Flores update, October 2005

This week's Nature is carrying a paper by Morwood, Brown, and colleagues (2005) presenting additional skeletal material from Liang Bua as well as a commentary by Daniel Lieberman. Thanks to a reader, I found the permission slip from Nature lifting the embargo, so I can let fly without bogarting the kind journalist who forwarded me the paper.

What is noteworthy about the new bones?

The paper discusses three important specimens. The first is the adult mandible LB6/1. In its overall size and morphology it is similar to the mandible of LB1, reported last year. Like LB1 it lacks a chin and Morwood et al. (2005:1013) compare its symphyseal morphology to Dmanisi D211. Overall, the mandible is slightly smaller in tooth size and corpus size compared to LB1, and its ramus is quite a bit shorter.

LB6/1 is part of a partial skeleton. The other elements are not described in the paper, but they are listed: a portion of proximal ulna, a partial right scapula, a foot bone, one each of finger and toe bones, and a complete radius 157mm long. That's a short radius -- barely more than 6 inches. It was broken during life and healed.

Wait a minute. Did you say a 6-inch long radius?

Funny how nobody else seems to have picked up on this yet.

The authors estimate a brachial index (radius to humerus) for LB1 of 78 percent, estimating likely radius length from the ulna. This would put LB1 within the range of "tropical" human populations. If the LB6 individual had the same "tropical" brachial index, its humerus would be around 200mm long. That's 43mm shorter than LB1.

This admits a couple of explanations:

  1. LB6 was simply a smaller individual than LB1. The mandible is more or less consistent with this hypothesis, which may therefore be the most likely.
  2. LB6 did not have the unusually long arms of LB1. Where LB1 is australopithecine-like, perhaps LB6 was more humanlike. This seems less likely, but it would be consistent with the idea that the proportions of LB1 represent some kind of pathology.

Wait a minute. Did you say australopithecine-like proportions?

Yes, the LB1 humerus and ulna are relatively long compared to the femur:

For example, the humerofemoral index of 85.4 is outside the range of variation for H. sapiens, but is the same as AL 288-1 A. afarensis, and midway between the indices for apes and humans. The more complete left ilium [pelvic bone] also indicates that the pelvis is flared antero-laterally, consistent with an australopithecine-shaped thoracic region. Body proportions of LB1 are the same as AL 288-1 A. afarensis, but differ from all other hominins for which they are reliable data, including H. erectus (Morwood et al. 2005:1016).

"Outside the range of H. sapiens" is also outside the range of any Pleistocene human, by the way.

Didn't you say this was an australopithecine when it came out a year ago?

Well, yes. My first post on the subject was titled, "Liang Bua: an australopithecine from Flores?" And I did present a rationale for believing that the skeleton was australopithecine rather than Homo. My point initially was that the combination of small body size and relatively small brain size was very simple to imagine as a descendant of an australopithecine, but very difficult to imagine in a descendant of Pleistocene Homo.

Some other features of the skeleton resemble Australopithecus. None of them individually is sufficient to label the skeleton as australopithecine, but together they are suggestive. For example, the pelvis is broad, with a very prominent anterior superior iliac spine. It's very similar to australopithecines like AL 288-1 (Lucy) or Sts 14.

But we don't know to what extent the breadth and morphology of australopithecine pelves are consequences of their phylogeny as opposed to allometric consequences of their small body sizes. In other words, LB1 might look australopithecine-like because it is small, instead of actually being an australopithecine.

The postcanine teeth are relatively large for a human, but they are far from australopithecine-like in size. Aside from their size, the first molars are the largest; just the opposite of the australopithecine condition. The roots of the premolars are completely uninformative, since both australopithecines and early Homo have the bifurcated roots found in both Liang Bua mandibles. So if this was an australopithecine-derived population, it had evolved considerably smaller teeth. Happily, this evolution of smaller teeth might also account for the gracile, Homo-like facial morphology.

OK, so it's an australopithecine, right?

Maybe. It would not only have to be an australopithecine; it might have to be a DWARF AUSTRALOPITHECINE.

Consider that the femur length of LB1 is just a millimeter shorter than Lucy and its body proportions are basically the same. Lucy (AL 288-1) is not only the most complete known australopithecine skeleton (barring STW 573, which is yet to be described), it has the smallest limbs. There are some individual bone fragments with smaller dimensions than Lucy's, but not very much smaller. At the same time, there are many larger specimens. Some of these, like the Sibilot radius KNM-ER 20419, are a whole lot larger.

Now at Liang Bua, LB1 is nearly the biggest specimen. Brown et al. (2004) do report another radius from an older part of the deposit with an estimated length of 210mm. Again assuming the same brachial index, this would correspond to a humerus of 269mm, around an inch longer than LB1.

But the other two adult long bones reported are the LB6 radius (157mm) and the LB8 tibia. At an estimated 216mm, this tibia is substantially shorter than the 235mm LB1 tibia. There is no comparably complete australopithecine femur, but if Lucy (missing around a third of the shaft) was around the same length as LB1, then LB8 would be shorter than any australopithecine.

Even worse, it is shorter than all but one of 47 chimpanzee tibiae in my comparative data. That's really short.

So as it stands, it appears that the Liang Bua sample is substantially shorter than australopithecines. At the same time, remember that the brain size of LB1 (estimated by Falk et al. 2005 as 417 ml) is smaller than all but three australopithecines (KNM-WT 17000, AL 162-28, and AL 333-105). Together with the facial and tooth reduction, this is good evidence for selection for smaller size in an australopithecine-like population.

OK, so that rules out any chance that it is a dwarf modern human population, right?

Probably. This would have to be an exceptionally short sample of an exceptionally short population. But then it partly depends on how accurate the stature regressions are. As you get to the bottom of the size range of skeletons from which a regression is calculated, you get less accurate body size estimates. And the proportions affect the estimates. These deserve to be gone over carefully.

And there is no reason in principle why a modern human population could not have been smaller than any pygmy populations of today.

There are two stumbling blocks to the hypothesis that this sample represents a dwarf modern human population. The first is the fact that neither mandible looks modern. It is very hard to argue that the features of these mandibles are the consequence of small body size alone; they genuinely appear archaic.

The second is the size of the brain.

Speaking of the brain, is it small enough to rule out descent from early Homo?

That's a very good question. Large-bodied early Homo appears to have an average endocranial volume around 800 ml. A reduction in body size to LB1 should not have cut the endocranial volume in half -- we would expect a volume closer to 600 ml. In fact, an average of around 600 ml is just about what we observe for small-bodied early Homo, including H. habilis in Africa, and possibly including the Dmanisi sample of early Homo from the Republic of Georgia.

To get from a habiline-sized hominid to LB1 body size would take relatively little size reduction. This means that the brain size of LB1 would be very surprisingly small for a habiline of its body size (particularly since many habilines are its size, yet have much larger brains).

So to go from any variety of early Homo to LB1 in brain size would require pretty substantial selection for smaller brains. It is hard for me to see that happening in a hominid population, because it would likely lead to functional compromise of some kind. In particular, if most of the selection for larger brains in hominids has been to promote social intelligence, it is hard to see how selection for smaller brains would happen.

On the other hand, who knows? There is, after all, the endocast shape, which is Homo-like (Falk et al. 2005). And the facial morphology. And the teeth.

In the face of all this, Morwood et al. (2005) appear to be persuaded most strongly by the limb proportions. Maybe an australopithecine-like limb ratio is a good phylogenetic indicator, but considering the recent spat over early hominid limb proportions in Current Anthropology, this might not be the best hook to hang your hat on.

Didn't you say six months ago that LB1 is pathological? Well, what do you say now, smarty-pants?

I still think it's pathological. We have so far seen a few of the details that point to that conclusion. For example, there's the low torsion of the humerus. "Torsion" refers to the angle between the axis of the head and the axis of the distal end of the bone. The humerus of LB1 is unlike any hominid. It's unlike any great ape. It's like monkeys and gibbons. That's weird. Then there is the bowed tibia, and the rotated premolars, all clearly in the reports so far.

We will see if these and other features can be combined into a single diagnosis of pathology, one that possibly includes the small size of the brain or details of its endocast morphology. For myself, I think there is sufficient evidence to question whether LB1 is characteristic of its population.

Now we have additional evidence of body size in the population, including the two very short elements described above. Is that enough to believe that the brain is characteristic of the population? If there has ever been a case to invoke the "extraordinary claims require extraordinary evidence" clause, it is this one. LB1 is not only small-brained for a human, it is small-brained for any hominid.

So I don't think that pathology is a magic wand that is going to make this population into ordinary modern humans. But I'm not ready to jump to the conclusion that this was a population of hominids with australopithecine-sized brains. There is, after all, the problem of how they got to that island in the first place.

Speaking of getting to the island, what about their technology?

I think the tools are a complete red herring. There is every reason to think that modern humans were on Flores throughout the Liang Bua sequence. After all, modern people were on Australia by 50,000 years ago, and out to New Britain by 35,000. Maybe they bypassed Flores on the way, but it seems more likely that it would have been occupied long before these more far-flung locations.

Therefore, it is simplest to assume that modern humans made the tools and hunted the stegodon. Maybe they hunted the hobbits. Maybe some of the bones at the site are modern humans. Maybe some of them were dwarf modern humans.

Seems all tangled up, doesn't it? Yet the behavior speaks to the presence of Homo, and from the character of the tools, modern human seems likely. If those modern humans weren't the hobbits, then they lived alongside the hobbits.

So what's next?

We should see before long at least two (and possibly more) papers that dispute the Homo floresiensis interpretation. Carl Zimmer reports at his weblog that one of these will come from Robert Martin:

"Regardless of one's stand on this issue," Dr. Martin wrote to me in an email, "it is about time that the message got out that there are serious grounds for doubt about current interpretation of the Flores remains."

I think that's right. Of course Martin's interest is allometry of the brain, which made the initial interpretation -- island dwarfing of an early Homo variant -- seem very unlikely from the start. An australopithecine origin is less problematic from the point of view of allometry, but introduces the problems of biogeography -- how did they get there, and why weren't they anywhere else? Pathology would help a lot to explain that brain....

The pathology work will be the most interesting; I know there are many people working very hard to find a single pathology explanation that is consistent with the anatomy of LB1. If they have succeeded (and we should find out before long) it will be a major accomplishment.

And there will be comparative anatomical and possibly genetic work on pygmy people in the region. Remember the Rampasasa Pygmy Somatology Expedition? It will be coming to a journal near you.

The stealth factor is whether Max Planck (or anybody else) has gotten any DNA out of the bones. Wouldn't it be interesting if part of the mtDNA sequence looked like one of the more ancient human-specific nuclear genome mtDNA inserts (numts)? If this is an australopithecine population, mtDNA would be enough to show it.

More information here

All the Flores files

Original discovery

Endocast study

The damage to the specimens

Myth of the ebu gogo


Brown P, Sutikna T, Morwood MJ, Soejono RP, Jatmiko, Saptomo EW, Due RA. 2004. A new small-bodied hominin from the Late Pleistocene of Flores, Indonesia. Nature 431:1055-1061.

Falk D, Hildebolt C, Smith K, Morwood MJ, Sutikna T, Brown P, Jatmiko, Saptomo EW, Brunsden B, Prior F. 2005. The brain of LB1, Homo floresiensis. Science 308:242-245. Full text (subscription)

Morwood MJ, Brown P, Jatmiko, Sutikna T, Saptomo EW, Westaway KE, Due RA, Roberts RG, Maeda T, Wasisto S, Djubiantono T. 2005. Further evidence for small-bodied hominins from the Late Pleistocene of Flores, Indonesia. Nature 437:1012-1017.