2015 Volume 123 Issue 1 Pages 19-39
The vertebrate fossil localities of the Chorora Formation, Ethiopia, comprise one of only a few sub-Saharan African paleontological research areas that illuminate Late Miocene African mammalian and primate evolution. Field work at Chorora since 2007 has resulted in the establishment of new vertebrate fossil localities and a revised chronostratigraphic framework. The new Chorora Formation fossils include the earliest known records of Cercopithecinae, Hippopotaminae, and Leporidae in Africa. Two lineages of hipparionins are recognized at Chorora, a larger and smaller morph, forming potential phyletic links between the earlier Samburu Hills hipparionins and later Eurygnathohippus turkanensis and E. feibeli from Lothagam, Kenya. The Chorora colobines are larger than the >9 Ma Microcolobus and morphologically conservative with only moderate molar cusp notches. The Chorora cercopithecines represent the earliest documented occurrence of the subfamily.
The Chorora Formation (Fm), located along the southeastern margin of the transition zone of the Afar and the Main Ethiopian Rifts, contains vertebrate fossils that represent the little-documented 7–12 Ma time period of African mammalian evolution. A small but important early Late Miocene vertebrate fossil assemblage has been known since the 1970s from what has been called the ‘type locality’ of the Chorora Fm (Sickenberg and Schönfeld, 1975; Tiercelin et al., 1979; Geraads et al., 2002). The age of this assemblage has been widely considered to be ~10.5 Ma (Geraads et al., 2002; Winkler, 2002; Bernor et al., 2004; Cote, 2004; Suwa et al., 2007; Werdelin and Sanders, 2010; Melcher et al., 2014). In the present study, we refer to this fossil locality as the Chorora Fm Type Locality (hereinafter TL). In 2006/07, another vertebrate fossil locality, the Beticha locality, was discovered (Suwa et al., 2007), including fragmentary fossils of Cercopithecidae and the hominoid Chororapithecus abyssinicus. No primates have been reported from the TL by the previous researchers.
In this paper, we describe the vertebrate fossil assemblages of the Chorora Fm, recently discovered from the TL, Beticha, and four other newly established fossil localities. These fossils are now considered to span the ~7–9 Ma time period (see below). This is a time period crucial in elucidating the early evolutionary history of the modern African great ape/human clade and many other extant mammalian groups. However, until now, vertebrate fossil assemblages of sub-Saharan Africa have been known to be either younger than ~7.4 Ma or older than ~9 Ma. Thus, the Chorora Fm fossils help to fill a largely unknown segment of the mammalian fossil record. Although our newly recovered fossils are few and fragmentary, they nevertheless contain key new information on a wide range of taxa including cercopithecid primates, hipparionin equids, leporids, hyaenids sensu lato, bovids, nyanzachoere suids, and hippopotamids.
Here, we provide a detailed descriptive account of the Chorora Fm cercopithecid fossils. These include the first sub-Saharan cercopithecid fossils known between 7.5 and 9 Ma. Fossils attributable to Hominoidea, their chronologies, and interpretations will be presented elsewhere. We also describe the associated non-primate mammalian fossils, in particular the equids. We have now more than doubled the Chorora equid collection, which we present here in some detail in the context of the revised chronology. The Chorora equids have been considered among the oldest examples of sub-Saharan African Equidae (Bernor et al., 2004), and are therefore important in relating the Chorora Fm faunas, including the primates, with other Late Miocene assemblages. Hippopotamids appear to be represented by at least one new primitive species of Hippopotaminae, and will be presented in more detail elsewhere.
The purpose of this paper is to provide the initial descriptive and interpretive accounts of the newly recovered Chorora Fm faunas. Although our evaluations are confined largely to comparisons with other eastern African Late Miocene fossil assemblages, our aim is to provide the basic information applicable to a wider range of considerations.
The locations of the Chorora Fm localities relevant to this paper are shown in Figure 1. Chorora Fm fossils collected by us are numbered in the form of CHO-XX #, with XX representing the locality names as follows: Type Locality (TL), Beticha (BT), Odakora North (ODN), Teso Tadecho (TTD), and Gutosadeen (GSD). These fossils are housed in the laboratory facilities of the Authority for Research and Conservation of Cultural Heritage (ARCCH), Ministry of Culture and Tourism, Addis Ababa, Ethiopia. Comparative specimens numbered KNM-XX # are housed in the National Museums of Kenya. Maxillary and mandibular dental elements are referred to by upper- and lower-case abbreviations, respectively, e.g. M1 (maxillary first molar) and p3 (mandibular third premolar). Maxillary and mandibular deciduous premolars are referred to as dP and dp, respectively.
Location of the Chorora Formation and localities investigated in the present study.
At the Chorora Fm TL, rhyolitic ignimbrites were reported to both over- and underlie the sedimentary sequence that contains the TL main vertebrate fossil-bearing unit (Sickenberg and Schönfeld, 1975; Tiercelin et al., 1979; Geraads et al., 2002). Repeated prior investigations of the (supposedly) overlying ignimbrites converged to a ~10.0 Ma K–Ar age (Geraads et al., 2002). However, our more comprehensive geological investigations necessitate major revisions of the Chorora Fm chronostratigraphy. According to our revised chronology, the TL sediments stratigraphically overlie (not underlie) ignimbrites that we redated to between 9 and 10 Ma. The details of our geological and geochronological revisions are being presented elsewhere. Our revised age estimates of the TL and Beticha (Suwa et al., 2007) fossils are as young as approximately 8.5 and 8.0 Ma, respectively. At several newly established paleontological localities, fossil-containing sediments occur higher in the stratigraphy, some dating to ~7.0 Ma.
Sub-Saharan Late Miocene faunas predating ~7.0 Ma are known from only a few localities in eastern Africa. Vertebrate fossils that date between 9 and 12 Ma have been reported from three areas in Kenya: the Ngorora Fm at Tugen Hills (9–13 Ma) (Hill, 1999; Hill et al., 2002; Pickford, 2001; Gilbert et al., 2010), Nakali (9.8–9.9 Ma) (Kunimatsu et al., 2007), and the Namurungule Fm at Samburu Hills (9.5–9.6 Ma) (Sawada et al., 2006). In sub-Saharan Africa, no vertebrate fossil sites were known to securely date between 7.5 and 9 Ma. At Lothagam, southwestern Lake Turkana Basin, Kenya, abundant fossils of the Lower Nawata Member (Mb) have been bracketed between 6.5 and 7.44 Ma, but most of these fossils are <~7 Ma (Leakey et al., 1996; Leakey and Harris, 2003).
The Sahelanthropus tchadensis-bearing Toros-Ménalla locality of Chad has been suggested to be around or slightly older than 7 Ma (Lebatard et al., 2008, 2010). However, these summary dates are based on weighted means of highly variable individual analyses, and corroboration of the inferred chronology by other methods is needed. Biochronological considerations leave open the possibility of a younger 6–7 Ma age, as suggested initially by Vignaud et al. (2002) and from similarities (e.g. Lihoreau et al., 2006) with the lower Sahabi fauna now best considered 6–7 Ma. The latter is inferred from the combination of an early Messinian geological setting of the lower Sahabi Fm and paleomagnetism; a reverse to normal transition occurs within the terrestrial mammalian fossil-bearing Member U (Boaz et al., 2008a, b).
The recently published Toros-Ménalla suid, Nyanzachoerus khinzir, is dentally more advanced than the Lothagam Ny. tulotos (Boisserie et al., 2014). If the two suid lineages were closely related, this would suggest a 5.5–6.5 Ma (Upper Nawata Mb equivalent) estimate for the Chadian fossils. However, such a direct biochronological reading of ‘dental morphological stage’ may not be applicable. This is because the craniomandibular morphological distinctions between the Lothagam and Toros-Ménalla nyanzachoere species suggest basin- or region-specific morphological evolution (Boisserie et al., 2014). An alternative possibility would be that the Toros-Ménalla suid was time equivalent of some segment of the Lower Nawata Mb Ny. tulotos lineage that represents an earlier dental morphological stage. This would then suggest a biochronological placement between 6.5 and 7.4 Ma.
Thus, the Chorora Fm fossil assemblages provide the first informative glimpse of sub-Saharan mammalian evolution spanning the 7–9 Ma time period.
Geraads et al. (2002) reported on the combined 1970s (Sickenberg and Schönfeld, 1975; Tiercelin et al., 1979) and 1990s Chorora TL fossil collection. They recorded ten mammalian taxa (not including rodents) identifiable to the genus or tribe levels, including an early form of Stegotetrabelodon (Proboscidea), a large hipparionin equid, a dicerotin rhinoceros (now considered Ceratotherium, see below), a chalicothere attributed to Ancylotherium, a small boselaphin bovid, a large giraffid probably sivathere, a hippopotamid perhaps Kenyapotamus, and a machairodontine felid (Table 1). This fossil assemblage was considered to be ~10.5 Ma, although Geraads et al. (2002) noted that aspects of the TL fauna suggest a somewhat more recent age. The latter observation was based largely on the Stegotetrabelodon evidence (initially presented by Coppens and Tassy in Tiercelin et al., 1979), which Geraads et al. (2002) considered more derived than the ~9.6 Ma Tetralophodon from Samburu Hills (Nakaya et al., 1987). Although the Chorora TL fauna has been long considered to date between 10 and 11 Ma (see above), our revised geochronology indicates a younger ~8.5 Ma age. The details of our revised chronostratigraphy will be presented elsewhere.
| Taxon | Type Locality | Beticha | Odakora North | Teso Tadecho | Gutosadeen |
|---|---|---|---|---|---|
| Primates | |||||
| Chororapithecus abyssinicus | ○ | ||||
| Colobinae sp. | ○ | cf ○ | |||
| Cercopithecinae sp. | ○ | ●large | |||
| Cercopithecidae indet. | ○* | ||||
| Cetartiodactyla | |||||
| Hippopotamidae indet. | ○ | ||||
| Hippopotaminae nov. sp. 1 | ○ | ||||
| Hippopotaminae nov. sp. 2 | ● | ● | ● | ||
| Nyanzachoerus sp. | (○)* | ○ | |||
| Nyanzachoerus cf. tulotos | ● | ||||
| Sivatheriini sp. | ○ | ○ | |||
| Paleotragus sp. | (○)* | ○ | |||
| Boselaphini sp. small | ○ | ○ | |||
| cf. Bovini | ○ | ||||
| Reduncini sp. | cf ○** | ● | ● | ||
| cf. Neotragini sp. | ○ | ||||
| Perissodactyla | |||||
| “Cormohipparion” sp. large | ○ | ○ | cf ○ | ||
| Hipparionini sp. small | ○ | ○ | ○ | ||
| Ceratotherium sp. | ○ | ||||
| cf. Dicerotini | ○ | ||||
| Ancylotherium tugenensis | ○ | ||||
| Proboscidea | |||||
| Stegotetrabelodon sp. | ○ | ○? | ○? | ||
| Deinotherium sp. | ○* | ||||
| Carnivora | |||||
| ?Herpestides afarensis | ○ | ||||
| Machairodus cf. aphanistus | ○ | ||||
| Felidae sp. | large | medium | |||
| Percrocutidae sp. large | ○ | ||||
| cf. Hyaenictis | ○* | ||||
| Lagomorpha | |||||
| Alilepus sp. | ● | ||||
| Rodentia | |||||
| Paraulacodus johanesi | ○ | ○ | ○ | ||
| Paraphiomys chororensis | ○ | ||||
| Nakalimys lavocati | ○ | ||||
| Gerbillinae sp. large cf. Abudhabia | ○ | ||||
| Afromys guillemoti | ○ | ||||
| “Dendromus” (?Saccostomus) spp. | ○ | ||||
| Praecomys kikiae | ○ | ||||
| aff. Stenocephalomys | ○ | ||||
| cf. Parapelomys | ○ | ||||
| cf. Tectonomys | ○ | ||||
| Xerus sp. | ○ | ||||
○ indicates presence of taxon, parenthesis based on photographic documentation (specimens not collected).
● indicates presence of taxon and a chronological age younger than Beticha/Type Locality.
Our post-2007 collection from the TL is less diverse in taxonomic representation (eight taxa identifiable to family or lower taxonomic levels) than the previous collection, but we nevertheless added three newly documented taxa (Table 1). One new taxon is Deinotherium sp. (Proboscidea) based on tooth fragments. Another is a hyaenid represented by a P3, probably Hyaenictis (Figure 2). This P3 is similar in size and morphology to the <6.0 Ma Hyaenictis wehaietu recorded in the Middle Awash, Ethiopia (Haile-Selassie and Howell, 2009). We significantly increased the TL collection of some of the previously known taxa, especially that of the hipparionin equid which dominates this fossil assemblage. The Chorora Fm hipparionin has been considered to be one of the earliest African equid occurrences (Pickford, 2001; Geraads et al., 2002; Bernor et al., 2004, 2010). The newly collected hipparionin fossils at the TL comprise 21 catalogued specimens. Fifteen are postcanine teeth or fragments, two are incisors, and four are postcranial elements. Together with the previous materials, the number of premolar and molars with crown lengths measured (or estimated) now totals 20 (Table 2). A notable addition to the TL collection is Cercopithecidae, represented by a single upper canine crown. The TL and other Chorora cercopithecid fossils are described below and summarized in Table 3 and Table 4.
CHO-TL 27 cf. Hyaenictis left upper third premolar (P3). From left to right: lingual, occlusal, and buccal views, crown length 19.0 mm, crown breadth 10.5 mm.
| Element | Specimen no. | Taxon | Crown length | Source |
|---|---|---|---|---|
| TYPE LOCALITY | ||||
| Maxillary | ||||
| dP3/4 | CHO-TL 55 | “Cormohipparion” sp. large | 31.0 | this study |
| P2 | no# | “Cormohipparion” sp. large | 35.0 | Geraads et al. (2002) |
| CHO1-1 | “Cormohipparion” sp. large | ((33)) | this study, Geraads et al. (2002) 31+ | |
| P3/4 | CHO-TL 42 | “Cormohipparion” sp. large | (27.0) | this study |
| M1/2 | CHO1-2 | “Cormohipparion” sp. large | 25.2 | Geraads et al. (2002) |
| no# | “Cormohipparion” sp. large | 28.0 | this study, Geraads et al. (2002) 28.5 mm | |
| NL(B)CHOR 2 | “Cormohipparion” sp. large | 24.0 | Bernor et al. (2004) | |
| CHO-TL 1 | “Cormohipparion” sp. large | 25.0 | this study | |
| M3 | CHO1-3 | “Cormohipparion” sp. large | 26.0 | this study, Geraads et al. (2002) 25 mm |
| Mandibular | ||||
| p3/4 | CHO1-5 | “Cormohipparion” sp. large | 29.6 | Geraads et al. (2002) |
| no# | “Cormohipparion” sp. large | 27.8 | Geraads et al. (2002) | |
| NL(B)CHOR 1 | “Cormohipparion” sp. large | 27.2 | Bernor et al. (2004) | |
| NL(B)CHOR 6 | “Cormohipparion” sp. large | 30.9 | Bernor et al. (2004) | |
| CHO-TL 4 | “Cormohipparion” sp. large | 28.5 | this study | |
| CHO-TL 6 (P3) | “Cormohipparion” sp. large | 27.5 | this study | |
| CHO-TL 6 (P4) | “Cormohipparion” sp. large | ((26.5)) | this study | |
| m1/2 | CHO1-8 | “Cormohipparion” sp. large | 25.5 | Geraads et al. (2002) |
| CHO-TL 2 | “Cormohipparion” sp. large | 25.5 | this study | |
| CHO-TL 4 | “Cormohipparion” sp. large | 26.0 | this study | |
| m3 | CHO1-7 | “Cormohipparion” sp. large | 28.0 | this study, Geraads et al. (2002) 27.4 mm |
| BETICHA | ||||
| Mandibular | ||||
| m1/2 | CHO-BT 97 | hipparionin sp. small | 22.0 | this study |
| p3/4 | CHO-BT 116 | “Cormohipparion” sp. large | 26.5 | this study |
| CHO-BT 122 | “Cormohipparion” sp. large | 27.5 | this study | |
| ODAKORA NORTH | ||||
| Maxillary | ||||
| P3/4 | CHO-ODN 1 | hipparionin sp. small | 24.0 | this study |
| GUTOSADEEN | ||||
| Maxillary | ||||
| dP3/4 | CHO-GSD 3 | hipparionin sp. large | 28.5 | this study |
| P3/4 cf 4 | CHO-GSD 2 | hipparionin sp. small | 24.5 | this study |
| Mandibular | ||||
| m3 | CHO-GSD 7 | hipparionin sp. small | 23.0 | this study |
| m1/2 | CHO-GSD 22 | hipparionin sp. small | (22.5) | this study |
Crown lengths: ( ) corrected for damage, (( )) rough estimate.
| Specimen no. | Age | Taxon | Element | Length | Breadth |
|---|---|---|---|---|---|
| CHO-TL 10 | 8.5 Ma | Cercopithecidea | left C | 7.0 | 4.8 |
| CHO-BT 1 | 8.0 Ma | Cercopithecidae? | left C root | ||
| CHO-BT 2 | 8.0 Ma | Cercopithecidae | partial proximal phalanx | ||
| CHO-BT 77 | 8.0 Ma | Cercopithecinae sp. indet. (Papionini? small species) | right m3 buccal fragment | ((9.0–9.5)) | |
| CHO-BT 78 | 8.0 Ma | Colobinae sp. | left m1/2 | 6.6 | 4.7, 4.8 |
| CHO-BT 112 | 8.0 Ma | Cercopithecinae sp. indet. (Papionini? small species) | partial right m1/2 | (7.0) | |
| CHO-BT 114 | 8.0 Ma | Colobinae sp. | left m2 with distal root | (7.7) | (5.9), (6.0) |
| CHO-BT 115 | 8.0 Ma | Colobinae sp. | right M1/2 | 6.8 | 7.0, 7.0 |
| CHO-BT 118 | 8.0 Ma | Colobinae sp. | left M1/2 | (6.8) | |
| CHO-BT 123 | 8.0 Ma | Cercopithecidae | right c | 6.4 | 4.7 |
| CHO-TTD 1 | 7.0–7.5 Ma | cf. Colobinae | left P3 | 5.35 | 6.7 |
| CHO-TTD 2 | 7.0–7.5 Ma | cf. Colobinae | left P4 | 5.3 | 6.4 |
| CHO-GSD 47 | 7.0–7.5 Ma | Cercopithecinae sp. indet. (Papionini? medium-sized species) | left m1/2 | ((9.0)) |
Lower canine metrics are maximum oblique basal diameter and breadth perpendicular to maximum diameter.
Length, mesiodistal crown length; Breadth, buccolingual (or labiolingual) crown breadth. For molars, mesial and distal breadths are shown.
( ) indicates estimated values corrected for minor damage and/or wear. (( )) indicates rough estimations.
| Specimen/sample | Age (Ma) | L (mm) | L/MW | MW/DW | MCP (mm) | MCP/MW | DCP (mm) | DCP/DW | NH (mm) | NR (mm) | NH/NR | MSL (mm) | MSL/L |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 1KNM-NK 305 (Microcolobus sp.) cf m1 | 9.8–9.9 | 6.4 | 133 | 102 | 1.8 | 1.8 | 100 | ||||||
| 1KNM-BN 1740 (M. tugenensis) m1 | ~9.0 | 5.5 | 125 | 102 | 1.2 | 1.5 | 80 | 1.6 | 29 | ||||
| 1KNM-BN 1740 (M. tugenensis) m2 | ~9.0 | 5.6 | 119 | 96 | 1.7 | 1.8 | 94 | 1.5 | 26 | ||||
| CHO-BT 78 m1 (or m2) | 8.0 | 6.6 | 140 | 98 | 2.95 | 63 | 3.12 | 65 | 1.85 | 1.78 | 104 | 1.73 | 26 |
| CHO-BT 114 m2 (or m1) | 8.0 | (7.7) | (130) | (98) | |||||||||
| 2Lothagam Lower Nawata (sp A, LT 24107) m2 | 6.5–7.4 | 6.6 | 132 | 98 | |||||||||
| 2Lothagam Lower Nawata (sp B, n = 2) m1 | 6.5–7.4 | 7.0–7.3 | 132–143 | 90–93 | |||||||||
| 2Tugen Hills, Mpesida (sp A, TH30975) m2 | 6.4–7.2 | 8.0 | 127 | 98 | 130 | ||||||||
| 1Mesopithecus pentelicus mean (n = 13) m1 | ~7.0 | 6.8 | 114 | 94 | 137 | 1.6 | 23.5 | ||||||
| 1Mesopithecus pentelicus mean (n = 15) m2 | ~7.0 | 7.5 | 112 | 94 | 125 | 1.85 | 25 | ||||||
| 1Modern colobine | |||||||||||||
| m1 mean (range) | 135 (114–140) | 96 (93–102) | 70 (64–80) | 71 (59–83) | 124 (83–143) | 24 (22–27) | |||||||
| m2 mean (range) | 125 (115–130) | 97 (93–102) | 70 (51–79) | 68 (45–85) | 150 (113–176) | 30 (28–31) | |||||||
| 1Modern cercopithecine | |||||||||||||
| m1 mean (range) | 125 (105–129) | 101 (97–102) | 60 (53–71) | 55 (49–63) | 54 (53–78) | 25 (24–27) | |||||||
| m2 mean (range) | 116 (100–122) | 106 (104–112) | 56 (41–62) | 53 (46–57) | 58 (35–65) | 31 (29–35) | |||||||
Metrics after Benefit and Pickford (1986) and Benefit (1993). L, crown length; MW, mesial width; DW, distal width; MCP, mesial cusp tip proximity; DCP, distal cusp tip proximity; NH, cusp height above lingual notch; NR, crown height below lingual notch; MSL, mesial shelf length.
Numbers in ( ) for CHO-BT 114 are estimated values. Numbers in ( ) for modern colobines and cercopithecines are range of species means.
The newly recovered hipparionin cheek teeth (premolars and molars) are in accord with previous morphological assessments (Geraads et al., 2002; Bernor et al., 2004) and help characterize the Chorora hipparionin. Crown size is large (Table 2). Molar plications are well-developed, as illustrated in Geraads et al. (2002). In the upper cheek teeth, pre- and postfossette plications tend to be both deep and extremely complicated. Pli-caballin enamel spurs are commonly as many as four (or more). In the lower permanent cheek teeth there are no instances of an ectostylid. Crown height appears moderately high. An unworn M2 height is ~60 mm buccally, and two p4s, well into wear, have crown heights of ~50 mm. Except for this moderately high crown, also shared by the ~9.6 Ma Samburu Hills homologues (Nakaya et al., 1984, 1987; Tsujikawa, 2005; Bernor et al., 2010) and approximated by the ~10.5 Ma Bou Hanifia “Cormohipparion” africanum (Bernor and White, 2009), the Chorora cheek teeth have been considered to reflect the ancestral primitive Eurasian morphology differing from that of the later African hipparionins (Geraads et al., 2002; Bernor et al., 2004, 2010). However, we note that the maximum crown height of the Bou Hanifia, Chorora, and Samburu Hills hipparionins all exceed that of Central European Vallesian Hippotherium primigenium (Bernor and White, 2009). Thus, contrary to Geraads et al. (2002), the Chorora cheek teeth are not referable to H. primigenium despite their highly ornamented occlusal enamel plications.
Among the newly collected fossils, three are well-preserved postcranial elements, an astragalus, a calcaneus, and an intermediate phalanx (Figure 3). These are the first known postcranial elements of the Chorora hipparionin, and are therefore of considerable interest. These are surprisingly large in size; they are as large as homologues of the larger of the later Mio-Pliocene African species of Eurygnathohippus. The Chorora astragalus is marginally larger than the two Lothagam Lower Nawata E. turkanensis examples reported by Bernor and Harris (2003), and about 10% larger than the largest example reported from Samburu Hills (Nakaya and Watabe, 1990). This larger size could be due to individual variation. The calcaneus is broken at its distal end but its maximum preserved length is 112 mm (its unbroken length was probably in the range of ~120 mm). The Chorora intermediate phalanx is as broad as it is long, as is also the case with the Samburu Hills hipparionin intermediate phalanges (Nakaya et al., 1984, 1987). The Samburu Hills phalanges are ~10% smaller than the Chorora phalanx in absolute size. The Lothagam E. turkanensis homologues are in turn slightly larger than the Chorora phalanx. The intermediate phalanx of all these species differs from those of the slender-limbed E. feibeli which are slightly longer than wide (Bernor and Harris, 2003).
Chorora Type Locality hipparionin postcranial elements. Left: CHO-TL 3 right astragalus, maximum length 66 mm. Right: CHO-TL 40 ray III intermediate phalanx, maximum length 41.5 mm, proximal and distal widths 41.5 and 37.5 mm, respectively.
The overall impression of the Chorora TL hipparionin is a large taxon, somewhat larger in both postcanine and postcranial sizes than the Samburu Hills large morph hipparionin (Nakaya et al., 1984, 1987). The Chorora large hipparionin may have had somewhat robust distal limbs, as suggested for the Samburu Hills larger specimens (Nakaya et al., 1984, 1987). However, they were probably not as derived as Lothagam Eurygnathohippus turkanensis (Bernor and Harris, 2003); the latter is characterized by extremely robust (broad) distal limb bones (metapodials and phalanges). More fossils are needed to confirm and clarify these comparative impressions.
Previous mesowear and isotopic analysis of teeth from the TL suggested that the Chorora hipparionin was a mixed-feeder, with average enamel δ13C values of four teeth ranging from −6.3 to −4.6‰ (Bernor et al., 2004). Two of these teeth were analyzed at three intervals along their length, extending the total δ13C range from −7.5 to −4.6‰. We measured enamel δ13C values on three of the newly collected teeth, following previous methods (White et al., 2009). Our new results average −4.6‰ (−5.3 to −3.7‰, Table 5) which is slightly less negative than that of −5.5‰ reported by Bernor et al. (2004). This difference between means is small, and the combined average of seven individuals is −5.1 ± 0.8‰ (1 SD). Assuming an enamel δ13C value of −12.5‰ for a 100% C3 browse diet and +2‰ for a 100% C4 diet, the total range of δ13C values reflects ~35–60% C4 grass consumption (mixed feeding as suggested by Bernor et al., 2004).
| Specimen no. | Tooth | Lab sample no. | δ13C‰ (pdb) | δ18O‰ (vsmow) | δ18O‰ (pdb) |
|---|---|---|---|---|---|
| CHO-TL 1 | Left M1 | KGTE 163 | −3.7 | 32.7 | 1.8 |
| CHO-TL 6 | Left p4 | KGTE 162 | −4.8 | 31.9 | 0.9 |
| CHO-TL 9 | Right p/m fragment | KGTE 164 | −5.3 | 33.3 | 2.3 |
| CHOR 1 | Left p4 | −5.0 | |||
| CHOR 1 | Left p4 | −6.4 | |||
| CHOR 1 | Left p4 | −7.5 | |||
| CHOR 2 | Left M1 | −4.6 | |||
| CHOR 3 | Right m1 | −5.6 | |||
| CHOR 3 | Right m1 | −5.7 | |||
| CHOR 3 | Right m1 | −6.1 | |||
| CHOR 5 | Left m2 | −5.1 |
Values are in international standards. PDB, Pee Dee Belemnitella fossil carbonate, for carbon; PDB and VSMOW, Vienna Standard Mean Ocean Water, for oxygen.
The CHOR number samples are from Bernor et al. (2004).
Although a formal mesowear study has not yet been undertaken with the currently available larger sample, many specimens exhibit steep or distinct molar wear slopes, consistent with mixed feeding. The average δ13C value of the Chorora TL hipparionin is midway between the Nakali (−8.6‰) and Namurungule Fm (Samburu Hills) (−2.0‰) averages (Uno et al., 2011). The Chorora TL equids apparently consumed more C4 grass than at Nakali (~9.8 Ma), but less than at Namurungule (9.5–9.6 Ma). This suggests that C4 grasses were not as abundant at ~8.5 Ma Chorora as they were at 9.5–9.6 Ma Samburu Hills. Lothagam Lower Nawata Mb (6.5–7.4 Ma) δ13C values average −0.9‰, which is only slightly lower than that of modern grazers (Uno et al., 2011). The low to intermediate δ13C values of the Chorora TL hipparionin, combined with the mesowear evidence, suggest that the transition from mixed feeding to obligate grazing in hipparionins occurred after ~8.5 Ma.
The Beticha locality fauna (~8.0 Ma)The Beticha locality is situated ~3 km SSW of the TL (see Figure 1). Its main fossiliferous horizon occurs toward the top of the exposed sequence and is correlated to the upper part of the Chorora TL sediments (Suwa et al., 2007). Fossils occur in a gravelly sand unit of limited lateral extent, and are fragmentary. Chororapithecus is numerically dominant, followed by what appears to be a low-crowned primitive Hippopotaminae. Cercopithecids, suids, and bovids are the next common, while equids are rarer. The Beticha hippopotamids are of particular interest because they seem distinctly primitive compared to the earliest known advanced hippopotamid (i.e. Hippopotaminae) postdating ~7.4 Ma (Boisserie et al., 2011). The Beticha hippopotamid fossils include several informative molars and other dental fragments, and will be presented elsewhere. The list of taxa so far identified at Beticha is presented in Table 1. The more complete specimens, aside from Chororapithecus and hippopotamids, are described below.
Nyanzachoerus is represented by a complete m1 (or 2) crown (CHO-BT 108), and half a crown of a large lower premolar (CHO-BT 66). These are compatible with a Ny. tulotos-sized nyanzachoere species, although the lower molar might be an m2 of a smaller species. This molar is extremely low-crowned (Figure 4), with a strong basal flare unusual in Nyanzachoerus lower molars postdating ~7 Ma. It is clearly a lower and not an upper molar, from its mesial cingular morphology (thin horizontal cingular shelf separated from the median mesial cusplet). This molar also exhibits a deep lingual cleft between the low main cusps, and a small distal accessory cusp with little distal protrusion. These features make this tooth more primitive-looking than the Lothagam Ny. tulotos homologues. It is morphologically compatible with the ~9.6 Ma Samburu Hills Nyanzachoerus molars (Nakaya et al., 1984, 1987; Tsujikawa, 2005; this study).
CHO-BT 108 Nyanzachoerus sp. right lower molar (m1/2). From left to right: lingual, occlusal, and mesial views, crown length 23.3 mm, crown breadth 18.7 mm.
The dominant bovid is represented by two premolars and other molar fragments that we attribute to a small species of Boselaphini (Figure 5), probably the same taxon (gen. et sp. indet. 1) as that described by Geraads et al. (2002) at the TL. The p4 is characterized by a slender crown with a mesiodistally expanded metaconid lingual face that is fused mesially with the paraconid and distally with the entoconid.
Boselaphini and Reduncini teeth. Upper row, Boselaphini sp. from Beticha and the Type Locality (from left to right): CHO-BT 36 left p3 occlusal view, crown length 12.1 mm; CHO-BT 72 right p4 occlusal view, crown length 12.6 mm; CHO-TL 22 right M1/2, occlusal and mesial views, crown length 14.0 mm. Lower row, Reduncini sp. from Gutosadeen (from left to right): CHO-GSD 4 left M2 mesial view; CHO-GSD 4 left M2 and M3 occlusal views, crown lengths 16.7 and 17.3 mm, respectively.
A large bovid p4, CHO-BT 127, was recovered splintering in situ, but restored into an almost complete crown (Figure 6). This is a large premolar, in the size range of fossil and modern Bovini. This p4 has an elongate crown with a buccally convex crown contour, a distinct and well-separated paraconid and parastylid complex, a bulbous metaconid deflected anteriorly, a capacious anterior lingual cleft, a distinct and deep posterior lingual cleft, and an entoconid–entostylid complex wearing into a transverse looped pattern. These features are often matched in Bovini p4s. However, its low crown height and developed (buccally projecting and anteroposteriorly long) hypoconid are unlike bovins. Some of these p4 crown features are also approximated by Hippotragini, for example those reported from Toros-Ménalla (Geraads et al., 2008). However, the Chadian Hippotragini p4s are smaller and appear more mesiodistally compressed. They also seem to have less distinct paraconid–parastylid separation, more constricted lingual clefts, buccally pinched hypoconids, and a higher crown. The morphology of the BT 127 p4 is in part approximated by large Eurasian forms, such as the Late Miocene protoryxoids (Kostopoulos, 2009; Kostopoulos and Bernor, 2011). However, these p4s tend to have a more elongate metaconid and weaker lingual clefts. Large boselaphin p4s from <6.0 Ma of Ethiopia, such as Tragoportax sp. (Haile-Selassie et al., 2009), are of similar size but differ in their mesiodistally expanded metaconid.
CHO-BT 127 cf. Bovini left lower premolar (p4). From left to right: buccal, occlusal, and lingual views, crown length 18.7 mm, crown breadth 10.2 mm, lingual crown height 14.3 mm (measured at metaconid as worn).
The BT 127 p4 is smaller and lower-crowned than teeth of Bovini from the ~5.8 Ma Asa Koma Member (Haile-Selassie et al., 2009), but about the same size as those of the early bovin teeth recorded from the ~6–9 Ma Siwalik deposits (Bibi, 2007). BT 127 differs from these early bovins in an anteriorly deflected (and not distally expanded) metaconid, a more constricted anterior lingual cleft, a more prominent hypoconid, and lower crown height. These features of BT 127 are most likely primitive, although the tooth is younger in age than the oldest Siwalik bovin. From the above comparisons, we consider it possible that BT 127 represents either a large ‘boselaphin’ or some form of early, primitive bovin. We tentatively attribute this tooth to cf. Bovini pending more definitive fossils.
Both large- and medium-sized giraffids are represented. The larger taxon is known by a small fragment of molar, comparable in size to Sivatheriini of the TL. The medium-sized morph is represented by a partial upper molar in the size range of Paleotragus of Lothagam Nawata Mb or Samburu Hills Namurungule Fm.
In contrast to the TL, equids are comparatively rare at Beticha, represented by only three lower cheek teeth and a small fragment of upper premolar or molar. Two lower premolars are large (Figure 7, Table 2), comparable to the TL homologues in both size and morphology (complicated flexid patterns and lack of ectostylid). The ectoflexids of these premolars are deeply incised, as is the case with the TL homologues, a primitive feature for Old World hipparionins. To the contrary, the single m2 (and the P/M fragment) is distinctly smaller than the TL hipparionin and exhibits simpler occlusal morphology (Figure 7), although it shares the lack of ectostylid with the larger TL hipparionin. We discuss the significance of these observations further below.
Hipparionin cheek teeth from the Type Locality, Beticha and Gutosadeen. Upper row (from left to right): CHO-TL 42 right P3/4 (buccal crown face and styles are broken off); CHO-GSD 2 left P4. Lower row (from left to right): CHO-BT 122 left p3; CHO-BT 116 left p4; CHO1-8 left m2 (Type Locality, Geraads et al., 2002); CHO-BT 97 left m2; and CHO-GSD 22 left m1/2. Other Type Locality hipparionin teeth are illustrated in Geraads et al. (2002).
The rhinocerotid at Beticha is represented by two premolar/molar fragments, which we tentatively attribute to Dicerotini. Although the TL rhinos (two upper premolars) were initially published as Dicerotini (Geraads et al., 2002), the protoloph of these courses somewhat obliquely, tending to lingually fuse with the metaloph. Recently, Geraads (2010) considered these TL rhinos as representing Ceratotherium sp. In contrast to the TL premolars, one of the Beticha rhinocerotid fragments exhibits a paracone style, and another preserves a less oblique protoloph, suggesting Dicerotini attributions.
Carnivores are represented by a large felid and a large hyaenid (sensu lato). The felid is represented by a Machairodus-sized distal tibia. The hyaenoid premolar (CHO-BT 107) is large and bulky-crowned (Figure 8). This tooth is in the size range of Dinocrocuta of the Eurasian and northern African Late Miocene (Howell and Petter, 1985; Werdelin and Peigné, 2010). A similarly large percrocutid has been reported from Nakali (Morales and Pickford, 2006) and attributed to the genus Percrocuta based on upper carnassial morphology. We provisionally allocate the BT 107 premolar to cf. Percrocutidae.
CHO-BT 107 percrocutid right lower premolar (p3). From left to right: distal, occlusal, and buccal views, crown length 26.0 mm, crown breadth 16.0 mm.
Proboscideans are represented by small fragments, which are tentatively considered Stegotetrabelodon. The cercopithecid fossils from Beticha are described in some detail below.
The Odakora North, Teso Tadecho, and Gutosadeen faunas (<~7.5 Ma)These fossil localities occur as isolated small sediment exposures that overlie the TL equivalent Chorora Fm sediments. Taxa identified at each locality are summarized in Table 1.
Odakora North is located ~4 km SW of the TL (see Figure 1). It was first briefly visited by some of us in January 1989, during a short survey for the Paleoanthropological Inventory of Ethiopia project led by one of us (B.A.) (Asfaw et al., 1990). At that time, a hipparionin molar was observed on site but not collected. We revisited the site in March 2011 and confirmed that the site is poorly fossiliferous. However, we recovered four hippopotamine and two hipparionin specimens, mostly fragments of teeth. The hippopotamine specimens include a fragmentary mandible with partial juvenile dentition. This fossil represents a species more advanced than the Beticha hippopotamine in crown height and in some details of molar structure. Its morphology approximates the ~7 Ma and later hippopotamine condition, amply represented by fossils from Toros-Ménalla, Lothagam (Upper and Lower Nawata Mbs), the Middle Awash, Sahabi, and Abu Dhabi (Gentry, 1999; Boisserie, 2005; Boisserie et al., 2011).
The hipparionin is represented by a cheek tooth that preserves most of the unworn crown. It has a slightly mesiodistally elongated crown and appears to be a P3 or 4. This premolar is distinctly smaller and lower crowned than the TL hipparionin (Figure 9). Its unworn lingual height is ~40 mm, and its maximum buccal height must have been <50 mm. As exposed on the unworn occlusal surface, the pli-caballin appears single.
Comparison of unworn hipparionin upper cheek teeth crown heights (lingual view). Left: CHO-ODN 1 right P3/4. Right: unnumbered Type Locality right M1/2 (Geraads et al., 2002). Both are broken just at the crown bases.
Teso Tadecho is located ~7 km SW of the TL (see Figure 1). We have so far recorded five taxa identifiable to the subfamily or lower levels. A large upper molar represents a low-crowned hippopotamine. As is the case with the Odakora North fossils, this molar is morphologically more advanced than the Beticha hippopotamid. Its morphology approximates that of the ~7 Ma and later eastern and central African hippopotamines, although retaining some primitive details of the crown. An ill-preserved horn core base may represent Reduncini. A leopard-sized felid premolar was recovered. Of note are two cercopithecid upper premolars. These are described below. Several rodent molars were recovered. The dominant rodent is Paraulacodus, indistinguishable from P. johanesi of the TL (Geraads, 1998), also represented at Beticha (Figure 10). A tiny fragment of mandible preserving the m2 (Figure 10) appears to be a gerbelline, perhaps a large species of Abudhabia.
Rodents and lagomorph. Tooth measurements are shown in parenthesis in the form of length × breadth in millimeters. Upper row (from left to right): Teso Tadecho gerbilline CHO-TTD 3 left m2 (2.1 × 2.3 mm) in small mandible fragment; Type Locality Paraulacodus johanesi CHO-TL 17 right M3 (3.1 × 3.7 mm), CHO-TL 47 right M2 (3.1 × 4.2 mm); Beticha P. johanesi CHO-BT 33 left M2 (3.2 × 4.0 mm); Teso Tadecho P. johanesi CHO-TTD 4 left M3 (3.0 × 3.7 mm). Lower row (from left to right): Gutosadeen Alilepus sp. CHO-GSD 11 mandibular fragment with left p3 (3.7 × 3.7 mm); Teso Tadecho P. johanesi CHO-TTD 11 left m2 (3.6 × 3.5 mm), CHO-TTD 6 left m2 (3.8 × 3.8 mm); Beticha P. johanesi CHO-BT 103 right m1 (3.3 × 3.2 mm); Type Locality P. johanesi CHO-TL 47 right mandible with dp4–m2, dp4 (3.4 × 2.8 mm), m1 (3.2 × 3.1 mm), m2 (3.7 × 3.6 mm).
Gutosadeen is located ~2.5 km NW of the TL (see Figure 1) and has produced a larger collection of identifiable fossils than Odakora North or Teso Tadecho. Large but low-crowned hippopotamine molars also occur here. A large Nyanzacheorus species is represented by an m1 and partial m3 (Figure 11). These are indistinguishable from the Lothagam Nawata Mb Ny. tulotos in both size and morphology. Bovids are represented by a set of medium-sized upper molars with reduncin-like combination of buccal ribbing, pinched lingual lobes, and a moderate crown height. The basal pillar is not expressed (Figure 5), but this might be the case in early reduncin species. A notable fossil is the leporid mandibular fragment with p3 (Figure 10). In all details, the p3 morphology adheres closely to that considered diagnostic of Alilepus (Winkler, 2003).
Gutosadeen fossils attributed to Nyanzachoerus tulotos. Left: CHO-GSD 1 left m1, crown length 20.5 mm, crown breadth 16.3 mm. Right: CHO-GSD 6 right m3 distal fragment.
Several measurable hipparionin premolar or molars were collected. A dP3 or 4 is large, in the size range of the Chorora TL and Namurungule large-morph hipparionin. Although plication patterns cannot be assessed, this tooth has four pli-caballin folds, similar to the TL hipparionin. The other three measurable molars, a P4, m1 or 2, and m3, are all small (Figure 7, Figure 12). They exhibit comparatively simple enamel patterns, and therefore differ from the larger TL homologues. In contradiction to the small-sized Eurygnathohippus feibeli of Lothagam Nawata Mb and the Middle Awash Valley (Bernor and Harris, 2003; Bernor and Haile-Selassie, 2009; Melcher et al., 2014), these lower molars lack an ectostylid including even small or basal expressions.
A comparison of hipparionin lower third molars (occlusal and buccal views). Upper left two: CHO-TL 43 right distal m3 (broken at the main flexid region). Lower left two: CHO-GSD 7 right m3. Right two: Type Locality CHO1-7 left m3 (Geraads et al., 2002).
Aside from Microcolobus of the 9–10 Ma deposits of Kenya (Benefit and Pickford, 1986; Kunimatsu et al., 2007; Nakatsukasa et al., 2010), cercopithecids that span the 7.5–10 Ma time period of eastern Africa were unknown until now (Gilbert et al., 2010). Here, we describe 13 cercopithecid fossils from the Chorora Fm that come from the 7–8.5 Ma time period (Table 3). One is ~8.5 Ma, nine are ~8.0 Ma, and three are <~7.5 Ma.
Chorora TL and Beticha: ~8.5 and ~8.0 Ma Canines (Figure 13, Table 3)CHO-TL 10 is a little-worn male upper canine of small size. Mesiodistal crown length is 7.0 mm. The crown tip is broken (postmortem) and its preserved crown height is 9.0 mm. A typical sharp honing wear occurs close to the broken edge for ~1 mm. Basally, the root is broken immediately above the cervix. The crown is strongly compressed buccolingually. The mesiolingual groove is deep and localized near the mesial margin of the crown. The groove terminates at the cervix. The root (just at the cervix) is not indented, suggesting only a weak mesiolingual groove development on the root (as is sometimes the case in modern cercopithecids).
Cercopithecid canines. Left three (from left to right): CHO-TL 10 left upper canine, lingual, mesial, and labial views. Right four (from left to right, and lower right): CHO-BT 123 right lower canine, mesial, lingual, distal, and occlusal views.
CHO-BT 1 is possibly a left upper canine root. Its surface is variably weathered. A shallow vertical groove is perceptible along one of the root faces. Its cross section shape is compatible with a tooth such as the CHO-TL 10 upper canine, but appears marginally larger (by 5–10%).
CHO-BT 123 is an unworn lower canine of a small male. The mesial cervical line and distal tubercle were formed, but the labial crown base was probably not yet entirely complete. The crown tip is broken (postmortem) and its preserved crown height is 9.8 mm. Maximum basal crown dimension is 6.4 mm, compatible in size with the TL 10 upper canine.
Colobine molars (Figure 14, Table 3, Table 4)CHO-BT 78 and CHO-BT 114 are lower molars that we attribute to Colobinae based on crown morphology. BT 78 is better preserved, whereas the crown surface of BT 114 is mostly weathered away. Nevertheless, the two molars are similar (where discernible) in overall crown morphology as outlined below.
Teeth attributed to Colobinae. Upper row left three (from left to right): CHO-BT 114 left m2 (or m1), lingual, mesial, and occlusal views. Bottom row left two (from left to right): occlusal views of CHO-BT 115 right M1/2, CHO-BT 118 left M1/2 (mesial towards the top in both specimens). Right two columns: CHO-BT 78 left m1 (or m2); top row, occlusal and lingual views; middle row, mesial and buccal views; bottom right, oblique mesiobuccal view.
In both BT 78 and BT 114, the crown is tall relative to breadth. In both molars, the buccal crown face is comparatively vertical and basal buccal crown flare is minimal. The lingual cusps of BT 78 are tall, angled slightly mesially, and delineate a deep median lingual notch. BT 78 is in initial wear, with small (pinhole-size) dentine islands exposed at the buccal cusp tips, and wear facets formed along the distal metaconid and mesial and distal entoconid slopes. A well-formed wear facet occurs along the mesiobuccal slope of the entoconid along the high transverse crest. Although the mesial crest of the entoconid itself is not yet faceted, a narrow lingual notch would have been formed and maintained with further wear. In BT 78, the base of the lingual notch is positioned at approximately half of the total crown height (Table 4). The main lophids are well-developed but not oblique. The BT 114 molar shares a similar occlusal structure with BT 78 although its lingual cusps are relatively lower due to slight wear (~0.6 mm diameter dentine islands occur on both metaconid and entoconid tips). This gives the impression of a less-developed lingual notch in BT 114, but the BT 114 molar shares with BT 78 high lophids that form a deep inter-lophid basin. The median buccal cleft is deep and wide in BT 78. Surface detail can be observed in BT 78, which shows that the wide median buccal cleft has a shelf-like floor. The lower margin of the buccal cleft takes the form of a thin crest-like structure, lacking expression of cingular-like thickenings. Only faint mesial and distal buccal grooves occur at the extreme corners of the buccal crown face. The median buccal cleft of BT 114 is narrower, but shares a shelf-like structure with BT 78.
CHO-BT 78 is distinctly smaller than BT 114. Because of its buccolingually narrow and mesiodistally elongated crown shape, BT 78 is possibly a first molar. Based on their size and morphology, the two molars may represent the first and second molars of the same species, given serial and individual variation.
CHO-BT 115 and CHO-BT 118 are upper molars (M1 or 2s) attributable to Colobinae. As was the case with the lower molars described above, the buccal notches of these upper molars are not as deep as they are in most modern colobines. However, crown morphology such as relatively vertical crown faces, retention of high mesial and distal lophs in moderate wear, and the presence of only a weak mesiolingual groove indicate a colobine taxonomic attribution. In both morphology and size, these upper molars are compatible as representing the same taxon as the lower molars described above.
Cercopithecine molars (Figure 15, Table 3)At Beticha, a second cercopithecid taxon is represented by two fragmentary molars. CHO-BT 77 is a buccal crown fragment of a right m3. It is broken longitudinally (mesiodistally) close to the buccal cusp tips. The mesial and distal breaks are right at the margins of very small protoconid and hypoconulid dentin wear patches, respectively. Crown and cusp heights are inferred to be low. Its buccal crown face is more curved (vertically) than in the colobines (BT 78 and BT 114), related to its strong buccal crown flare. The BT 77 median buccal cleft takes form of an ill-defined depression, different from the cleft and shelf structure seen in BT 78 and BT 114.
Teeth attributed to Cercopithecinae. Upper row, occlusal views (from left to right): CHO-BT 77 right m3 buccal fragment, CHO-BT 112 right partial m1/2, CHO-GSD 47 left m1/2 (mesial towards the top). Lower row, buccal views (from left to right): CHO-BT 77, CHO-BT 112, CHO-GSD 47 (mesial towards the right, except for GSD 47). Crown surface of CHO-GSD 47 is abraded exposing parts of dentine.
CHO-BT 112 is a partial right m1 or 2 lacking much of the lingual crown. Crown height is low, with a strong buccal crown flare. The buccal cleft is deeper than in BT 77 and opens occlusally (V-shaped in lateral view). Enough of the entoconid is preserved to show that the main lophids were much lower than in the colobine BT 78 or BT 114. Despite crown surface abrasion, distinct remnants of the mesial and distal buccal grooves are preserved. The mesial groove appears strong and delineates a fold-like mesiobuccal ridge. Judging from the preserved buccal cusp sizes, BT 112 represents a comparably sized or slightly smaller individual than BT 77. Its length is >6.8 mm (crown faces abraded away).
We attribute these molars to Cercopithecinae, probably papionin, based on their low cusps, strong buccal flare, and expressed buccal grooves (BT 112). Despite incomplete preservation, enough of the buccal cusps are preserved to enable rough assessments of size. They are smaller than the Lothagam Upper and Lower Nawata Mb Parapapio lothagamensis (Leakey et al., 2003), and comparable to small modern papionins such as Lophocebus, Cercocebus, or small species of Macaca. It is possible that they represent a species of Parapapio smaller than P. lothagamensis, although other attributions are also possible.
The CHO-BT 2 phalanxAt Beticha, a distal half of a proximal phalanx shaft was recovered. This partial phalanx is somewhat abraded and lacks the articular end, making its evaluation difficult. It is probably the shaft of a proximal pedal phalanx (M. Nakatsukasa, personal communications), compatible in size with either the Beticha colobine or cercopithecine represented by the molars.
RemarksThe relatively well-preserved CHO-BT 78 lower molar is close in size and morphology to the molar from Nakali, KNM-NK 305, described in detail by Benefit and Pickford (1986). Metrics that evaluate lingual notch development (cusp/lophid saliency) are known to numerically differentiate modern and fossil colobines from cercopithecines (Benefit and Pickford, 1986; Benefit, 1993; Gilbert et al., 2010; Rossie et al., 2013). BT 78 has values well within Colobinae and close to KNM-NK 305. One difference between the two molars is crown shape; mesial crown breadth is smaller than distal breadth in BT 78, while the reverse is the case in KNM-NK 305 (Table 4). However, this difference is slight and easily encompassed within the variation of a single taxon. It is tempting to suggest that BT 78 and KNM-NK 305 represent the same or closely related species. However, the Beticha monkey is almost 2 million years younger. Furthermore, if the few Beticha colobine specimens represent a single taxon, then its dental size would be considerably larger than Microcolobus tugenensis (Benefit and Pickford, 1986). Although the recently recovered Nakali Microcolobus sp. fossils are mostly undescribed (Kunimatsu et al., 2007), the partial skeleton of the Nakali Microcolobus sp. (Nakatsukasa et al., 2010) has an m3 as small as that of M. tugenensis. Thus, although it is possible that BT 78 represents a large individual of the Nakali Microcolobus lineage, it is likewise possible that the Beticha colobine represents a different, larger species, with or without Microcolobus affinities. Only more fossils will enable a resolution of this matter.
Few fragmentary dental remains of early African colobines that are comparably small-sized are known from Lothagam Lower Nawata Mb (Leakey et al., 2003), Tugen Hills Lukeino Fm (Gilbert et al., 2010), and Lemudong’o (Hlusko, 2007). The second (not first) molars of these small colobines are about the size of the CHO-BT 78 molar, which suggests that they were smaller than the Beticha colobine. However, these post ~7.4 Ma colobines have greater cusp relief, and are thus more modern-looking. At Lothagam Lower Nawata Mb (6.5–7.4 Ma) and Mpesida (6.4–7.2 Ma), colobine species larger than the Beticha colobine are also known from fragmentary teeth. Thus, the Beticha colobine probably represents a morphologically conservative lineage intermediate in size between the smaller and larger morphologically more advanced post-7 Ma species.
Finally, it is possible that the fragmentary canines recovered from the Chorora TL and Beticha represent the same colobine species as the molars. If that were the case, canine size might have been relatively smaller than in modern African colobines. Considerable variation in relative canine size is seen among taxa, with modern Asian colobine species tending to have relatively smaller canine sizes (Swindler, 2002).
The upper Chorora levels: Teso Tadecho and Gutosadeen Probable colobine premolars (Figure 16, Table 3)CHO-TTD 1 and TTD 2 are upper premolars (P3 and P4, respectively) found on the sediment surface within a few meters of each other. The P3 was just coming into wear, and the P4 lacks any sign of occlusal wear or polish. Because the P4 is smaller than the P3, a size relationship not usually encountered in cercopithecid dentitions, these premolars probably represent two individuals.
Probable colobine upper premolars from Teso Tadecho. From left to right: CHO-TTD 1 left P3 occlusal and mesial views, CHO-TTD 2 left P4 occlusal and mesial views.
The P3 exhibits a tall and asymmetric paracone with moderate mesiobuccal basal asymmetry. Although the protocone is much lower in height, it is nevertheless distinctly developed, and not rudimentary as is the case in many modern colobines. The P4 is characterized by a moderately tall paracone and a well-developed protocone. Because paracone height is not as exaggerated as in the P3, the two P4 cusps are subequal in height. This upper P4 appears close in size and morphology to the ~6 Ma cf. colobine premolar from Lukeino described by Gilbert et al. (2010). They considered the Lukeino tooth as probably colobine, but cautiously stated that this identification was uncertain. However, the non-bulbous cusps and moderately sharp occlusal ridges of the Teso Tadecho premolars are more typical of colobine than cercopithecine monkeys. They also resemble Mesopithecus homologues in a relatively projecting P3 and P4 protocone, probably a primitive retention.
In size, the Teso Tadecho premolars are compatible with the Beticha colobine species represented by the upper and lower molars. Together with the conservative notch development of the Beticha molars, also shared by Mesopithecus, these primitive features might have been characteristic of a wide range of early colobine species lineages.
Large cercopithecine molar (Figure 15, Table 3)CHO-GSD 47 is a lower molar (m1 or 2) of a larger-sized cercopithecid than any of the above-described taxa. The crown is fluvially eroded, so that the original tooth surface is lost. Abrasion exposes large parts of the internal dentine structure. Although damage makes metric and morphological comparisons difficult, its crown and cusp heights appear low, and buccal flare is undoubtedly strong. This molar most likely represents a cercopithecine monkey, decidedly larger than the Beticha cercopithecine. Its size is in the range of Parapapio lothagamensis, which it may represent.
The taxonomic representations of the mammalian fossils from the five Chorora Fm localities are summarized in Table 1. According to our revised chronostratigraphy of the Chorora Fm, the TL and Beticha faunas are ~8.5 and ~8.0 Ma, respectively. Therefore, the large Chorora hipparionin is no longer among the oldest of African equid representatives. Abundant hipparionin fossils are reported from Nakali (Kunimatsu et al., 2007; Uno et al., 2011) close to 10 Ma. Pickford (2001) reported hipparionin fossils from Member E (Bed E4) of the Ngorora Fm, the age of which is best considered ill-constrained between 10 and 11 Ma. Thus, the sub-Saharan first appearance datum (FAD) of Equidae is either <10 Ma at Nakali or slightly earlier at Ngorora Fm Member E.
Some commonalities are seen between the Chorora TL (~8.5 Ma) and Beticha (~8.0 Ma) faunal assemblages. The large hipparionin that dominates the TL fauna also occurs at Beticha. In both assemblages, a morphologically similar small boselaphin is the common bovid, and Nyanzachoerus, Paleotragus, a sivathere, and the rodent Paraulacodus occur. However, whereas equids (40%) and rodents (20%) dominate the TL assemblage, they are much rarer at Beticha (3% each). On the contrary, the small hipparionin is documented at Beticha, but so far not at the TL. The rhinocerotid representation of the two localities also differs. Cercopithecids are more common at Beticha, and Chororapithecus occurs at Beticha but not at the TL. These differences may stem from habitat sampling biases more so than age; the Chorora TL is probably indicative of a more open environment than Beticha. However, the Chorora TL hipparionin consumed less C4 grass than the earlier Samburu Hills (Namurungule Fm) hipparionins, which could indicate even a more open habitat at Samburu Hills.
That the Chorora TL and Beticha faunas postdate the Namurungule and Nakali faunas is corroborated by several lines of evidence. The cercopithecids from the Chorora Fm appear more diverse, and the Beticha colobine is larger-sized than Microcolobus of Nakali and Ngeringerowa. Furthermore, at Beticha, a primitive hippopotamine and a possible early bovin occur; neither has been recorded from Nakali or Samburu Hills so far. Finally, the TL Stegotetrabelodon appears more advanced than the Samburu Hills Tetralophodon (Geraads et al., 2002).
At Chorora TL a hippopotamid premolar was initially reported as Hippopotamidae gen et sp. indet. by Sickenberg and Schönfeld (1975), and later as cf. Kenyapotamus sp.? by Geraads et al. (2002). Geraads et al. (2002) suggested resemblances with Ngeringerowa Kenyapotamus (9–9.5 Ma), including an inferred lack of anterior diastema. At the same time, they also noted that the Chorora premolar exhibits some advanced features such as a higher crown. They considered the generic attribution to Kenyapotamus as inconclusive. Kenyapotamus has so far been recorded from the Kenyan sites ranging from ~9 to 15 Ma. The 9–10 Ma occurrences are represented at Nakali, Samburu Hills, and Ngeringerowa, while the earliest occurrence of Hippopotaminae has been considered ~7.4 Ma (the lower levels of Lothagam Lower Nawata Mb) (Boisserie et al., 2010, 2011). Thus, as noted by Geraads et al. (2002), it is important to establish the identity of the Chorora TL hippopotamid, especially now that its age is considered ~8.5 Ma. At Beticha, a hippopotamine with low crown height and primitive molar structure is present. The occurrence of a primitive hippopotamid at Beticha, but nevertheless attributable to Hippopotaminae, is in accord with our geochronological revisions of the Chorora Fm fossils at <9 Ma. However, it is possible that hippoppotamines arose from ~12 Ma or earlier Kenyapotamus (Boisserie et al., 2010).
Further support for a 7–9 Ma time range of the Chorora Fm sediments comes from similarities of the Chorora Fm upper-level fauna with that of the Lothagam Lower Nawata Mb. These include the presence of a nyanzachoere indistinguishable from Lothagam Ny. tulotos, the occurrence of the leporid Alilepus, and the presence of a moderately high-crowned Reduncini. Occurrence of Parapapio hinted by the large cercopithecine molar from Gutosadeen, if confirmed, would further strengthen the ~7.0–7.5 Ma placement of the upper Chorora fauna. Alilepus indicates that the Chorora Fm upper levels postdate 7.5–8 Ma (Flynn et al., 2013). Nevertheless, based on the marginally more primitive hippopotamine and hipparionin (see below), the upper Chorora fauna probably predates most of the Lothagam Lower Nawata Mb fauna. It is also worthy of note that an anterolabial cuspid is present on the lower molars of Paraulacodus johanesi, even in those from the upper Chorora levels (e.g. Teso Tadecho). This is a primitive feature absent in the closely related Protohummus dango (Kraatz et al., 2013) from the Baynunah Fm, Abu Dhabi. The latter is best considered 6.5–7.5 Ma (Bibi et al., 2013), in turn suggesting that the upper Chorora fauna is older than this. Further geochronological refinement is needed for a better resolution of these issues.
Late Miocene equid evolutionGeraads et al. (2002) conservatively assigned the large Chorora TL hipparionin to Hipparion sp. cf. primigenium. They noted the similarity of the Chorora hipparionin with the Bou Hanifia “Hipparion” africanum (quotation ours, or “Cormohipparion” following Bernor and White, 2009) in occlusal morphology such as protocone shape, demarcated hypocone, and hipparionin double knot. At the same time, Geraads et al. (2002) noted that the Chorora hipparionin cheek teeth were slightly larger and had stronger plications. According to Bernor et al. (2004, 2010), these and other morphologies, such as the lack of ectostylids, are best considered primitive retentions from their early Late Miocene Eurasian ancestry. This primitive dental pattern is broadly shared by the other medium- to large-sized early Late Miocene African forms represented at Ngorora, Nakali, and the Samburu Hills (Nakaya and Watabe, 1990; Geraads et al., 2002; Bernor et al., 2010). Taking into account the facial morphology of the Samburu Hills hipparionin (Nakaya and Watabe, 1990), Bernor and colleagues referred the Chorora hipparionin to “Cormohipparion” sp. (Bernor et al., 2004, 2010) and more recently to “C.” aff. africanum (Bernor et al., 2012; Melcher et al., 2014).
Bernor et al. (2010) also suggested that the early Late Miocene hipparionin of eastern Africa, represented by Namurungule Fm “Cormohipparion” aff. africanum (~9.6 Ma), while retaining primitive morphology in common with northern African “C.” africanum (~10.5 Ma), appears phylogenetically close to Siwalik Sivalhippus perimensis (7–9 Ma). This was based on the observation that the Namurungule hipparionin and the Siwalik S. perimensis share a dorsally positioned preorbital fossa, perhaps related with increased hypsodonty. In the Namurungule hipparionin (large morph), measured tooth crown height has been recorded to range in excess of 60 mm (Nakaya et al., 1984, 1987; Tsujikawa, 2005), suggesting maximum crown heights of ~70 mm (Bernor et al., 2010). Wolf et al. (2013) suggested that the Namurungule hipparionin may have had close relationships with the penecontemporaneous Siwalik Sivalhippus nagriensis (~9–10.5 Ma). Despite some differences in facial morphology, S. nagriensis was considered to share similarities with the Namurungule hipparionin in cheek tooth morphology and in moderate crown heights, at least in the specimens postdating 10 Ma (Wolf et al., 2013). Wolf et al. (2013) furthermore referred the Lothagam Eurygnathohippus turkanensis materials (<7.4 Ma) to Sivalhippus turkanensis, based on shared robust distal limb bone proportions of the Lothagam hipparionin and S. perimensis (in addition to the previously noted cranial similarities, Bernor and Harris, 2003).
Prior to this, Nakaya and colleagues (Nakaya and Watabe, 1990; Nakaya, 1994; Tsujikawa, 2005) had also attributed the Namurungule hipparionin to “Hipparion” aff. africanum or “H.” africanum (quotations ours) based on shared preorbital morphology, dental similarities, and the presence of postcranial elements that suggest moderately slender limb bones. However, as discussed below, if two taxa, one larger and the other smaller, were represented at Namurungule, it is not clear if the larger morph with known cranial and dental similarities with the Bou Hanifia “H.” africanum actually had slender limb bones. The slender metapodial and phalanges at Namurungule tend to be small and may represent the smaller species, while Nakaya et al. (1984, 1987) noted that the larger postcranial specimens look more robust (see also above).
Nakaya et al. (1984, 1987) initially suggested the presence of two taxa at Namurungule: a larger morph with a more complex occlusal enamel pattern and relatively robust distal limb bones, and a smaller morph perhaps with simpler molar morphology and slender metapodials and phalanges. However, in their more detailed publication (Nakaya and Watabe, 1990), the Namurungule hipparionin materials were not separated into two species. This was because, although they observed a bimodal distribution of tooth crown dimensions, the degree of dental size and postcranial variation was not sufficient to reject a single taxon interpretation.
The Chorora and Namurungule fossils collectively provide new insights into Late Miocene African equid evolution. We suggest that both larger and smaller hipparionin lineages were present in eastern Africa perhaps by ~9.6 Ma at Samburu Hills (Namurungule Fm). The Namurungule large-morph hipparionin combines primitive cheek tooth morphology, moderately high crown, and facial features reminiscent of earlier (10.5 Ma) northern African “Cormohipparion” africanum. By Chorora times, at ~8.5 Ma, the large-morph hipparionin is even larger and has more complicated cheek tooth plication patterns. It is tempting to suggest that “Cormohipparion” sp. of Chorora was a descendent of the Namurungule large-morph hipparionin. In such a scenario, Lothagam Eurygnathohippus turkanensis may be a further descendent of this Namurungule–Chorora large-sized hipparionin lineage.
Although one of us (R.B.) prefers the hypothesis that Lothagam Eurygnathohippus turkanensis (<7.4 Ma) (alternatively Sivalhippus turkanensis) is an immigrant species from southern Asian, the Chorora fossils that partially fill the 8–9 Ma gap of knowledge in sub-Saharan Africa indicate that there are other possibilities. One possibility is a sister lineage relationship between E. feibeli and E. turkanensis in eastern Africa, that occurred in parallel with diversification of Sivalhippus spp. in southern Asia. Here, it is imperative to recognize that the Chorora fossils correlate closely in time with the Siwalik horizons that contain Sivalhippus perimensis (7–9 Ma). If a sub-Saharan African lineage had evolved from moderate to large size accompanied by increasingly robust distal limbs, as suggested by the combined Namurungule and Chorora evidence (Figure 17), then it is not clear which way the robust-limbed form might have dispersed; it could have dispersed from eastern Africa to the Siwaliks. The >9 Ma eastern African (Ngorora) and slightly younger Siwalik S. perimensis proximal phalanges are robust but marginally less so than the Lothagam E. turkanensis homologues (Wolf et al., 2013). This suggests a parallel timing of emergence of derived limb morphology, or perhaps a slightly earlier derivation in eastern Africa. Alternatively, there may have been a common hipparionin species (or superspecies) pool between the Siwaliks and sub-Saharan Africa extending back to >9 Ma, therefore giving rise to the similarities. Further informative fossils from Chorora, Samburu Hills, the Siwaliks, and elsewhere are needed to better assess the situation. In either case, Bernor et al.’s (2010) suggestion of a close bioprovincial relationship between eastern African and southern Asia is supported.
Late Miocene hipparionin phalanges. Upper row, dorsal (anterior) views of ray III proximal phalanx (from left to right): Namurungule Formation KNM-SH 18008, KNM-SH 12299; Lothagam Lower Nawata Member Eurygnathohippus feibeli KNM-LT 25472, E. turkanensis KNM-LT 26294. Lower row, dorsal (anterior) views of ray III intermediate phalanx (from left to right): Namurungule Fm KNM-SH 14777; Chorora Type Locality CHO-TL 40; Lothagam Lower Nawata Member E. feibeli KNM-LT 25472, E. turkanensis KNM-LT 25440.
The presence of a small-morph hipparionin at Chorora is equally of interest. Although even less well known than the large-morph Chorora hipparionin, its cheek teeth are broadly similar to both the Samburu Hills small-morph hipparionin and Lothagam Eurygnathohippus feibeli. However, they differ from E. feibeli homologues in lacking ectostylids or the distinctly ‘sharp’ parastyle considered characteristic of Lothagam and later E. feibeli (Bernor and Harris, 2003; Bernor and Haile-Selassie, 2009). It is possible that the Namurungule and Chorora small-morph hipparionin represents a continuous sub-Saharan Late Miocene lineage, ancestral to the Lothagam (and later) E. feibeli. The latter is characterized by advanced emphasis of slender limb bones, and retention of facial fossa and low postcanine crown height (Bernor et al., 2010; Melcher et al., 2014). We hypothesize that E. feibeli might have evolved from the Namurungule–Chorora small hipparionin lineage. The limb bone similarities recognized between the Namurungule hipparionin and northern African “Cormohipparion” africanum (Nakaya and Watabe, 1990) may reflect the primitive condition of this slender-limbed lineage, which was further enhanced (lengthening of metapodials and proximal phalanges) in E. feibeli (Figure 17). On the contrary, the large-morph lineage appears to have increased their limb robusticity through time. Such an early lineage divergence within the sub-Saharan Late Miocene hipparionins needs to be confirmed by further fossils spanning 7 to >10 Ma.
Late Miocene cercopithecid evolutionSignificant cercopithecid fossils have been recently described from the late Middle and Late Miocene of sub-Saharan Africa (Leakey et al., 2003; Hlusko, 2007; Frost et al., 2009; Gilbert et al., 2010; Nakatsukasa et al., 2010; Rossie et al., 2013). In northern Africa, Late Miocene materials from ~7 Ma and later are known from Sahabi and elsewhere (Jablonski, 2002; Benefit et al., 2008). Recently, a possible early Cercopithecini molar was reported from the 6.5–7.5 Ma Baynunah Fm, Abu Dhabi (Bibi et al., 2013; Gilbert et al., 2014). The Baynunah cf. Cercopithecini molar differs from the Beticha cf. Papionini molars in its smaller size and narrower crown with less buccal flare. It differs from the Beticha colobine in lower crown and cusp notch heights. These fossils, although highly fragmentary, are steadily illuminating the little-known early evolutionary trajectories of the modern cercopithecid subfamilies and tribes.
Nakatsukasa et al. (2010) analyzed a Microcolobus partial skeleton from Nakali and demonstrated that arboreal adaptation was fully established in those colobines close to 10 Ma. They furthermore pointed out that the Nakali early colobine shares some details of postcranial morphology specifically with the modern African colobines. These observations suggest that the Asian and African colobine clades separated prior to ~10 Ma, and that their common ancestor was likely arboreal. The Chorora colobine appears to differ from Microcolobus spp., although the few described Microcolobus specimens from Nakali (~9.8 Ma) and Ngeringerowa (9– 9.5 Ma) share with the Beticha colobine a conservative molar shearing capacity relative to modern colobines (Table 4). In these Late Miocene colobines, the lingual notch base of the lower molar lies at approximately mid-crown height. This is also the case with the broadly contemporaneous ~7– 8.5 Ma Mesopithecus spp. of Eurasia. On the contrary, known Late Miocene African colobines postdating ~7.4 Ma have taller lingual cusps with deep lingual notches, comparable to the modern colobine condition (Leakey et al., 2003; Hlusko, 2007; Frost et al., 2009; Gilbert et al., 2010). The acquisition of greater molar shear, perhaps indicating more committed folivory, seems to have occurred across a wide range of body sizes and lineages by the Late Miocene, as inferred from the diversity of advanced colobines known in eastern Africa postdating ~7 Ma.
Despite conservative molar-shear shared by the Beticha colobine, known Microcolobus, and Mesopithecus spp. (Benefit and Pickford, 1986; de Bonis et al., 1990, 1997; Koufos, 2009), these early colobines differ morphologically and apparently represent distinct lineages. The implication is that there was probably a diversity of colobine lineages in sub-Saharan Africa by ~9 Ma that are yet to be discovered and recognized. Compared to the Beticha lower molars, the Mesopithecus homologues are not only broader crowned, but tend to be broader distally than mesially (Table 4). Mesopithecus also tends to exhibit stronger buccal cingular expressions (Benefit and Pickford, 1986) than the Beticha colobine. The Beticha colobine further differs from early Mesopithecus spp. in an asymmetric lower molar lingual notch (with concave mesial border).
While Rossie et al. (2013) reported a probable colobine lower molar at 12.5 Ma, until now, the oldest reported cercopithecine was the Lothagam Lower Nawata Mb Parapapio lothagamensis at <~7.4 Ma and the cf. Cercopithecini from the 6.5–7.5 Ma Baynunah Fm. The Chorora fossils described above record at least three cercopithecid species lineages between ~7 and ~8.5 Ma, and securely establish the presence of both Colobinae and Cercopithecinae at ~8.0 Ma, slightly earlier than at Lothagam. The Chorora ~8 Ma cercopithecine is probably a papionin that was smaller than P. lothagamensis. Other Late Miocene cercopithecines so far known are the few papionins from northern Africa (<~7 Ma) (Benefit et al., 2008). These are in the P. lothagamensis size range but differ morphologically. Some of these have been attributed to Macaca presumably mainly from biogeographic considerations, while Benefit et al. (2008) conservatively assigned the few Sahabi mandibular fragments and teeth to cf. Macaca or Parapapio sp.
Recent fieldwork at the Late Miocene Chorora Fm, Ethiopia, has resulted in new mammalian fossils and new geochronological frameworks. The Chorora Fm faunas occur in several chronostratigraphic intervals: the ~8.5 Ma TL fauna, the ~8.0 Ma Beticha fauna, and the ~7–7.5 Ma upper Chorora fauna. These are the first mammalian fossils reported from sub-Saharan Africa that span the 7–9 Ma time period. The Chorora TL and Beticha faunas share some time-distinctive taxa that indicate a biochronological position in between the ~9.6 Ma Samburu Hills and the <~7.5 Ma upper Chorora faunas. The newly recovered Chorora Fm fossils include the earliest known records of Cercopithecinae and Hippopotaminae in Africa (and elsewhere). In contrast, compared with the TL or Beticha faunas, the upper Chorora fauna exhibits more evolved forms with increasing similarities with the 6.5–7.4 Ma Lothagam Lower Nawata Mb fauna. These include a nyanzachoere suid indistinguishable from Nyanzachoerus tulotos, a hippopotamine more advanced than at Beticha, and Alilepus representing a new African FAD of Leporidae at ~7–7.5 Ma. The Chorora hipparionins are a potential link between the earlier Samburu Hills and later Lothagam counterparts. They appear to represent two lineages, perhaps ancestral to Lothagam Eurygnathohippus turkanensis and E. feibeli, respectively. The Chorora colobines are larger than the >9 Ma Microcolobus, and morphologically conservative in exhibiting only moderately developed molar cusp notches. The Chorora cercopithecines represent the earliest documented occurrence of the subfamily. The emerging evidence suggests the hidden presence of considerable cercopithecid diversity in sub-Saharan Africa prior to 8–9 Ma.
We thank the Culture and Tourism Office Western Hararge Chiro Zone and the administration of the Mieso Woreda for fieldwork support, the Bureau of Culture and Tourism of Oromia Region for facilitation, and the Authority for Research and Conservation of Cultural Heritage, Ministry of Culture and Tourism of Ethiopia for permissions and facilitation of the project. We thank all participants in the fieldwork, especially the Gololcha people, who were essential to the success of the project. We thank M. Nakatsukasa, Y. Kunimatsu, D. Shimizu, M. Takai, T. White, Y. Haile-Selassie and H. Saegusa for suggestions and/or comparative materials. This project was supported primarily by the Japan Society for the Promotion of Science (Kakenhi grant numbers 21255005 and 2400005).
