Implications of A.L. 444-2 for the Taxonomic and Phylogenetic Status of Australopithecus afarensis

A.L. 444-2 is the first specimen to preserve the cranium and mandible of a single adult individual of A. afarensis. Pairing this specimen with A.L. 417-1, which includes a mandible and maxilla, enables us to compare comprehensively the craniofacial morphology of male and female individuals of the species for the first time. The occluded mandibles and maxillae of A.L. 444-2 and A.L. 417-1 reveal a distinctive hominoid snout contour, combining a strongly inclined, convexly sloping nasoalveolar clivus with a relatively upright mandibular symphysis, a straight to slightly rounded anterior symphyseal outline, and an anteriorly placed gnathion. Both A. afarensis specimens feature a very deep mandibular corpus, whose height occupies close to 70% of the orbitoalveolar height of the face. In the African great apes, this value ranges from 36% to 54%, and in modern humans, it is 66%. The high value in humans is due to a short orbitoalveolar region rather than to a deep mandible. A. afarensis appears to share a relatively deep corpus with A. robustus (the only robust species in which the feature can be determined for a single individual) but not with A. africanus. Relative to the calvarial length, the A.L. 444-2 braincase height is apelike, falling between the tall modern human braincase and the low braincase of A. boisei and A. aethiopicus. In A. africanus (Sts. 5) and H. habilis (KNMER 1813) the relative braincase height is like that of A.L. 444-2 and the great apes. According to Le Gros Clark’s (1950) index expressing the height of the calvaria above the roof of the orbit as a percentage of total calvarial height, Sts. 5 and KNM-ER 1813 have tall, “humanlike” braincases, whereas A.L. 444-2, A. boisei, A. aethiopicus, and the African great apes group together with low braincases. In contrast to the rounded, nearly circular midsagittal outline of the chimpanzee calvaria, the posterior parietal/ occipital arc in A.L. 444-2 is steep and deviates anteriorly from the circle. This is also true of the A. boisei calvaria. As expected from the calvarial height comparison, the slope of the A.L. 444-2 frontal squama is smaller than that of A. africanus and H. habilis.

A.L. 444-2: The Skull as a Whole

Among the largest Plio-Pleistocene hominin skulls found to date, A.L. 444-2 is bigger, though not by much, than an average female gorilla’s skull. At first glance, A.L. 444-2 assumes a somewhat simian appearance, the outcome of a relatively small braincase combined with an inclined frontal squama and prognathic jaws. However, this apelike appearance is offset by several distinctive hominin features: a very tall face that is much less prognathic than would be expected from the skull’s general simian-like appearance; a deep, vertical mandibulosymphyseal profile; delicate supraorbital elements; and the absence of a supratoral sulcus intervening between the frontal squama and the forward-jutting supraorbital element. Nevertheless, the characteristics that account for the skull’s hominin appearance demonstrate a certain uniqueness, which is manifested in the disproportion between the considerable total height of the face and the great size of its constituent elements (primarily the zygomatic and maxillary bones), on the one hand, and the delicateness of the supraorbital element and the almost negligible degree of its anterior projection, on the other. An apparent unevenness emerges along the vertical axis of the face between its upper portion—the orbits, including the elements above and between them—and its lower portion, that is, the elements below the level of orbitale down to gnathion. Undoubtedly, part of this appearance stems from the heavy, somewhat vertical, deep, and anteriorly bulbous symphyseal region of the mandible. The corresponding region in the African apes, in contrast, is transversely pinched, as its two sides converge downward toward the midline. Furthermore, the region slopes inferoposteriorly; in anterior view, it is tucked under the alveolar element and hence is less exposed than in A.L. 444-2. The preservation of the mandible of A.L. 444-2 and its occlusion with the upper dental arcade afford a unique opportunity to evaluate some of the characteristics of an entire A. afarensis skull. Two standard measurements can be recorded: the distance between gnathion and the estimated site of nasion—a measure of the total height of the face—which is 150 mm, and the distance between gnathion and basion, estimated at 157 mm.

Elements of the Disarticulated Skull

The 1970s collection of hominin cranial remains from Hadar is notoriously weak in its representation of the frontal bone. Besides the complete but distorted frontal of the A.L. 333-105 juvenile (Kimbel et al., 1982), only two very incomplete adult specimens provided glimpses of frontal morphology: A.L. 288-1 (Johanson et al., 1982b) and A.L. 333-125 (Asfaw, 1987). With the recovery of the almost complete frontal bone of A.L. 444-2, we are able to fill one of the last remaining gaps in our knowledge of the Hadar hominin adult skull. Another frontal specimen, A.L. 438-1b, contributes important information on the glabellar and supraglabellar regions, which are missing or poorly preserved in A.L. 444-2. The A.L. 444-2 frontal bone features prominent, laterally projecting supraorbital bars, strongly convergent temporal lines, and a transversely broad squama with only moderate postorbital constriction. The minimum distance between the temporal lines (30 mm) in the plane of the postorbital constriction is much smaller than the postorbital constriction itself (77 mm), creating on each side an extensive, almost horizontally inclined facies temporalis that, in coronal section, slopes gradually from the inferior temporal lines to the medial wall of the temporal fossa. In between the temporal lines, the supraglabellar region bears a mild hollow that grades smoothly onto the superior surface of the supraorbital bars. Neither a supratoral sulcus nor a trigonum frontale is present. The supraorbital bars are wide anteroposteriorly, measuring 16 mm at the right lateral break, about 42 mm lateral to the midline. The preserved portions of the anterior supraorbital margins are aligned coronally, forming right angles with the midsagittal line. At the lateral break on each side, the margin actually occupies a slightly more anterior plane than the middle part of the margin, suggesting an anteriorly prominent superolateral corner of the orbit. At the medial break through the left supraorbital, about 22 mm lateral to the midline, the anterior margin begins to swing out toward glabella (this area is damaged on the right side). The extent of anterior glabellar protrusion is suggested by the preserved supraglabellar plate, whose superior surface projects in the midline about 5 mm beyond the anterior supraorbital margins.

Endocranial Morphology of A.L. 444-2

The original endocast of A.L. 444-2 consisted of a single plastic cast, colored to show the original fragments (light brown) and the reconstructed missing parts (black). This we label the Rak-Kimbel endocast, which was based on the reconstruction of cranial and facial fragments. Because distortion was severe enough to interfere with morphological description and measurements, and especially the assessment of endocranial capacity, a plaster endocast was received from Yoel Rak in 1998 for purposes of modification. This newer plaster endocast formed the basis for the original endocast reconstruction done by R.L.H., who based the reconstruction on the less distorted side (left) and then doubled its water-displaced volume to achieve the final endocranial volume. As will become clear in our descriptions, this first method required several additions and subtractions to compensate for missing portions, for flash lines left from the casting process, and for distortion remaining in the reconstruction. We concluded that a more accurate reconstruction would result if the portions of the original endocast were separated from reconstructed elements and approximated on a plasticene “core” so that distortion could be effectively eliminated. The second method, which was accomplished mostly by M.S.Y. with minimal guidance from R.L.H., permitted a range of possible reconstructions of the actual brain endocast pieces and provided a range of endocast volumes. This reconstruction methodology, referred to as the “dissection method,” eliminated most of the distortion and obviated the need to correct for flash lines. Although both methods provide a final endocranial capacity very close to what must have been the actual living brain volume of A.L. 444-2, we consider the dissection method to be the more accurate one. Distortion of the endocranial cast mirrors that of the cranium. While the right parietotemporal area appears to be depressed, the left parietotemporal area shows signs of bulging in compensation. In addition, due to a gap that runs anteroposteriorly along the left temporal lobe, there is an artificial increase in the distance between the base of the endocast and its apex of about 3–8 mm on the left side.

Recovery and Reconstruction of A.L. 444-2

The A.L. 444-2 skull was found on 26 February 1992, during a strategic paleontological survey of Kada Hadar Member sediments that are stratigraphically situated between BKT-1 and BKT-2 tephras, on the eastern edge of the Awash River’s Kada Hadar tributary. Yoel Rak discovered two fragments of hominin occipital bone (A.L. 444-1) at the base of a steep hill composed of Kada Hadar Member silts and clays capped by a weathered sandstone remnant. Subsequent examination of the upslope surface revealed additional hominin skull fragments (the temporal bones and maxillae) clustered together and partially exposed in a narrow gully that dissected the face of the hill. During the next seven days, probing and dry sieving of the gully infill and hillside colluvium over a 77 m2 area led to the recovery of fragments representing about 75%–80% of a single hominin skull. It was immediately apparent that the upslope finds duplicated the anatomical parts represented by the two A.L. 444-1 occipital fragments and therefore constituted a second hominin individual, cataloged as A.L. 444-2. In addition, the lambdoidal suture of the A.L. 444-1 occipital is completely unfused, suggesting subadult status, whereas fused cranial sutures and extreme dental occlusal wear indicate an advanced ontogenetic age for A.L. 444-2. In February–March 1993 the A.L. 444 hillside was excavated in an effort to locate missing parts of the A.L. 444-2 skull and to determine its precise stratigraphic provenance. No further remains of the hominin skull were encountered in situ, but a complete viverrid cranium and indeterminate fragments of large mammal bone with preservation and patina (mottled dark gray, white, and yellowish gray) identical to those of the hominin were excavated in an unstratified, cemented carbonate silt that exactly matches the matrix adhering to A.L. 444-2. We are confident that the hominin skull is from this sedimentary horizon. It is approximately 10.5 m stratigraphically below the BKT-2 tephra, which outcrops in the immediate vicinity of A.L. 444 Single-crystal laser fusion (SCLF) 40Ar/39Ar ages for BKT-2 and Kada Hadar Tuff (KHT) bracket the geological age of A.L. 444-2 between 2.94 and 3.18 Myr (Kimbel et al., 1994; Walter, 1994; Semaw et al., 1997).

Background

Australopithecus afarensis is a fossil hominin species known from at least four East African Rift Valley sites ranging from northern Ethiopia in the north to northern Tanzania in the south and bridging the time period between approximately 3.6 and 3.0 million years ago (Ma). First identified in the late 1970s as the bipedal but craniodentally apelike rootstock from which later Australopithecus and Homo evolved (Johanson et al., 1978; Johanson and White, 1979), A. afarensis constituted the first substantial record of unequivocal human ancestors older than 3.0 million years (Myr). An array of more recently made discoveries have placed A. afarensis in a pivotal position in early hominin phylogeny, bracketed in time between, on the one hand, two temporally successive species, A. anamensis and Ardipithecus ramidus, that jointly extend the hominin record back to 4.4 Ma (M. Leakey et al., 1995, 1998; White et al., 1994, 1995), and, on the other hand, the earliest strong (stratigraphic) evidence for hominin lineage diversification, with the first known records of A. africanus (ca. 2.7 Ma) in southern Africa, and of A. aethiopicus (ca. 2.7 Ma) and A. garhi (2.5 Ma) in eastern Africa (Walker et al., 1986; Asfaw et al., 1999).2 The task of sorting out the relationships among all of these species hinges on the interpretation of A. afarensis itself, from its alpha taxonomy and phylogenetic role to its pattern of evolution over time. A prerequisite to achieving this goal is a more complete knowledge of the A. afarensis fossil record, narrowing gaps in our knowledge of anatomy and variation, as well as of distributions in space and time. On sample size alone, A. afarensis is the best-known hominin species in the eastern African fossil record. The vast majority of fossils in the A. afarensis hypodigm, some 360 specimens, or approximately 90% of the total, have been recovered at the Hadar site, from the 200+ meter sequence of silts, sands, and clays that comprise the Hadar Formation, which is exposed along the drainages of the Awash River in the Afar Depression of northern Ethiopia (Johanson et al., 1982a; Kimbel et al., 1994).