The fragile bones of infants rarely survive long enough to make it into the hominin fossil record. But if they do, they provide precious evidence about the growth and development of the individual and its species. This helps researchers not only to understand how such processes have changed during hominin evolution, but also to interpret the function and the taxonomic significance of the better-sampled adult specimens. In this respect, the remarkably complete 3.3-million-year-old skeleton of a three-year-old Australopithecus afarensis female, found in Dikika, Ethiopia, is a veritable mine of information about a crucial stage in human evolutionary history. In this issue, the fossil is described by Alemseged et al. (page 296)1, and its geological and palaeontological context is reported by Wynn et al. (page 332)2.

Thanks to efforts in Ethiopia and elsewhere, we already know a good deal about A. afarensis. It has been called an 'archaic' hominin for at least two reasons. First, it is old: its fossils date from between 4 million and 3 million years ago. Second, its morphology is archaic, in the sense that its brain case, jaws and limb bones are much more ape-like than those of later taxa that are rightly included in our own genus, Homo. When adjusted for its body size, the brain of A. afarensis is not much larger than that of a chimpanzee, and although it has lost the large canines that distinguish apes from hominins, other aspects of its dentition, such as its relatively large chewing teeth, are still primitive (Fig. 1).

Figure 1: A hominin taxonomy.
figure 1

Species are ordered according to the period of their fossil record and, left to right, according to their resemblance to modern humans: those with large brains, small chewing teeth and jaws similar to those of Homo sapiens are found to the left, those with large chewing teeth and jaws to the right. Australopithecus afarensis, an infant female specimen of which has been found in Dikika, Ethiopia1,2, lived between 4 million and 3 million years ago. Its small brain is not much larger than that of a chimpanzee, but its dentition has features akin to those found in more modern hominins.

There remains a great deal of controversy regarding the posture and locomotion of A. afarensis. Most researchers accept that it could stand upright and walk on two feet, but whether it could climb up and move through trees is still disputed. Some suggest that its adaptations to walking on two feet preclude any significant arboreal locomotion, and interpret any limb features that support such locomotion as evolutionary baggage without any useful function3. Others suggest that a primitive limb morphology would not have persisted unless it served a purpose4.

The Dikika infant is not the first early hominin infant to be found. That distinction belongs to the Taung child, whose discovery was reported just over 80 years ago5. What makes the Dikika infant remarkable is its unprecedented completeness for such a geologically ancient specimen. The infant was found in sediments that formed the bottom of a small channel close to where a river discharged into a lake2. This was not a turbulent stream or river. The flow was sluggish, typical of the type of braided streams that make up a river delta. The corpse of the infant was buried more or less intact, and the sediment in flood waters must have swiftly covered it.

Some parts of the specimen — the pelvis, the lowest part of the back and parts of the limbs — are still missing, but what is preserved is remarkably complete. The face, the brain case and the base of the cranium, the lower jaw, all but two of the teeth (including unerupted adult teeth still in the jaw), both collar bones, the vertebrae down to the lower back, many ribs, both knee caps and the delicate bone that holds open the throat, the hyoid, are all there. Even the medial epicondyle of the humerus has survived. This is the bony projection on the inside of your elbow against which your left thumb rubs if you hold your right elbow with your left hand. In a three-year-old infant, this tiny piece of bone is still separate from the main shaft of the humerus. One must travel forward in time more than three million years, to a Neanderthal infant from Dederiyeh6, Syria, to find a comparably complete hominin infant skeleton.

This anatomical cornucopia was not evident when the specimen was found in 2000: most of the Dikika infant was invisible, hidden within a slab of sandstone. Zeresenay Alemseged has devoted many thousands of hours over a five-year period to removing, painstakingly, the cement-like matrix that surrounds the delicate bones. The patience, time, skill and effort required to preserve and expose the morphology of this and other similar early hominin fossils7 should not be underestimated.

But why are Alemseged et al.1 so sure that the infant belongs to A. afarensis, and can we have confidence in its age — both the geological age of the fossil and the age of the child it represents? The geological age is secure. The Dikika sediments contain crucial evidence of the same layers of ash that have provided reliable argon–argon isotope ages at other East African fossil sites2. There are also subtle and not so subtle differences between the faces of A. afarensis and the other hominin taxa known from similarly aged rocks, and the Dikika infant already shows signs of the type of upper jaw and nasal morphology that is seen only in A. afarensis. These signs are a rounded area above the upper teeth; a separation between the bone covering the roots of the upper canine teeth and the edge of the opening for the nose; and hourglass-shaped nasal bones that fit into a recess in the frontal bone much like a tenon fits into a mortise.

The second of the two age estimates, the chronological age of the infant, is less secure. All one can do is use the kind of computed-tomography imaging familiar from modern hospitals to compare the development of the yet-to-emerge permanent tooth germs of the Dikika infant with the teeth of modern human and chimpanzee infants of known ages8. The best match is with three-year-old chimpanzees. But it is highly unlikely that the pace of development of A. afarensis was exactly the same as that of modern chimpanzees. So, for now, the chronological age of the Dikika infant must remain an informed guess.

The discoverers of the Dikika fossil have only just begun the task of capturing all the data contained in the specimen, but already these preliminary data1 are informing the controversy of how A. afarensis moved. If its mode of locomotion was exclusively on two legs, one would expect that the limb bones and the organs that help it to balance would be more similar to those of the only living bipedal higher primate (that is, us) than to those of chimpanzees and gorillas. These primates walk on two feet only rarely, if at all.

Alemseged et al.1 pay careful attention to the shoulder, hand and the semicircular canals of the inner ear, the morphologies of which record the motion of the body. The shoulder-bone (scapula) of the fossil is more like that of a gorilla than a modern human, and the bones of the only complete finger are curved like those of a chimpanzee. Chimpanzee finger bones begin life only slightly curved, but become more curved when the hands are used to climb branches9; this is what seems to have happened in the case of the Dikika infant. Lastly, images of the inner ear of the specimen show it to have semicircular canals more like those of chimpanzees than of modern humans10. The fluid-filled semicircular canals are crucial in maintaining balance, and so all three lines of evidence suggest that the locomotion of A. afarensis was unlikely to have been restricted to walking on two feet.

I am especially intrigued by the detailed morphology of the hyoid bone in the throat of the fossil. Does the open space in the body of the hyoid mean that A. afarensis had air sacs in its neck? In the absence of large canines, these air sacs might have been a way in which males established a dominance hierarchy, and females judged the quality of a potential mate.

Whatever the answers to such questions, the Dikika infant has the potential to provide a wealth of information about the growth and development, function and taxonomy of A. afarensis.