Published online 16 February 2011 | Nature | doi:10.1038/news.2011.98


The con of convergence

Sifting early human from non-human fossils is complicated by convergent evolution.

bonesThe hominid Oreopithecus bambolii shares some characteristics with modern humans and was originally mistakenly identified as a human ancestor.Bernard Wood & Terry Harrison

Teasing out the fossils of our early ancestors from those of related but non-human species has proven an extreme scientific challenge. Now, a study suggests that the evolutionary phenomenon of convergence, in which two unrelated organisms acquire the same trait, has added to the difficulties.

The review, published in Nature today1 by palaeoanthropologist Bernard Wood at George Washington University in Washington DC and Terry Harrison, an anthropologist at New York University, says it is misleading to use key features that distinguish modern humans from great apes to try to sort hominin fossils - those in the human lineage - from non-human hominid fossils. Key traits in hominins might have also been selected for in unrelated hominids living in a similar environment.

One such feature is canine teeth, which are much smaller in modern humans than in apes. Yet Wood and Harrison point out that during the late Miocene epoch, there were hominids that were not on the path towards humanity but that had reduced canines. The Eurasian hominids Oreopithecus, Ouranopithecus, and Gigantopithecus all evolved small canines, presumably because this helped them to feed more effectively in their habitat.

Another key trait often used to sort early human from non-human fossils is the position of a hole in the cranium known as the foramen magnum. In great apes it is farther back in the skull to allow the apes to bend, and move through trees, more easily. In humans it is more or less centred beneath the skull to support upright posture and make bipedal movement possible. But, Wood and Harrison say, this interpretation might be too simplistic.

The review notes, for instance, that gibbons have foramen magnums that are similar in placement to those of humans – another example of convergence. In the case of gibbons, however, the more central placement of the hole is there to allow for different skull characteristics rather than a bent posture. For this reason, the authors suggest that using the foramen magnum as a method for diagnosing bipedal behaviour may be misleading.

They also point to evidence in past palaeontology of how this kind of analysis can misfire. The late Miocene fossil ape Oreopithicus, found in Italy, is only distantly related to modern humans, yet initially it was incorrectly identified as a hominin because it shares many features traditionally used when assigning fossils to the human lineage.

Off target

Wood and Harrison use their analysis to take aim at the recent assignment of the hominid Ardipithecus ramidus to the human lineage, arguing the decision is marred by convergence.

They suggest that because the hands and feet of Ardipithecus are ape-like, its hominin characteristics, primarily its small canines, are the result of convergent evolution, explains Jeremy DeSilva, a palaeoanthropologist at Boston University in Massachusetts. "It is a unique foot that has a big grasping toe, but is stiffer than a chimpanzee's foot. Convergence has to be happening at some level either in the apes, in A. ramidus or in both, but we just don't know enough yet to be certain," says DeSilva.

Tim White at the University of California, Berkeley, one of the palaeoanthropologists who worked with A. ramidus, argues that the review is deeply flawed.

"This paper may seem 'researchy' but it mysteriously ignores nearly all of our 2009 analyses published in Science2. We showed that the extensive and independent character complexes of the teeth, skull and postcrania shared by Australopithecus and Ardipithecus did not result from convergence. Without addressing our results and conclusions with any new data, analyses or ideas, this 'review' boils down to an illustrated six-page polemic without the scholarship normally expected in a peer-reviewed Nature paper," says White.

Rampant convergence


DeSilva argues that the main point of the review remains strong even if A. ramidus is ignored.

"With all the convergent traits seen in [Miocene hominid] Sivapithecus and Oreopithecus, this problem is already rampant in anthropology. Of course it is important to include A. ramidus: it's on everybody's minds these days. But this review would have been just as relevant had A. ramidus never been dug up," says DeSilva.

Yet convergence confusion may not remain a serious issue for long. Techniques for imaging fossils are improving quickly and might one day allow us to see how even identical convergent specimens came to be, explains Wood. But until that time, he says, "we should be more cautious". 

  • References

    1. Wood, B. & Harrison, T. Nature 470, 347-352 (2011). | Article
    2. White, T. D. et al. Science 326, 75-86 (2009). | ISI |
Commenting is now closed.