Andrew Hill New cercopithecoids and a hominoid from Department of Anthropology, 12·5 Ma in the Tugen Hills succession, Yale University, P.O. Box 208277, New Haven, CT 06520, U.S.A. E-mail: [email protected] The early evolutionary history of the cercopithecoids is poorly under- stood, primarily due to a lack of material from between 15 and about 9 Ma. Cercopithecoid specimens from a fossil site in Meave Leakey the Ngorora Formation of the Tugen Hills, Kenya, belong to the Department of Palaeontology, genus Victoriapithecus, possibly a new species. These are National Museums of Kenya, associated with a hominoid specimen that resembles , and P.O. Box 40658, Nairobi, another tooth of a catarrhine, also probably hominoid. The locality is Kenya. BPRP#38, in the Kabasero type section of the Ngorora Formation, E-mail: [email protected] and well dated at 12·5 Ma. If the hominoid specimen is confirmed as Proconsul, it would be one of the most recent recorded. The relatively John D. Kingston diverse fauna from the site in some ways resembles that of Department of Anthropology, Fort Ternan. The site is underlain, and not far removed in time, from Emory University, one of the best fossil macro-floras in , which indicates lowland 1557 Pierce Drive, Atlanta, rainforest conditions in this portion of the Rift Valley at 12·6 Ma. GA 30322, U.S.A.  2002 Academic Press E-mail: [email protected]

Steve Ward Department of Anatomy, Northeastern Ohio Universities College of Medicine, Rootstown, OH 44272, U.S.A. E-mail: [email protected]

Received 15 December 2000 Revision received 12 June 2001 and accepted 14 June 2001

Keywords: Hominoidea, Cercopithecoidea, , Journal of Evolution (2002) 42, 75–93 Ngorora, Tugen Hills, doi:10.1006/jhev.2001.0518 Kenya. Available online at http://www.idealibrary.com on

Introduction (Le Gros Clark & Leakey, 1951), a man- dible fragment and an isolated lower The early history of the Cercopithecoidea in from Loperot, Kenya (Szalay & Delson, Africa is sparse (Gundling & Hill, 2000). A 1979) and 16 specimens including man- handful of primitive Old World monkeys is dibular and maxillary fragments and isolated known from a number of early Miocene sites teeth from Buluk, northern Kenya (Leakey, in East Africa ranging in age from approxi- 1985). In north Africa, early cercopithecoids mately 19–17 Ma. These include an M2 are known from two early Miocene locali- from Napak, Uganda (Pilbeam & Walker, ties. The specimens include three damaged

1968), an isolated M3 from Ombo, Kenya mandibles from Wadi Moghara, Egypt

0047–2484/02/010075+19$35.00/0  2002 Academic Press 76 .  ET AL.

(Simons, 1969), and a mandible fragment within the Cercopithecidae. Benefit’s re- from Gebel Zelten, Libya (Delson, 1979). analysis of this material and a later collection From the supposed Middle Miocene at made by Pickford, along with her detailed

Ongoliba, Congo, an M3 of a has comparisons with a large number of modern been reported (Hooijer, 1963). In contrast cercopithecoids, indicate that the Maboko to these sparse and isolated occurrences, a monkeys represent a single species, V. large collection of monkey cranial and post- macinnesi (Benefit, 1987). Harrison (1987) cranial fragments belonging to the genus reached a similar conclusion. The extended Victoriapithecus has been recovered from the time interval from which the collection is 15 Ma, Middle Miocene site of Maboko derived resulted in a slightly high level of (von Koenigswald, 1969; Benefit, 1987, variation in the specimens. Benefit identified 1993; Benefit & McCrossin, 1989, 1993). a number of primitive dental traits Another recently discovered occurrence unique to these early monkeys, and at least of Victoriapithecus material is from the three derived dental traits (and possibly Kipsaramon site complex in the Tugen many more) common to the modern sub- Hills. At site BPRP#89, for example, dated families Colobinae and Cercopithecinae, at 15·5 Ma (Hill et al., 1991; Behrensmeyer but exclusive of the Victoriapithecinae. She et al., 2002) there is a partial left mandible concluded that the modern subfamilies are with M2 and M3, as well as isolated teeth more closely related to one another than (e.g., KNM-TH 31013, a left P4; and either is to the Victoriapithecinae. She KNM-TH 31014, a right P4). therefore proposed that the Victoriapitheci- This early cercopithecoid material has nae be raised to family rank (Benefit, been attributed to two genera, Prohylobates 1987, 1993, 1994, 1999; see also Harrison, and Victoriapithecus, which share several 1987). primitive dental traits and which differ The large collection of Maboko monkeys, largely in their absolute size and molar pro- although providing exceptional evidence for portions (Leakey, 1985; Benefit, 1987, the primitive characteristics of the Victoria- 1993). Prohylobates is known from few pithecidae, leaves many questions un- occurrences in the early Miocene of east and answered concerning the mode and timing north Africa (e.g. from Buluk, Wadi of the origin of modern monkeys. What was Moghara, Gebel Zelten) whereas Victoria- the origin of the Cercopithecidae, and the pithecus is well known from the exceptional modern cercopithecid subfamilies? The ear- Maboko collection. Detailed studies by liest cercopithecid reported from Africa is a Benefit (1987, 1993, 1994, 1999, 2000; colobine from Ngeringerowa, a site complex Benefit & McCrossin 1989, 1993, 1997)of within the Ngorora Formation of the Tugen the numerous specimens from Maboko have Hills succession (Benefit & Pickford, resulted in a good understanding of these 1986), which has now been dated by the early monkeys which provides a sound basis Baringo Paleontological Research Project for comparison of new discoveries of Middle to between 9·5 Ma and 8·8 Ma (Deino, Miocene cercopithecoids. personal communication). There is also an Earlier studies identified two species of isolated colobine tooth from the site of the genus Victoriapithecus in the Maboko Nakali (Aguirre & Leakey, 1974; Benefit & collection, V. macinnesi and V. leakeyi (von Pickford, 1986). An isolated cercopithecoid

Koenigswald, 1969; Delson, 1973; Simons right P4 (KNM-BN 1251) from the & Delson, 1978; Szalay & Delson, 1979). Ngorora site discussed here (BPRP#38, These species were regarded as forming a =2/1) was also described in Benefit & separate subfamily, the Victoriapithecinae, Pickford (1986). Later, Benefit (1999)    77 mistakenly equates this Ngorora site with Pickford, 1997), and the Lukeino molar, Nakali. However, Nakali is not in the Tugen also from the Tugen Hills (see Hill, 1999)]. Hills, but on the eastern side of the Rift, and The present paper does not rectify this is several million years younger than deficiency significantly, but adds one, BPRP#38. and possibly two, hominoid teeth to the

The single right P4 from BPRP#38 has inventory. been the only evidence of African cerco- pithecoids yet reported from the crucial six million year interval between the last occur- Geology and dating rence of the Victoriapithecidae, around 15 Ma, at Maboko and Kipsaramon, and The Ngorora Formation was first investi- the first occurrence of the Colobinae at gated and mapped by Chapman, working Ngeringerowa at 9·5–8·8 Ma. The further with the East African Geological Research specimens described here from 12·5 Ma Unit (EAGRU), and first described by deposits in the Tugen Hills succession Bishop & Chapman (1970). Further notes represent an interval in time which could on the fauna appeared in Bishop et al. provide answers to important questions (1971) and the unit was discussed in more concerning the origin and radiation of the detail in Chapman (1971). Aguirre exca- two modern cercopithecid subfamilies. vated vertebrate fossil localities in the The record of African Miocene hominoids Ngorora Formation for one season (Aguirre is also poorly known after 14 Ma until the & Leakey, 1974), and Pickford undertook appearance of hominids in the Pliocene. more detailed investigations (Bishop & Again, the Ngorora Formation in the Tugen Pickford, 1975; Pickford, 1975a, 1978). Hills provides the next evidence in time. Initial age determinations are summarized in A molar tooth of uncertain attribution Chapman & Brook (1978). The work of the (KNM-BN 1378) is known from site Baringo Paleontological Research Project BPRP#60 in the Bartabwa section of the (BPRP) on the formation is reported in Hill Formation, which is not yet precisely dated, et al. (1985, 1986; Hill, 1995, 1999) and but is probably about 12 Ma (Bishop & more specific information concerning radio- Chapman, 1970; Hill & Ward, 1988; Hill, metric dating and paleomagnetic stratigra- 1994, 1999). A premolar similar to Proconsul phy appear in Tauxe et al. (1985) and Deino (KNM-BN 10489) comes from site et al. (1990). BPRP#65 (Hill et al., 1985; Hill & Ward, The unit is defined as lying between the 1988; Hill, 1994, 1999), and is dated at Tiim Phonolites beneath, and the Ewalel about 12·4 Ma (Deino et al., 1990). It is Phonolites above, and in the type section at associated with a canine (KNM-BN 10556) Kabasero there are about 370–400 m of in very poor condition, that could also be sediments. Volcaniclastic deposits of the hominoid, but if it is, it is unlikely to be from Ngorora Formation are extensive geographi- the same species as the premolar (Hill & cally, occurring in disjunct fault bounded Ward, 1988). Apart from these few speci- basins, located mainly in the north of the mens there is little from later in time than Tugen Hills range, but significant outcrops, the Fort Ternan site, dated at 14 Ma, which such as at Ngeringerowa, occur nearly has produced a variety of hominoids, but no 40 km further south. monkeys, until the Pliocene [additional The formation is also extensive in time, hominoid specimens in this interval include representing over 4 Ma, a remarkably Otavipithecus (Conroy et al., 1992), Sambu- long period for continental sediments in rupithecus from the Samburu Hills (Ishida & Africa. Deino et al. (1990) report on the 78 .  ET AL. lithostratigraphy, the paleomagnetic stra- series of epiclastic, feldspathic sandstones tigraphy and a series of single crystal laser and primary airfall tuffs. Sediments in this fusion 40Ar/39Ar age determinations. That portion of the Kabasero succession are com- paper revises and adds to the work published posed exclusively of pyroclastic material flu- in Tauxe et al. (1985) and Hill et al. (1985) vially reworked by alluvial fan processes and by applying single crystal laser fusion tech- exhibit channeling and cross bedding fea- niques to samples analyzed in these earlier tures. Preservation of glass shards and abun- studies, as well as to additional material. dant subangular pumice fragments in these Dates for tuffs cited here are weighted mean accumulations indicate limited transport ages of inferred primary populations of and weathering. Substantial local lateral single crystal feldspar populations (Deino variations in thickness, continuity and facies et al., 1990). Age determinations on the characterize the reworked beds. Intercalated Tiim phonolite flow underlying the base of with these sediments are subaerial and the type section give a date of 13·15 Ma, and water-laid tuffs with minimal evidence of sedimentation began shortly after this. Con- reworking, which form distinctive marker cordant dates on a series of tuffs through the horizons at Kabasero, providing a means of section document more or less continuous correlating between isolated or structurally sedimentation in the type section to younger and stratigraphically complex exposures. than 10·5 Ma. Elsewhere in the Formation, These tuffs include one with fossil plants to the south, we have dates as young as at site BPRP#55, an accretionary lapilli 8·8 Ma (Deino personal communication), tuff, and a crystal lithic tuff (RD53; Figure and dates on the capping Ewalel Phonolite 2) which bracket the BPRP#38 fossil local- are 7·6 Ma and 7·2 Ma (Deino et al., 1990). ity and provide a secure stratigraphic con- The series of radiometric determinations text for the site within the Ngorora from the Formation provide great control on sequence. the ages of fossil occurrences through this Euhedral feldspar crystals were collected relatively long time span. The relation of by Robert Drake from a poorly sorted, the radiometric information and the paleo- coarse-grained epiclastic sandstone directly magnetic stratigraphy, correlated to the overlying the accretionary lapilli tuff (RD- geomagnetic reversal time scale (GRTS), 54), a well cemented tuffaceous sandstone has implications for deep sea spreading about 7 m stratigraphically above the lapilli rates, and the nature and quality of the data tuff with in situ fossils, comprising the suggest revisions to GRTS for this portion of BPRP#38 site horizon (RD-51), and the time (Deino et al., 1990). crystal lithic tuff (RD-53). The original The particular site discussed here 40K/40Ar radiometric analyses on sanidine (BPRP#38; EAGRU 2/1; see Figure 1) was phenocrysts from these three tuff horizons discovered by Chapman in 1968 during the provided dates of 12·37 Ma, 12·7 Ma and course of his geological mapping and is the 11·57 Ma respectively (Hill et al., 1985; first fossil locality recorded from the For- Tauxe et al., 1985). More recent single- mation. It is located in the Kabasero type crystal, laser fusion, 40Ar/39Ar analyses on section of the Formation at UTM 36 the same samples have given corresponding 8116718 N 0099510 (N 054 E3551). dates of 12·560·01 Ma, 12·490·02 Ma Subsequent collections from the site were and 12·260·02 Ma (see Figure 2) which made by Aguirre, by Pickford, and by our are preferred here (Deino et al., 1990). The own expedition. Fossils at BPRP#38 are overall homogeneous nature, extremely poor associated with pumice and lithic pebble sorting, and generally massive character of conglomerate lenses interbedded with a the sediments between the accretionary    79

Figure 1. Map of the Tugen Hills, showing the location of site BPRP#38.

lapilli tuff and the RD-53 tuff suggests rapid Paleontology emplacement. The average sedimentation rate is 11·2 cm/ka for the interval between There is a diverse fauna from the Ngorora RD-54 and RD-51, and 9·9 cm/ka between Formation, ranging through the entire time RD-51 and RD-53. span of the unit, from 13 Ma to less than RD-51 was sampled from the same strati- 9 Ma. Although BPRP#38 is spatially graphic horizon as the BPRP#38 fossils, restricted (<12 m2), and most finds are which therefore can be dated at about isolated teeth, it has produced a great 12·49 Ma. taxonomic diversity (Table 1), several of the 80 .  ET AL.

12.26 + 0.07 Ma (RD-53) 40 m

30

20 Site BPRP #38 - Fossil horizons 12.49 + 0.06 Ma (RD-51)

12.56 + 0.04 Ma (RD-54) 10 Accretionary lapilli tuff

Site BPRP #55 - Fossil plant horizon

0 Tuff

clay Conglomerate silt fss mss Intercalated volcaniclastic siltstone cs ss and sandstones with minor cg conglomerate lenses Figure 2. Geological section through site BPRP#38 near the type section of the Ngorora Formation, indicating dates and relation of the fossil level to the plant tuff. (RD samples were collected by Robert Drake in 1982).

specimens being significant enough to merit level of geological formation. This list is particular taxonomic description in a range confined to one narrow time horizon and of publications (Pickford, 1975b, 1986; provides a good idea of an East African Gentry, 1978; Hamilton, 1978; Patterson, fauna tightly constrained to 12·5 Ma; a time 1978; Thomas, 1981a,b; Benefit & period otherwise unknown in sub-Saharan Pickford, 1986; Pickford & Fisher, 1987; Africa. Winkler, 1990, 1994, 2002). Most pre- It is interesting to compare the BPRP#38 viously published faunal lists from the fauna with that from Fort Ternan Tugen Hills have been assembled at the (Shipman, 1986; Harrison, 1992), a site    81

Table 1 Faunal list for site BPRP#38 in the Kabasero section of the Ngorora Formation

PISCES REPTILIA Crocodilia gen. et sp. indet. MAMMALIA Victoriapithecidae Victoriapithecus sp. indet. (23136–23143, BN 1251) (1) Hominoidea family indet. gen. et sp. indet. (BN 1461) (2) possibly gen. et sp. indet. #2 (23144) (2) Carnivora ?Ptolmaiidae cf. Kelba guadeemae (BN 10036) ?Mustelidae ?Mellivorinae gen. et sp. indet. (BN 2101) Hyaenidae gen. et sp. indet. (possibly a new species) (BN 10040) Felidae gen. et sp. indet. (BN 1441?) Tubulidentata Orycteropodidae ‘‘Orycteropus’’ chemeldoi (3) Hyracoidea Pliohyracidae Parapliohyrax ngororaensis (BN 232, BN 1243) (4) Proboscidea Family indet. gen. et sp. indet. (BN 1438) Perissodactyla Rhinocerotidae gen. et. sp. indet. (BN 554, BN 135, BN 145)

Continued overleaf. about 1·5 Ma older. At Fort Ternan no communication). A genet is known from crocodiles have been found. At BPRP#38 Fort Ternan; a felid, a possible arctocyonid there are no macroscelidids, or other insec- (Kelba), and a possible mustelid come from tivores. At Fort Ternan there are no mon- BPRP#38. The aardvark ‘‘Orycteropus’’ keys. A primate similar to Proconsul is chemeldoi is common to both sites. No hyrax present at both sites, but there is only one is known from Fort Ternan, whereas Para- additional possible catarrhine present at pliohyrax ngororaensis occurs at BPRP#38. BPRP#38 (see below), whereas, according There is an indeterminate proboscidean to Harrison (1992) Fort Ternan has, in from BPRP#38; Choerolophodon ngorora and addition to Proconsul, wickeri, Prodeinotherium come from Fort Ternan. Simiolus, something like Kalepithecus, and a Similarly, an unidentified rhinocerotid is species of oreopithecid. Among carnivores, known from BPRP#38; Chilotheridium and possibly three creodont taxa occur at Fort Paradiceros mukirii are recorded from Fort Ternan; none has been recorded from Ternan. BPRP#38. A hyaenid is present at both As many more fossil specimens have been localities, although the one at BPRP#38 recovered from Fort Ternan than from may be a new species (J. Barry personal BPRP#38, absences of taxa from Fort 82 .  ET AL.

Table 1 Continued

Artiodactyla Anthracotheriidae gen. et sp. indet. (BN 526) Suidae Lopholistriodon kidogosana (BN 862, BN 1162, BN 1254, BN 1259) (5) Palaeomerycidae Climacoceros gentryi (6) Giraffidae Paleotragus sp. (BN 172, BN 200) Tragulidae Dorcatherium sp. (BN 1159) Bovidae Boselaphini Protragocerus sp. (BN 52, 54, 55, 58, 61–63, 65, 67–69) (7) ?Cephalophini gen. et sp. indet. (BN 96) (8) ?Neotragini Homoiodorcas tugenium (BN 92) (9) ?Caprinae Pseudotragus? gentryi (BN 51, 350, 540, 1516, 1757) (10) Probably other species Rodentia (11) Muridae Cricetomyinae gen. et sp. nov. (BN 10990)

For vertebrates, names are followed by the accession numbers of examples of specimens that we believe establish that taxon at the site. All should be prefixed with KNM-TH except where otherwise indicated. The following are the taxonomic references that describe specimens from this site: (1) This paper [see also Benefit & Pickford (1986)]. (2) This paper. (3) Pickford (1975b) [Patterson (1978) believed the generic status to be uncertain]. (4) Pickford & Fischer (1987) describe two isolated teeth from this site (BN 232, BN 1243). Specimen KNM-BN 215 is listed by them as coming from site 2/11 (BPRP#39), also in the Kabasero section, but younger. All the data regarding early Baringo collections that still remain in the archives of the Kenya National Museum indicate that KNM-BN 215 comes from BPRP#38. If so it is interesting, as the specimen is a symphysis with left and right mandibular bodies, broken off behind P3. It suggests that specimens more complete than isolated teeth may be recoverable from BPRP#38 in the future. (5) Pickford (1986). (6) Hamilton (1978). (7) Gentry (1978). (8) Gentry (1978) [see also Thomas (1981a)]. (9) Thomas (1981a) [see also Gentry (1978)]. (10) Thomas (1981b). (11) Winkler (1990, 1994, 2001).

Ternan are more significant than absences Harrison (1992) documents a number of from the Ngorora site. The absence of changes between the faunas from Maboko victoriapithecids from Fort Ternan, for and Fort Ternan and suggests that this ‘‘. . . example, is probably biologically rather than could indicate that the slender estimated age statistically meaningful. On the other hand, difference between the sites actually requires the presence of only one or two hominoid revising, with Fort Ternan, perhaps, being species at BPRP#38 can be explained by the somewhat younger than 14 Ma’’. In this he relatively small sample size at the site, rather follows Pickford (1986), but Pickford’s than suggesting a genuine lower diversity. criterion is largely that the Fort Ternan    83 fauna is ‘‘more evolved’’—whatever that thonous, with leaf fall occurring soon after means—than that of Nyakach, which he eruption, and deposition was close to the dates at younger than 13·4 Ma, quoting fis- parent plant. Consequently the deposit sion track dates. Biostratigraphy is a respect- represents the composition and diversity of able technique for providing relative dates the original plant community very well. In on geological horizons, but when it is addition to whole leaf morphology, the employed in the context of understanding specimens allow the study of fine structure the nature and rate of faunal change itself, by scanning electron microscopy. More than then the circularity in argument can cause 55 species have been identified, which indi- problems. In addition, there is no real cate a tropical moist or wet forest, in the reason at present to doubt radiometric esti- tropical lower montane or premontane mates for the Fort Ternan fossiliferous forest category (Jacobs & Kabuye, 1987)in horizons of about 14 Ma. Moreover this this portion of the Rift Valley during the late corresponds with interpretations of the over- Middle Miocene. all geological history of the site and the dating of overlying and underlying strata Primates (Shipman et al., 1980). Possibly a real Primates from the site consist of eight teeth ff change may have a ected the east African belonging to a species of victoriapithecid fauna between 15 Ma and 14 Ma, and monkey, one incisor which we believe tentatively it also appears, on the basis belongs to a large hominoid, and one canine of the information provided about site tooth of less specific catarrhine attribution. BPRP#38 here, that the new state inaugu- The incisor specimen (KNM-BN 1461) rated by the change persisted until at least was collected by Aguirre’s expedition in 12·5 Ma. 1969, but is not mentioned in his publi- An interesting feature of BPRP#38 is its cation (Aguirre & Leakey, 1974) and appar- proximity to one of the best Neogene fossil ently was not identified. Benefit & Pickford macrofloras in Africa. Only 15 m below the (1986) describe a single right P (KNM-BN ff 4 primate horizon is a tu aceous unit that 1251) from the site. In 1990 BPRP found preserves abundant fossils of branches and additional specimens on the surface at leaves (BPRP#55). Jacobs & Kabuye are BPRP#38 and we retrieved others on studying this flora (Kabuye & Jacobs, 1986; subsequent dry sieving of the site later that Jacobs & Kabuye, 1987, 1989; Jacobs & year. Winkler, 1992). Originally published as 12·2 Ma in age (Jacobs & Kabuye, 1987), Description of the material our subsequent redating of the sequence Cercopithecoidea: Victoriapithecus sp. indet. now permits a more accurate estimate. The tuff dated at 12·56, RD54 (Deino et al., KNM-TH 23143, right P4 (Figure 3). This 1990), occurs just above, suggesting, on the complete isolated P4, with roots, is fairly basis of interpolation between dated hori- worn, so that detail on the occlusal surface is zons, an age for the plant horizon of lost. A lozenge shaped dentine area is 12·59 Ma. exposed on the paracone and a smaller bell The plant fossils occur in a 20-cm-thick shaped area on the protocone. Although it is accretionary lapilli tuff in the form of whole not always possible to confidently differenti- leaves which are extremely well preserved. ate isolated upper premolars, features listed Jacobs & Winkler (1992) provide detailed by Benefit (1993), including the absence of information on the taphonomy and mode of a rootward extension of enamel onto the origin of the deposit. The deposit is autoch- mesiobuccal aspect of the root (Delson, 84 .  ET AL.

Figure 3. Upper teeth of Victoriapithecus. Above, from left to right, occlusal, mesial, lingual and distal ?2 views of right M , KNM-TH 23137. Below left, occlusal and mesial views of right P4, KNM-TH 23143. Below right, occlusal views of right M?1, KNM-TH 23138. The teeth have been coated with ammonium chloride for greater clarity of the occlusal views.

1973), little disparity between the heights of length is greater than the paracone height the protocone and paracone and lack of a (Benefit, 1993). sharply angled mesiobuccal corner to the occlusal surface, indicate that this tooth is KNM-TH 23138, right M?1 (Figure 3). This almost certainly a P4. well-preserved upper molar crown, possibly The tooth is indistinguishable from the P4 M1, lacks roots and is slightly worn. Small of V. macinnesi. Relative to extant monkeys circles of dentine are exposed on the buccal the Victoriapithecus P4s are relatively wider cusps and larger bell shaped areas on the buccolingually and have a relatively long lingual cusps (Benefit’s wear stage 5) mesial shelf (Benefit, 1993). This P4 is rela- (Benefit, 1987). Characteristic of all V. mac- tively wide buccolingually (W/L=1·36, and innesi upper molars, and in contrast to many the mean for Victoriapithecus is 1·39). The other Old World monkeys, the tooth is protocone is set just mesial to the paracone, wider than it is long, the mesial and distal and there is no fissure separating the two widths are subequal and both are greater cusps. The height of the paracone of this than the length. The buccal and lingual worn tooth is equal to the crown length sides of the crown flare from the cusp tip to (4·58 mm) indicating that in the unworn the cervix and this is particularly pro- state the paracone height would have nounced lingually. A crista obliqua links the exceeded the crown length, the condition protocone and metacone and there is no typical of V. macinnesi and Colobus.InCer- transverse distal loph linking the metacone cocebus, Cercopithecus and Papio, the crown and hypocone as is characteristic of the    85

Cercopithecidae. The metacone portion of the M1 terminates well above the cervix. the crista obliqua forms a slight oblique The mesial lingual groove is distinct and the crest, partially restricting the trigon basin. buccal lingual groove absent although there The protocone portion (postprotocrista) is a hint of a depression in the position of the intersects the metacone portion at a shallow distal lingual groove. groove, but occlusal wear has obscured mor- phological details. The trigon basin is small KNM-BN 1251, right P4 (Figure 4). This relative to the crown area and crown length, little worn P4 crown has a small dentine partly due to the close proximity of the circle exposed on the protoconid. It is rela- buccal cusps to the lingual cusps, and partly tively long compared to its buccolingual due to the crista obliqua restricting its size. width and the enamel extends towards the Both cusp proximity and the presence of a root on the mesiobuccal aspect of the proto- crista obliqua are features of Victoriapithecus. conid, so that buccally the protoconid Among other features of this tooth is a small appears high. The mesial shelf is distinctly cuspule at the base of a deep median lingual shorter than the distal shelf. The anterior cleft which terminates high above the cervix. fovea is small compared to the posterior The mesial lingual groove is distinct and fovea and a lophid links the protoconid and V-shaped, and both the mesial and distal metaconid. There is a small but distinct buccal grooves, although short, are clearly cuspule just distal to the metaconid cusp defined. There is a slight depression which apex. The protoconid is the larger cusp and hints at the presence of a distal lingual is higher than the metaconid, but in contrast groove. to V. macinnesi, this is true both when the occlusal relief is expressed as the distance KNM-TH 23137, right M?2 (Figure 3). This from the cusp tip to a point on the cervix well-preserved right upper molar, possibly and as the height of the metaconid above the M2, lacking most of the roots, is only slightly lingual notch relative to that of the proto- worn with no dentine exposed (Benefit’s conid above the buccal notch). This is wear stage 2). The tooth is much wider than characteristic of some colobines and in it is long and the mesial width is greater than contrast to cercopithecines and Victoria- the distal. Characteristic of Victoriapithecus, pithecus (Benefit, 1993). The tooth also the unworn buccal and lingual cusp tips are differs from that of V. macinnesi in its in close proximity and the buccal and lingual dimensions. The V. macinnesi P4 is relatively sides of the crown flare even more markedly wider than those of most extant species, than those of the M1, KNM-TH 23138, being most similar to Cercocebus, and the described above. The flare is particularly metaconid and protoconid are set far apart, pronounced lingually and beneath the pro- whereas the cusps of the BPRP#38 P4 are tocone there is a distinct bulge. The trans- more closely proximated and the tooth verse distal loph is absent and the crista significantly longer than wide. ff obliqua is similar in position and develop- The V. macinnesi P4 di ers from modern ment to that of the M1, although, because of cercopithecids in its oblique orientation to the lack of occlusal wear, it is more distinct. the tooth row, a feature considered to be The postprotocrista is clearly visible and primitive. Although it is difficult to assess intersects the obliquely oriented metaconid the orientation of an isolated tooth, several portion of the crista obliqua at a slight angle. features of KNM-BN 1251 suggest that its The triangular trigon basin is relatively small orientation was also skewed. When the tooth and restricted by the crista obliqua. The is held so that the lingual side would be median lingual cleft is deep and like that of parallel to that of the molars, the anterior 86 .  ET AL.

Figure 4. Lower canine of gen. et sp. indet. (above left) and lower cheek teeth of Victoriapithecus. Above left, mesial, distal and occlusal views of right male lower canine, KNM-TH 23144.

Above right, occlusal and buccal views of left M?1, KNM-TH 23142. Centre left, occlusal and buccal views of right P4, KNM-BN 1251. Centre right, occlusal and buccal views of right M?2, KNM-TH 23141. Below left, occlusal, buccal and mesial views of right M3, KNM-TH 23136. Below right, occlusal and buccal views of right M3, KNM-TH 23139. The teeth have been coated with ammonium chloride for greater clarity of the occlusal views. fovea is lingual to the midline of the crown dentine patches are exposed on the two and the protoconid is mesial to the meta- buccal cusps (Benefit’s wear stage 3). The conid. In addition, the distal contact facet is tooth has a rather square outline: the mesial on the mesial buccal border of the tooth and distal widths are almost identical and rather than on the midline. These typical only slightly less than the mesiodistal length. victoriapithecid features suggest that the The molar is bilophodont and has a very natural orientation of the tooth was oblique small but distinct hypoconulid. Although to the molar row. the small size of this cusp argues against it being a true hypoconulid, its position

KNM-TH 23142, left M?1 (Figure 4). This is is identical to that of V. macinnesi another well-preserved lower molar crown hypoconulids and it is almost certainly which retains only a small portion of the homologous. The height of the metaconid distal root. There is slight wear, so that small is greater than that of the entoconid, a    87 character also shared with V. macinnesi. groove on KNM-TH 23136 (this area is There is a shallow but distinct median lin- missing on KNM-TH 23139). gual notch and deep median buccal notch. The mesial buccal groove is similar in size to Hominoidea: gen. et sp. indet. the distal buccal groove, although the latter KNM-BN 1461, right I . This specimen is a incises the crown more deeply. Both mesial 1 crown of an I . Wear on the incisal edge has and distal lingual grooves are absent. 1 removed the mammelons. Slight chipping of the mesial and distal surfaces of the crown KNM-TH 23141, right M?2 (Figure 4). This has eradicated interproximal contact facets. well-preserved lower molar crown, possibly Overall the crown is relatively tall and nar- M2, is slightly more worn than the M?1 row. In size it is comparable to the mandibu- described above. Small dentine pits can be lar central incisors of Proconsul hesloni. seen on all the cusps except the metaconid The labial surface is essentially featureless which shows slight signs of wear (Benefit’s except for an inter-mammelon groove that wear stage 4). There is no sign of a descends from the incisal edge to a point hypoconulid. There is marked flare on the about half way to the labial cervix. The buccal face from cusp tip to cervix with lingual surface is almost devoid of surface distinct bulging below the median buccal detail. There is a poorly expressed lingual cleft, whereas the lingual face is rather steep pillar. Neither a basal tubercle nor any sign sided. The mesial and distal shelves are of a lingual cingulum is evident. Compari- short and the mesial shorter than the distal. sons of KNM-BN 1461 with modern and The mesial and distal buccal grooves are fossil cercopithecoids and hominoids sup- distinct and short, and on both, the mesial port attribution of the Ngorora incisor to and distal lingual grooves are absent. Hominoidea, gen. et sp. indet. In particular, the height of the crown relative to its basal KNM-TH 23136 and 23139, right M (Figure diameter, in addition to its robusticity, 3 relative to that of cercopithecoids and 4). These two M crowns are similar in 3 archaic or ‘‘stem’’ hominoids clearly justify morphology and at a similar stage of wear this assignment. Other characters support- with distinct dentine circles exposed on the ing assignment to hominoidea include a two buccal cusps (Benefit’s wear stage 4). vertical, rather than curved labial crown KNM-TH 23136 has the roots intact. surface, and absence of lingual cingulum KNM-TH 23139 lacks the anterior root and elements. the crown is broken mesially and lingually and the metaconid lost. Catarrhini: gen. et sp. indet. The hypoconulid is well-developed and positioned slightly buccally. That of KNM-TH 23144, lower right C (Figure 4). KNM-TH 23139 contributes to a rather The root of this lower male canine is pointed distal margin in contrast to broken off just beneath the cervix and is KNM-TH 23136 which is rounded. A small largely missing. The crown, which lacks a tuberculum sextum, represented by a raised small part of the tip, is otherwise well area of enamel, can be observed between the preserved and has only slight wear on the metaconid and hypoconid on each tooth. mesial surface close to the tip. Like V. The mesial width of KNM-TH 23136 is macinnesi the crown is low relative to the greater than the distal width. There is a width. This is in contrast to those of most distinct mesial buccal groove on both teeth other cercopithecoids, which have relatively and there is no trace of a mesial lingual higher crowned lower male canines. In size 88 .  ET AL. the canine is absolutely larger than any and differentially expressed along the molar Victoriapithecus specimen recorded from row, being most frequent on M1 and least Maboko (Table 2). frequent on M3. Both the M1 and M2 from The canine differs from that of V. macin- BPRP#38 retain a crista obliqua. True distal nesi and other cercopithecoids in several lophs or hypolophs are rare among V. significant features, which suggest it is cer- macinnesi upper molars, and when they tainly not Victoriapithecus and it is unlikely to occur they are frequently in combination be a cercopithecoid. There is a distinct cin- with the crista obliqua or other crests gulum that extends lingually from the base (Benefit, 1993). Both of the BPRP#38 of the mesial ridge around a prominent heel upper molars lack distal lophs. V. macinnesi onto the distal aspect of the crown. Cerco- unworn molars have lingual and buccal pithecoids generally lack a cingulum and cusps more closely approximated than those when present it is short and restricted to the of most Cercopithecidae. The close prox- base of the mesial ridge. V. macinnesi lacks a imity of the cusps of the lightly worn M2, prominent heel. The mesial ridge and the KNM-TH 23137, is consistent with those of parallel running mesial sulcus are long, V. macinnesi. This results in marked flaring straight, distinct and evenly developed. The of the buccal and lingual sides of the crown mesial ridge and sulcus of cercopithecoids from tip to cervix, and a bulge is often are most prominent towards the base. At apparent on the buccal side of the crown the base of the mesial ridge, the enamel below the base of the median lingual cleft, a margin is only slightly raised, whereas in feature common to both V. macinnesi and cercopithecoids it is markedly raised to pro- the BPRP#38 molars. The lower molars duce a tight V with the apex at the base of also show similarities in the presence of the the ridge. There are two additional sulci, hypoconulid on the M1 and in the relative both absent on Victoriapithecus lower dimensions of this tooth which is only canines, on the buccal aspect, either side of slightly longer than wide. The retention of a the buccal margin. Both are deeper towards hypoconulid on M1 and M2 is variable in V. the base of the crown and peter out towards macinnesi, and the M1 is more square than the tip. that of modern extant cercopithecids The canine differs from known cerco- (Benefit, 1987). The lower third molars, like pithecoid and hominoid lower canines. It those of V. macinnesi, show a high disparity probably has more features in common with between the mesial and distal widths. The ff hominoids, but in the absence of associated di erence between the M3 mesial and distal posterior teeth it is impossible to refer it with widths of the Maboko monkeys is more any certainty to one or the other. extreme than that observed for the majority of extant species with the exception of Cercocebus albigena (Benefit, 1987). These Discussion characters of the molars indicate affinities Comparison of the eight cercopithecoid with the Victoriapithecidae. teeth with V. macinnesi reveals strong simi- In contrast, the isolated P4 is distinguish- larities between the upper cheek teeth and able from V. macinnesi. Although it shares lower molars, which are indistinguishable primitive traits typical of the Victoriapithe- from comparable teeth from Maboko and cidae it is distinguished by other features. show the same diagnostic primitive char- The primitive traits relate to its skewed acters. Benefit (1987) found that in V. orientation relative to the long axis of the macinnesi, the development, orientation and molar row, such that the mesial portion of presence of the crista obliqua was variable the tooth is buccal to the distal portion. This Table 2 Measurements of cercopithecoid teeth, and of Catarrhini gen. et sp. indet. (KNM-TH 23144)

CP4M1M2M3 Upper teeth LWLW LAWDWLAWDWLAWDW

KNM-TH 23143 4·58 6·24 KNM-TH 23138 6·47 7 6·9 KNM-TH 23137 6·9 8·4 7·87 Victoriapithecus Mean 4·65 6·5 6·17 6·59 6·3 7.28 8·06 7·41 Range 4·25–4·8 6·0–7·0 5·45–6·75 5·9–7·05 5·25–6·85 6·55–8·5 6·7–9·5 6·2–8·65

S.D. 0·2734 0·29775 0·2711 0·2767 0·269 0·4666 0·5832 0·5    CV 5·88 4·57 4·34 4·2 4·3 6·4 7·23 6·76 n 9 9 38 34 37 57 55 57

CP4M1M2M3 Lower teeth L W OL W L AW DW L AW DW L AW DW

KNM-TH 23144 7·54 4·97 KNM-BN 1251 5·52 4·22 KNM-TH 23142 6·37 5·43 5·56 KNM-TH 23141 7·62 7·16 7·14 KNM-TH 23136 8·98 5·56 4·7 KNM-TH 23139 8·9 Victoriapithecus Mean 6·62 4·89 5·58 4·72 6·12 5·26 5·3 7·38 6·72 6·4 8·87 6·33 5·34 Range 5·7–7·35 4·0–5·85 3·65–6·8 3·75–6·5 5·55–6·9 4·25–5·6 4·5–6·3 6·2–8·85 5·7–8·65 5·3–7·75 6·8–10·85 5·15–7·65 4·1–6·6 S.D. 0·372 0·3608 0·738 0·5284 0·2957 0·2795 0·3642 0·5187 0·5854 0·5332 0·9383 0·5568 0·5176 CV 5·62 7·4 13·2 11·2 4·83 4·7 7 7 8·71 8·33 10·57 8·8 9·7 n 47 47 41 49 40 39 29 86 85 89 75 72 70

The Victoriapithecus measurements are from Benefit (1993). Abbreviations are as follows: L: maximum mesiodistal length, W: maximum buccolingual width, AW: maximum mesial buccolingual width, DW: maximum distal buccolingual width, OL: maximum length of the P4 between mesial and distal margins of the occlusal surface. 89 90 .  ET AL. is expressed in a number of features them being associated with one or other described above which the BPRP#38 P4 of these three individuals. shares with the P4 of V. macinnesi. The As for the noncercopithecoid teeth, ff BPRP#38 P4 di ers from that of V. macin- KNM-BN 1461 is morphologically almost nesi in other significant ways including identical to the mandibular incisors of the reduced width relative to the mesio- KNM-RU 7290, Proconsul heseloni (Walker distal length (occlusal length/width=1·3), et al., 1993). However, it is also very similar and the higher protoconid relative to the in the features described here to KNM-MB metaconid. 11830, attributed to Equatorius africanus Prohylobates is not well represented in (Pickford, 1985; Ward et al., 1999). Based the fossil record so that characters that on the description of a newly erupted, are diagnostic of Prohylobates are not unworn E. africanus I2 (McCrossin & well defined. More complete Prohylobates Benefit, 1993), the Ngorora incisor mor- remains are needed to fully understand phology would seem compatible with the the morphological characteristics of the lateral incisors of Equatorius. In the final genus. analysis, beyond assigning KNM-BN 1461 The precise taxonomic identification of to Hominoidea, specifically to a large isolated teeth is often problematic since hominoid taxon, little more can be done to many diagnostic characters relate to the rela- identify the affinities of the specimen. tive size and proportions of the teeth. It is The mandibular canine KNM-TH 23144 often impossible to identify the serial pos- is also possibly attributable to a hominoid ition of isolated molars with any confidence. primate. As this canine is clearly not Procon- The isolated teeth described above clearly sul, and as mandibular canines are poorly demonstrate the presence at BPRP#38 of diagnostic generally, there is a possibility (1) a victoriapithecid with upper and lower that a large hominoid taxon hitherto molar teeth which are indistinguishable unknown was present in the Tugen Hills from those of V. macinnesi, and (2) a area about 12·5 Ma. A similar situation victoriapithecid which can probably be dis- exists at site BPRP#65, in the Bartabwa area tinguished from V. macinnesi on the mor- of the Ngorora Formation, and dated at phology of the P4. The question remains 12·42 Ma (Hill et al., 1985; Deino et al., as to whether the BPRP#38 collection 1990), where a large hominoid P4 represents one or more species. While more (KNM-BN 10489) was recovered at the complete material with associated anterior same locality as KNM-BN 10556, a poorly and posterior teeth is required to resolve this preserved canine (Hill & Ward, 1988). question, it is possible to consider the speci- As noted by Alpagut et al. (1990), fossil mens as representing a single new species hominoid P4s are notoriously variable, and of Victoriapithecus, distinguished from V. difficult to orient when isolated. Hill & macinnesi on the morphology of the P4. But Ward (1988) were inclined to assign the at present the material is considered too BPRP#65 P4 provisionally to Proconsul, incomplete to be confident of this and to while recognizing the tenuous nature of the permit the naming and diagnosis of a new attribution. If the enigmatic canine from the species. This collection of eight teeth could Bartabwa locality is a hominoid, however, it represent as few as three individuals. The is clearly not attributable to Proconsul (Hill & two right lower M3s are very lightly worn Ward, 1988). Whether or not Proconsul whereas the upper M?2 is unworn so must exists at these Ngorora sites, there is cer- represent a third individual. The wear on the tainly another large hominoid present, remaining teeth does not preclude any of which is so far little known.    91

Summary (BPRP), based at Yale University, and car- The collection of eight isolated cercopithecoid ried out in collaboration with the National teeth from the Ngorora Formation, Tugen Museums of Kenya. We thank the Govern- Hills, shows that victoriapithecids were still ment of the Republic of Kenya for present 12·5 Ma ago, two and a half million permission to carry out research in Kenya years later in time than their previously estab- (Permit OP/13/001/C 1391/ issued to AH), lished last appearance. However, they retained and permission to excavate from the primitive characters typical of their 15 Ma Minister for Home Affairs and National predecessors at Maboko. These characters Heritage. BPRP has been supported by include: relatively wide upper molars with grants to AH from NSF (most recently closely approximated buccal and lingual SBR-9208903), the Louise H. and David S. cusps, marked buccal and lingual flare, vari- Ingalls Foundation, the Louise Brown able occurrence of a crista obliqua and ab- Foundation, Clayton Stephenson, and Yale sence of distal lophs; skewed orientation of the University. John Kingston received funding

P4 relative to the long axis of the molar row; from NSF (# BNS 9014433), from the M1s that are relatively square; variable reten- Bill Bishop Memorial Fund, Boise Fund, tion of M1 and M2 hypoconulids; and M3s that GSA, Sigma Xi and the L.S.B. Leakey are distally constricted. Although there are Foundation. Thanks for field help are some morphological similarities with colob- due to Boniface Kimeu, Kiptalam Cheboi ines in the P4, other characters in common and Sally McBrearty. We had useful with the victoriapithecids are shared with the discussions with Al Deino concerning the cercopithecids (Benefit, 1987, 1993). Overall, geochronology. John Barry generously pro- the affinities are clearly with Victoriapithecus. vided unpublished notes on some of the There is no evidence in this collection of Carnivora, and gave us helpful comments isolated teeth to suggest that at 12·5 Ma the on the manuscript. Alan Walker very kindly colobine or cercopithecine subfamilies had took the photographs that form Figures 3 evolved. However, this does not constitute evi- and 4. dence that they had not. Colobines are known from about 9·5 Ma at Ngeringerowa, in References the Tugen Hills (Benefit & Pickford, 1986; Gundling & Hill, 2000), and it is by no means Aguirre, E. & Leakey, P. (1974). Nakali: nueva fauna necessary that they evolved from the Victori- de Hipparion del Rift Valley de Kenya. Estudios Geologicos 30, 219–227. apithecidae, nor that, even if they did, victori- Alpagut, B., Andrews, P. & Martin, L. (1990). New apithecids should not exist after they had hominoid specimens from the Middle Miocene site at diverged. This cercopithecoid at BPRP#38 Pasalar, Turkey. J. hum. Evol. 19, 397–422. Behrensmeyer, A. K., Deino, A., Hill, A., Kingston, occurs in association with at least one large J. K. & Saunders, J. (2002). Geology and geo- hominoid species and a diverse mammalian chronology of the middle Miocene Kipsaramon fauna reminiscent in some ways of that from Site Complex, Muruyur Beds, Tugen Hills, Kenya. J. hum. Evol. 1/2, 11–38. Fort Ternan. The horizon is not far removed Benefit, B. (1987). The molar morphology, natural in time from a varied fossil macroflora, history, and phylogenetic position of the middle indicating a tropical moist or wet forest. Miocene monkey Victoriapithecus. Ph.D. Disser- tation. New York University. Benefit, B. (1993). The permanent dentition and phylogenetic position of Victoriapithecus from , Kenya. J. hum. Evol. 25, 83–172. Acknowledgements Benefit, B. (1994). Phylogenetic, paleodemographic, and taphonomic implications of Victoriapithecus This research is part of the work of the deciduous teeth from Maboko, Kenya. Am. J. phys. Baringo Paleontological Research Project Anthrop. 95, 277–331. 92 .  ET AL.

Benefit, B. (1999). Victoriapithecus: the key to Old Hamilton, W. R. (1978). Fossil giraffes from the World Monkey and Catarrhine origins. Evol. Miocene of Africa and a revision of the phylogeny of Anthrop. 7, 155–174. the Giraffoidea. Phil. Trans. R. Soc. Lond. B 283, Benefit, B. (2000). origins and 165–229. diversification: an evolutionary study of diet and Harrison, T. (1987). The phylogenetic relationships of dentition. In (P. F. Whitehead & C. J. Jolly, Eds) Old the early catarrhine primates: a review of the current World Monkeys, pp. 133–79. Cambridge: Cambridge evidence. J. hum. Evol. 16, 41–80. University Press. Harrison, T. (1992). A reassessment of the taxonomic Benefit, B. & Pickford, M. (1986). Miocene fossil and phylogenetic affinities of the fossil catarrhines cercopithecoids from Kenya. Am. J. phys. Anthrop. from Fort Ternan, Kenya. Primates 33, 501–522. 69, 441–464. Hill, A. (1994). Late Miocene and early Pliocene Benefit, B. & McCrossin, M. (1989). New primate Hominoids from Africa. In (R. S. Corruccini & R. L. fossils from the middle Miocene of Maboko Island, Ciochon, Eds) Integrative Paths to the Past: Paleo- Kenya. J. hum. Evol. 18, 493–497. anthropological Advances in Honor of F. Clark Howell, Benefit, B. & McCrossin, M. (1993). Facial anatomy of pp. 123–145. Englewood Cliffs, NJ: Prentice Hall. Victoriapithecus and its relevance to the ancestral Hill, A. (1995). Faunal and environmental change in cranial morphology of Old World monkeys and . the Neogene of east Africa: evidence from the Tugen Am. J. phys. Anthrop. 92, 329–370. Hills sequence, Baringo District, Kenya. In (E. S. Benefit, B. & McCrossin, M. (1997). Earliest known Vrba, G. H. Denton, T. C. Partridge & L. H. Old World Monkey skull. Nature 388, 368–371. Burckle, Eds) Paleoclimate and Evolution, With Bishop, W. W. & Chapman, G. R. (1970). Early Emphasis on Human Origins, pp. 178–193. New Pliocene sediments and fossils from the northern Haven: Yale University Press. Kenya Rift Valley. Nature 226, 914–918. Hill, A. (1999). The Baringo Basin, Kenya: from Bill Bishop, W. W. & Pickford, M. H. L. (1975). Geology, Bishop to BPRP. In (P. Andrews & P. Banham, Eds) fauna and palaeoenvironments in the Ngorora For- Late Cenozoic Environments and Hominid Evolution: a mation, Kenya Rift Valley. Nature 254, 185–192. Tribute to Bill Bishop, pp. 85–97. London: Geological Bishop, W. W., Chapman, G. R., Hill, A. & Miller, Society of London. J. A. (1971). Succession of Cainozoic vertebrate Hill, A. & Ward, S. (1988). Origin of the : assemblages from the northern Kenya Rift Valley. the record of African large hominoid evolution Nature 233, 389–394. between 14 My and 4 My. Yearb. phys. Anthrop. 31, Chapman, G. R. (1971). The geological evolution of 49–83. the northern Kamasia Hills, Baringo District, Kenya. Hill, A., Drake, R., Tauxe, L., Monaghan, M., Barry, Ph.D. Dissertation, University of London. J. C., Behrensmeyer, A. K., Curtis, G., Fine Jacobs, Chapman, G. R. & Brook, M. (1978). Chronostratig- B., Jacobs, L., Johnson, N. & Pilbeam, D. (1985). raphy of the Baringo Basin, Kenya Rift Valley. In Neogene paleontology and geochronology of the (W. W. Bishop, Ed.) Geological Background to Fossil Baringo basin, Kenya. J. hum Evol. 16, 749–773. Man, pp. 207–223. Edinburgh: Scottish Academic Hill, A., Curtis, G. & Drake, R. (1986). Sedimentary Press. stratigraphy of the Tugen Hills, Baringo District, Conroy, G. C., Pickford, M., Senut, B., Van Kenya. In (L. E. Frostick, R. W. Renaut, I. Reid & Couvering, J. & Mein, P. (1992). Otavipithecus J-J. Tiercelin, Eds) Sedimentation in the African Rifts, namibiensis, first Miocene hominoid from southern pp. 285–295. Oxford: Blackwell, Geological Society Africa. Nature 356, 144–148. of London Special Publication No. 25. Deino, A., Tauxe, L., Monaghan, M. & Drake, R. Hill, A., Behrensmeyer, K., Brown, B., Deino, A., (1990). Single crystal 40Ar/39Ar ages and the litho Rose, M., Saunders, J., Ward, S. & Winkler, A. and paleomagnetic stratigraphies of the Ngorora (1991). Kipsaramon: a lower Miocene hominoid site Formation, Kenya. J. Geol. 98, 567–587. in the Tugen Hills, Baringo District, Kenya. J. hum. Delson, E. (1973). Fossil colobine monkeys of the Evol. 20, 67–75. circum-Mediterranean region and the evolutionary Hooijer, D. A. (1963). Miocene mammalia of the history of the Cercopithecidae. (Primates, Mam- Congo. Ann. Mus. R. Afr. Cent. (Ser. 8, Sci. Geol.) malia). Ph.D. Dissertation, Columbia University. 46, 1–71. Delson, E. (1979). Prohylobates from the early Miocene Ishida, H. & Pickford, M. (1997). A new Late Miocene of Libya: a new species and its implications for hominoid from Kenya: kiptalami cercopithecid origins. Geobios 12, 725–733. gen. et sp. nov. C.r. Acad. Sci, Paris 325, 823– Gentry, A. (1978). The fossil Bovidae of the Baringo 829. Area, Kenya. In (W. W. Bishop, Ed.) Geological Jacobs, B. F. & Kabuye, C. H. S. (1987). Environ- Background to Fossil Man, pp. 293–308. Edinburgh: ments of early hominoids: evidence for middle Scottish Academic Press. Miocene forest in East Africa. J. hum. Evol. 16, Gundling, T. & Hill, A. (2000). Geological context of 147–155. fossil Cercopithecoidea from eastern Africa. In (P. F. Jacobs, B. F. & Kabuye, C. H. S. (1989). An extinct Whitehead & C. J. Jolly, Eds) Old World Monkeys, species of Pollia Thunberg (Commelinaceae) from pp. 180–213. Cambridge: Cambridge University the Miocene Ngorora Formation, Kenya. Rev. Pal- Press. aeobot. Palynol. 59, 67–76.    93

Jacobs, B. F. & Winkler, D. A. (1992). Taphonomy of Shipman, P. (1986). Paleoecology of Fort Ternan a middle Miocene autochthonous forest assemblage, reconsidered. J. hum. Evol. 15, 193–204. Ngorora Formation, central Kenya. Palaeogeogr., Shipman, P., Walker, A., Van Couvering, J. A., Palaeoclimatol., Palaeoecol. 99, 31–40. Hooker, P. & Miller, J. A. (1980). The Fort Ternan Kabuye, C. H. S. & Jacobs, B. F. (1986). An interesting hominoid site, Kenya: geology, age, taphonomy and record of the genus Leptaspis, Bambusoidea from paleoecology. J. hum. Evol. 10, 49–72. Middle Miocene flora deposits in Kenya, East Africa. Simons, E. (1969). A Miocene monkey (Prohylobates) In Abstracts: International Symposium on Grass System- from northern Egypt. Nature 223, 687–689. atics and Evolution, pp. 32. Washington, DC: Simons, E. & Delson, E. (1978). Cercopithecidae and Smithsonian Institution. Parapithecidae. In (V. J. Maglio & H. B. S. Cooke, Leakey, M. G. (1985). Early Miocene cercopithecoids Eds) Evolution of African , pp. 100–119. from Buluk, northern Kenya. Folia primatol. 44, Cambridge: Harvard University Press. 1–14. Szalay, F. S. & Delson, E. (1979). Evolutionary History Le Gros Clark, W. E. & Leakey, L. S. B. (1951). The of the Primates. New York: Academic Press. Miocene Hominoidea of East Africa. In Fossil Mam- Tauxe, L., Monaghan, M., Drake, R., Curtis, G. & mals of Africa: No. 1, pp. 1–117. British Museum Staudigel, H. (1985). Paleomagnetism of Miocene (Natural History). east African Rift sediments and the calibration of the McCrossin, M. & Benefit, B. (1993). Recently recov- Geomagnetic Reversal Timescale. J. geophys. Res. 90, ered Kenyapithecus mandible and its implications for 4639–4646. great and human origins. Proc. natl. Acad. Sci. Thomas, H. (1981a). Les bovide´s mioce`nes de la U.S.A. 90, 1962–1966. formation de Ngorora du bassin de Baringo Patterson, B. (1978). Pholidota and Tubulidentata. In (Rift Valley, Kenya) II. Proc. Konink. Ned. Akad. (V. J. Maglio & H. B. S. Cooke, Eds) Evolution of Wetensch. B 84, 357–375. African Mammals, pp. 268–278. Cambridge, MA: Thomas, H. (1981b). Les bovide´s mioce`nes de la Harvard University Press. formation de Ngorora du bassin de Baringo Pickford, M. H. L. (1975a). Stratigraphy and palaeo- (Rift Valley, Kenya) III. Proc. Kon. Ned. Acad. Wet. B ecology of five late Cainozoic Formations in the 84, 377–392. northern Kenya Rift Valley. Ph.D. Dissertation, University of London. von Koenigswald, G. H. R. (1969). Miocene Cerco- Pickford, M. H. L. (1975b). New fossil Orycteropodi- pithecoidea and Oreopithecoidea of East Africa. In dae (Mammalia, Tubulidentata) from East Africa. (L. S. B. Leakey, Ed.) Fossil Vertebrates of Africa: Orycteropus minutus sp. nov. and Orycteropus chemeldoi Volume 1, pp. 39–52. New York: Academic Press. sp. nov. Netherlands J. Zool. 25, 57–88. Walker, A., Teaford, M. F., Martin, L. & Andrews, P. Pickford, M. H. L. (1978). Geology, palaeoenviron- (1993). A new species of Proconsul from the early ments and vertebrate faunas of the mid-Miocene Miocene of Rusinga-Mfangano Islands, Kenya. J. Ngorora Formation, Kenya. In (W. W. Bishop, Ed.) hum. Evol. 25, 43–46. Geological Background to Fossil Man, pp. 237–262. Ward, S., Brown, B., Hill, A., Kelley, J. & Downs, W. London: Scottish Academic Press. (1999). Equatorius: a new hominoid genus from the Pickford, M. H. L. (1985). A new look at Kenyapithecus Middle Miocene of Kenya. Science 285, 1382–1386. based on recent discoveries in western Kenya. J. Winkler, A. (1990). Systematics and biogeography of hum. Evol. 14, 113–143. Neogene rodents from the Baringo District, Kenya. Pickford, M. H. L. (1986). A revision of the Miocene Ph.D. Dissertation, Southern Methodist University. Suidae and Tayassuidae (Artiodactyla, Mammalia) Winkler, A. J. (1994). The middle/upper Miocene of Africa. Tertiary Research: Special Paper No. 7, dispersal of major rodent groups between southern 1–83. Asia and Africa. In (Y. Tomida, C. Li & Pickford, M. H. L. & Fischer, M. S. (1987). Paraplio- T. Setoguchi, Eds) Rodent and Lagomorph Families of hyrax ngoroaensis, a new hyracoid from the Miocene Asian Origins and Diversification, pp. 173–184. of Kenya, with an outline of the classification of National Science Museum Monographs No. 8. Neogene Hyracoidea. N. Jb. Geol. Pala¨ont. Abh. 175, Tokyo: National Science Museum. 207–234. Winkler, A. (2002). Neogene paleobiogeography and Pilbeam, D. & Walker, A. C. (1968). Fossil monkeys Baringo paleoenvironments: contributions from the from the Miocene of Napak, N.E. Uganda. Nature Baringo rodents and lagomorphs. J. hum. Evol. 1/2, 270, 657–660. 237–256.