Australopithecus afarensis Scapular Ontogeny, Function, and the Role of Climbing in Human Evolution

See allHide authors and affiliations

Science  26 Oct 2012:
Vol. 338, Issue 6106, pp. 514-517
DOI: 10.1126/science.1227123


Scapular morphology is predictive of locomotor adaptations among primates, but this skeletal element is scarce in the hominin fossil record. Notably, both scapulae of the juvenile Australopithecus afarensis skeleton from Dikika, Ethiopia, have been recovered. These scapulae display several traits characteristic of suspensory apes, as do the few known fragmentary adult australopith representatives. Many of these traits change significantly throughout modern human ontogeny, but remain stable in apes. Thus, the similarity of juvenile and adult fossil morphologies implies that A. afarensis development was apelike. Additionally, changes in other scapular traits throughout African ape development are associated with shifts in locomotor behavior. This affirms the functional relevance of those characteristics, and their presence in australopith fossils supports the hypothesis that their locomotor repertoire included a substantial amount of climbing.

Scapular morphology corresponds closely with locomotor habits, often irrespective of phylogeny (17). However, our understanding of this important element in hominin evolution is limited by the paucity of scapular fossil remains. Upon its discovery, the right scapula associated with the juvenile Australopithecus afarensis skeleton from Dikika, Ethiopia (DIK-1-1, “Selam”) represented the most complete such fossil known for this well-documented early hominin species (8). Furthermore, comparison of this complete juvenile with adult australopith fossils promised to shed light on A. afarensis growth and development (8, 9). Continued preparation has since freed both scapulae from the matrix encasing much of the axial skeleton (Fig. 1).

Fig. 1

The DIK-1-1 scapulae; top and middle row images show dorsal, lateral, and ventral views of the recently prepared right and left scapulae, respectively. The left scapula was more recently prepared and some rib and vertebral elements are still adhering to it (this did not impede the measurements presented). Images along the bottom row are scapulae of comparably aged Homo, Pan, Gorilla, and Pongo individuals.

Before DIK-1-1’s discovery, the limited number of available fossil scapulae provided only tentative clues that the australopith shoulder was apelike (10). In addition, we lack a clear understanding of what the scapular morphology of the last common ancestor (LCA) of Pan and Homo looked like, making it difficult to determine whether australopiths retained apelike features from the LCA or if these features evolved independently (1114). Furthermore, limited information on the postcranial architecture, developmental pathways, and the manner in which behavioral variation contributes to morphological diversity among extant hominoids presents a challenge for reconstructing locomotor patterns in extinct taxa. Here, we describe further the DIK-1-1 scapulae and infer the locomotor behavior of Australopithecus through comparisons with other fossil hominins—including the new specimen from Woranso-Mille, Ethiopia (KSD-VP-1/1) (15)—and modern apes and humans (16). We track the ontogeny of scapular shape among extant hominoids to evaluate how juvenile scapular morphology compares with the adult form. We also evaluate functionally relevant characters throughout development to identify various genetic and epigenetic influences on hard-tissue morphology. These approaches consider how ontogenetic shifts in locomotor behavior (e.g., in Pan and Gorilla) influence scapular shape, providing context for evaluating the morphology of more fragmentary adult fossils and a more comprehensive view for inferring the locomotor implications of australopith shoulder anatomy.

The original analysis of the right DIK-1-1 scapula showed it to be most similar to that of juvenile Gorilla (8), but the two principal component axes describing its shape explained only ~7% of variance, drawing criticism (15). We performed two canonical variates analyses (CVAs) among juvenile and adult representatives of modern Homo, Pan, Gorilla, and Pongo, as well as DIK-1-1 and the immature H. ergaster (early H. erectus) scapula of the Turkana Boy (KNM-WT 15000) (17). In the first CVA, Homo and Pongo separated from Pan and Gorilla along the first root axis, which accounted for 70.3% of the variation; Pongo and Pan separated from Homo and Gorilla, respectively, along the second root axis (16.0%; Fig. 2A). The DIK-1-1 scapulae did not significantly differ from one another (P = 0.81) and were most similar to those of Gorilla juveniles (table S6); KNM-WT 15000 fell among the juvenile Homo data (Fig. 2A). The second CVA considered fewer variables to include the less complete KSD-VP-1/1, but did not distinguish the groups as effectively. Homo separated from the African apes along the first root, which explained 84.2% of the variation, and Pongo fell between the two groups with considerable overlap (Fig. 2B). The two DIK-1-1 scapulae did not significantly differ (P = 0.42) and fell among the Pongo and Gorilla data (table S8). KNM-WT 15000 was again most similar to Homo juveniles, whereas KSD-VP-1/1 fell near the intersection of adult Homo and Pongo.

Fig. 2

Canonical variates analysis (CVA) plots. (A) The first CVA considered three angular and 10 size-corrected, linear measures (table S5). Homo, KNM-WT 15000, and Pongo separated from Pan, Gorilla, and DIK-1-1 positively along the first root. Homo could be further distinguished from Pongo along the second root, as could Pan from Gorilla. The DIK-1-1 scapulae were most similar to those of Gorilla juveniles, whereas KNM-WT 15000 fell among the juvenile Homo data. (B) A second CVA considered five angular measures and also the less complete Woranso-Mille specimen, KSD-VP-1/1 (table S7). Although this CVA did not distinguish the extant taxa as effectively as the previous analysis, Homo separated from the African apes along the first root and Pongo fell intermediately between the two groups. The two DIK-1-1 scapulae did not significantly differ from one another and fell among the Pongo and Gorilla data. KNM-WT 15000 was most similar to Homo juveniles, whereas KSD-VP-1/1 fell near the intersection of adult Homo and Pongo. See also tables S6 and S8.

These multivariate analyses confirm that there are two distinct scapular shapes among living and extinct hominoids (tables S5 and S6). The scapulae of the African apes, and to a lesser extent, Pongo, differ from those of Homo in possessing more cranially oriented glenoid fossae, which may be an adaptation to more effectively distribute strain over the joint capsule during climbing and reaching when the upper limb is loaded (Fig. 3) (18). Suspensory great apes also possess obliquely oriented scapular spines (fig. S1) with superoinferiorly narrow infraspinous fossae and relatively broader supraspinous fossae (Fig. 4). The orientation of the scapular spine is associated with the relative size and shape of the dorsal scapular fossae and the corresponding muscles, as a more obliquely oriented spine provides a direct line of action for these muscles in preventing displacement of the humeral head during suspensory behaviors (1921).

Fig. 3

Box plots of ventral bar/glenoid angle across extant taxa and fossil individuals for juvenile, adolescent, and adult age groups. All of the Australopithecus fossils differ significantly from modern human scapulae and are more similar to the suspensory apes with cranially oriented shoulder joints. In contrast, modern humans, KNM-WT 15000, and LB6/4 display more laterally oriented joints.

Fig. 4

Box plots of relative infraspinous fossa breadth across extant taxa for juvenile, adolescent, and adult age groups and the DIK-1-1 and KNM-WT 15000 fossils. The two DIK-1-1 scapulae differ from one another, but both possess relatively narrow infraspinous regions that are more similar to those of the suspensory apes, whereas KNM-WT 15000 possesses a very broad infraspinous fossa that exceeded even the modern human range.

Other fragmentary Australopithecus fossils (A. afarensis: A.L. 288-1; A. africanus: Sts 7 and Stw 162) were included in bivariate comparisons (table S1). All australopiths possessed more cranially oriented shoulder joints relative to modern humans (Fig. 3) (17). Both DIK-1-1 scapulae fell within the Gorilla confidence interval (CI), whereas KNM-WT 15000’s shoulder joint was most similar to that of modern humans (Fig. 3 and table S9). Shoulder joint orientation does not significantly change in Pan or Gorilla throughout ontogeny, and it becomes slightly more cranially oriented in Pongo during the middle ontogenetic stages, but returns to the juvenile configuration in adulthood. In contrast, Homo shoulder joints become significantly more cranially oriented throughout ontogeny (P < 0.001), but remain more laterally oriented than those of the other hominoids at all stages (Fig. 3, fig. S2, and table S10). Starting from DIK-1-1, a humanlike ontogenetic pattern would imply that adult A. afarensis individuals should have more cranially oriented shoulder joints than those displayed by either A.L. 288-1 or KSD-VP-1/1. However, both juvenile and adult A. afarensis representatives have comparably oriented shoulder joints, suggesting that this trait remained relatively stable during ontogeny. This implies a developmental pattern for A. afarensis similar to that exhibited by the living African apes, but both developmental scenarios point toward a distinctly apelike shoulder joint configuration for A. afarensis throughout ontogeny (Fig. 3 and table S9).

It has been debated whether the cranially facing shoulder joint of A.L. 288-1 (Lucy) is an allometric result of the specimen’s diminutive size, rather than an indicator of arboreal adaptations (15, 22, 23). Our results support the functional inference: Both Sts 7 and Stw 162 are larger than A.L. 288-1, yet possess more cranially oriented shoulder joints. Additionally, the Lucy-sized LB6/4 scapula (H. floresiensis) has a “hyperhuman,” laterally facing shoulder joint (Fig. 3 and table S9) [(24), p. 725; (25)]. Moreover, the youngest modern humans had the most laterally positioned shoulder joints, further distinguishing them from juvenile great apes and DIK-1-1 (Fig. 3 and table S9). These findings contradict the hypothesis that cranially oriented shoulder joints are a by-product of small size. Thus, we conclude that A. afarensis possessed an apelike, cranially oriented scapula, distinct from the configuration seen in modern and fossil Homo.

Both DIK-1-1 scapular spines are oriented significantly more obliquely than in Homo, with angle values within the Pongo CI. In contrast, KNM-WT 15000 has a significantly more transversely oriented spine that even exceeded the modern human range (fig. S1 and table S9). The KSD-VP-1/1 scapula is described as having a more transversely oriented spine (15), whereas the spine of A.L. 288-1 is more obliquely oriented, falling just above the Pongo CI (table S9). The Sts 7 spine is the most oblique of the australopiths and fell within the Gorilla CI (table S9). Scapular spine orientation does not change significantly in the great apes throughout ontogeny, but modern human scapular spines shifted significantly more obliquely (P < 0.01; fig. S3 and table S10). As observed for shoulder joint orientation, the relative orientation of juvenile and adult A. afarensis scapular spines might be partially explained by a more apelike ontogenetic trajectory than that exhibited by modern humans (figs. S1 and 3 and tables S9 and S10).

Scapular spine orientation is a principal determinant of dorsal scapular fossa shape (21). In particular, the infraspinatus muscle has been shown to be primarily involved in shoulder joint stabilization during suspensory activities (20). DIK-1-1’s infraspinous fossae are narrow relative to glenoid size and most similar to those of Gorilla and Pongo juveniles, whereas KNM-WT 15000 has an extremely broad fossa (Fig. 4 and tables S1 and S9). Supraspinous breadth generally increases in all taxa throughout ontogeny, whereas infraspinous breadth does not show any significant increase from stage to stage in Homo or Pongo (table S10). In contrast, infraspinous breadth increases significantly throughout both Pan and Gorilla ontogeny (P < 0.03; fig. S4 and table S10). The ratio of supraspinous:infraspinous breadth (SIB) increases throughout ontogeny in Homo, does not significantly change in Pongo, but significantly decreases in both Pan and Gorilla. Given the increase in relative supraspinous breadth in Pan and Gorilla, the decrease in the SIB ratio similarly highlights the relative increase in infraspinous breadth (table S10).

These developmental patterns further inform the link between shoulder morphology and locomotor behavior. Arboreal hominoids possess narrower infraspinous regions, in contrast to the broad fossae displayed by modern humans (6, 19, 26). Further, the increase in infraspinous breadth during Pan and Gorilla ontogeny corresponds with a behavioral shift from a principally arboreal lifestyle at younger ages to an adult locomotor repertoire predominated by terrestrial knuckle-walking (27, 28). The infraspinatus muscle is consistently recruited to stabilize the shoulder joint during both suspensory and knuckle-walking behaviors in chimpanzees (20, 29), so the change in African ape infraspinous fossa shape might represent an adaptive optimization of the scapular blade. A narrow infraspinous region with an obliquely oriented scapular spine is a more effective configuration for infraspinatus’ role in stabilizing the shoulder joint during suspensory activities (19, 20). In contrast, an enlarged infraspinous fossa allows the muscle to pass broadly behind the humeral head, which might facilitate joint integrity when the arm is loaded from below as individuals engage more regularly in knuckle-walking activities (19).

The change in infraspinous fossa shape during African ape ontogeny may represent a response to the changing loading regimes of a dynamic locomotor repertoire. This interpretation is supported by experimental evidence, where differences in shoulder activity during growth corresponded with significant infraspinous fossa shape changes in mice (30). Thus, in addition to a more cranially oriented shoulder joint and an oblique scapular spine, we propose that DIK-1-1’s relatively narrow infraspinous region is a functionally meaningful characteristic. This configuration further highlights its overall apelike appearance while also distinguishing it from juvenile modern humans and the considerably more derived KNM-WT 15000 adolescent.

Comparing the DIK-1-1 scapulae to those of adult conspecifics suggests that growth of the A. afarensis shoulder may have followed a developmental trajectory more like that of African apes than modern humans. This conclusion is consistent with evidence purporting that A. afarensis dental development was also apelike (31). Additionally, behavioral changes that occur throughout African ape ontogeny could be linked with morphological shifts, indicating that some scapular blade characteristics track locomotor habits, even during an organism’s lifetime. The apelike appearance of the most complete A. afarensis scapulae strengthens the hypothesis that these hominins participated in a behavioral strategy that incorporated a considerable amount of arboreal behaviors in addition to bipedal locomotion.

Supplementary Materials

Materials and Methods

Supplementary Text

Figs. S1 to S4

Tables S1 to S10

References (3341)

References and Notes

  1. The new A. sediba shoulder blade from Malapa, South Africa (32) was not included in the present study, but represents another significant addition to the scapular fossil record.
  2. Materials and methods are available as supplementary materials on Science Online
  3. Acknowledgments: We thank C. Kiarie and the staff at the National Museum of Ethiopia for help during the preparation of these fragile fossils. We greatly appreciate critical comments offered by B. Richmond, B. Wood, R. Bernstein, M. Hamrick, L.P. Hernandez, and three anonymous reviewers on this manuscript and A. Gordon for analytical assistance. We thank D. Hunt, L. Gordon, E. Westwig, I. Tattersall, G. Garcia, J. Chupasko, M. Omura, Y. Haile-Selassie, L. Jellema, M. Harman, A. Gill, E. Mbua, S. Muteti, M. Yilma, P.V. Tobias, B. Zipfel, S. Potze, and T. Perregil for coordinating museum visits. We also acknowledge the NSF IGERT grant (9987590), NSF Doctoral Dissertation Improvement Grant (BCS-0824552), NSF (BCS-0914687), The Leakey Foundation, the Wenner-Gren Foundation, The George Washington University, Midwestern University, and the California Academy of Sciences for funding support. This paper was written by D.J.G. and Z.A. Fossil data were collected and described by D.J.G. and Z.A. Extant primate data were collected and analyzed by D.J.G. The data reported in this paper are summarized in the supplementary materials; raw data are available on request to D.J.G.
View Abstract

Navigate This Article