Thoughts on the Sahelanthropus reconstruction

I am at the AAPA meeting in Milwaukee this week, and so posting is by necessity very light. However, the news of the new Sahelanthropus remains and CT reconstruction have come out this week. I have been thinking about them since I got a hold of the proofs last week, so I can post some comments about them. There are some thoughts I'm holding on to for now, however, since I have a manuscript that covers some of them. It's bad enough to be scooped by other people; I surely don't want to scoop myself!

BBC News story, with artist rendition from Nature cover.

The lead story seems to be the reconstruction, probably because it was intended to sort out many of the problems with the distortion in the original fossil. To some extent it succeeds in simplifying the interpretation. For example, the reconstruction clearly places the foramen magnum in a more anterior position than the original. It is not clear to me how the anatomy of the original could conform to the reconstructed base, but doubtless working with a CT is better than working with photographs.

Actually, the article does not place a great emphasis on the anterioposterior position of the foramen magnum. This is sensible, because chimpanzees and australopithecines overlap considerably in this position compared to other basicranial landmarks like the bicarotid line. TM 266 is within the region of overlap, both in the original distorted version and in the reconstructed version.

Instead, Zollikofer and colleagues make two complementary arguments for why the skull is hominid. The first concerns the angulation of the foramen magnum (characterized by the basion-opisthion line) compared to a line tangent to the upper and lower orbital margins.

Despite substantial differences in neck orientation, humans and non-human primates tend to locomote with their orbital planes (the line joining the superior and inferior margins of the orbits) approximately perpendicular to the ground. In addition, primates orient the upper cervical vertebrae approximately perpendicular to the plane of the foramen magnum, and with only a limited range (about 10 degrees) of flexion and extension possible at the cranio-cervical joint. The combined effect of these angular constraints is that the angle between the foramen magnum and the orbital plane is nearly perpendicular in Homo sapiens (103.2 +- 6.9 degrees, n = 23) but more acutely angled in Pan troglodytes (63.7 +- 6.2 degrees, n = 20), and other species with more pronograde postures. The foramen magnum angle relative to the orbital plane in the TM 266 reconstruction is 95 degrees, similar to that in humans and later bipedal hominids such as Australopithecus afarensis (AL 444-2) and A. africanus (Sts 5). TM 266-01-060-1 as a quadruped would requier an unusually extended angle of the neck relative to the plane of the foramen magnum (Zollikofer et al. 2005:757).

A weakness in this argument is that this angle is exquisitely sensitive to the reconstruction. That is, a small difference in the vertical position of either basion or opisthion (the front and rear points on the foramen magnum border, respectively) will have a large effect on the angle of the line passing through these points. But assuming the reconstruction is correct, it is fairly compelling evidence that the habitual posture of the head in Sahelanthropus was not like chimpanzees or gorillas.

The second argument concerns the downward lip of the nuchal crest, which they argue indicates the directionality of the nuchal muscles. It is true that some other hominids have a downward lip on this crest, but I would like to go through a large ape sample to see the range of variation in this trait. In any event, this feature cannot be isolated from the exceedingly unique nuchal morphology in this specimen; the orientation and function of the nuchal musculature cannot be assumed to be like that of other apes whether it had vertical posture or not.

So was it a biped? From the reconstruction alone it may not be possible to confirm or deny the hypothesis. A more vertical habitual posture might or might not imply facultative bipedality. One possibility that would not imply bipedality is that Sahelanthropus had long arms, on the scale of Dryopithecus or longer. In this case, a quadrupedal stance would involve a more vertical trunk position. The distinction between this adaptation and that of gibbons or dryopithecines would be the larger body size and consequent greater degree of terrestriality. This hypothesis might also explain other Miocene hominoids that have been suggested to be like bipeds in certain characters, including Ouranopithecus and Oreopithecus. A test of the relationship of trunk position, limb length, and cranial base morphology might be informative.

Setting aside the question of whether it was a biped, was it a hominid? These are different questions if we assume that the advent of hominid bipedalism followed after some significant time the divergence of hominids from chimpanzees. Aside from Sahelanthropus, the earliest comparably complete hominid cranial remains are less than half its age. The closest is the as-yet-undescribed StW 573 skull. Then is KNM-WT 40000, followed by the cranial remains from Hadar, including the AL 444-2 specimen. A. afarensis and later A. africanus both have extensive adaptations to masticatory force. The extensive nuchal plane of TM 266 is long, narrow, and flat, and it is unlike any early hominid. The browridges are larger (especially in proportion to its relatively small overall cranial size) than in any australopithecine. Thus, it is a challenge to explain exactly what this skull represents in adaptive terms. I think an explanation of its anatomy is in order before it is accepted as being phylogenetically close to the australopithecines.

The paper by Brunet et al. (2005) presents new mandibular and dental remains of Sahelanthropus, including a lower canine with apical wear.

The new material presented here is important for several reasons. . . . The S. tchadensis hypodigm now includes a minimum of six individuals (a maximum of nine) from three sites in a small area of the Anthracotheriid Unit. Second, these new fossils now present a more complete and reliable understanding of this earliest known hominid taxon. S. tchadensis shares major derived features with other recognized hominids that are consistent with its position in the hominid clade, close to the last common ancestor of chimpanzees and humans. In the dentition these anatomical characters are a non-honing C/P3 complex; no diastema between C and P3; a vertical symphysis with weak transverse tori; canines with a small crown and long root; a lower canine crown with a large distal tubercle, both shoulders being very low; an upper P3 with a steeply sloping buccal surface; postcanine teeth with maximum radial enamel thickness intermediate between chimpanzees and australopithecines; and bulbous, slightly crenulated postcanine occlusal morphology. All the hominid mandubular premolar specimens from Toros-Menalla have the same root pattern, with two roots and three separate pulp canales in each premolar (one mesial and two distal) retaining the presumed primitive condition for the Pan/Homo clade (Brunet et al. 2005: 754).

This is a bit of a confused list, since very few of these characters are actually both derived and shared with later hominids. For example, a character that retains "the presumed primitive condition for the Pan/Homo clade" clearly is not a "major derived feature" shared with "other recognized hominids."

The most persuasive similarity with hominids is the reduced canine. But to my eyes, the Sahelanthropus lower canine is distinct from later hominids, especially considering the prominent ridge, or shoulder, around the base of the crown. This feature is found among dryopithecines, and it may simply be a primitive feature retained in an otherwise reduced canine. So the idea that this is intermediate between a larger, ape-like canine and the canines of later hominids is possible, but not demonstrated.

So in my view, the hypothesis that Sahelanthropus is in fact an early hominid has not been strongly substantiated. In many of its features it is basically plesiomorphic, and shares the morphology of a number of Miocene apes. In a few features, it shares a derived (or partially derived) morphology with australopithecines. It also has cranial features such as its long flat nuchal torus and hulking browridge that are derived, not shared with later hominids, and would therefore tend to indicate a separate evolution for this taxon. In my opinion, we probably have entered a time period early enough that the relationships of early hominids, early chimpanzees, gorillas and their ancestors may not be readily resolved with morphological comparisons.


Brunet M et al. 2005. New material of the earliest hominid from the Upper Miocene of Chad. Nature 434:752-755.

Zollikofer CPE et al. 2005. Virtual cranial reconstruction of Sahelanthropus tchadensis. Nature 434:755-759.