What, if anything, is Australopithecus sediba?

Today we finally get to learn about the exceptional discovery of four partial hominin skeletons from Malapa Cave, South Africa. Two of the fossil skeletons are described by Lee Berger and colleagues in the current issue of Science, descriptions of two more are still forthcoming.

A kind journalist sent me a copy of the research papers a few days ago, so my graduate students and I have had a chance to think about them a little bit and compare them with other material.

Berger and colleagues have named a new species to contain the fossils, Australopithecus sediba. For anybody who follows paleoanthropology, the new species won't be surprising -- if I found a fossil, I'd surely make up a new name for it, even if I thought it was my great-great-grandmother. In this case, the morphological reasons for naming a new species aren't trivial, but I'll begin by approaching them skeptically, especially in comparison with the large samples of South African fossils both earlier and later than Malapa. I'll conclude that a new species within Australopithecus was probably the right call, but not an easy one.

The press is running with a "new fossils provoke debate" storyline -- are they possible ancestors of Homo or not?

The simple answer to that question is that the Malapa skeletons are too late to be ancestors of Homo. After all, we have early Homo nearly a half-million years earlier.

A more complicated answer is that it depends what we mean by Homo. My feeling is that these skeletons don't comport with what most of us mean when we say "Homo". Most of us have in mind an adaptive shift from Australopithecus to Homo that included larger brain size as a significant element, and the MH1 skeleton has a small endocranial volume.

But if we accept that model of Homo, we have to accept its consequences, as the Malapa skeletons now make clear. One important consequence is that, if we assume that MH1 isn't Homo, we can no longer say have any skeletal evidence of Homo from before 1.95 million years ago. Because the Malapa specimens are more like Homo in their dental and mandibular features than are earlier specimens that have usually been called Homo.

And if we throw out all those earlier Homo specimens...well, then suddenly Malapa isn't too old to be an ancestor of Homo after all.

How old are they?

The fossils lay above a flowstone with a U-series and paleomagnetic date consistent with an age just around 2 million years ago. That's a maximum age for the fossils; they must be younger than that.

The hominins are in water-deposited sediments, which are inferred to represent ancient washes of subterranean water flows through the cave system. Two elements above the flowstone contain the hominin specimens, called facies D and E, and both have normal magnetic polarity. The most likely interpretation is that they belong to the Olduvai paleomagnetic subchron, which occurred between 1.95 and 1.78 million years ago. A specimen of the sabertooth cat Megantereon in one of these facies has a last appearance elsewhere in Africa at 1.5 million years ago. So it appears that 1.78 million years is a very likely minimum age for the fossils.

That's about as good as dating gets in South Africa, where we're used to seeing very wide age brackets on hominin-bearing localities. It means that the Malapa hominins lived at around the same time as KNM-ER 1470 in the Turkana basin, or OH 24 at Olduvai Gorge. Until today, I think we could justly claim that the only australopithecines still known to occur in this time interval were the robust species A. boisei and A. robustus -- although the first appearance of A. robustus might (might) be later than Malapa.

Why aren't they A. africanus?

To me, this is the hardest question to answer.

The Sterkfontein Member 4 sample of A. africanus is tremendously variable. The postcrania of both Malapa skeletons are tremendously informative, but fall within the range of variation at Sterkfontein for almost every feature that the authors reported. The few exceptions (such as humeral torsion and femur neck/shaft angle) are right at the edge of the Sterkfontein range.

Malapa skeletons

In other words, it's my impression that the postcrania of the Malapa skeletons fit within A. africanus. The limits of my impression are that there are a whole lot of observations here, and the paper generally does not report metrics for the postcrania. Maybe the sequel will give us some more surprises.

I would have added a comparison with the Swartkrans A. robustus sample, which overlaps nearly totally in body size with Sterkfontein and contains elements that are in some cases more comparable to the Malapa skeletons. In particular, the os coxa of MH1 looks a lot like SK 3155, and the proximal femur looks like SK 82 to me, at least in the tiny picture provided with the paper. On the whole, I don't think that the Malapa hominins are particularly like A. robustus, I just think that if you put together a reasonably large sample of australopithecine postcrania, these two skeletons don't stand out.

I'll take up the discussion of proportions of the different elements below. My feeling is that the proportions aren't exceptional for Australopithecus, either, but we have to temper that against the observation that really only AL 288-1 (Lucy) is comparable, and it's more than a million years older.

What about the teeth? Generally speaking, the teeth of MH1 and MH2 are both at the small end of the A. africanus range. In a couple of cases (the lower canine of MH1, the lower second molar of MH2), the teeth are absolutely smaller than any Sterkfontein individual. The canines are within the range of A. robustus (remember that the robust australopithecines have small anterior teeth), but the premolars are nothing like the large, molarized Swarktrans sample of premolars.

They're a little small but within the range of those known for Homo habilis at Olduvai Gorge. For example, OH 7 -- the type specimen of Homo habilis has molars that are 1.5 mm larger than MH1 in both dimensions.

But then, Homo habilis really doesn't differ much in tooth size from Sterkfontein.

In size, the Malapa teeth are exactly what you would expect for Homo erectus. The first molars are smaller than those of Dmanisi D2700/D2735, for example. But unlike H. erectus dentitions, the molars of the Malapa hominins get bigger toward the back -- M3>M2>M1.

The Malapa mandibles are strikingly gracile. The MH1 mandible has a relatively vertical symphysis with a small cross-section. The long, parallel upper and lower corpus borders really strike me like a mandible of Homo erectus, something like KNM-ER 993 or OH 22 -- but this impression may be exaggerated considering the M3 of MH1 has yet to erupt. Metrically, the corpus breadth and height are most like OH 13. There are small australopithecine specimens that compare to this, such as AL 277-1, and it is worth remembering that MH1 is a juvenile mandible. I can't compare the ramus heights with those of other samples because the authors don't report those measurements.

An interesting question: If these mandibles had been found in isolation, would we call them Australopithecus? The Olduvai H. habilis mandibles OH 7 and OH 13 have M3>M2>M1, while OH 16 has M2>M3>M1. The Malapa mandibles look much more like later Homo than do early Turkana basin mandibles like KNM-ER 1801, KNM-ER 1802, or KNM-ER 1482, all of which are much more robust and have larger, more molar-shaped premolars than MH1, and all of which have M3>M2>M1 except KNM-ER 1802 which lacks M3. This is a quick comparison on my part, but I think the Malapa mandibles look more like Homo than does the existing hypodigm of Homo habilis. It's hard to imagine that the mandibles in isolation would have been referred to Australopithecus. More on that below.

Compared to the mandibles, the cranium of MH1 looks more like its counterparts from Sterkfontein. To be sure, it is an 11-13-year-old juvenile and more gracile in some respects than any of the Sterkfontein crania. But take a look at it next to Sts 71:

MH1 next to Sts 71, frontal view

MH1 (left) next to Sts 71 (right)

They're not identical, naturally. Sts 71 has higher temporal lines, a slightly smaller vault, and more prominent cheeks. It also has more postorbital constriction compared to MH1, though that isn't obvious from this angle. MH1 has a true superorbital torus, Sts 71 has at best a shade of one. But you can see the similarities -- the angle of the zygomatic process of the maxilla, the narrow and concave interorbital region, the tall and narrow orbits. MH1 has no prominent anterior pillars (bony swellings on either side of the nasal aperture), but Sts 71 is not very different in this region. Sts 71 has bigger teeth.

Consider also Sts 52:

MH1 next to Sts 71, frontal view

MH1 (left) next to Sts 52 (right)

Again, Sts 52 has anterior pillars and bigger teeth, but the shape of the face is very comparable between these two. The nasal bones in particular are similar in this pair, almost "pinched" at the midline, with a lateral expansion both superiorly and inferiorly.

We can do a similar exercise for most of the features of the MH1 cranium. What is exceptional, in the context of the Sterkfontein sample, is the overall gracility of the masticatory apparatus.

One important thing that is not in the least bit exceptional: Its brain. An endocranial volume estimate of 420 ml (from CT reconstruction) puts MH1 at the bottom of the range of variation at Sterkfontein -- the best-known skull from Sterkfontein, Sts 5, has a volume of 485 ml, while STW 505 has a volume larger than 550 ml. Before MH1, the smallest of the South African crania were estimated to have volumes of 428 ml. This one seems to be smaller mainly by being flatter -- a shape that it shares with early Homo, but I wouldn't say it was without parallel in Australopithecus.

But the smallest endocranial volume known for early Homo is KNM-ER 1813, at 510 ml. That specimen is extreme: the next smallest is 585.

The vault fits in A. africanus, most of the facial features have comparable specimens in the Sterkfontein sample, with some exceptions, and the postcranial skeleton is unexceptional. The teeth are mostly within the range at Sterkfontein with some exceptions. But the mandible -- like those few facial characters -- stands out.

Australopithecus sediba -- a new species within Australopithecus -- then seems like a fair diagnosis. The craniodental derived features are of the sort that would usually justify a new species. Heck, in the case of Kenyanthropus, even more minor differences in the face and size of teeth from contemporary A. afarensis caused Leakey and colleagues (2001) to name a new genus.

Is MH1 really a male?

Berger and colleagues (2010) infer that the MH1 skeleton (the one with the skull) is a male. It is large and more robust than the MH2 skeleton: Its teeth are bigger than the MH 2 skeleton, its mandible is more robust with a taller ramus, the articular ends of its limb bones are a bit larger. In addition, the greater sciatic notch on its preserved os coxa is narrower than other australopithecines like Lucy and Sts 14, and the pelvic inlet may (based on the anterior position of the auricular surface) have been smaller.

But the skeleton isn't really very big. Its endocranial volume is small, its long bones are not nearly so robust as some australopithecines. There are large male australopithecine skeletons -- STW 431, for example -- and MH1 doesn't seem so large as these. Again, it's hard to tell without postcranial measurements, but the sex of this specimen isn't a clear call either way.

The sex of the specimen is important to the way we interpret it, because the features that make it stand out from A. africanus concern masticatory gracility. If it's a female, it doesn't seem quite so different from A. africanus as if it's a male.

Are they Homo?

Let's start with the brain size, which at 420 ml seems to be the most obvious thing separating MH1 from our genus. Well, except for Liang Bua 1 -- with its endocranial volume of, um, 420 ml. Is brain size fundamental to Homo? Maybe. Maybe not.

Alan Boyle's report on the fossils ("Fossils shake up our family tree") has an excellent letter from Don Johanson, who makes the argument that the Malapa fossils should be assigned to Homo. Of course, Johanson and Bill Kimbel in 1996 described a 2.33-million-year-old fossil from Hadar as the earliest clear maxilla of Homo. That maxilla, AL 666-1, resembles Homo in having a more vertical subnasal profile, a parabolic dental arcade, molars that are long relative to their breadth, and a "squared-off" jaw that is relatively straight across the anterior dentition. In other words, basically the dental features seen in the MH1 maxilla.

We've got two choices. Maybe these are genuine shared derived features with these specimens and Homo -- in which case, we should probably name them Homo, as Kimbel and colleagues did for AL 666-1.

Or, there were several australopithecines after 2.5 million years ago with these dental and maxillary (and for the Malapa hominins, we can add mandibular) characters. In which case, they're not signs of Homo at all. They may reflect parallel dental reduction in several australopithecine lineages, all of which faced niche differentiation from the emerging robust australopithecines. One of those lineages may have given rise to Homo, but we don't know which. Maybe it was South African, but it need not have been. It could even have been Asian.

The question is just how important we think brain evolution was to the origin of our genus. If the brain was the key adaptation, then Malapa shows that the dental features are irrelevant to the brain -- because these skeletons have more dental reduction than most of the East African Homo habilis sample, but MH1 has a much smaller brain.

What about tool manufacture?

Part of the logic of pre-2-million-year-old Homo is the emergence of stone tool manufacture 2.6 million years ago. It stands to reason that this major shift in behavior and diet might have given rise to a new adaptive plateau for early hominins, and that would have been tied to the evolution of larger brains. The problem is that we don't have larger brains in any fossil remains until after 2 million years ago -- KNM-ER 1470 remains the earliest hominin with a brain larger than 600 ml. Up to now, people have conjectured that large-brained hominins may have existed earlier, even to the point of arguing about the brain size reflected by the otherwise-robust temporal bone from Chemeron. But it's worth pointing out that none of these pretenders to the Homo throne have smaller teeth than A. africanus. The diet shift that should have been made possible by a meat-eating stone tool economy didn't lead to smaller teeth until much later.

And now we know that at least one small-toothed hominin also was a small-brained one.

We don't know whether the Malapa hominins would have been toolmakers. The fact that they weren't found with tools isn't really evidence either way. Dirks and colleagues (2010) suggest that the skeletons were deposited by water washing them from an initial death trap into a secondary location. If true, it would be a miracle beyond belief for stone artifacts to have made the trip with them.

We do know that stone tools are present in Sterkfontein Member 5 and Swartkrans Member 1, and cutmarked fauna are in the latter. Both these may be roughly contemporaneous with the Malapa hominins, depending on their date. So toolmaking hominins were in the immediate area, around the time that the Malapa hominins lived.

SK 847 is from Member 1 of Swartkrans, and could be as old as the Malapa skeletons. Its endocranial volume isn't known, but facially it looks even more like Homo erectus than does MH1. It seems plausible that this skull represents the local toolmaking population, but even so, this skull does resemble MH1 in several respects, and again we don't know its volume. STW 53, probably a bit older than Sterkfontein Member 5, has also often been referred to Homo but it definitely doesn't have a substantially larger endocranial volume than MH1.

So again, we seem to be faced with two choices: Broaden the definition of Homo to include this very australopithecine-like sample, or restrict it to later large-brained hominins. In either case, brain size and tool manufacture do not necessarily go together.

What's the single most obvious thing that the paper doesn't describe?

Which brings me to the fingertip. MH2 has a distal phalanx. The paper doesn't describe it, even though this bone element has taken on such importance in the evolution of Homo compared to Australopithecus. Big fingertips are supposed to be adaptations to force transfer through the fingertip grip used in tool manufacture.

The picture of the thing is so tiny -- I mean, literally we're talking about two pixels of finger -- that I can't make anything out of it. Does it have a large apical tuft, like OH 7? Or is it like the Hadar distal phalanges, with narrow, apelike apical tufts?

If one was wondering about whether the thing was Homo or not, I would think this is one of the first things you would check....

What about those limb proportions?

For fifteen years, a bunch of otherwise sensible paleoanthropologists have been engaged in a debate about the limb proportions of A. africanus and H. habilis. The reason why this particular question may not be sensible is because the debate is about the length of the arm relative to the leg, but there's no specimen of A. africanus that preserves both the length of the arm and the length of the leg.

What there are: OH 62, a skeleton apparently of H. habilis that has a complete humerus and more than half the length of one femur, STW 431, which has an acetabulum and mostly complete humerus, and Sts 14, which has a partial femur, an acetabulum, and a piece of distal radius. On the basis of these fossils, we've seen some intense debate about the reconstruction of the OH 62 femur length, and a lot of discussion about whether the sizes of articular surfaces are relevant to the function of the limbs. Indirectly, it has appeared that A. africanus and H. habilis shared longer arms than were present in AL 288-1 (Lucy).

Well, now we have two fossil skeletons with both hindlimb and forelimb elements. The paper doesn't address the issue directly, nor does it provide raw measuremnets that would lead to a quick answer. But the humerus is short relative to the size of the femur head, compared to earlier hominins, while a bit long relative to Homo by the same comparison. So it looks like the Malapa skeletons may be somewhere in between.

The authors do argue that OH 62 is an odd skeleton in one respect: They consider the "diaphysial strength" of the humerus and femur. This is a cross-sectional measure of the area of cortical bone, and reflects the robusticity of both forelimb and hindlimb elements. In their estimation, OH 62 has a much stronger arm relative to its leg strength than the Malapa skeletons.

It's not obvious how to interpret this observation. Is OH 62 more apelike in its locomotor pattern than Malapa? Or does the strength ratio vary allometrically with body size, and OH 62 is just at the smallest end of the comparison? Hard to tell without the length measurements.

OK, what's the bottom line?

Here's the important thing. From today forward, there are a bevy of features of the face, teeth and jaw that are no longer "derived characters" of Homo.

Some people will want to fix this by broadening the definition to Homo to include the Malapa skeletons. Others will want to narrow the definition of Homo to include only large-brained specimens.

Every specimen attributed to Homo before 2 million years ago is now up for grabs. Maybe they're Homo, or maybe their resemblances to Homo are just masticatory parallelism. We already know that parallelism explains many of the derived locomotor and masticatory resemblances of great apes, and many strongly suspect that parallelism explains the derived masticatory resemblances of robust australopithecines. If the dental reduction that once was a marker of Homo joins this list, it will hardly be surprising.

If we follow the logic that connects tool use to dental reduction, however slowly and indirectly, then I think we have to conclude that A. sediba was likely a toolmaker and meat-eater. This hypothesis is testable, both by bone chemistry and dental morphology and wear.

Malapa suggests the hypothesis that brain evolution followed the appearance of stone tool manufacture by a considerable delay. If so, I wonder what exactly caused the brain to expand. Did the diet shift to higher-quality foods unfold over a long time, allowing brains to expand only after 3/4 million year delay? Or was brain evolution caused mostly by non-dietary factors, such as social dynamics or climate instability?

Or did the evolution of our genus happen somewhere else, far from the places where we currently have fossil samples? The Rift Valley and South African cave systems may have been wonderful for preserving fossils, but who's to say they weren't relative backwaters when it came to the evolution of Homo?

Well, I'm running out of gas for this installment. More later....

References:

Berger LR, de Ruiter DJ, Churchill SE, Schmid P, Carlson KJ, Dirks PHGM, Kibii JM. 2010. Australopithecus sediba: A New Species of Homo-Like Australopith from South Africa. Science 328:195. doi:10.1126/science.1184944

Dirks PHGM, Kibii JM, Kuhn BF, Steininger C, Churchill SE, Kramers JD, Pickering R, Farber DL, Mriaux A-S, Herries AIR, King GCP, Berger LR. 2010. Geological Setting and Age of Australopithecus sediba from Southern Africa. Science 328:205. doi:10.1126/science.1184950