width, though the width is still some- what greater. Rugose surfaces are prominent on the medial and lateral sides of the shaft. DISCUSSION. All of this material is clearly felid and is too large to represent any taxon other than Homotherium. The ml matches the lower carnas- sial of most other Homotherium in size and pro- portions, though it is distinctly smaller than ml Werdelin: Carnivorans ■ 129 from a Homotherium mandible from Koobi Fora (KBS Member), as well as that of H. problemati- cum from Makapansgat. However, none of the Kanapoi material can be considered diagnostic among species of Homotherium and the material must be left as indeterminate species for the time being. The Kanapoi material represents the hitherto oldest described material of Homotherium in east- ern Africa. Felis Linnaeus, 1758 Fossils of the genus Felis are very rare in the fossil record of eastern Africa. In fact, aside from the Kanapoi record, only a single specimen from the Denen Dora Member of the Hadar Formation can be unequivocally referred to Felis sensu stricto (per- sonal observations). Felis sp. KANAPOI MATERIAL. 30546, fragments of P4 of one or two individuals. This material comes from a small feline, smaller than ""Felis small species” from Laetoli (Barry, 1987). It is the size of the extant F. lyhica Forster, 1780. The main cusp is taller and shorter antero- posteriorly than in the Laetoli specimen. DISCUSSION. Given the fragmentary nature of the material, as well as the almost complete lack of knowledge of fossil African Felis at the present time, it is inadvisable to put a specific name to this material. Family Herpestidae Apart from the notable exceptions of Laetoli and Olduvai, sites that have been excavated for micro- mammals, herpestids are rare in the fossil record of eastern Africa. Because of the lack of screen- washed localities, it is not at present possible to establish whether this is a sampling artifact, wheth- er it reflects a biased sample of localities vis-a-vis environment, or whether it is a real phenomenon. Helogale Gray, 1861 Dwarf mongooses are among the more common herpestids in the Plio-Pleistocene of eastern Africa, with several species described from Laetoli and the Shungura Formation (Wesselman, 1984; Petter, 1987). Helogale sp. KANAPOI MATERIAL. 32826, fragments of a right mandibular ramus with broken p4, damaged ml, roots of m2; 31034, lower canine. This material belongs to a very small carnivore I species. The horizontal ramus is slender but rela- tively deep. In the carnassial, the paraconid is by far the largest and tallest cusp, making up about half of the trigonid in occlusal view. The protoco- nid is small and set buccally. The carnassial notch is relatively shallow. The metaconid is set directly behind the paraconid and lingual to the protoconid. It is separated from both by shallow valleys. The talonid is low and short. The hypoconid is promi- nent, the entoconid less so. The m2 is single rooted, while the p4 is too damaged to provide any useful morphological information. DISCUSSION. To the extent that comparisons can be made, this material strongly resembles Hel- ogale species. It is a little larger than H. palaeogra- cilis (Dietrich, 1942) from Laetoli by the same amount that that species is larger than the extant H. hirtula (Thomas, 1904). The Kanapoi material clearly is not adequate for specific identification, and I prefer to leave it as Helogale sp. herein. Family Viverridae Viverrids are not uncommon in the Plio-Pleistocene of eastern Africa, but very little of the material has as yet been published. However, most of the ma- terial pertains to species larger than any living vi- verrid. Such species are found in three lineages, Viv- erra Linnaeus, 1758, with V. leakeyi Petter, 1963, known from a number of localities, Pseudocivetta, with the single species P. ingens Petter, 1973, of uncertain affinities, and a third species from Koobi Fora that may be related to Civettictis (Petter, 1963, 1973; Hunt, 1996). Smaller viverrids are less com- mon, and almost none of the material has been studied. Genetta Cuvier, 1816 Genetta sp. nov. (Figure 8) Material that can be referred to Genetta is known from a number of localities in the Plio-Pleistocene of eastern Africa. Kanapoi is the oldest of these, though material referred to cf. Genetta (two spe- cies) is known from Lothagam (Werdelin, 2003). Younger localities with material of Genetta sp. in- clude Laetoli and the Shungura Formation, mem- bers B and C. The first record of the extant G. ge- netta (Linnaeus) is from the Upper Burgi Member of the Koobi Fora Formation (personal observa- tions). KANAPOI MATERIAL. 32565, left maxilla fragment with posterior half of P3 and complete P4-M1; 32815, left mandibular ramus fragment with ml and roots of p4 and m2 (Fig. 8); 30222, left lower canine. The maxilla fragment represents a small carni- vore species. The P3 is damaged anteriorly. It has a small but relatively tall posterior accessory cusp. The upper carnassial is elongated, with a small but sharp parastyle. The protocone is large and reaches further anteriorly than the parastyle. It is separated from the paracone by a deep valley. The paracone is large but short and pyramidal in shape. The me- tastyle is long and low, longer than the paracone. The Ml is broad but short and set at about 60° to P4. The parastyle wing is large and well developed. 130 ■ CS 498, Harris and Leakey: Kanapoi 10 mm Figure 8 Genetta n. sp., KP 32815, left mandibular ramus fragment in (top to bottom) buccal, lingual, and occlusal view while the metastyle wing has been reduced. Both the paracone and the metacone are present, with the paracone being the larger of the two. The tooth tapers gradually in length to the protocone, which is the largest single cusp of the three cusps on Ml. There is a deep basin between the paracone-meta- cone and protocone. The horizontal ramus of the mandible is fairly thick and deep, becoming deeper but thinner at the level of the ascending ramus. The masseteric fossa reaches to the posterior end of ml. The lower car- nassial has a trigonid with tall cusps and a short, narrow talonid. In occlusal view, the paraconid is the largest cusp, but the protoconid is taller. The long axis of the paraconid is set at nearly right an- gles to the main axis of the tooth. The carnassial notch is deep, while the notch separating the para- conid from the metaconid is shallower but wider. The paraconid-protoconid shearing blade is set at about 45° to the main axis of the ramus. The pro- toconid is set buccally, overhanging the ramus to some extent. The metaconid is set directly posterior to the paraconid and lingual to the posterior end of the protoconid. It is separated from the proto- conid by a shallow transverse valley. The talonid is very low and short compared to the trigonid. There are two distinct cusps, which can be homologized with the entoconid and hypoconid. The m2 was small and single rooted. The lower canine KP 30222 is small and re- curved. The root is robust and relatively straight, though broken off part way down. The crown shows no accessory cusps or grooves. DISCUSSION. The morphology of the teeth readily identify these specimens as belonging to the genus Genetta. They are similar in size and most features to the extant G. genetta, but there are dif- ferences that indicate that the Kanapoi material represents a separate species. These differences in- clude the less reduced protocone, broader P4 blade, and less reduced Ml. These are all features in which the extant G. genetta is more derived than the Kanapoi form. Carnivora Family Indet. The following specimens have not been identified to family, mainly because of their incomplete na- ture. In view of the relative abundance of the spe- cies in the identified material, it seems likely that most, if not all, the material of “medium species” should probably be referred to Parahyaena sp. nov. 29289, distal metapodial fragment, medium spe- cies; 32827, proximal phalanx, medium species; 32517, distal fragment of left? MC VP, small spe- cies (may not be carnivore); 32549, proximal me- tapodial fragment (may not be carnivore); 31738, distal metapodial fragment, medium species; 32808, proximal phalanx, medium species; 32569, fragment of anterior premolar, possibly PI, medium species; 32540, left lower canine, small species (possibly mustelid); 478, fragment of astragalus, large species (Machairodontinae indet. in Behrens- meyer 1976); 30478, vertebral fragments and distal metapodial fragment, medium species; 30494, ver- tebral centrum; 32883, vertebral fragments includ- ing dens of axis, medium species; 30432, distal right humerus condyle; 29291, fragment of distal left femur; 30469, fragment of proximal left femur; 29298, right upper canine. SUMMARY The Kanapoi carnivore fauna, with its eight species in as many genera, representing five families, is a substantial addition to the early Pliocene record of Carnivora in Africa. It shares a number of genera and species with other African early-middle Plio- cene localities, such as Langebaanweg and Laetoli, but overall has a unique mixture of species. Simi- larities with Langebaanweg, which is somewhat older and relatively distant, are at the generic level {Enhydriodon, Dinofelis, Homotherium), while similarities with Laetoli, which is closer both in age and geography, lie at the species level {Parahyaena howelli, D. petteri). The small number of taxa from the Apak Member at Lothagam makes compari- sons with that site difficult. On the other hand, differences between Lange- baanweg and Kanapoi show that the former still includes Miocene relicts (taxa such as Hyaenictis Gaudrey, 1861, Plesiogulo Zdansky, 1924, and Werdelin: Carnivorans ■ 131 Machairodus Kaup, 1833), while the latter is more typically Pliocene and lacks these Miocene forms. The Nawata Formation at Lothagam includes a number of forms whose affinities lie outside Africa (mostly in western Eurasia, but also on the Indian subcontinent). The Kanapoi fauna, on the other hand, includes only forms whose immediate fore- bears can be found in Africa. Thus, the Kanapoi fauna represents the currently best-known evidence for the first post-Miocene radiation of endemic Af- rican Carnivora. ACKNOWLEDGMENTS I would like to express my thanks to the government of the Republic of Kenya and to the National Museums of Kenya for allowing me to carry out this study. My thanks also to M.G. Leakey for her invitation to work on the Kanapoi material, to M.E. Lewis for discussions, to the many curators who have allowed me to study comparative material in their care, and to all the field crew and mu- seum staff of the Department of Palaeontology, National Museums of Kenya, without whose efforts there would be no material to study. Inger Wikman-Backstrom made the specimen drawings. This work was financed by the Swed- ish Science Council. LITERATURE CITED Barry, J. C. 1987. Large carnivores (Canidae, Hyaenidae, Felidae) from Laetoli. In Laetoli: A Pliocene site in northern Tanzania, eds. M. D. Leakey and J. M. Harris, 235-258. Oxford: Clarendon Press. Behrensmeyer, A. K. 1976. Lothagam Hill, Kanapoi and Ekora: A general summary of stratigraphy and fau- na. In Earliest man and environments in the Lake Rudolf Basin, eds. Y. Coppens, F. C. Howell, C. E. Isaac, and R. E. E Leakey, 163-170. Chicago: Uni- versity of Chicago Press. Collings, C. E., A. R. I. Cruickshank, J. M. Maguire, and R. M. Randall. 1976. Recent faunal studies at Mak- apansgat Limeworks, Transvaal, South Africa. An- nals of the South African Museum 71:153-165. Dietrich, W. O. 1942. Altestquartare Saugetiere aus der siidlichen Serengeti, Deutsch-Ostafrika. Palaeonto- graphica 94:43-133. Erxleben, J. C. P. 1777. Systema regni animalis per classes, ordines, genera, species, varietates, cum synonymia et historia animalium. Classis I. Mammalia. Wey- gandianis: Lipsiae. Evans, H. E. (ed.). 1993. Miller’s anatomy of the dog, 3rd ed. W. B. Saunders Company. Forster, C. 1780. Herrn von Buff on’s Naturgeschichte der vierfussigen Thiere. Mit vermehrungen, aus dem franzdsischen iibersetzt. Vol. 6. Berlin: J. Pauli. Harris, J. M., F. H. Brown, and M. C. Leakey. 1988. Stra- tigraphy and paleontology of Pliocene and Pleisto- cene localities west of Lake Turkana, Kenya. Con- tributions in Science, Natural History Museum of Los Angeles County 399:1-128. Hendey, Q. B. 1974. The late Cenozoic Carnivora of the south-western Cape Province. Annals of the South African Museum 63:1-369. . 1978a. Late Tertiary Hyaenidae from Langebaan- weg. South Africa, and their relevance to the phy- logeny of the family. Annals of the South African Museum 7 6:165-197 . . 1978b. Late Tertiary Mustelidae (Mammalia, Carnivora) from Langebaanweg, South Africa. An- nals of the South African Museum 7 6:519-557 . Howell, F. C., and C. Petter. 1976. Carnivora from Omo Croup formations, southern Ethiopia. In Earliest man and environments in the Lake Rudolf Basin, eds. Y. Coppens, F. C. Howell, C. L. Isaac, and R. E. F. Leakey, 314-331. Chicago: University of Chi- cago Press. . 1980. The Pachycrocuta and Hyaena lineages (Plio-Pleistocene and extant species of the Hyaeni- dae). Their relationships with Miocene ictitheres: Palhyaena and Hyaenictitherium. Geobios 13:579- 623. Hunt, R. M., Jr. 1996. Basicranial anatomy of the giant viverrid from ‘E’ Quarry, Langebaanweg, South Af- rica. In Palaeoecology and palaeoenvironments of Late Cenozoic mammals: Tributes to the career of C. S. (Rufus) Churcher, eds. K. M. Stewart and K. L. Seymour, 588-597. Toronto: University of Toron- to Press. Jenks, S., and L. Werdelin. 1998. Taxonomy and system- atics of living hyaenas (family Hyaenidae). In Hy- aenas: Status survey and conservation action plan, eds. M. C. L. Mills and H. Hofer, 8-17. Cland, Swit- zerland: lUCN. Linnaeus, C. 1758. Systema Naturae, 10th ed. Stockholm: Laurentii Salvii. Lydekker, R. 1884. Siwalik and Narbada Carnivora. Pa- leontologica Indica, sen 10, 2:178-351. Petter, C. 1963. Etude de quelques Viverrides (Mammi- feres. Carnivores) du Pliocene inferieur de Tangan- yika (Afrique orientale). Bulletin de la Societe Geo- logique de Prance, sen 7, 5:267-274. . 1973. Carnivores Pleistocenes du Ravin d’Olduvai. In Possil vertebrates of Africa, Vol. 3, eds. E. S. B. Lea- key, R. J. C. Savage, and S. C. Coryndon, 43-100. Lon- don: Academic Press. . 1987. Small carnivores (Viverridae, Mustelidae, Canidae) from Laetoli. In Laetoli: A Pliocene site in northern Tanzania, eds. M. D. Leakey and J. M. Harris, 194-234. Oxford: Clarendon Press. Petter, C., and F. C. Howell. 1988. Nouveau felide ma- chairodonte (Mammalia, Carnivora) de la faune pli- ocene de I’Afar (Ethiopie): Homotherium hadarensis n. sp. Gomptes Rendus de I’Academie des Sciences, Paris 306:731-738. Petter, C., M. Pickford, and F. C. Howell. 1991. La loutre piscivore du Pliocene de Nyaburogo et de Nkondo (Ouganda, Afrique occidentale): Torolutra ougan- densis n. g., n. sp. (Mammalia, Carnivora). Comptes rendus de I’Academie des Sciences, Paris 312:949- 955. Pickford, M. H. L. 1978. Stratigraphy and mammalian palaeontology of the late-Miocene Lukeino forma- tion, Kenya. In Geological background to fossil man, ed. W. W. Bishop, 263-278. Edinburgh: Scottish Ac- ademic Press. Savage, R. J. C. 1978. Carnivora. In Evolution of African mammals, eds. V. J. Maglio and H. B. S. Cooke, 249-267. Cambridge, Massachusetts: Harvard Uni- versity Press. Stromer, E. 1931. Reste siisswasser- und landbewohnender Wirbeltiere aus den Diamantenfeldern Klein-Nama- qualandes (Siidwest-afrika). Sitzungsberichte der bayerischen Akademie der Wissenschaften zu Miin- chen 1931:17-47. Turner, A., L. Bishop, C. Denys, and J. K. McKee. 1999. Appendix: A locality-based listing of African Plio- 132 ■ CS 498, Harris and Leakey: Kanapoi Pleistocene mammals. In African biogeography, cli- mate change & human evolution, eds. T. G. Brom- age and F. Schrenk, 369-399. Oxford: Oxford Uni- versity Press. Thomas, O. 1904. On a collection of mammals obtained in Somaliland by Major H. N. Dunn. Annals and Magazine of Natural History 14(7):94-99. Thunberg, C. P. 1820. Beskrifning och techning pa ett nytt species, Hyaena brunnea. Kongliga Vetenskapsaca- demiens Handlingar 1820:59-65. Werdelin, L. 2003. Mio-Pliocene Carnivora from Lotha- gam, Kenya. In Lothagam: Dawn of humanity in eastern Africa, eds. M. G. Leakey and J. M. Harris, 261-328. New York: Columbia University Press. Werdelin, L., and J. Barthelme. 1997. Brown hyena {Para- hyaena brmmea) from the Pleistocene of Kenya. Journal of Vertebrate Paleontology 17:758-761. Werdelin, L., and M. E. Lewis. 2001. A revision of the genus Dinofelis (Mammalia, Felidae). Zoological Journal of the Linnean Society 132:147-258. Werdelin, L., and N. Solounias. 1991. The Hyaenidae: Taxonomy, systematics and evolution. Fossils and Strata 30:1-104. Wesselman, H. B. 1984. The Omo micromammals. Sys- tematics and paleoecology of early man sites from Ethiopia. Basel: Karger. Willemsen, G. F. 1992. A revision of the Pliocene and Quaternary Lutrinae from Europe. Scripta Geologi- ca 101:1-115. Received 26 December 2002; accepted 23 May 2003. I ;■' , ■ i . . r>-^’ f« '.V :•; :-v. - "V J y y -v p: • ■■■(.f <«■' yr< „»i»- V fe- 't j J ( r 9 1 ', r)‘ ■ '/ It cMtTHRONlAN INSTITUTION LIBRARIES 3 9088 01402 9672 Natural of Los Angeles County 900 Exposition Boulevard Los Angeles, California 90007