Discoveries in Chad by the Mission Paleoanthropologique Franco-Tchadienne from the Late Miocene Toros-Menalla 266 fossiliferous locality (≈7 million years ago) have substantially added to our understanding of early human evolution in Africa (1–3). Among the material recovered is a nearly complete cranium (TM 266-01-60-1) assigned to a new hominid∥ species Sahelanthropus tchadensis (1). Despite its relative completeness, the TM 266-01-60-1 specimen exhibits partial distortion from fracturing, displacement, and plastic deformation. The hypothesis that TM 266-01-60-1 is an ape (4–6) has been once again refuted by the recent virtual reconstruction (2), which confirms the presence of many hominid features in S. tchadensis and indicates that it was likely some kind of biped. However, questions remain about the species' systematic relationship to extant apes, other known Late Miocene hominids, and later hominids from the Pliocene. Although we do not believe it yet possible to resolve reliably the phylogenetic position of S. tchadensis and other early hominids, a first step is to consider in what ways the TM 266-01-60-1 cranium shares primitive features with extant apes and derived features with various known hominids. Thus, we assess here the morphological similarities and differences of the reconstructed TM 266-01-60-1 cranium with ape and other hominid taxa by using geometric morphometric methods (7, 8) along with comparative quantitative data on variables that reflect key similarities and differences in hominid craniofacial morphology.

Our analysis uses the maximum number of three-dimensional landmarks we could reliably obtain from the reconstruction of TM 266-01-60-1 (2) (see Table 2, which is published as supporting information on the PNAS web site). Landmark coordinates (26 facial, 11 neurocranial, and 15 basicranial) were acquired by using a Microscribe digitizer on a stereolithographic replica of the virtual reconstruction. The landmarks were also digitized on cross-sectional ontogenetic samples of Pan troglodytes (n = 40), Gorilla gorilla (n = 41), and Homo sapiens (n = 24) (see Table 3, which is published as supporting information on the PNAS web site). In addition, we digitized as many of the same landmarks as possible on a sample of available relatively complete fossil hominid crania: the stereolithograhic replica of AL 444-2 (Australopithecus afarensis) (9); CT scans of Sts 5 and Sts 71 (Australopithecus africanus) and KNM-ER 1813 (Homo habilis); and a cast of OH 5 (Paranthropus boisei). Rigorously, it has to be noted that the restorations, reconstructions, and/or deformations associated with some fossil hominid crania (e.g., AL 444-2, OH5, and Sts 71) are potential sources of bias because the changes represent approximations, even if they are accurate, of the proper morphology.

To compare overall cranial shape in TM 266-01-60-1 with that of other taxa, we first performed a principal components analysis of shape by using the maximum shared set of 29 cranial landmarks (see Table 4, which is published as supporting information on the PNAS web site) for TM 266-01-60-1, the above fossil sample, and pooled-sex samples of adult P. troglodytes, G. gorilla, and H. sapiens crania following a generalized least squares (GLS) superimposition using morphologika software (10) (Fig. 1A ). The first principal component (PC1), which accounts for 78% of total variance, quantifies shape changes that primarily describe facial length and projection, the relative anteroposterior position of the foramen magnum and mandibular fossae, and the length and height of the cranial vault. This component discriminates humans, early hominids, and great apes, and it places TM 266-01-60-1 within the shape space occupied by the australopiths. Compared with Pan and Gorilla, TM 266-01-60-1 has a more vertical face with a relatively anteroposteriorly shorter rostrum, and a posterior cranial base** in a more anterior position. The second principal component (PC2), which explains only 5% of the variance, separates TM 266-01-60-1, AL 444-2 (A. afarensis), OH 5 (P. boisei), and Sts 5 (A. africanus) from other crania in the analysis. PC2 quantifies shape changes that primarily describe the midsagittal contour of the neurocranium and the relative height and projection of the upper face.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

S. tchadensis cranial reconstruction and comparative fossil hominid/African ape morphology. (A) PC1 (78% of variance) versus PC2 (5% of variance) of adult P. troglodytes (triangles), G. gorilla (circles), H. sapiens (squares), and various hominids including TM 266-01-60-1. The wire-framed crania (lateral view) indicate pattern and magnitude of shape change associated with PC1. Dotted symbols indicate females. (B) PC1 (74% of variance) vs. centroid size of cross-sectional ontogenetic samples of H. sapiens (squares), P. troglodytes (triangles), G. gorilla (circles), and various hominids including TM 266-01-60-1. Red symbols indicate adults, white symbols indicate subadults, orange symbols indicate juveniles, and green symbols indicate infants. The shaded areas encompass adult P. troglodytes and G. gorilla, with male adult G. gorilla in the area within the dashed line. For G. gorilla, the dashed line is the female reduced major axis regression; arrows indicate the reduced major axis regression line when males are included.

To quantify the allometric differences between TM 266-01-60-1 and other taxa, Fig. 1B graphs the first principal component of shape variation following a GLS superimposition of the same 29 landmark configurations against centroid size in an expanded sample that includes juvenile apes and humans (see Tables 3 and 4). The shape changes described by PC1, which explains 74% of the variance, are essentially the same as PC1 in Fig. 1 A . TM 266-01-60-1 falls within the same range of values on PC1 as the other adult hominids, but well above the ontogenetic allometries (calculated by reduced major axis regression) between size and shape for the African apes. This result suggests that, like other hominids, TM 266-01-60-1 is paedomorphic for PC1, primarily because of its shorter rostrum and its projected upper face relative to overall cranial size.

To explore further the key morphological similarities and differences between TM 266-01-60-1, African apes, and other hominid taxa for which sufficient data are available, we compared in lateral view the TM 266-01-60-1 reconstruction with G. gorilla (consensus adult), P. troglodytes (consensus adult), AL 444-2 (A. afarensis), Sts 5 (A. africanus), and KNM-ER 1813 (H. habilis). Each comparison, shown in Fig. 2, was computed by using a GLS superimposition of the maximum possible set of landmarks. In terms of cranial vault shape, TM 266-01-60-1 resembles P. troglodytes and G. gorilla in having a low, long superior contour of the neurocranium, particularly anterior to the apex of the vault. The nuchal plane in TM 266-01-60-1, however, is relatively long, flat, and rotated toward the horizontal as in A. africanus and A. afarensis (11). Relative facial height (from nasion to prosthion) in TM 266-01-60-1 is similar to that of AL 444-2 and Sts 5 (respectively, A. afarensis and A. africanus) and KNM-ER 1813 (H. habilis) but shorter than that in G. gorilla and P. troglodytes.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

GLS superimposition (left column), lateral view, of TM 266-01-60-1 with female adult consensus of G. gorilla (A), pooled-sex adult consensus P. troglodytes (B), AL 444-2 (A. afarensis)(C), Sts 5 (A. africanus)(D), and KNM-ER 1813 (H. habilis)(F). TM 266-01-60-1 configuration of landmarks is figured in red. Dots indicate location of main cranial landmarks deviations. Each comparison is associated to lateral view (right column) of G. gorilla (female) cranium (A), P. troglodytes cranium (B), AL 444-2 (A. afarensis)(C), Sts 5 (A. africanus)(D), virtual reconstruction of the TM 266-01-60-1 cranium (E), and KNM-ER 1813 (H. habilis) (F). Inferior and superior dashed line, respectively, correspond to alveolare–porion and nasion–porion; the solid line is alveolare–porion minus nasion–porion distance (see Table 1 for data on relative facial projection).

The GLS superimposition in Fig. 2 also suggests that TM 266-01-60-1 exhibits some possibly derived differences from crania of Australopithecus and African apes, notably in the upper face (including the orbital superstructures, glabella and nasion), which projects substantially in front of the neurocranium. Upper facial projection in the TM 266 cranium most likely accounts for its long and flat frontal squama and its elongated supraorbital torus (12, 13). In addition, the Sahelanthropus face has a relatively short premaxilla with no diastema and reduced subnasal prognathism compared not only to Pan and Gorilla but also to some cranial representatives of Australopithecus (e.g., AL 444-2). Subnasal prognathism, partly a function of premaxilla length, varies within Australopithecus (9, 14), but no known crania attributed to A. afarensis or A. africanus show evidence of the association of a shortened rostrum with a substantial projection of the upper face relative to neurocranium seen in TM 266-01-60-1. Aspects of the overall structure of the TM 266-01-60-1 face (i.e., projecting upper face and reduced subnasal prognathism) resemble the configuration seen in KNM-ER 1813 (H. habilis) (Fig. 2F ). However, further research is necessary to determine the structural bases of this apparent similarity. For instance, TM 266-01-60-1 contrasts with KNM-ER 1813 in its frontal squama morphology, which is flat, long and low, possibly because of its small endocranial capacity relative to the length and flexure of the cranial base (15). The TM 266 cranium also has a distinctive midfacial conformation with an anteroinferiorly sloping midfacial contour in the midsagittal plane (see ref. 2, Fig. 2).

To clarify the similarities and differences between the reconstructed S. tchadensis cranium and other hominid crania, Table 1 summarizes some of the features that TM 266-01-60-1 shares with African apes, and those it shares with a selected sample of several Pliocene hominid representatives (all of which are at least 3 million years younger). We currently lack sufficient data to compare the overall cranial morphology of TM 266-01-60-1 with specimens close in age such as Ardipithecus (16–19), and there are no cranial remains attributed to Orrorin (20).

Table 1 indicates that, as might be expected for a Late Miocene hominid (21), TM 266-01-60-1 shares many apparently primitive features with Pan and Gorilla, particularly in the palate andregions of the cranial vault other than the nuchal plane (e.g., Table 1, n, o, and w). TM 266-01-60-1 also has an estimated endocranial volume of 360–370 ml (2), which is the smallest yet documented for an adult hominid but within the chimpanzee range (22).

The TM 266-01-60-1 reconstruction exhibits several widely recognized hominid synapomorphies, including a long, flat nuchal plane that is more horizontally oriented than in African apes (contra ref. 4); a shortened basioccipital; and a more anteriorly positioned foramen magnum (1) (e.g., Table 1, t, x, and y). In addition, as discussed in Brunet et al. (1, 6, 30), the dentition of TM 266-01-60-1 and referred specimens resembles those of Ardipithecus and later hominids compared with Pan in several respects such as reduced canines, a C/P₃ nonhoning complex, and postcanine teeth with thicker enamel. TM 266-01-60-1 also shares several derived cranial features with later hominids, especially in the posterior cranial vault and the cranial base (e.g., Table 1, r, s, t, v, and z). However, as noted above, some aspects of the overall facial structure of TM 266-01-60-1 are unlike those of any known published Australopithecus, including AL 444-2 (A. afarensis). Several of these features, such as subnasal height, palatal protrusion, and relative upper facial projection (Table 1, b, c, and j), may be derived differently than those in Australopithecus. Others are probably either primitive or convergent with Homo. For example, the upper and middle portions of the face in TM 266-01-60-1 are similar in width, as in Pan and Homo, but unlike the hexagonal-shaped configuration in Australopithecus, in which the midface is wider than the upper face because of masticatory-related expansion of the zygomatic region (Table 1, k) (9).

The mosaic of primitive and derived features in TM 266-01-60-1 poses some interesting questions about early hominid systematics. A preliminary study, derived from the three-dimensional cranial shape data (31), suggests that S. tchadensis groups phenetically with later hominids rather than Pan but is not overall more similar to any other, more derived hominid species (Fig. 3). This affinity stems from the many basicranial and neurocranial similarities discussed above and is supported by previous results derived from more classical data (32). However, further research is needed to determine the evolutionary relationship between Sahelanthropus and known Miocene and Pliocene hominids. First, we currently lack sufficient cranial data to include other key taxa such as Ardipthecus (16–19), Orrorin (20), and Australopithecus bahrelghazali (3) in a comprehensive analysis. Second, we do not yet understand the developmental bases of most cranial features that allow us to define independent characters known to be phylogenetically informative (33–36), and it has sometimes proven difficult to derive robust, reliable phylogenetic relationships from morphological data (31, 32). As suggested above, a reasonable possibility is that Sahelanthropus is part of a clade of primitive hominids that is ancestral to all later genera, including Australopithecus and Homo. Other members of this clade potentially include Ardipithecus and Orrorin, but further research is necessary to test the hypothesis. Derived features that support this phylogenetic hypothesis include the many basicranial and neurocranial similarities between TM 266-01-60-1 and Australopithecus (see Table 1, r–t, v, and x–z), along with the additional derived features of the vault, palate, and expanded brain shared exclusively by Australopithecus and Homo (see Table 1, d, n, o, w, and ecv).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

Phenetic analysis of affinities between S. tchadensis (TM 266-01-60-1), living apes, and representative hominids generated from three-dimensional cranial shape data. Raw data consist of Euclidean distances calculated from Procrustes coordinates of each taxon in the overall generalized Procrustes analysis. Mean Procrustes coordinates have been used for G. gorilla (pooled sex), P. troglodytes (pooled sex), H. sapiens (pooled sex), and A. africanus (Sts 5 and Sts 71). Taxa were clustered by using the neighbor-joining method (31).

However, the picture is more complicated and less easy to resolve. Critical answers will probably come from detailed study of the upper Miocene hominids biogeography. Hence, although Ardipithecus and Orrorin are both known from eastern Africa (respectively, Ethiopia and Kenya), Sahelanthropus is known from central Africa (Chad), 2,500 km west of the Rift Valley. This pattern raises intriguing questions about hominid phylogeny with, for instance, potamophile mammals (Anthracotheriid and Hippopotamid) attesting, around 7 million years ago, to faunal exchanges between Libya and Chad but not with eastern Africa (37).

Although Sahelanthropus tchadensis is clearly a hominid, its complex mosaic of features poses some interesting systematic questions about early hominid evolution that can be resolved only with more data of four types. First, more information is needed to compare Orrorin, Ardipithecus, and Sahelanthropus in detail. Second, additional hominid fossil material is needed from Late Miocene deposits in different parts of Africa. Third, we need a better understanding of the developmental and functional bases of cranial morphology that are necessary to hypothesize phylogenetically informative characters. Fourth, we particularly need a better understanding of the vertebrate and hominid biogeographical relationships through Africa during the upper Miocene.