Early Homo at 2.8 Ma from Ledi-Geraru, Afar, Ethiopia

See allHide authors and affiliations

Science  20 Mar 2015:
Vol. 347, Issue 6228, pp. 1352-1355
DOI: 10.1126/science.aaa1343

Finding Homo nearly 3 million years ago

The fossil record of humans is notoriously patchy and incomplete. Even so, skeletal remains and artifacts unearthed in Africa in recent decades have done much to illuminate human evolution. But what is the origin of the genus Homo? Villmoare et al. found a fossil mandible and teeth from the Afar region in Ethiopia. The find extends the record of recognizable Homo by at least half a million years, to almost 2.8 million years ago. The morphological traits of the fossil align more closely with Homo than with any other hominid genus. DiMaggio et al. confirm the ancient date of the site and suggest that these early humans lived in a setting that was more open and arid than previously thought.

Science, this issue p. 1352, p. 1355


Our understanding of the origin of the genus Homo has been hampered by a limited fossil record in eastern Africa between 2.0 and 3.0 million years ago (Ma). Here we report the discovery of a partial hominin mandible with teeth from the Ledi-Geraru research area, Afar Regional State, Ethiopia, that establishes the presence of Homo at 2.80 to 2.75 Ma. This specimen combines primitive traits seen in early Australopithecus with derived morphology observed in later Homo, confirming that dentognathic departures from the australopith pattern occurred early in the Homo lineage. The Ledi-Geraru discovery has implications for hypotheses about the timing and place of origin of the genus Homo.

Fifty years after the recognition of the species Homo habilis as the earliest known representative of our genus (1), the origin of Homo remains clouded. This uncertainty stems in large part from a limited fossil record between 2.0 and 3.0 million years ago (Ma), especially in eastern Africa. Some taxa from this time period, such as Australopithecus africanus (~2.8 to 2.3 Ma) and the less well known A. garhi (~2.5 Ma) and A. aethiopicus (~2.7 to 2.3 Ma), appear too specialized cranially and/or dentally to represent the probable proximate ancestral conditions for Homo species known in Africa by ~2.0 Ma (H. habilis and H. rudolfensis). This leaves a thin scatter of isolated, variably informative specimens dated to 2.4 to 2.3 Ma as the only credible fossil evidence bearing on the earliest known populations of the genus Homo (2, 3).

Here we describe a recently recovered partial hominin mandible, LD 350-1, from the Ledi-Geraru research area, Afar Regional State, Ethiopia, that extends the fossil record of Homo back in time a further 0.4 million years. The specimen, securely dated to 2.80 to 2.75 Ma, combines derived morphology observed in later Homo with primitive traits seen in early Australopithecus. The discovery has implications for hypotheses concerning the timing and place of Homo origins.

The LD 350 locality resides in the Lee Adoyta region of the Ledi-Geraru research area (Fig. 1). Geologic research at Lee Adoyta (4) identified fault-bounded sedimentary packages dated 2.84 to 2.58 Ma. The LD 350-1 mandible was recovered on the surface of finely bedded fossiliferous silts 10 m conformably above the Gurumaha Tuff (Fig. 1). The matrix adherent to the specimen is consistent with it having eroded from these silts [for details on stratigraphy and depositional environment, see (4)]. The Gurumaha Tuff is radiometrically dated to 2.822 ± 0.006 Ma (4), a date that is consistent with the normal magnetic polarity of the Gurumaha section, presumably the Gauss Chron. An upper bounding age for LD 350-1 is provided by an adjacent downfaulted younger block that contains the 2.669 ± 0.011 Ma Lee Adoyta Tuff. A magnetostratigraphic reversal 12 m conformably above the Lee Adoyta Tuff is inferred to be the Gauss/Matuyama boundary at 2.58 Ma (4). Because no significant erosional events intervene between the Gurumaha Tuff and the fossiliferous horizon, the age of LD 350-1 can be further constrained by stratigraphic scaling. Applying a sedimentation rate of either 14 cm per thousand years (ky) from the Lee Adoyta fault block or 30 cm/ky from the Hadar Formation (5) provides age estimates of 2.77 and 2.80 million years (My), respectively, for LD 350-1. Based on the current chronostratigraphic framework for Ledi-Geraru, we consider the age of LD 350-1 to be 2.80 to 2.75 My.

Fig. 1 The geographic and geologic setting of the LD 350-1 hominin mandible.

(A) Map of the Lee Adoyta region in the Ledi-Geraru research area. (B) Stratigraphic section of the Lee Adoyta sequence indicating the provenience of LD 350-1. (C) The LD 350 locality, with the position of the LD 350-1 mandible on the surface in relation to the Gurumaha Tuff. Resistant ledges in the center of the photo are localized cemented sand and carbonate nodule pebble conglomerate. The section in (B) is indicated by the black line at left in (C). Details in (6).

The hominin specimen, found by Chalachew Seyoum on 29 January 2013, comprises the left side of an adult mandibular corpus that preserves the partial or complete crowns and roots of the canine, both premolars, and all three molars. The corpus is well preserved from the symphysis to the root of the ascending ramus and retromolar platform. Surface detail is very good to excellent, and there is no evidence of significant transport. The inferior margin of the corpus and the lingual alveolar margin are intact, but the buccal alveolar margin is chipped between P3 and M1. The P4, M2, and M3 crowns are complete and well preserved, but the C, P3, and M1 crowns are incomplete (Fig. 2 and text S2). The anterior dentition is represented by the broken root of the lateral incisor and the alveolus of the central incisor.

Fig. 2 The LD 350-1 mandible.

(A) Medial view. (B) Lateral view. (C) Occlusal view. (D) Basal view. (E) Enlarged occlusal view of dentition. Scale bars, 1 cm.

Given its location and age, it is natural to ask whether the LD 350-1 mandible represents a late-surviving population of A. afarensis, whose youngest known samples date to ~3.0 Ma at the neighboring Hadar site (6). Indeed, in overall dental and mandibular size, LD 350-1 matches smaller specimens of A. afarensis (figs. S1 to S4 and tables S1 to S3). In addition, LD 350-1 shares with A. afarensis a primitive anterior corpus, including an inclined symphyseal cross section, a bulbous anterior symphyseal face, and a projecting inferior transverse torus that is only slightly elevated above the corpus base (Fig. 2A). There are, however, substantial differences between LD 350-1 and the mandibles and teeth of A. afarensis. LD 350-1 lacks the lateral corpus hollow in which the mental foramen is located, which distinguishes A. afarensis mandibles across their size range. Instead, the LD 350-1 corpus is slightly convex inferosuperiorly, with the mental foramen located lateralmost on this arc (fig. S5). Unlike in A. afarensis, the anterior corpus of LD 350-1 bears a distinct symphyseal keel, accentuated by a flattened area between it and a pronounced canine jugum (Fig. 2). It lacks both an incisura mandibulae anterior and a mentum osseum, and thus does not have a chin. The P3 buccal crown profile is symmetrical in occlusal view, lacking the prominent mesiobuccal extension that skews the P3 crown outline in A. afarensis. The P3 occlusal enamel is worn through to dentine buccally, yet the moderately worn molars retain considerable occlusal relief. In A. afarensis, a reversed wear pattern is typical: The canine and mesial moiety of P3 remain relatively unscathed by wear even when molar occlusal enamel is exhausted (6). The wear disparity in LD 350-1 may signal modification of the relatively primitive C/P3 complex of A. afarensis, which, while non-honing, retained apelike aspects of crown shape and occlusal wear (6, 7). The basally expanded hypoconid that creates the characteristic “bilobate” buccal crown contour in A. afarensis M1s and M2s is not evident in LD 350-1 (7). Finally, there is a distinct C7 cusp on the LD 350-1 M1 (Fig. 2), yet no such cusp has been recorded for an A. afarensis M1 (n = 18) (8, 9).

Despite a well-developed lateral prominence that swells the corpus anterior to the ascending ramus, LD 350-1 lacks the great corpus thickening, inflated corpus contours beneath P3-M1, and anterior dental arch constriction seen in robust australopith mandibles, including Omo 18-1967-18 (holotype of A. aethiopicus, ~2.6 Ma). Postcanine dental expansion, including the differential enlargement of lower molar entoconids typical of robust australopiths, is also absent (8) (figs. S1 to S3 and table S4).

LD 350-1 also differs from Australopithecus in features that align it with early Homo. The inferior and alveolar margins of the corpus are subparallel, resulting in similar corpus depths at P3 and M2 (Fig. 2), whereas in A. afarensis, A. africanus, and A. sediba, the corpus is typically deepest under the premolars (10, 11). Some specimens of early Homo, such as SK 45 and KNM-ER 60000 (10, 12), have a deeper corpus anteriorly, but LD 350-1 resembles many others that are relatively uniform in depth along the corpus (10, 1316) (figs. S5 and S6 and table S5). In LD 350-1, the mental foramen opens directly posteriorly into a short groove on the corpus. A posteriorly opening foramen is the most common condition in early and modern Homo (13, 17). The foramen always opens anteriorly or anterosuperiorly in A. afarensis mandibles, whereas in A. africanus its orientation ranges from anteroinferior to directly lateral. The anterior margin of the ascending ramus in LD 350-1 becomes independent from the corpus opposite the middle of the M3 crown, and in lateral view only the distalmost part of this crown is hidden (Fig. 2 and fig. S5). In the great majority of Australopithecus mandibles, the anterior margin of the ramus is positioned further anteriorly, between mid- and distal M2. A posterior origin of the ramus margin is most common in specimens of early Homo (fig. S7 and table S6).

The LD 350-1 molars have simple occlusal surfaces, with salient, marginally positioned cusp apices and broad occlusal basins. With the exception of a tiny protostylid on M3, the crowns are devoid of cingular remnants. Estimated M1 length is small (fig. S4). The breadth across the mesial aspect of M2 and M3 measures less than the distal breadth, giving these crowns a mesially tapered occlusal outline. This distinctive outline is due in part to a relatively confined mesiolingual cusp (metaconid) on these teeth (Fig. 2 and table S4). In A. afarensis and A. africanus, and also in A. sediba specimen MH1, M2 and M3 crown outlines are square or distally tapered (especially M3) (fig. S8). Although some early Homo distal molars show reduced mesial breadths, there is significant overlap with Australopithecus in mesial:distal breadth ratio, and none approaches the uniquely tapered outline of LD 350-1 (table S7). Whereas Australopithecus M2 and M3 crowns usually have a sloping buccal face up to the point of advanced occlusal wear, the buccal walls of these teeth in LD 350-1 descend almost vertically from the occlusal rim, as frequently observed in early and later Homo (fig. S8). LD 350-1 departs from the A. africanus dental pattern, which includes buccolingually expanded postcanine teeth (especially M2-3) and flattened molar occlusal surfaces in specimens with similar degrees of premolar wear (6, 7). LD 350-1 possesses a mesiodistally shorter M3 than M2, which is common in Homo but occurs only in 1 out of 16 A. afarensis and 3 out of 14 A. africanus specimens in our sample. In crown length, the LD 350-1 M3 falls outside the distribution of 20 A. africanus and 19 A. afarensis M3s (fig. S3).

With an age of 2.80 to 2.75 My, the Ledi-Geraru mandible is younger than the youngest known A. afarensis fossils from Hadar, ~3.0 Ma. Although it shares with A. afarensis primitive anterior corpus morphology, the specimen falls outside the range of variation for most other diagnostic mandibular and dental traits of this species. In the majority of traits that distinguish it from this species, LD 350-1 presents morphology that we interpret as transitional between Australopithecus and Homo.

The Homo postcanine dental pattern (18) is present in members E through G of the Shungura Formation of Ethiopia (~2.4 to 2.0 Ma) and is best represented by mandible Omo 75-14, which has a reduced M3 crown and a C7 cusp on M1 (18), features shared with LD 350-1. The A.L. 666-1 maxilla from the Busidima Formation at Hadar also establishes the Homo dentognathic pattern by ~2.3 Ma (19). A small collection of postcanine teeth from members B and C (~2.9 to 2.6 Ma) of the Shungura Formation evinces more derived morphology than A. afarensis and has been affiliated with A. africanus/Homo (18, 20). Although A. africanus is usually thought to have been endemic to southern Africa, the corresponding time period in the East African record is notoriously poor in hominin fossils. At ~2.80 to 2.75 Ma, LD 350-1 falls within the member B-C temporal interval and is broadly contemporaneous with A. africanus samples from Makapansgat and Sterkfontein (21). Morphologically, however, it is distinguished from A. africanus by its subequal anterior and posterior corpus heights, posterior position of the anterior ramus margin, posteriorly oriented mental foramen, steeper premolar-molar wear gradient, vertical buccal walls of M2-3, and reduced M3. In the combination of these features, LD 350-1 resembles younger East African Homo specimens more than it does geologically contemporaneous or older Australopithecus. Summarizing the Shungura Formation dental evidence, Suwa et al. [(18) pp. 270–271] identified the 2.9- to 2.7-Ma interval as “the transitional period when evolution occurred from an A. afarensis-like to a more advanced species…close in overall dental morphology to the A. africanus condition but also mostly within the A. afarensis and/or early Homo ranges of variation.” We consider the Ledi-Geraru mandible to sample a population from this transition and to point to a close phyletic relationship with Homo at 2.4 to 2.3 Ma.

A set of teeth comprising the isolated left and right P3-M2 crowns of a single individual (KNM-ER 5431) from the upper Tulu Bor Member at Koobi Fora, Kenya (22), is germane to the status of LD 350-1. Constrained in age between ~2.7 and ~3.0 Ma (23), these teeth show a mix of australopith (premolar cusp morphology) and early Homo-like (C7 cusp on M1 and M2) characters (15), whereas its P3 form has been characterized as “intermediate between A. afarensis…and A. africanus or early Homo” [(24), p. 199]. The KNM-ER 5431 teeth could represent the same East African transitional form as the LD 350-1 mandible.

Arguments for the role of 2.5-Ma A. garhi as the direct ancestor of Homo are relevant to interpretations of LD 350-1 (25). The tremendous postcanine dental size and differentially enlarged P3 of A. garhi holotype cranium BOU-VP 12/130 are a poor match for the modest-size teeth of the Ledi jaw (in resampling analyses, the LD 350-1 lower second molar and the BOU-VP 12/130 upper second molar were found to be diminishingly unlikely to sample the same species; see text S3). To include LD 350-1 and BOU-VP 12/130 in the same species lineage and simultaneously postulate an ancestor-descendant relationship between A. garhi and Homo would require a dramatic increase in tooth size between ~2.8 and ~2.5 Ma, and then an equally dramatic decrease in tooth size in a ~2.5- to 2.3-Ma transition to Homo. Although nonrobust mandibles with derived corpus morphology and smaller teeth are approximately contemporaneous with A. garhi in the Middle Awash of Ethiopia (25, 26), it is unclear whether these represent A. garhi or a second lineage, potentially of Homo. For all of these reasons, we consider the hypothesis that posits LD 350-1 as representing a population ancestral to ~2.4- to 2.3-Ma Homo, to the exclusion of A. garhi, to be more parsimonious than ones that include this taxon in the Homo lineage [consistent with the phylogenetic analysis in (27); see text S3].

By ~2.0 Ma, at least two species of the genus Homo were present in Africa, H. habilis and H. rudolfensis (3, 28), but primitive anterior corpus morphology distinguishes LD 350-1 from both of them. These species are distinct from one another in dental arcade shape (12, 29), and LD 350-1 suggests the primitively arched canine/incisor row seen in H. habilis sensu stricto (fig. S9). For the present, pending further discoveries, we assign LD 350-1 to Homo, species indeterminate. The identification of the 2.80- to 2.75-Ma Ledi-Geraru mandible as representing a likely phyletic predecessor to early Pleistocene Homo implies that phylogenetic schemes positing the origin of the Homo lineage from A. sediba as late as 1.98 Ma are likely to be incorrect [contra (30); see text S3].

The time period 2.8 to 2.5 Ma witnessed climatic shifts that are frequently hypothesized to have led to the origin of the Homo lineage (3, 3133). Although the open habitats reconstructed for the Lee Adoyta faunal assemblages provide a new window on these changes (4), too little is known of the pattern of hominin evolution during this period to forge causal links to specific evolutionary events. The Ledi-Geraru specimen confirms that divergence from australopith dental and mandibular anatomy was an early hallmark of the Homo lineage. Additional discoveries are needed to determine whether or not these dentognathic changes were accompanied by neurocranial expansion, technological innovation, or shifts in other anatomical/behavioral systems that are familiar components of the Homo adaptive pattern.

Supplementary Materials

Text S1 to S3

Figs. S1 to S10

Tables S1 to S7

References (3447)


  1. Acknowledgments: We thank the Authority for Research and Conservation of Cultural Heritage, Ethiopian Ministry of Culture and Tourism, for permission to conduct field work at Ledi-Geraru and laboratory research in the National Museum of Ethiopia, Addis Ababa, in which the LD 350-1 mandible is housed. Research funding provided by the National Science Foundation (NSF) (grant no. BCS-1157351), and the Institute of Human Origins at Arizona State University is gratefully acknowledged. B.A.V. also thanks NSF (BCS-0725122 HOMINID grant) and the George Washington University Selective Excellence Program for support during this research project. We thank Z. Alemseged, B. Asfaw, L. Berger, F. Brown, C. Feibel, D. Johanson, F. Spoor, G. Suwa, C. Ward, T. White, and B. Wood for discussion, assistance, and/or permission to examine fossils. Constructive comments by F. Spoor, B. Wood, and three anonymous referees substantially improved the manuscript. Supporting data for this paper are presented in the supplementary materials.
View Abstract

Stay Connected to Science

Navigate This Article