Human evolution

From forelimbs to two legs

Article metrics

A reanalysis of the wrist bones of early human fossils provides the first good evidence that humans evolved from ancestors who 'knuckle-walked', as chimps and gorillas do today.

Most people think that palaeoanthropologists spend their lives in exotic places prospecting for fossils. Not so — researchers working on human evolution generally spend most of their time in the laboratory seeking new ways of testing ideas about the existing fossil record. At least as many advances in our understanding of human evolution have come from reanalyses of familiar fossils as from the discovery of new ones. The study described by Richmond and Strait on page 382 of this issue1 is an excellent case in point.

Richmond and Strait reanalysed several well-known early hominid fossils, including the famous skeleton known as Lucy. Comparing the wrists of these fossils with the wrists of modern humans and several other primate species, Richmond and Strait found that two of our earliest fossil relatives, Australopithecus afarensis and Australopithecus anamensis, exhibit characteristics of the wrist that are seen today only in the African apes. These features are thought to be associated with knuckle-walking, an unusual mode of quadrupedal locomotion in which the fingers are bent and weight is supported on the backs of the second of the three rows of finger bones. In contrast, the wrists of the other early hominid species in their sample, Australopithecus africanus and Paranthropus robustus, are like those of modern humans in that they lack the putative knuckle-walking characteristics ( Fig. 1).

Figure 1: Approximate time ranges of known hominid species.

Those species shown in the red circle are robust ‘australopithecines’ belonging to the genus Paranthropus and represent an evolutionary dead-end. Those in the blue circle belong to the genus Australopithecus, and those in the green circle are of the genus Homo. There is no consensus over the precise phylogeny linking these species. Richmond and Strait1 have recognized knuckle-walking features in the wrists of those species indicated by dark orange colouring; species indicated in light orange do not have these features. Evidence for the presence or absence of such features in the other species is either non-existent or not yet analysed.

Richmond and Strait's study is significant for several reasons. First, it bears on the long-standing debate over the evolutionary history — the 'phylogenetic' relationships — of modern humans (Homo), chimpanzees (Pan) and gorillas (Gorilla). Genetic analyses overwhelmingly indicate that chimpanzees and humans are more closely related to one another than either is to gorillas2. Until the study by Richmond and Strait, however, the anatomical evidence largely ran counter to this conclusion. Many palaeoanthropologists took the knuckle-walking characters of Pan and Gorilla to be ‘shared derived’ characters which showed that the two groups of African ape were each other's closest relative3.

The anatomical evidence could be reconciled with the molecular evidence in only two ways. Some researchers thought that the knuckle-walking characters exhibited by Pan (Fig. 2, overleaf) and Gorilla were independently derived or ‘convergent’ characters, resulting from natural selection under similar environmental and ecological conditions4. This perspective implied that the knuckle-walking characters are of no phylogenetic significance. Others argued that the knuckle-walking characters were present in the common ancestor of Pan, Gorilla and Homo, and were simply lost in the lineage leading to Homo5. According to this view, the knuckle-walking characters of Pan and Gorilla are ‘primitive’ features, which, like convergent characters, tell us nothing about phylogeny. Richmond and Strait's results offer strong support for the idea that knuckle-walking characters were present in the common ancestor of modern humans and the African apes, and provide grounds for reconciliation between the molecular and anatomical evidence.

Figure 2


The common ancestor of modern humans and African apes, including gorillas, bonobos and chimps (pictured), may have used a form of locomotion called ‘knuckle-walking’ when on the ground.

The second point of significance in the new study1 relates to the phylogenetic relationship between A. afarensis and A. africanus . For many years, palaeoanthropologists thought that A. afarensis was the ancestor of A. africanus and the later hominids because its skull has more chimpanzee-like features than that of A. africanus. In 1995, however, the discovery at Sterkfontein, South Africa, of a collection of foot bones from a single A. africanus individual6 complicated the picture. Unexpectedly, the bones indic-ated that the foot of A. africanus was more ape-like than the foot of its putative ancestor A. afarensis. This finding was supported by analyses of a newly discovered shin bone from Sterkfontein, which suggested that the lower leg of A. africanus was more ape-like than that of A. afarensis7. It was also supported by an assessment of hominid joint sizes, from which it seemed that A. africanus had more ape-like limb proportions than A. afarensis8.

So it looked like the skulls of these species were telling one story (that A. afarensis was the ancestor of A. africanus and the later hominids), and their limbs were telling another (that either A. afarensis was a side-branch in human evolution and A. africanus was the ancestor of the later hominids, or vice versa). The work by Richmond and Strait further complicates the picture: it suggests that A. afarensis retained some knuckle-walking features, whereas A. africanus did not. It is no longer a case of the skull pointing to one set of phylogenetic relationships, and the postcranial skeleton — everything but the skull — to another. Rather, different parts of the postcranium may not support the same phylogenetic hypothesis.

The third notable aspect of Richmond and Strait's study concerns the interpretation of fossil morphology in terms of function, especially locomotion. Ever since Lucy was discovered in the late 1970s, there has been debate about the locomotor repertoire of A. afarensis, the species to which she is assigned.

The basic facts are not in dispute. A. afarensis has a combination of traits that is not seen among living primates. In some respects, A. afarensis is quite human-like (for instance in the foot structure, non-opposable big toe, and pelvis shape). In others, it is quite ape-like (relatively long and curved fingers, relatively long arms, and funnel-shaped chest).

What can be made of these features? For some researchers, the ape-like characteristics of A. afarensis are non-functional retentions from the common ancestor of hominids and the African apes. Here, emphasis is put on the human-like characteristics, and A. afarensis is seen as a hominid that walked on two legs and got about in no other way9. For others, the ape-like traits are functionally important, and A. afarensis is interpreted as using a ‘mixed’ locomotor repertoire, in which a form of terrestrial bipedalism was combined with an ability to move around effectively in trees10.

Richmond and Strait add a twist to this debate. They propose that the knuckle-walking features of A. afarensis are non-functional retentions from the common ancestor of hominids and African apes. This seems an entirely reasonable position, given that A. afarensis shows many traits that are thought to be associated with bipedal locomotion. The alternative idea — that A. afarensis combined knuckle-walking, bipedalism and climbing — is somewhat counterintuitive, because it implies the use of two entirely different modes of terrestrial locomotion.

By the same token, however, Richmond and Strait's argument undermines the idea that A. afarensis combined bipedalism with climbing, which many researchers have hitherto considered to be the best interpretation of the evidence. Can we assert that one set of ape-like characters indicates that A. afarensis was an able climber, while at the same time arguing that another, equally good, set of ape-like characters is indicative of nothing except the phylogenetic history of A. afarensis? If the knuckle-walking characters are considered to be primitive retentions, must not the same hold for the other ape-like characters?

Are our only choices to accept that A. afarensis was a striding biped with a large number of non-functional primitive retentions, or to have to take seriously the counterintuitive idea that the locomotor repertoire of A. afarensis included forms of bipedalism, climbing and knuckle-walking?

It is difficult to predict how palaeo-anthropologists will react to Richmond and Strait's study. But one thing is certain. It will encourage many researchers to reconsider their assumptions about the phylogenetic and functional implications of bone shape and size in the primates, and most especially in the early hominids.


  1. 1

    Richmond, B. G. & Strait, D. S. Nature 404, 382–385 (2000).

  2. 2

    Ruvolo, M. Mol. Biol. Evol. 14, 248–265 (1997).

  3. 3

    Andrews, P. & Martin, L. B. J. Hum. Evol. 16, 101–118 (1987).

  4. 4

    Dainton, M. & Macho, G. A. J. Hum. Evol. 36, 171–195 (1999).

  5. 5

    Pilbeam, D. R. Mol. Phyl. Evol. 5, 155–168 (1996).

  6. 6

    Clarke, R. & Tobias, P. V. Science 269, 521–524 (1995).

  7. 7

    Berger, L. R. & Tobias, P. V. J. Hum. Evol. 30, 343–348 (1996).

  8. 8

    McHenry, H. M. & Berger, L. R. J. Hum. Evol. 35, 1–22 (1998).

  9. 9

    Latimer, B. & Lovejoy, C. O. Am. J. Phys. Anthropol. 78, 369–386 (1989).

  10. 10

    Susman, R. L., Stern, J. T. & Jungers, W. L. Folia Primatol. 43, 113– 156 (1984).

Download references

Author information

Correspondence to Mark Collard or Leslie C. Aiello.

Rights and permissions

Reprints and Permissions

About this article

Further reading


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.