ramidus and evolution William H. Kimbel Institute of Human Origins School of and Social Change Arizona State University

Small canine teeth and bipedal locomotion have been central to explanatory hypotheses about human origins at least since the publication of Charles Darwin’s Descent of Man and Selection in Relation to Sex in 1871. Darwin postulated a feedback model for the evolution by natural selection of terrestrial bipedality, a large brain, small canine teeth, tool use/manufacture, and /defense. His hypothesis was, however, unconstrained by a “chronicle of events,” as no fossils linking the few then-known Neandertals to the last common ancestor (LCA) of African apes and had yet been discovered. In the absence of an early fossil record, all Darwin could do was to project linked groups of differences between modern humans and the African great apes back in time to hypothetical ancestors. A little more than half a century later, Darwin’s hypothesis was challenged by the recognition of Australopithecus as a human ancestor. The fossils from South Africa showed that it was possible to be a hominin biped in spite of having an ape-size brain and lacking sophisticated tool-making capabilities. In the mid- 1960s, the habilis finds showed that only a modest increase in had accompanied the onset of patterned stone-tool making and the routine scavenging or hunting for meat. The more recent recovery of the Australopithecus lineage (i.e., the A. anamensis – A. afarensis evolutionary species) demonstrated that both relatively monomorphic non-honing canines and musculoskeletal structures associated with bipedal locomotion had evolved by ca. 4.0 Ma, but were imprecise matches for modern human counterparts — still sharing clear, derived similarities with Homo sapiens but in much more generalized anatomical and adaptive milieux. No serious scholar doubts that all of these extinct species are more closely related to humans than to despite their significant departures from modern human form and behavior. As these examples illustrate, the answer to the fundamental question “What is a hominin?” is not a fixed list of traits, no matter how significant we think those traits may be. In fact, the answer is phylogenetic: a hominin is any taxon more closely related by descent to Homo sapiens than to Pan. Both of these branches have accumulated many changes from their last common ancestor over at least 6 myr of separate evolution. Naturally, we can expect that the closer we get in time to the -human LCA, the more challenging will be the job of disentangling these branches, but the same principles and methods for doing so still apply. The characteristics of the LCA of chimpanzees and humans have until recently been a matter of informed speculation. Only within the last 20 years have new discoveries extended the fossil record back to a time close to the molecular estimates of the age of the chimpanzee-human divergence, ca. 8-6 Ma. Beginning

1 in the mid-1990s, a spate of discoveries in east and north-central Africa revealed a group of taxa that bear on the identity of the chimpanzee-human LCA. Spanning the time period ca. 7 to 4.4 Ma, these species — Ardipithecus ramidus, Ar. kadabba, tchadensis, and tugenensis — have been the focus of debate about the morphological diagnosis of earliest hominins and whether our closest living relatives, gorillas and chimpanzees, tell us very much about what the LCA was like. Of the recently identified late -early taxa, the 4.4 myr-old Ardipithecus ramidus from is by far the best represented anatomically. The fragmentary initial sample, reported in 1994, displayed non-honing canine teeth and a foreshortened cranial base. Mainly on the basis of these features, its discoverers (a team working in the Middle Awash valley, Ethiopia, led by T.D. White of UC Berkeley) were able to identify Ar. ramidus as a phyletic hominin, more closely related to Australopithecus + Homo than to Pan. But later finds, now numbering more than 100 specimens, described by White’s team in 2009, include a partial (though crushed) skull and skeleton that have made it clear that Ar. ramidus was a very unusual species. The skeleton’s low, broad pelvis accommodated hip abductor muscles that helped balance the trunk in bipedal locomotion, as in humans. But its strongly divergent great toe (for grasping or climbing) and the long ischium of its pelvis (enhancing the leverage of the hip extensor muscles in quadrupedal postures) suggest that its locomotor repertoire also featured a substantial arboreal component. There are, however, few indications of African-apelike below-branch suspension (or knuckle-walking) in the postcranial skeleton of Ar. ramidus. In addition, its skull, with an endocranial volume (ca. 300 cc) smaller than an average chimpanzee’s, notably lacks the exaggerated lower facial prognathism of the chimpanzee, the great midfacial height of Australopithecus and the gorilla, and the distinctive supraorbital torus seen in all extant African apes. In short, the ways in which Ar. ramidus differs morphologically from undoubted hominins are not, in key respects, particularly African-apelike. Another way of putting this is that Ar. ramidus is not very similar to what we have come to expect the LCA to look like based on Darwin- style comparisons of modern humans and chimpanzees (and gorillas).

This unusual mix of attributes has led some paleoanthropologists to argue that, rather than being related exclusively to the Australopithecus+Homo , Ar. ramidus may instead represent a descendant population of one of several late Miocene ape lineages in which parallel evolution resulted in the widespread distribution of humanlike features by the time of the chimpanzee-human split. Parallel evolution is a widespread phenomenon, and its prevalence has led to many debates regarding the historical path of human evolution. Is it possible that the non-honing canines, shortened basicranium, and skeletal traits related to bipedality seen in Ar. ramidus are the products of parallel evolution, rather than of phylogenetic affinity with hominins?

Recently, I, together with my colleagues Gen Suwa, Berhane Asfaw, Yoel Rak, and Tim White, addressed this question by conducting a detailed comparative study of the Ar. ramidus basicranium. The cranial base is a valuable resource for phylogenetic study because its anatomical complexity and association with the

2 neural, locomotor and masticatory systems provide numerous opportunities for evolutionary change. The human basicranium differs profoundly from that of apes and other in shape and morphological detail. In humans, the foramen magnum (the large opening through which the spinal cord exits the braincase) and occipital condyles (the vertebral column’s articulation with the head) are more forwardly located than in apes, the cranial base in the midline is relatively short from front to back and strongly “flexed” internally, and the bilateral openings in the base for blood vessels and nerves are more widely separated. These changes affect the arrangement of the basicranial bones in such a way that it is fairly easy to tell apart even isolated fragments of ape and human basicrania. My previous work with Yoel Rak has shown that many of these human peculiarities were already present in the earliest known Australopithecus skulls (A. afarensis) ca. 3.0-3.4 myr ago. Studies by Gen Suwa and colleagues, published in the journal Science in 2009, demonstrated that Ar. ramidus was humanlike in having a forwardly positioned foramen magnum associated with a short basicranium. Does Ar. ramidus share additional basicranial features with undoubted hominins? The primary subject of our analysis was a beautifully preserved but incomplete cranial base of Ar. ramidus that was part of the series of fossils published in 1994. We compared this specimen to modern ape and human crania as well as to a diverse set of Australopithecus specimens. What we found expands the basicranial pattern Ar. ramidus shares with undoubted hominins. For example, as in modern humans and Australopithecus, the central part of the base in Ar. ramidus is very broad in relation to overall skull size. In effect, this broadening accommodates the forward shift of the craniovertebral articulation on the shortened basicranial axis. We also made a second estimate of cranial base length, using different methodology from that employed by Suwa and his colleagues, and confirmed the humanlike short basicranium in Ar. ramidus. A few other primates show a slightly forward location of the foramen magnum (, for instance), but they do not show the same broad, short cranial base shape that Ar. ramidus shares with Australopithecus and Homo. Many detailed changes in the morphology of the temporal and occipital bones, which in undoubted hominins accompany these fundamental shifts in shape, are also seen in Ar. ramidus. Our work on the cranial base reinforces the hypothesis that Ar. ramidus shares unique features with Australopithecus and Homo due to common ancestry rather than parallel evolution. The question then becomes, What was driving these early evolutionary changes in the skull base? Were they due to a shift in the poise of the head on the vertebral column associated with the adoption of bipedality? If so, does the humanlike cranial base of Ar. ramidus confirm postcranial evidence for partial bipedality in this species? Or, do the changes tell us about the shape of the brain (and of the base on which it sits), perhaps a manifestation of early changes in neocortical organization in the hominin lineage? Both alternatives will need to be studied carefully in light of the finding that, despite its unusual features, Ar. ramidus is indeed more closely related to you than to a chimpanzee.

3