Hominoid Teeth with Chimpanzee- and Gorilla-Like Features from the Miocene of Kenya: Implications for the Chronology of Ape-Huma

ANTHROPOLOGICAL SCIENCE Vol. 000, 000-000, 2004

Hominoid teeth with chimpanzee- and gorilla-like features from the Miocene of Kenya: implications for the chronology of ape-human divergence and biogeography of Miocene hominoids MARTIN PICKFORD1*, BRIGITTE SENUT2

1Chaire de Paléoanthropologie et de Préhistoire du Collège de France, GDR 983 and UMR 5143 du CNRS, 8, rue Buffon, 75005 Paris 2Département Histoire de la Terre du Muséum National d’Histoire Naturelle, Paris

Received 30 June 2003; accepted 16 March 2004

Abstract One of the major lacunae in our knowledge of African hominoid evolution concerns the origins of the chimpanzee and gorilla. Several thousand specimens from the Plio–Pleistocene of Africa have been attributed to Hominidae (sensu stricto) of which only a few, including Ardipithecus ramidus, have been re-interpreted by some authors as possibly representing an ape rather than a hominid (Senut, 1998). Four recently discovered ape-like specimens from the late Middle Miocene (12.5 Ma) and Late Miocene (5.9 Ma) of Kenya partly fill the gap in the fossil record of African apes, and show some morphological and metric affinities with teeth of chimpanzees and gorillas. If these few specimens from Kenya are indeed more closely related to chimps and gorillas than to hominids, then this implies that the dichotomy between African apes and hominids occurred several million years earlier than is cur- rently estimated by most researchers. Furthermore these ape teeth from Ngorora and Lukeino suggest that extant African apes evolved in Africa, and did not immigrate into the continent from Europe or Asia.

Key words: Miocene, Kenya, Orrorin, Pan, Gorilla

Introduction trally directed crests which close off a low walled mesial fovea. Anterior crests from these same cusps meet mesially This paper concerns four ape-like teeth from Kenya, one to wall off the anterior side of the mesial fovea, which is from the Ngorora Formation (12.5 Ma), and three from the buccolingually wide and mesiodistally short. The hypoconid Lukeino Formation (5.9 Ma), collected by the Kenya Palae- has low crests descending from its apex mesially and distally ontology Expedition. but not crossing the grooves, which separate the cusp from In this paper we restrict the term ‘hominid’ to encompass its neighbors. There is a basal cingular enamel fold between only those hominoids that possess skeletal morphology the protoconid and hypoconid and a tiny fold between the indicative of habitual or obligate bipedal locomotion. We do hypoconid and hypoconulid. The hypoconulid has a crest not employ it in the much wider sense that has recently been descending obliquely anteriorly into the talonid basin, but it used by some authors, in which even the genera Pan and is low and does not separate the basin from the distal fovea. Gorilla are included in the family Hominidae. The postcristid reaches lingually towards the low cusplet present between the hypoconulid and the entoconid. The Ngorora Lower Molar The apices of the metaconid and entoconid are strongly buccolingually compressed and peripherally positioned, a Morphological description (Figure 1A) disposition that results in a voluminous sunken talonid basin Bar 91’99 is an unworn right lower molar, probably the bordered by cusps that are somewhat trenchant in appear- second (11.4 mm mesiodistal P 9.4 mm buccolingual), ance. The posterior crest of the metaconid descends towards although a case could be made for it being a third molar. The a low, peripherally positioned cusplet between the meta- apices of the protoconid and hypoconid are buccolingually conid and entoconid. The hypoconid extends beyond the compressed and not very voluminous. They are located midline of the crown and reaches this small cusplet. As a towards the buccal margin of the tooth and they express consequence, the metaconid and entoconid do not touch minor buccal flare. The apices of the protoconid and meta- each other. A postentoconid crest descends obliquely conid are 5.2 mm apart and the tooth is 9.2 mm broad at this towards another low cusplet located between the entoconid level. The protoconid and metaconid have low-relief, cen- and hypoconulid from which it is separated by narrow but short vertical grooves on the posterolingual surface. * Corresponding author. e-mail: [email protected] The occlusal basin is wide, deep, and long. The wall phone: +33-1-40-79-30-03; fax: +33-1-40-79-35-80 between the anterior fovea and the talonid basin is low and Published online 13 July 2004 the posterior fovea is not clearly demarcated from it, and in J-STAGE (www.jstage.jst.go.jp) DOI: 10.1537/ase.04S014 thus extends for almost the entire length of the tooth. The

Figure 1. Selected hominoid teeth from Africa and Europe (scale bar, 10 mm). (A) Bar 91’99, right lower molar (cast); (A1) stereo occlusal view; (A2) mesial view to show buccal flare. (B) Pan paniscus right m/2, occlusal view. (C) Pan troglodytes, cast of right m/2, occlusal view. (D– F) Dryopithecus brancoi casts of lower molars from Europe, occlusal views (D, Trochtelfingen; E, Salmendingen; F, Ebingen). (G) Gorilla gorilla lower m/2, occlusal view.

occlusal surface of the enamel in the basin is wrinkled. chtelfingen, and Ebingen (Figure 1D–F). It is not impossible that the Ngorora species is related to the genus Dryop- Comparisons with Pan ithecus. The main differences from molars of Pan paniscus (Fig- ure 1B) concern the buccal cusps, which are slightly more Comparisons with Miocene African hominoids internally positioned in the fossil. In P. paniscus, the buccal Comparison of the Ngorora tooth (Figure 1A) with lower cusps are even more peripherally located than they are in the molars of African fossil hominoids reveals that it does not Ngorora tooth. Another difference between the Ngorora belong to any of the described Miocene forms, none of tooth and P. paniscus lies in the height of the entoconid, which possess the large occlusal basin and peripheralized which is taller and more trenchant in P. paniscus. The ante- cusps that distinguish this tooth. The only specimen with rior fovea is buccolingually narrower in the fossil than in which it accords in morphology is the Ngorora upper molar Pan. However, these are relatively minor differences. (KNM-BN 1378), which has wide occlusal fovea and a large The Ngorora molar resembles lower molars of Pan trogl- protocone, but this tooth is appreciably larger than what odytes from Mahale, Tanzania (Figure 1C) more closely than would be expected to occlude with Bar 91’99. it does those of P. paniscus. This is mainly because the buccal cusps are slightly more internally positioned in the com- Implications of the Ngorora lower molar mon chimpanzee, and they thus have moderate buccal flare Until now, no fossil chimpanzees and gorillas have been of the cusp walls. A minor difference is the extent of the reported and there has thus been a lack of evidence to test the mesial fovea, which is greater in P. troglodytes than it is in molecular clock estimates of the dichotomy between African the Ngorora specimen, and most chimpanzee molars do not apes and hominids. If the derived characters of the Ngorora have any trace of a buccal cingulum. molar represent homologies shared with chimpanzees, then it would indicate that the Pan clade has its roots in the latter Comparison with molars of Dryopithecus part of the Middle Miocene, sometime prior to 12.5 Ma. Lower molars of the European genus Dryopithecus (Fig- The dichotomy between chimpanzees and humans is usu- ure 1D–F) have peripheralized cusps, and broad but shallow ally estimated by molecular biologists to be more recent than occlusal basins (Begun, 2002). They also possess slight buc- 6 Ma (Gagneux et al., 1999; Stauffer et al., 2001) and the cal cingula and some specimens have a small accessory cus- split between chimpanzees and gorillas has been estimated plet separating the metaconid and entoconid in the lingual at 8–9 Ma by Wrangham and Pilbeam (2001) and 7.7 Ma by notch. The most significant differences between the Kenyan Gagneux et al. (1999). The only molecular biologists who and European specimens are the shallower occlusal basins have proposed an earlier age for African ape origins are and the less buccolingually compressed lingual cusps that Arnason and co-workers (Arnason et al.,1996, 1998, 2000; occur in Dryopithecus. In particular, the Ngorora tooth Janke and Arnason, 2002), but their results are usually con- resembles some of the central European D. brancoi, espe- sidered suspect by others who appear to favor appreciably cially specimens such as those from Salmendingen, Tro- later divergence times (Bailey et al., 1992; Adachi and Vol. 112, 2004 HOMINOID TEETH FROM THE MIOCENE OF KENYA 3

Hasegawa, 1995; Gagneux et al., 1999; Stauffer et al., 2001; are the entire paracone and metacone, part of the hypocone Pilbeam, 2002). and the distobuccal extremity of the protocone (the crista The Ngorora specimen thus runs counter to the recent obliqua). The distal fovea is complete. The tooth is 14 mm ideas of Pilbeam (1996) who wrote that “the common ances- mesiodistal. It is low crowned (7.3 mm from cervix to tip of tor of humans and chimpanzees was probably chimpanzee- metacone) and the cusps are endowed with low but broad like, a knuckle-walker with small thin-enamelled cheek wrinkles. teeth,” and Wrangham and Pilbeam (2001) who postulated The paracone and metacone are separated from each other that the “6 mybp ancestor…would have been thin- by a narrow slit-like buccal groove (Figure 2B). The part of enamelled, knuckle-walking, and females would have had the trigon basin preserved is relatively large and open, the black body coats.” It is already known that 6 Ma hominids main cusps being peripherally positioned and the cusps not such as Orrorin tugenensis had thick-enamelled molars with bulbous. The mesial fovea is not preserved. Two low crests restricted occlusal basins and were fully bipedal (Senut et descend from the apex of the paracone towards the midline al., 2001; Pickford et al., 2002). Instead the Ngorora fossil of the crown (i.e. towards the mesial fovea). The metacone ape tooth accords with the scenario published by Arnason et sends a crest lingually and slightly anteriorly towards the al. (1996, 1998, 2000) based on molecular evidence, of an protocone, from which it is separated by a groove. The meta- early divergence (ca. 13.5 Ma) between Pan and Homo. cone has another, smaller crest that descends obliquely distally directly towards the hypocone, bifurcating on its way The Kapsomin Ape Teeth down the cusp. The anterior branch of the crest forms a low wall between the trigon basin and the distal fovea. The distal Bar 1757’02, upper molar cingulum reaches from the base of the metacone to the hypo- Description cone and walls off the distal fovea on the distal aspect of the The Kapsomin tooth is a partial left upper molar (Figure tooth. The metacone and hypocone are spaced far apart, and 2A–D). It lacks the roots but the cervical line is partly pre- as a consequence, the distal fovea is buccolingually wide served, allowing estimation of cusp height. Also preserved and slightly obliquely oriented with the buccal end more

Figure 2. Kapsomin upper molar and comparisons. (A–D) Bar 1757’02, Kapsomin large ape, cast of left upper molar; (A) occlusal; (B) buccal; (C) enlarged internal view to show dentine enamel junction and measuring point (oblique black line) for enamel thickness determination; (D) oblique view. (E) Pan troglodytes upper M1/–M3/, occlusal view. (F) Orrorin tugenensis upper molar row, occlusal view. NB The toothrow of O. tugenensis comprises two teeth (M1/ and M2/) found after the description of the M3/ (Senut et al., 2001). Concordant interstitial facets reveal that the three teeth belong to a single individual. (G) Gorilla gorilla M2/, oblique view. (H) G. gori ll a upper M2/, occlusal view. 4 M. PICKFORD AND B. SENUT ANTHROPOLOGICAL SCIENCE mesial than the lingual one. The apex of the hypocone is Pan are extensive and the crests separating the basin from missing so it is not possible to provide detailed measure- the mesial and distal foveae are quite low. In any case, the ments of the distance between its apex and those of the para- distal fovea in Pan is reduced in buccolingual extent. A fur- cone and metacone. We estimate the distance between the ther difference is the greater buccolingual compression of tips of the hypocone and metacone at 7.5 mm. The distance the apices of the paracone and metacone in Pan. between the tips of the paracone and metacone is 6 mm. The distal margin of the crown is not markedly curved, suggest- Comparisons with Sahelanthropus tchadensis ing that it is not a third upper molar, but more likely a second Bar 1757’02 is larger than the upper molars in or first. The surface of the enamel is wrinkled but not as Sahelanthropus tchadensis (Brunet et al., 2002) (mesio- heavily as is usually the case in chimpanzee molars. Enamel distal length is 14 mm in Bar 1757’02, compared to 10.9– thickness measured radially (Schwartz, 2000) 2 mm below 11.5 mm in M1/, 13 mm in M2/, and 10.7–10.8 mm in M3/ the tip of the preserved part of the hypocone is 1.6–1.7 mm of Sahelanthropus). Brunet et al. (2002) describe the molars (line in Figure 2C), but this figure is likely to be an overesti- as having “low rounded cusps…,” in which case they are mate of the true thickness, as it is taken slightly obliquely different from the Kapsomin tooth. The enamel in the upper due to the orientation of the fractured surface of the cusp. molars of Sahelanthropus has been reported to be 1.71 mm thick in the paracone of M3/ and 1.79 mm at the hypocone of Comparisons with Orrorin tugenensis and Pliocene homi- M2/, which suggests that enamel in Sahelanthropus is nids approximately the same as in the Kapsomin ape in which it Bar 1757’02 is highly divergent from the teeth in the is 1.6–1.7 mm on the hypocone. The enamel thickness of upper tooth row of Orrorin (Figure 2F), not only in its Orrorin was originally (Senut et al., 2001) said to be 3.1 mm greater dimensions, but also in its morphology. The trigon on the protoconid of m/2 but this was a typographical error basin and distal fovea in the upper molars of Orrorin are that unfortunately carried through into both translated restricted in size; the buccal cusps are more flared from apex versions of the text. The thickness measured radially near to cervix, the buccal slit is non-existent, the main cusps are the tip of the protoconid on the original fossil was 2.1 mm. lower crowned, the dentine penetrance is low, and the The enamel thickness decreases towards the cervix such that enamel is less wrinkled. The Kapsomin ape tooth is closer in it is about 1.7 mm thick about 2 mm below the apex of the size and some aspects of morphology to upper molars of cusp. Further study with a scanner is planned. Australopithecus afarensis and Praeanthropus africanus. However, the higher and less-inflated cusps, the steep, flat, Comparisons with other Miocene hominoids and unflared buccal surface, thinner enamel, and the greater Bar 1757’02 differs in various features from most of the dentine penetrance suggest that Bar 1757’02 is not from an known medium and large hominoids from the Miocene of australopithecine. Africa. Molar morphology in Kenyapithecus, Nacholap- ithecus, Otavipithecus, and Afropithecus is quite different, Comparisons with Gorilla gorilla these taxa possessing non-peripheralized bulbous main Upper molars of gorillas are generally longer than Bar cusps and restricted occlusal basins (Conroy et al., 1992; 1757’02 but there are small individuals that overlap in size; Ward and Duren, 2002). In Proconsul and Ugandapithecus upper M1/s of Gorilla gorilla range in mesiodistal length the cusps are not peripheralized (Harrison, 2002; Senut et from 12.7 to 16.7 mm, M2/s from 13.4 to 17.9 mm, and al., 2000). The cusps in the upper molars of Samburupi- M3/s from 12.0 to 17.0 mm (Pilbeam, 1969). thecus are inflated, and the occlusal basins are extremely The crown morphology of Bar 1757’02 is similar in sev- restricted (Pickford and Ishida, 1998), which makes them eral respects to that of molars of Gorilla gorilla (Figure 2G, very divergent from the Kapsomin tooth. H). In particular the mesio-distally short but bucco-lingually Ouranopithecus and Ankarapithecus from European Late wide distal fovea of the Lukeino tooth resembles that of Miocene deposits have thick-enamelled molars with Gorilla as does its oblique orientation. Another resemblance restricted occlusal basins and low dentine penetrance, inter- lies in the degree and kind of enamel wrinkling and the high preted by de Bonis et al. (1981) as features shared with dentine penetrance. In G. gorilla enamel is generally consid- Hominidae, from which we infer that the genus is markedly ered to be thin (ranging from 0.98 to 1.63 mm on the proto- different from the Kapsomin ape. cone and between 0.7 and 1.33 mm on the paracone: Schwartz, 2000), and in Bar 1757’02 it is 1.6–1.7 mm on the Discussion hypocone, which falls slightly above the range of variation It seems clear that the fragment of upper molar from Kap- of Gorilla. This measure was taken on a slightly oblique nat- somin (Bar 1757’02) represents a species distinct from O. ural section of the tooth (Figure 2C), and the radial thickness tugenensis. Apart from its greater dimensions, it has mor- would be slightly less than this figure. phology that is different from the bunodont crown with inflated main cusps and restricted trigon basin and foveae of Comparisons with Pan the latter taxon. Whereas the molars of Orrorin recall those Bar 1757’02 differs markedly from molars of P. troglo- of later hominids (Australopithecus, Praeanthropus, and dytes and P. paniscus, not only in size, but also in morphol- even Homo) in overall crown shape, bunodonty, cusp infla- ogy (Figure 2E). M1/s and M2/s of P. troglodytes range in tion, and basin restriction, Bar 1757’02 stands out as anom- mesiodistal length from 9 to 12 mm, and M3/s from 8.2 to alous, with its more peripheralized cusps, widely separated 10.6 mm (Pilbeam, 1969). The trigon basins in molars of metacone and hypocone, wide, obliquely oriented distal Vol. 112, 2004 HOMINOID TEETH FROM THE MIOCENE OF KENYA 5 fovea, vertical and flat buccal surface, high dentine pene- any close relationship with Pan, which, apart from being trance, and thinner enamel with light wrinkling. The tooth smaller and having peripheralized main cusps, has quite dif- has high dentine penetrance as shown by the fact that if it ferent molar crown morphology. were worn to the same level as the M2/ in the Orrorin tooth row, which shows no dentine exposure, the Kapsomin ape Bar 1001’00, upper central incisor tooth would have a large dentine exposure (the shadow at the Bar 1001’00 is a right upper central incisor that has been apex of the protocone of the M2/ of Orrorin in Figure 2F is abraded mesially and distally (Figure 3B). The preserved a small patch of chemical etching floored by enamel and is part of the crown is relatively low compared to root length. not an exposure of dentine). The morphology of the dentine- In lateral view the labial and lingual surfaces form a wedge- enamel junction in the Kapsomin tooth is best appreciated shape; the tooth is 8.7 mm thick labio-lingually near the cer- by examining the broken inner surface of the Kapsomin vix and narrows rapidly incisally. It has lightly wrinkled tooth where it is clearly visible, showing the dentine pene- enamel. The crown is unlike the corresponding teeth of aus- trating high into the hypocone (Figure 2C). In most of these tralopithecines and humans because there is no fossa on the features the tooth is closest to G. gorilla, yet it is by no lingual side (Figure 3A). In hominids the crown is higher means a perfect fit with this species. from cervix to incisal edge and the lingual surface has a large Metrically, the Kapsomin tooth is appreciably larger than fossa, sometimes with lingual ridges or pillars, but usually any of the Orrorin molars, but it falls within the range of possessing a scoop-shaped profile. In gorillas, in contrast, metric variation of the gorilla. It also falls within the range of the upper central incisors are often wedge-shaped, with no variation of Pliocene fossil hominids, including Ardip- sign of a lingual fossa, or if one is developed it is not very ithecus, Australopithecus, and Praeanthropus. Its relation- prominent. Comparison of the Kapsomin upper incisor with ships to Sahelanthropus are not clear as published those of gorillas reveals close similarities (Figure 3C) not photographs of the latter do not reveal enough detail about only in size but also in shape and crown height relative to crown morphology, but descriptions of the fossil suggest root length. Even though these similarities are enhanced by major morphological differences. the wear stage of the Kapsomin incisor and that of the gorilla We conclude that Bar 1757’02 reveals the presence of a used for comparison, this is because the underlying mor- second hominoid in the Lukeino Formation, Kenya. Its phology is similar. Chimpanzee and hominid teeth worn as affinities appear to lie with gorillas rather than hominids. Its much as the Kapsomin fossil do not develop a comparable dimensions fall within the range of variation of gorilla upper wedge-shaped appearance in lateral view, because the lin- molars and well above that of chimpanzees, and we discount gual surface of their teeth is basin-like.

Figure 3. Kapsomin upper incisor and comparisons. (A) Praeanthropus africanus cast of upper central incisor from Laetoli; (A1) lingual view; (A2) distal view. (B) Bar 1001’01, cast of upper central incisor, Kapsomin large ape; (B1) lingual view; (B2) distal view. (C) Gorilla gorilla upper central incisor worn to the same level as Bar 1001’01; (C1) lingual view; (C2) distal view. 6 M. PICKFORD AND B. SENUT ANTHROPOLOGICAL SCIENCE

An anonymous referee considered that this tooth is similar Cheboit Lower Molar to A.L. 198-17a (Johanson et al., 1982) and on this basis suggested that the Kapsomin tooth may belong to an austral- Bar 2000’03 from Cheboit, Lukeino Formation, is an opithecine. However, comparison of the specimens reveals unworn right lower molar lacking the roots, but preserving that they are divergent in morphology. The Hadar fossil has much of the cervical line (Figure 4). The apices of the two a restricted basal lingual pillar bordered mesially and dis- lingual cusps are buccolingually compressed and peripher- tally by hollows, as in other australopithecine teeth, whereas ally located. The tips of the protoconid and metaconid are Bar 1001’00 does not, its basal lingual part being inflated 5.4 mm apart and the tooth is 10.5 mm broad at this level. right across the preserved part of the tooth, as in gorillas with The two buccal cusps are slightly in advance of the lingual which we compared the tooth. Furthermore, the wear facet ones, and there is minor buccal flare. Because of the periph- in A.L. 198-17a is apical, flat, and at right angles to the long eral positions of the main cusps, the occlusal basin is large axis of the tooth, whereas that in the Kapsomin tooth is and elongated. The mesial fovea is wide but mesiodistally steeply angled from incisal edge to cervix on the lingual short. The hypoconulid is small and is located slightly to the side, again as in gorillas. In apical view the Hadar tooth is a lingual side of the centre line of the tooth and in a very distal mesiodistally elongated rectangle, slightly concave lingually position, and as a result the tooth has an elongated trapezoi- with slight projection of the basal tubercle, whereas the dal outline (Figure 4). Because of this the tooth could be a Kapsomin tooth is much broader labiolingually. This is lower third molar. The distal fovea is thus small, but is not reflected in the dimensions of the teeth, the Kapsomin spec- separated from the main occlusal basin by the crests from the imens measuring 8.7 mm labiolingually near cervix and the hypoconulid or entoconid, as these do not reach each other. Hadar specimen only 7.1 mm. Larger upper central incisors The tooth measures 12.7 mm mesiodistal by 11.1 mm buc- from Hadar (A.L. 200-1a, A.L. 333x-4, and A.L. 333x-20) colingual. measure 8.5, 8.6, and 8.6 mm labiolingually, respectively. This tooth is morphologically compatible with Bar 1757’02, the upper molar from Kapsomin. We consider it Discussion likely that the two specimens belong to a single taxon. The upper central incisor (Bar 1001’00) was initially attributed to O. tugenensis (Senut et al., 2001) because at the Discussion time of the discovery it was assumed that only a single hom- This lower molar differs radically from those of O. tugen- inoid was represented at the site. With the discovery of Bar ensis of which it is a contemporary. The occlusal outline is 1757’02, a partial upper molar, it became clear that a second, long, narrow, and trapezoidal compared with a wider, more larger, hominoid taxon was present at Kapsomin, leading to rectangular outline in Orrorin. The main occlusal basin is a re-evaluation of all the specimens from the site. Because of deep, wide, and long, compared with the shallow, narrow, its dimensions (labiolingual breadth 8.7 mm) and its some- and short basin in Orrorin. The distal fovea is not separated what gorilla-like morphology and wear pattern, we now con- from the main basin, whereas in Orrorin it is. The buccal sider that the upper central incisor belongs to this second flare is not as marked as it is in Orrorin and the lingual cusps hominoid rather than to Orrorin. are more peripherally located. The main cusps are not as inflated as those of Orrorin, suggesting that the tooth has

Figure 4. Cheboit lower molar and comparisons. (A) KNM-LU 335, Orrorin tugenensis, cast of left m/3, stereo occlusal view. NB KNM-LU 335 was originally interpreted to be an m/2 (Andrews in Pickford, 1975), but comparison with teeth in the holotype mandible of O. tugenensis reveals that it is in fact an m/3. (B) Bar 2000’03, Cheboit large ape right lower molar; (B1) stereo occlusal view; (B2) mesial view. Vol. 112, 2004 HOMINOID TEETH FROM THE MIOCENE OF KENYA 7 thinner enamel and probably greater dentine penetrance, but Africa during the latter part of the Middle Miocene and the we have not yet had the opportunity to determine these Late Miocene. They thus refute the statement by Begun parameters. In Orrorin, the hypoconulid is located to the (2002) that “In actual fact, none of the many late Miocene buccal side of the midline, and is thus close to the hypo- African fossil localities has any hominoids….” When we conid, whereas in Bar 2000’03 it is to the lingual side of add them to Samburupithecus from the Late Miocene of midline, and far from the hypoconid. Indeed this cusp is Samburu Hills, Kenya (9.5 Ma) (Ishida and Pickford, 1997), closer to the entoconid than it is to the hypoconid, the oppo- Orrorin from Lukeino, Kenya (6–5.7 Ma) (Senut et al., site of the situation in Orrorin. In the latter genus, there is a 2001), Sahelanthropus from Toros-Menalla, Chad (ca. 7– clear valley buccally between the hypoconid and the ento- 6 Ma) (Brunet et al., 2002), and Ardipithecus from Ethiopia conid, whereas in Bar 2000’02, there is not even an indenta- (White et al., 1994, 1995), it is clear that Late Miocene tion in the distobuccal wall of the tooth. The latter tooth is Africa was not devoid of hominoids until they reintroduced also somewhat bigger than any of the Orrorin lower molars. themselves from Europe (Begun, 2002). Rather, it is more likely that chimp-sized Dryopithecus was originally an Afri- Discussion and Conclusions can lineage that invaded western Europe about 12.5–12 Ma, and while the evidence is scant, some of the large gorilla- Four ape-like teeth from the Miocene of Kenya reveal sized hominoids from the Late Miocene of Greece and Tur- greater similarities to extant chimpanzee and gorilla teeth key could also be of African origin rather than autoch- than to thick-enamelled Miocene apes and Mio–Plio– thonously evolved descendents of Dryopithecus or Pleistocene to recent hominids. The specimen from Ngorora Sivapithecus as envisaged by Begun (2002). Despite the rel- (12.5 Ma) is similar in size and some morphological details ative poverty of the African fossil record, the new discover- to Pan but also has resemblances to the European Miocene ies reveal that hominoids were more diverse in the Late genus Dryopithecus, with which it could be congeneric, Miocene of Africa than they were in Europe (five genera whereas the Lukeino specimens (5.9 Ma), recall, but are not now known in Africa compared to three or perhaps four in identical to, the teeth of gorillas. Europe). The morphology of the Ngorora tooth suggests that the Dryopithecus lineage may have evolved in Africa and then Acknowledgments invaded Europe about 12–12.5 Ma, rather than evolving within Europe from a thick-enamelled lineage such as We thank members of the Kenya Palaeontology Expedi- Griphopithecus (Begun, 2002). If it is part of the Pan clade, tion for their help in the field, in particular Mr Kiptalam then it would push back the split between hominids and Afri- Cheboi. Research permission was accorded by the Kenya can apes to the Middle Miocene. If this is so, then thick- Ministry of Education, Science and Technology. Funds were enamelled apes such as Kenyapithecus possibly take on a provided by the Collège de France (Professor Y. Coppens), renewed significance for throwing light on the earliest stages the Département Histoire de la Terre (Professor Ph. Taquet), in the evolution of hominids, as thought by L. Leakey in the the French Ministry of Foreign Affairs (Commission des 1960s (Leakey, 1962, 1967, 1969, 1970) even though the Fouilles), and the CNRS (Projet PICS). We are particularly supposedly hominid features employed by Leakey in his keen to thank the Community Museums of Kenya (Mr E. proposals have subsequently been interpreted as being Gitonga) for their help and cooperation and Professor H. related to sexual dimorphism and to plesiomorphic features Ishida for inviting MP to spend time in his laboratory as vis- found in several Middle Miocene hominoids, rather than to iting professor at Kyoto University where he was able to derived morphology shared with hominids (Pickford, 1985). examine the extensive cast collection of extant and fossil It is more parsimonious to consider that thick-enamelled hominoids that has been amassed. We also thank the Primate hominids descended from thick-enamelled precursors rather Research Institute, Inuyama, Japan and Yutaka Kunimatsu than to hypothesize a thin-enamelled intermediate stage, as for discussions. has apparently become the fashion (Wrangham and Pilbeam, 2001). What is required is a fresh look at the problem, including the relationships between diet on the one hand and References enamel thickness and dentine penetrance on the other. Adachi J. and Hasegawa M. (1995) Improved dating of the human- If the Kapsomin ape teeth belong to the gorilla clade, then chimpanzee separation in the mitochondrial DNA tree: heter- they would indicate that by about 6 Ma the lineage was a ogeneity among amino acid sites. Journal of Molecular Evo- separate entity from the Pan Homo clade. Taken together, lution, 40: 622–628. the Ngorora and Kapsomin ape teeth, and those of the early Arnason U., Gullberg A., Janke A., and Xiufeng Xu (1996) Pattern bipedal hominid Orrorin, plead for considerably earlier split and timing of evolutionary divergences among hominoids based on analyses of complete mtDNAs. Journal of Molecu- times between the gorilla, chimpanzee, and hominid clades lar Evolution, 43: 650–651. than most molecular biologists have considered possible for Arnason U., Gullberg A., and Janke A. (1998) Molecular timing of the past three decades (Gagneux et al., 1999; Stauffer et al., Primate divergences as estimated by two nonprimate calibra- 2001) but more in accord with the results of Arnason and his tion points. Journal of Molecular Evolution, 47: 718–727. colleagues (Arnason et al., 1996, 1998, 2000; Janke and Arnason U., Gullberg A., Schweitzer Burguette A., and Janke A. Arnason, 2002). (2000) Molecular estimates of Primate divergences and new hypotheses for Primate dispersals and the origin of modern The four teeth from Baringo district, Kenya, described humans. Hereditas, 133: 217–228. here reveal the presence of ape-like hominoids in East 8 M. PICKFORD AND B. SENUT ANTHROPOLOGICAL SCIENCE

Bailey W.J., Hayasaka K., Skinner C.G., Kehoe S., Sieu L.C., Sciences of the United States of America, 67: 746–748. Slightom J.L., and Goodman M. (1992) Reexamination of the Pickford M. (1975) Late Miocene sediments and fossils from the African hominoid trichotomy with additional sequences from Northern Kenya Rift Valley. Nature, 256: 279–284. the primate -globin gene cluster. Molecular Phylogenetics Pickford M. (1985) A new look at Kenyapithecus based on recent and Evolution, 1: 97–135. collections from Western Kenya. Journal of Human Begun D. (2002) European hominoids. In: Hartwig W.C. (ed.), The Evolution, 14: 113–143. Primate Fossil Record. Cambridge University Press, Cam- Pickford M. and Ishida H. (1998) Interpretation of bridge, pp. 339–368. Samburupithecus, an upper Miocene hominoid from Kenya. Brunet M., Guy F., Pilbeam D., Mackaye H., Likius A., Ahounta Comptes Rendus de l’Académie des Sciences de Paris, Série D., Beauvilain A., Blondel C., Bocherens H., Boisserie J.-R., 2A, 326: 299–306. de Bonis L., Coppens Y., Dejax J., Denys C., Duringer P., Pickford M., Senut B., Gommery D., and Treil J. (2002) Bipedal- Eisenmann V., Fanone G., Fronty P., Geraads D., Lehmann T., ism in Orrorin tugenensis revealed by its femora. Comptes Lihoreau F., Louchart A., Mahamat A., Merceron G., rendus Palevol, 1: 191–203. Mouchelin G., Otero O., Campomanes P., Ponce de Leon M., Pilbeam D.R. (1969) Tertiary Pongidae of East Africa: evolution- Rage J.-C., Sapanet M., Schuster M., Sudre J., Tassy P., Val- ary relationships and taxonomy. Bulletin of the Peabody entin X., Vignaud P., Viriot L., Zazzo A., and Zollikofer C. Museum of Natural History, 31: 1–185. (2002) A new hominid from the Upper Miocene of Chad, Pilbeam D. (1996) Genetic and morphological records of the Hom- Central Africa. Nature, 418: 145–151. inoidea and hominid origins: a synthesis. Molecular Phyloge- Conroy G., Pickford M., Senut B., Van Couvering J., and Mein P. netics and Evolution, 5: 155–168. (1992) Otavipithecus namibiensis, first Miocene hominoid Pilbeam D. (2002) Perspectives on the Miocene Hominoidea. In: from Southern Africa (Berg Aukas, Namibia). Nature, 356 : Hartwig W.C. (ed.), The Primate Fossil Record. Cambridge 144–148. University Press, Cambridge, pp. 303–310. de Bonis L., Johanson D., Melentis J., and White T. (1981) Varia- Schwartz G.T. (2000) Taxonomic and functional aspects of the pat- tions métriques de la denture chez les Hominidés primitifs: terning of enamel thickness distribution in extant large-bod- comparaison entre Australopithecus afarensis et Ourano- ied hominoids. American Journal of Physical Anthropology, pithecus macedoniensis. Comptes Rendus des Séances de 111: 221–244. l’Académie des Sciences, Série 2, 292: 373–376. Senut B. (1998) Les grands singes fossiles et l’origine des homi- Gagneux P., Wills C., Gerloff U., Tautz D., Morin P., Boesch C., nidés: mythes et réalités. Primatologie, 1: 93–134. Fruth B., Hohmann G., Ryder O., and Woodruff D. (1999) Senut B., Pickford M., Gommery D., and Kunimatsu Y. (2000) Un Mitochondrial sequences show diverse evolutionary histories nouveau genre d’hominoïde du Miocène inférieur d’Afrique of African hominoids. Proceedings of the National Academy orientale: Ugandapithecus major (Le Gros Clark and Leakey, of Sciences of the United States of America, 96: 5077–5082. 1950). Comptes Rendus de l’Académie des Sciences de Paris, Harrison T. (2002) Late Oligocene to Middle Miocene catarrhines Série 2A, 331: 227–233. from Afro-Arabia. In: Hartwig, W.C. (ed.), The Primate Fos- Senut B., Pickford M., Gommery D., Mein P., Cheboi K., and sil Record. Cambridge University Press, Cambridge, pp. 311– Coppens Y. (2001) First hominid from the Miocene (Lukeino 333. Formation, Kenya). Comptes Rendus de l’Académie des Sci- Ishida H. and Pickford M. (1997) A new late Miocene hominoid ences de Paris, Série 2A, 332: 137–144. from Kenya: Samburupithecus kiptalami gen. et sp. nov. Stauffer R. L., Walker A., Ryder O., Lyons-Wailer M., and Hedges Comptes Rendus de l’Académie des Sciences de Paris, Series S. (2001) Human and ape molecular clocks and constraints on 2A, 325: 823–829. paleontological hypotheses. Journal of Heredity, 92: 469–474. Janke A. and Arnason U. (2002) Primate divergence times. In: Ward S. and Duren D. (2002) Middle and Late Miocene African Galdikas B., Briggs N., Sheeran L., Shapiro G., and Goodall hominoids. In: Hartwig W.C. (ed.), The Primate Fossil J. (eds.), All Apes Great and Small. Volume 1: African Apes. Record. Cambridge University Press, Cambridge, pp. 385– Kluwer Academic/Plenum Publishers, New York, pp. 18–25. 397. Johanson D., White T., and Coppens Y. (1982) Dental remains White T., Suwa G., and Asfaw B. (1994) Australopithecus rami- from the Hadar Formation, Ethiopia: 1974–1977 collections. dus, a new species of early hominid from Aramis, Ethiopia. American Journal of Physical Anthropology, 57: 545–603. Nature, 371: 306–312. Leakey L.S.B. (1962) A new lower Pliocene fossil primate from White T., Suwa G., and Asfaw B. (1995) Australopithecus rami- Kenya. Annals and Magazine of Natural History, Series 13, 4: dus, a new species of early hominid from Aramis, Ethiopia. 689–696. Nature, 375: 88. Leakey L.S.B. (1967) An early Miocene member of Hominidae. Wrangham R. and Pilbeam D. (2001) African apes as time Nature, 213: 155–163. machines. In: Galdikas B., Briggs N., Sheeran L., Shapiro G., Leakey L.S.B. (1969) Fort Ternan hominid. Nature, 222: 1202. and J. Goodall J. (eds), All Apes Great and Small. Volume 1: Leakey L.S.B. (1970) The relationships of African apes, man and African Apes. Kluwer Academic/Plenum Publishers, New Old World monkeys. Proceedings of the National Academy of York, pp. 5–17.