The shrinking youth

Yesterday the Journal of Human Evolution released a new paper by Rhonda Graves and colleagues, titled Just how strapping was KNMWT 15000? Graves:2010. The paper challenges almost 25-year-old estimates for the body size of this important 1.5 million year old skeleton.

For all this time, the textbooks have reported that early Homo in Africa had the same tall and elongated physique as current East African people like the Maasai. The new paper says that the textbooks are wrong -- the skeleton doesn't represent an individual who would have grown to be 6'1" (185 cm), instead it was near the end of its growth trajectory, for an adult height of around 5'4" (163 cm).

That's a pretty massive change, and when the authors presented this work at the AAPA meetings last spring, it wasn't without controversy. So naturally we should look closely at the paper, understand its conclusions, and assess what this new estimate means for our understanding of early Homo. As you might guess from reading some of my earlier posts, I've been thinking that the body sizes of the rest of the Pleistocene record add up to a fairly simple picture. One of the few outliers from this picture was KNM-WT 15000. I'm inclined to think that the new estimate fits the bigger picture -- for example, I wrote this spring about "Shrinking erectus". Which means, of course, that I should be even more skeptical.

KNM-WT 15000 was a juvenile at the time of death, and so any estimate of body size involves some assessment of the skeleton's state of development. This has presented a problem for assessing how much the individual had still to grow at the time of its death. The eruption and development of the teeth appear to be consistent with a fairly young age at death, by most estimates younger than 11 years, and by some as young as eight. That's using a human frame of reference. If we turn to a frame of reference drawn from chimpanzees or other apes, the estimated age at death from tooth development is even a bit younger. In contrast, the state of bone development seems to indicate a somewhat higher age at death: older than 11, and by some estimates as old as 15 years.

KNM-WT 15000 skeleton

Graves and colleagues, looking at this apparent mismatch between dental and skeletal development in this specimen, suggests that we need to look at a broader range of possible developmental models for early Homo erectus. A modern human developmental model is not a good fit, and neither is an ape developmental model. So their study involves creating a range of possible developmental trajectories for early Homo. These trajectories are based on data from living apes and humans, but altered by accelerating some phases or changing the intensity of the adolescent growth spurt.

The growth spurt is very important to this issue, because it's one way that humans and most other primates differ greatly. Growth during that phase of development contributes disproportionately to the tall stature of modern humans. If Homo erectus didn't have the same kind of growth spurt as we do, then the stature of this specimen would have been a lot shorter than we would estimate for a human of the same age.

The section of Graves and colleagues' discussion that covers the adolescent growth spurt is, to my mind, the central issue in the paper. Their review begins with a survey of literature on why a growth spurt exists. Most assume that there is some kind of trade-off between early weaning in humans, brain growth, and a large adult body sizewith the optimal solution being slow juvenile somatic growth, fast juvenile brain growth, and they catch up of somatic growth during adolescence. Graves and colleagues assert that this pattern was not present in early Homo erectus, and that a more chimpanzee like growth spurt may be a better model.

The velocity growth curves for human stature and chimpanzee total body length (summed length of crown-rump, femur, and tibia) highlight the difference between modern human and chimpanzee growth and development (Fig.1). Both species exhibit growth spurts, but these spurts differ in rate, timing, and duration (Leigh, 1996). Pre-pubertal growth spurts in mass have been documented in many primates ([Tanner, 1962], [Laird, 1967], [Timiras and Valcana, 1972], [Leigh, 1996], [Leigh and Shea, 1996] and [Hamada etal., 1996]), but to date only slight increases in crown-rump length and total body length have been observed in chimpanzees (Hamada and Udono, 2002). Male chimpanzees (and possibly macaques) undergo a small growth spurt in length during the period between emergence of the first and third molars ([Watts and Gavan, 1982] and [Tanner etal., 1990]), but peak velocity is not as high and the growth spurt not as extended as in modern human adolescence. The velocity of chimpanzee growth decreases slightly between the ages of four and eight, and then begins to decline rapidly until adult total body length is reached at between 12 and 13 years of age. Chimpanzee growth spurts therefore differ in their onset, offset, and intensity compared to the modern human adolescent growth spurt (see Fig.1; [Bogin, 1993] and [Bogin, 1996]). The growth spurts in the ALH 12.3/25% and ALH 12.3/50% curves approximate the juvenile pre-pubertal growth spurt exhibited by chimpanzees, which is of shorter duration and lesser magnitude than the full-blown modern human adolescent growth spurt. We contend that these curves most closely match what is currently known about growth and development in H. erectus but acknowledge that the data currently available limit our ability to choose a single curve. It is also possible that future studies documenting growth in wild chimpanzee length may provide evidence to support a different set of growth curves.

Their small stature estimate for KNM-WT 15000 doesn't entirely hang on this point, but this assumption about the growth spurt makes more difference than any other single factor.

We can reasonably ask: is there any other support for this assumption?

The apparent mismatch between dental and skeletal developmental patterns in the specimen is consistent with the lack of a humanlike growth spurt. But evidence from the skeleton itself is weakened by the fact that KNM-WT 15000 appears to have suffered from some kind of growth pathology, as argued by Latimer and Ohman Latimer:Ohman:2001. The pathology argument has mostly come into play over the issue of vertebral canal size in the specimen, but anything that affected skeletal growth may well have affected the relation between epiphyseal closure and dental eruption. Naturally, if the developmental pathology was a significant influence on growth, then we shouldn't be using WT 15000 as a model for early Homo erectus stature anyway.

A more relevant argument is that KNM-WT 15000 is really an outlier when we assume that it would have grown to a very tall stature. On first appearance, this seems correct. We have quite a number of femora from Homo erectus, both inside and outside of Africa. Only two of them approach the length that had been estimated for the Nariokotome adult stature estimate. KNM-WT 15000's former adult estimate is the extreme.

But looking more closely, both those tall individuals come from generally the same time and place as KNM-WT 15000. KNM-ER 1808 and KNM-ER 736 both preserve partial femur shafts with estimated lengths above 480 mm. Both specimens are a bit older than Nariokotome, between 1.6 and 1.7 million years old. KNM-ER 1808 in particular contributed heavily to the argument that early Homo erectus had a very tall stature, because the partial skeleton includes a fragment of pelvis, argued to be female. A tall woman makes for a very tall species.

Still, these two specimens don't seem as significant in 2010 as they did twenty years ago. The Gona pelvis suggests that we don't really know the sex of KNMER 1808. Its pelvic fragment looks female in the context of living human dimorphism, but quite possibly male compared to the Gona individual. Henry McHenry McHenry:1991 estimated adult statures for the KNMER 1808 and KNMPR 736 femurs, both around 5'10" (180 cm). Those are the tall end of stature estimates for Homo erectus, both taller than average for living humans. But perhaps neither is surprising when taken as the largest and of the distribution that on the whole is relatively small bodied. An estimate of 163 cm for the adult height of KNM-WT 15000, as suggested by Graves and colleagues, would not be an outlier in this population, but neither would an estimate as large as 180 cm.

So I think the comparative evidence is equivocal. Revisiting the specimen with a smaller estimate is reasonable, but I think our ability to assess the accuracy of any estimate is very limited. In light of the pathology of KNM-WT 15000, it may not be very relevant to understanding body size evolution in early Homo, anyway.

The main problem facing us with understanding body size in early Homo is deciding which specimens should be included in which taxa. If we exclude everything except the relatively tall ones, like KNMER 1808 and KNMER 736, then we are going to end up with a tall stature estimate for a population, putatively H. erectus. But if we include some of the smaller specimens, like KNMER 993, or KNMER 803 both contemporaries of the Nariokotome skeleton than the average for this more inclusive population will be a lot lower. In East Africa 1.5 million years ago we can't assign an isolated femur to a species, and we won't have a good answer for this issue until we have many more associated specimens.

I tend to think that small stature is the null hypothesis now, given our knowledge of the small stature of the Dmanisi hominins, and the moderate body size of middle Pleistocene Homo everywhere else. There are a few specimens that represent individuals as tall as those indicated by KNM-ER 736 and KNM-ER 1808, but none taller, and many much shorter.

It's a much deeper topic than one skeleton, but the problems assessing stature in that skeleton help to highlight the difficulty of the problem in a global sense.

UPDATE (2010-09-18): A reader suggests that I give a link to a 2004 paper by Shelley Smith, which compared the dental and skeletal maturation of KNM-WT 15000 to a large growth series of modern Canadians Smith:WT15000:2004. She found cases in the sample with comparable mismatches of dental and epiphyseal age estimates, and argued that we can't exclude a humanlike growth spurt for early Homo. That's one reason why I think this issue can't be resolved -- the variation in humans is great enough to encompass the known fossil specimens.

A similar lack of resolution applies to enamel growth increments in KNM-WT 15000 ("Dental growth in early Homo"). The specimen can't be distinguished from Australopithecus, but the range in modern humans is very extensive.

At the moment, skeletal correlates of growth don't give us the resolution to answer these questions definitively about early Homo. If we had more specimens, we could at least reduce the component of error from sampling, which would help considerably. But we can't expect that anytime soon.