1. Evolutionary Biology
Download icon

Homoplasy in the evolution of modern human-like joint proportions in Australopithecus afarensis

  1. Anjali M Prabhat
  2. Catherine K Miller
  3. Thomas Cody Prang
  4. Jeffrey Spear
  5. Scott A Williams
  6. Jeremy M DeSilva  Is a corresponding author
  1. Anthropology, Dartmouth College, United States
  2. Ecology, Evolution, Ecosystems, and Society, Dartmouth College, United States
  3. Department of Anthropology, Texas A&M University, United States
  4. Center for the Study of Human Origins, Department of Anthropology, New York University, United States
  5. New York Consortium in Evolutionary Primatology, United States
Research Article
  • Cited 0
  • Views 1,198
  • Annotations
Cite this article as: eLife 2021;10:e65897 doi: 10.7554/eLife.65897

Abstract

The evolution of bipedalism and reduced reliance on arboreality in hominins resulted in larger lower limb joints relative to the joints of the upper limb. The pattern and timing of this transition, however, remains unresolved. Here, we find the limb joint proportions of Australopithecus afarensis, Homo erectus, and Homo naledi to resemble those of modern humans, whereas those of A. africanus, Australopithecus sediba, Paranthropus robustus, Paranthropus boisei, Homo habilis, and Homo floresiensis are more ape-like. The homology of limb joint proportions in A. afarensis and modern humans can only be explained by a series of evolutionary reversals irrespective of differing phylogenetic hypotheses. Thus, the independent evolution of modern human-like limb joint proportions in A. afarensis is a more parsimonious explanation. Overall, these results support an emerging perspective in hominin paleobiology that A. afarensis was the most terrestrially adapted australopith despite the importance of arboreality throughout much of early hominin evolution.

Introduction

Among extant hominoids, modern humans (Homo sapiens; hereafter, ‘humans’) are the only habitually bipedal species. Adaptation to upright walking and running in humans is evidenced by the presence of a host of postcranial morphologies functionally related to saving mechanical and metabolic energy (Lovejoy, 1988; Bramble and Lieberman, 2004; Pontzer, 2017). These include relatively long legs, arched feet (Venkadesan et al., 2020), and adaptations to protect the joints of the lower limbs from excessive stress by increasing their surface areas relative to the mass of the body (Ruff, 1988; Jungers, 1988; Lovejoy, 2005). These morphological traits are most strongly expressed in recent modern humans, which are nearly exclusively terrestrial in their locomotor adaptation. In contrast, the body plan of extant, nonhuman apes (hereafter, simply ‘apes’) reflects an adaptation to orthogrady and suspension (Keith, 1923; Gebo, 1996; Williams and Russo, 2015) with relatively long arms and large upper limb joints (Ruff, 1988), elongated, curved phalanges (Deane and Begun, 2008), and other morphological features suitable for arboreal behaviors (Gebo, 1996). Although chimpanzees, bonobos, and gorillas possess adaptations to terrestrial quadrupedalism (Gebo, 1996), including their knuckle-walking hand posture and heel-strike plantigrade foot posture (Gebo, 1992; Prang, 2019), they retain traits linked to an ancestry characterized by vertical climbing and suspension in some form (Gebo, 1996).

The relatively larger upper limb joints of apes compared to humans reflect disparate joint loading regimes associated with forms of quadrupedalism, climbing, and suspension (Gebo, 1996). Additionally, the relatively larger surface areas of the convex side of conarticular joints may contribute to increased range of motion, providing benefits to the arboreal locomotor performance of apes (Ruff, 1988; Godfrey et al., 1991; Hammond, 2014; Prang, 2016). Therefore, the relative size of postcranial joints and the relationship between the joints of the upper and lower limbs are important correlates of positional and locomotor behavior among hominoids (Ruff, 1988; Jungers, 1988; Godfrey et al., 1995; McHenry, 1992; McHenry and Berger, 1998; Green et al., 2007; Haeusler and McHenry, 2007).

The timing and pattern of the complicated, nonlinear evolutionary loss of adaptations to arboreality and the transition to a form of nearly exclusive terrestrial bipedalism among hominins has been debated for decades (Stern, 2000; Ward, 2002). The study of limb joint proportions initially focused on the preserved partial skeletons A.L. 288-1 (Australopithecus afarensis) and StW 431 (Australopithecus africanus) (McHenry and Berger, 1998; Green et al., 2007), along with OH 62 and KNM-ER 3735 (Homo habilis; Haeusler and McHenry, 2007; Johanson et al., 1987; Leakey et al., 1987). Previous studies have shown that the geologically younger A. africanus possessed relatively large upper limb joints and metaphyseal dimensions in comparison to A. afarensis (McHenry and Berger, 1998; Green et al., 2007). The OH 62 and KNM-ER 3735 partial skeletons are more fragmentary, but morphological comparisons of external morphology (Haeusler and McHenry, 2007; Hartwig-Scherer and Martin, 1991) and cross-sectional geometry (Ruff, 2009) suggest that the upper limbs of H. habilis bore similarities to extant chimpanzees and gorillas, implying greater reliance on forelimb-dominated behaviors, and may show a similar pattern to that observed in A. africanus (McHenry and Berger, 1998; Green et al., 2007; Haeusler and McHenry, 2007). The observed pattern of joint size proportions among extant hominoids implies that the relatively larger lower limb joints of A. afarensis are a reflection of increased terrestriality compared to A. africanus.

Cladistic analyses of hominin phylogeny based on craniodental characters consistently position A. africanus as more closely related to Homo than is A. afarensis (Dembo et al., 2016; Strait et al., 2015). Therefore, either (1) A. afarensis and H. sapiens independently evolved relatively larger lower limb joints (i.e., their similarities are homoplastic), (2) A. africanus and H. habilis evolved more ape-like joint proportions from an ancestor with more human-like limb proportions (i.e., the similarities between A. africanus, H. habilis, and apes are homoplastic), or (3) A. afarensis is more closely related to Homo than are A. africanus and H. habilis (Figure 1). Limited taxonomic sampling of fossil hominins in previous studies has rendered these competing scenarios exceedingly difficult to differentiate. Over the past few decades, however, the recovery of new fossil hominin partial skeletons preserving both upper and lower limb joints has provided an expanded sample that can be used to evaluate these hypotheses more rigorously (Table 1). Here, we re-examine the upper and lower limb joint proportions of multiple species of Australopithecus, Paranthropus, and Homo to evaluate these long-standing alternative hypotheses for patterns of postcranial evolution in hominins.

Alternative hypotheses to explain the pattern of limb joint proportions observed in the human fossil record.

Previous work interpreted the human-like ratio of upper to lower limb joint size (relative limb size index [RLSI]) in Australopithecus afarensis to indicate either (A) homoplasy between A. afarensis and Homo sapiens or (B) evolutionary reversals to a more ape-like body form in A. africanus and H. habilis.

Table 1
Fossil hominin and extant hominoid measurements.
SpecimenTaxonGHBURFSubATSac
Homo sapiens29.5 ± 2.7 (N = 67)42.3 ± 3.5 (N = 67)59.1 ± 4.7 (N = 52)21.8 ± 1.9 (N = 51)21.7 ± 2.1 (N = 51)44.5 ± 3.5 (N = 67)28.6 ± 2.1 (N = 52)51.2 ± 3.5 (N = 67)28.2 ± 2.0 (N = 66)38.6 ± 3.1 (N = 67)
Pan26.6 ± 2.4 (N = 113)38.2 ± 3.1 (N = 113)62.5 ± 5.5 (N = 95)22.5 ± 2.8 (N = 95)24.5 ± 1.8 (N = 94)32.8 ± 2.5 (N = 120)25.0 ± 2.0 (N = 98)38.6 ± 3.3 (N = 116)18.1 ± 34.1 (N = 116)28.4 ± 3.9 (N = 109)
Gorilla39.2 ± 5.5 (N = 119)54.9 ± 7.3 (N = 122)93.0 ± 13.0 (N = 94)33.4 ± 5.7 (N = 89)31.7 ± 4.4 (N = 91)46.6 ± 5.9 (N = 125)35.7 ± 4.9 (N = 93)53.1 ± 6.9 (N = 114)24.9 ± 5.6 (N = 108)37.7 ± 5.8 (N = 102)
Pongo29.2 ± 3.6 (N = 47)40.0 ± 4.8 (N = 49)63.9 ± 7.1 (N = 45)22.1 ± 3.2 (N = 46)22.8 ± 2.8 (N = 46)32.9 ± 4.0 (N = 49)20.8 ± 2.6 (N = 45)39.0 ± 4.7 (N = 49)18.0 ± 2.7 (N = 46)28.0 ± 4.1 (N = 43)
Hylobatids13.2 ± 1.7 (N = 62)18.5 ± 2.3 (N = 66)28.0 ± 3.1 (N = 66)11.0 ± 1.5 (N = 66)12.6 ± 1.5 (N = 69)16.4 ± 2.1 (N = 65)11.0 ± 1.5 (N = 65)20.6 ± 3.0 (N = 66)7.5 ± 1.0 (N = 59)14.7 ± 2.3 (N = 58)
A.L. 288-1A. afarensis21.628.941.116.115.128.620.837.018.025.3
KSD-VP-1/1A. afarensis30.158.849.032.4
DIK-1-1A. afarensis13.513.1
StW 573A. prometheus (?); A. africanus25.931.354.024.321.935.224.543.018.0
StW 431A. africanus59.025.721.945.027.5
MH1A. sediba57.018.933.023.222.0
MH2A. sediba24.630.152.417.418.832.718.123.6
BOU-VP-12/1A. garhi(?)21.423.7
TM 1517P. robustus54.022.018.9
OH 80P. boisei26.326.5
KNM-ER 1500P. boisei21.420.224.219.2*
KNM-ER 1503/1504P. boisei57.030.622.2
KNM-ER 3735H. habilis55.020.025.3
KNM-WT 15000H. erectus27.631.655.019.046.028.825.0*33.6
LES 1H. naledi33.216.136.024.224.5
LB 1H. floresiensis19.531.022.136.019.5
  1. *Estimated from tibial plafond width.

  2. G: glenoid size (geomean of SI height and AP width); H: humeral head diameter (geomean of SI height and AP width); B: humeral biepicondylar breadth; U: ulna olecranon width; R: radial head diameter; F: femoral head diameter; Sub: femoral subtrochanteric width (geomean of ML width and AP breadth); A: acetabulum height; T: talar mediolateral width; Sac: sacral size (geomean of ML width and AP breadth). ±1 standard deviation is given with sample size (N=#).

Results

We quantified limb joint proportions per individual using the relative limb size index (RLSI) (Green et al., 2007). The RLSI is the logged ratio of geometric means calculated from upper (forelimb) and lower (hindlimb) limb measurements and quantifies whether a given specimen has relatively larger forelimb or hindlimb joints (Green et al., 2007). We calculated a series of RLSIs to accommodate the differential preservation of postcranial elements among the 16 hominin partial skeletons sampled here. When the full upper to lower limb dataset is used, there is clear separation between humans, with their proportionally larger lower limbs, and modern apes, with their proportionately larger upper limbs, with no overlap. Importantly, there remains clear separation between humans and great apes in cases where truncated datasets were used to quantify the limb joint proportions of less complete hominin skeletons. The ape data, however, do not always accord with degree of arboreality (hylobatid > Pongo > Pan > Gorilla).

5 of the 16 partial hominin skeletons are human-like in their limb joint proportions (Figure 2, Figure 2—figure supplements 126). The RLSI of A.L. 288-1 (Lucy) is far outside the ape range and falls squarely within the range of modern humans (Figure 2, Figure 2—figure supplement 1). Likewise, the larger, presumed male A. afarensis partial skeleton KSD-VP-1/1 is positioned within the human range, though it overlaps with the low end of the hylobatid distribution (Figure 2—figure supplement 2). The infant partial skeleton of A. afarensis (DIK-1-1), as well as Lucy, has a human-like glenoid:talus ratio (Figure 2—figure supplement 3). KNM-WT 15000 (Homo erectus) has even larger relative lower limb joint proportions than the humans sampled in this study and is well outside the ape range (Figure 2—figure supplement 4). LES 1 (Homo naledi) falls within the human interquartile range, outside any modern ape distribution (Figure 2—figure supplement 5).

Figure 2 with 26 supplements see all
Relative limb size index (RLSI) in modern apes, humans, and fossil hominins.

Notice that A.L. 288-1 (Lucy) falls within the modern human distribution for RLSI no matter which combination of upper to lower limb joint proportions is examined (A–D). (A) Human-like upper to lower limb joint proportions remain human-like on the basis of preserved elements in a second partial skeleton of A. afarensis, KSD-VP-1/1. However, all other partial skeletons of Australopithecus (A–C), Paranthropus (B) and early Homo (C) are more ape-like. A high, ape-like RLSI is present even in the late Pleistocene hominin H. floresiensis (D).

All of the other hominin skeletons studied fall outside the human range, indicating that they are more ape-like in their joint proportions (Figure 2). StW 431 (A. africanus) has limb proportions positioned within the hylobatid interquartile range and within the distributions of Pan, Gorilla, and Pongo (Figure 2—figure supplement 6). MH1 (juvenile Australopithecus sediba) falls within the interquartile range of hylobatids and Pan and within the ranges of Gorilla and Pongo (Figure 2—figure supplement 7). MH2 (adult A. sediba) occupies the space between great apes and humans, positioned only within the range of hylobatids (Figure 2—figure supplement 8). StW 573 is similar to MH2 in having relatively larger upper limb joints than modern humans but smaller than extant apes, positioned only near a hylobatid outlier (Figure 2—figure supplement 9).

Partial skeletons attributed to Paranthropus all possess ape-like joint proportions. TM 1517 (Paranthropus robustus) and KNM-ER 1500 (Paranthropus boisei) fall within the ranges of all four extant apes (Figure 2—figure supplements 10 and 11). OH 80 (P. boisei) falls within the interquartile range of Pan and the range of Gorilla and hylobatids (Figure 2—figure supplement 12). Associated fossils KNM-ER 1503/1504 (tentatively attributed to P. boisei) have joint proportions within the interquartile range of Pan, Gorilla, and hylobatids (Figure 2—figure supplement 13).

BOU-VP-12/1 (Australopithecus cf. garhi) falls squarely within the Gorilla interquartile range and within the lower range of Pan (Figure 2—figure supplement 14). There is a single human outlier overlapping with the limb joint proportions of BOU-VP-12/1.

KNM-ER 3735 (H. habilis) has joint proportions in the hylobatid interquartile range and within the ranges of all extant great apes (Figure 2—figure supplement 15). The joint proportions of LB 1 (Homo floresiensis) fall within the ranges for all of the apes, though there is a single human outlier similar in joint proportions to LB 1 (Figure 2—figure supplement 16).

Parsimony reconstructions suggest that the human-like limb joint proportions in A. afarensis and modern humans are homoplastic, regardless of the phylogenetic hypothesis used (Figure 3). The only phylogenetic hypothesis that does not require either homoplasy or multiple reversals is one in which A. afarensis is more derived than H. habilis, H. floresiensis, and all other species of Australopithecus and Paranthropus examined in this study. Given that an A. afarensis-later Homo clade (to the exclusion of early Homo) has not been supported by any phylogenetic analysis, we consider this last scenario unlikely in the extreme.

Relative limb size index (RLSI; high in black; low in red) for the taxa examined in this study.

The phylogenies in (A) and (B) are from Dembo et al., 2016 (A) and Mongle et al., 2019 (B). The phylogenies in (CE) presented above are informed by various hypotheses about the relationships of Australopithecus and Paranthropus taxa that have been published but not recovered in formal phylogenetic analyses. These include the hypothesis that Australopithecus garhi is a unique ancestor of Homo (C; Asfaw et al., 1999), that Australopithecus sediba is a unique ancestor of Homo (D; Berger et al., 2010; Irish et al., 2013), and the hypothesis that Paranthropus is actually polyphyletic (E; topology based on hypothetical tree presented in Wood and Schroer, 2017). A hypothetical phylogeny in which Australopithecus afarensis is more derived than two species of Homo as well as all other Australopithecus and Paranthropus species (such as shown in F) would need to be correct for the pattern of RLSI in hominins to be best explained as anything other than homoplasy between A. afarensis and some later Pleistocene Homo.

Discussion

Apes have relatively larger upper limb than lower limb joints as reflected by their higher RLSI than modern humans (Green et al., 2007). With musculoskeletal anatomies adapted for climbing and suspension, apes possess larger upper limb muscles and joints with greater surface areas, which has the effect of limiting excessive stresses and strains arising from large joint reaction forces. Enlarged upper limb joint surface areas in apes may also contribute to increased ranges of motion (e.g., at the glenohumeral joint). In contrast, humans are characterized by relatively larger lower limb joints, which act to reduce stresses and strains on the joints and nonrenewable cartilage of the hip, knee, and ankle arising from repetitive high-magnitude ground and joint reaction forces during heel-striking bipedal walking and running. This pattern accords with expectations based on the posture and locomotion of apes and humans. Apes possess heavily built upper limbs associated with orthograde climbing and suspension, whereas modern humans have robust lower limbs adapted to terrestrial bipedalism. This morphological pattern provides a framework for interpreting the functional and evolutionary implications of joint proportions in fossil hominins.

It is noteworthy, however, that the ape RLSI did not always align with degree of arboreality (hylobatid > Pongo > Pan > Gorilla). As reported elsewhere (Gordon et al., 2020), the preserved anatomical elements used to calculate RLSI can impact where a taxon is positioned along a locomotor continuum. The relatively narrow great ape sacrum, hypothesized to facilitate entrapment of the lumbar vertebrae and stiffen the lower back during climbing, further increases the RLSI of Pan, Gorilla, and Pongo relative to the hylobatids (Figure 4—figure supplements 14). Additionally, while RLSI calculations that included the radial head and ulnar trochlear width separate the ape species by locomotor mode, those that include the glenoid size and the biepicondylar breadth do not. In fact, a post hoc examination of limb joint scaling found that while great apes and humans exhibit isometric scaling of the glenoid and humeral biepicondylar breadth relative to femoral head diameter, the hylobatids scale with negative allometry (glenoid m = 0.75; biepicondylar breadth m = 0.76). It is likely that the differing sizes of these apes and the functional demands on the limb joints in arboreal apes across this size range are driving some of the unexpected RLSI results reported here.

Our study provides a fresh perspective on alternative hypotheses for the evolution of limb joint proportions introduced by previous workers (McHenry and Berger, 1998; Green et al., 2007; Haeusler and McHenry, 2007). The reconstruction of patterns of hominin evolution relies on phylogeny, and, since the early adoption of cladistics, no quantitative analysis of hominin phylogeny has recovered a sister taxon relationship between A. afarensis and Homo (Dembo et al., 2016; Strait et al., 2015). The recovery of purported Homo fossils significantly predating the appearance of A. africanus and A. sediba may falsify hypotheses of exclusive ancestry and descent (Du and Alemseged, 2019). However, the consistent placement of A. africanus and A. sediba near Homo implies that they share a more recent common ancestor than do Homo and A. afarensis, despite their temporal and geographic distance (Dembo et al., 2016; Berger et al., 2010; Irish et al., 2013; Pickering et al., 2011). The inclusion of Paranthropus and early Homo fossils here helps alleviate the evolutionary implications of uncertainty surrounding the phylogenetic positions of Australopithecus species. The homology of a low RLSI in A. afarensis and later Homo can only be explained by an increasingly large number of evolutionary reversals. The independent evolution of similar limb joint proportions in A. afarensis and later Homo is a more parsimonious interpretation of the data.

There exists one additional piece of evidence that supports our interpretation that the low RLSI of A. afarensis and modern humans is homoplastic. Interestingly, the low RLSI of A. afarensis was achieved with a different morphological pattern compared to H. erectus. We found that H. erectus possessed a relatively smaller sacral body (like A. africanus and A. sediba) but a large femoral head relative to the upper limb than do modern humans; however, A. afarensis possessed a relatively small femoral head and large sacral body (Figure 4). This finding may further imply parallel evolution in limb joint proportions between A. afarensis and H. sapiens. The relatively small sacral body of the female H. erectus sacrum from Gona (Simpson et al., 2008, personal observation) demonstrates that these results are not the result of the juvenile status of KNM-WT 15000.

Figure 4 with 4 supplements see all
Additional evidence for homoplasy in relative limb size index (RLSI) between A.afarensis and H. sapiens is presented here.

Only femoral head diameter, sacral width, and humeral biepicondylar width are considered in this analysis and all extant apes and hominin fossils are shown relative to the modern human condition (vertical-colored stripes). Horizontal bars are 95% confidence intervals (with those of human highlighted in vertical colored bars). Note that as in apes, A. africanus (StW 431) and A. sediba (MH1 and MH2) have a relatively large humeral biepicondylar width and relatively small sacrum. A. afarensis (A.L. 288-1 and KSD-VP-1/1) has a slightly larger biepicondylar breadth and sacral width with a slightly smaller femoral head relative to modern humans, though as already demonstrated, the overall RLSI is human-like. However, while H. erectus (KNM-WT 15000) also possesses a human-like RLSI, it is accomplished in a different anatomical manner. Notice that the colored dots (blue and green) are reversed in H. erectus relative to both A. afarensis and H. sapiens, meaning that in H. erectus the sacrum is smaller than expected (as in other australopiths) and the femoral head larger than expected. The BSN49/P27 H. erectus pelvis possesses a similarly small sacrum, indicating that this result is not solely a result of the juvenile status of KNM-WT 15000.

The morphology and functional anatomy of the axial skeleton, pelvis, and lower limb display unambiguous evidence for bipedal posture and locomotion in Australopithecus and later hominins. Furthermore, the presence of traits potentially signifying the importance of arboreality among fossil hominins does not necessarily imply reduced bipedal competency. However, the distributions of RLSI data (Figure 2—figure supplements 1726), along with observations of other regional anatomies, imply differences among hominins in their adaptation to terrestrial, heel-striking bipedality. A. afarensis has relatively larger lower limb joints than any other early hominin currently known and possesses features of the foot and ankle that imply bipedal performance capabilities exceeding those of later early hominins. These traits include a more robust calcaneal tuber, a flatter subtalar joint, and a more plantarly oriented fourth metatarsal diaphysis and talonavicular joint (reviewed in DeSilva et al., 2019). The available morphological evidence suggests that, compared to other Plio-Pleistocene hominins, A. afarensis was better able to withstand the stresses and strains induced by the repetitive loading of the lower limb in frequent terrestrial bipedalism.

Over the past three decades, significant emphasis has been placed on the retention of ape-like characters in Australopithecus and Paranthropus since they could have been maintained through stabilizing selection if arboreality was a significant part of their positional repertoires (Stern, 2000). However, many researchers have repeatedly noted the difficulty of distinguishing the effects of stabilizing selection from those of evolutionary inertia (or ‘lag’). This critique is rooted in a maximum likelihood, character-based cladistic framework, which implicitly excludes information about the evolutionary process (e.g., evolutionary rates as a function of time). In other words, primitive retentions have the same meaning across varying temporal ranges in a traditional cladistic framework. The presence of presumed primitive, ape-like features in late Australopithecus and early Homo c. ~2 Ma implies that evolutionary processes, whether neutral (i.e., genetic drift) or non-neutral (i.e., directional selection), had not yet substantially modified them over a 4- to 5 million-year period given current estimates of the Pan-Homo divergence date. Therefore, in our view, primitive retentions in Australopithecus, Paranthropus, and Homo can be meaningful when interpreted within the context of the evolutionary process.

Regardless of assumptions underlying evolutionary processes, primitive retentions in late Australopithecus and early Homo occur alongside indirect evidence for arboreal activity in their trabecular bone density patterns and long bone diaphyseal properties. A recent study of trabecular bone density of the non-pollical metacarpal heads of A. sediba showed a close morphometric affinity with extant orangutans, despite having human-like hand proportions (Dunmore et al., 2020). The external structure and internal trabecular morphology of the A. sediba hand are consistent with the use of forceful metacarpophalangeal joint flexion, which is a requisite of forelimb-dominated, below-branch locomotion (Dunmore et al., 2020). The purportedly more ape-like limb joint and length proportions of the OH 62 partial skeleton are supported by a more chimpanzee-like humeral cross-sectional geometry, implying that the H. habilis upper limb was heavily built (Ruff, 2009). The femoral head trabecular bone density pattern of StW 311 from Sterkfontein attributed either to P. robustus or Homo sp. implies the more habitual use of flexed hip postures, which occurs during climbing (Georgiou et al., 2020).

In light of the congruence between the functional interpretations derived from the external and internal morphology of hominin postcranial fossils, we consider the limb joint proportions data presented here, and in previous studies, to be a reliable indicator of adaptation to arboreal locomotion. The relatively larger hindlimb joints of A. afarensis and later Homo are consistent with a more pronounced terrestrial component of their positional repertoires, whereas the relatively larger upper limb joints of most Australopithecus, Paranthropus, and early Homo individuals indicate a more pronounced arboreal component. The low RLSI of A. afarensis does not imply a lack of arboreal activity given the evidence it climbed trees (Stern and Susman, 1983; Ruff et al., 2016; Green and Alemseged, 2012; DeSilva et al., 2018b). The glenoid:talus proportions of the DIK-1-1 juvenile are human-like and distinct from this ratio in modern apes or in other species of Australopithecus, despite the presence of a Gorilla-like scapula (Green and Alemseged, 2012) and medial cuneiform (DeSilva et al., 2018b) indicating increased arboreal competency among juveniles. Furthermore, although A. afarensis had a more modern human-like RLSI, it possessed relatively longer arms and shorter legs than modern humans (Holliday, 2012), with more chimpanzee-like humeral-femoral strength proportions (Ruff et al., 2016), suggesting that the body plan and positional repertoire of A. afarensis were unique and unlike any living taxon.

We acknowledge that one of the limitations of our study includes uncertainty surrounding the taxonomic affinity of partial skeletons such as KNM-ER 1500. However, our evolutionary interpretation would not be altered by accepting the alternative interpretation of KNM-ER 1500 as H. habilis. Our finding that Paranthropus had a high RLSI is consistent with recent evidence suggesting the presence of heavily built, somewhat ape-like, distal humeri in this genus (Lague et al., 2019). Additionally, for some specimens (e.g. KNM-ER 1503/1504, TM 1517), uncertainty remains about whether they represent a single partial skeleton, but recent work supports the single-individual hypothesis for TM 1517 (Cazenave et al., 2020). Finally, two skeletons used in this study are juveniles (MH1 and KNM-WT 15000), though they are near skeletal maturation. Despite their juvenile status, MH1 has more ape-like limb joint proportions, whereas KNM-WT 15000 is more modern human-like. Fortunately, the limb joint proportions of A. sediba are represented by the MH2 adult specimen. A future study could evaluate the relative size and ontogenetic scaling of limb joint proportions across hominoids to evaluate the morphometric and functional affinities of juvenile hominin specimens in greater detail (e.g., DIK-1-1).

Despite these minor caveats, the pattern of limb joint proportions in hominins is clear. Partial skeletons belonging to A. afarensis, H. erectus, and H. naledi are human-like, with a low RLSI, whereas all others are more ape-like, with a high RLSI. These data strongly suggest that A. afarensis was a committed terrestrial biped that evolved adaptations to limit the larger lower limb stresses and strains characteristic of bipedal locomotion, as also occurred in later Pleistocene Homo (but not H. floresiensis). Other species of Australopithecus, Paranthropus, and early members of the genus Homo appear to have been less committed terrestrial bipeds that retained adaptations to the arboreal milieu. Overall, our analysis provides resolution on a long-standing hypothesis that A. afarensis evolved its low RLSI independently of some later Pleistocene hominins.

Materials and methods

The comparative sample includes hylobatids (all four genera are represented, total N = 69; Hoolock: N = 11; Hylobates: N = 39; Nomascus: N = 5; Symphalangus: N = 14), Pongo spp. (total N = 50; Pongo abelii: N = 13; Pongo pygmaeus, N = 37), Gorilla spp. (total N = 131; Gorilla beringei: 55; Gorilla gorilla: N = 76), Pan spp. (total N = 124; Pan paniscus: N = 26, Pan troglodytes: N = 98), and H. sapiens (N = 67). We measured adult specimens from the Harvard Museum of Comparative Zoology, the American Museum of Natural History, and the Cleveland Museum of Natural History. Data for hominin partial skeletons (N = 16) were acquired from published literature and/or measured on original fossils using Mitutoyo calipers. In some cases, casts from the Dartmouth Paleoanthropology lab were used to confirm published measurements.

Seven measurements were taken at the shoulder and elbow joints to represent the upper body (Figure 5): scapular glenoid superoinferior (SI) height and maximum mediolateral (ML) width, humeral head SI height and anteroposterior (AP) width, humeral biepicondylar breadth, ulnar olecranon width, and radial head semimajor axis diameter. Seven measurements were taken at the hip, lumbosacral, and talocrural joints to represent the lower body (Figure 5): acetabulum SI height, SI femoral head diameter, femoral subtrochanteric ML width and AP breadth, sacral (S1) body maximum ML width and AP diameter at midline, and width of the talar trochlear apex taken at the midpoint. Mean scapular glenoid and humeral head joint size was calculated using a geometric mean of the SI and ML dimensions. Mean femoral subtrochanteric and sacral body size was calculated using a geometric mean of the AP and ML dimensions.

Linear measurements were taken on the upper limb (top) and lower limb (bottom).

Limb joint proportions were calculated using the relative limb size index, which is the logged ratio of geometric means calculated from forelimb and hindlimb measurements shown above (Green et al., 2007).

We quantified limb joint proportions per individual using the RLSI (Green et al., 2007). The RLSI is the logged ratio of geometric means calculated from forelimb and hindlimb measurements (Green et al., 2007). Geometric means are typically used as size proxies over arithmetic means because they accommodate measurements with different ranges, which is common for morphometric measurements, and therefore normalize the weight of individual measurements (Jungers et al., 1995). Logging the ratio is necessary because ratios of normally distributed data cannot be normally distributed and thus violate the assumptions of statistical tests (e.g., Green et al., 2007; Smith, 1999). The measurements used to calculate the limb joint proportions of hominin specimens varied depending on which measurements were preserved in the fossil (Supplementary file 1). Separate comparative analyses including different ratios were conducted for each hominin partial skeleton to maximize the fossil sample.

To visualize evolutionary scenarios, we conducted ancestral states using parsimony on a variety of phylogenetic hypotheses. These hypotheses included both formal cladistic analyses (Dembo et al., 2016; Strait et al., 2015; Mongle et al., 2019) and published hypotheses that have not been recovered in phylogenetic analyses (Asfaw et al., 1999; Wood and Schroer, 2017; Villmoare, 2018).

Data availability

All data generated during this study are included in the manuscript (Table 1). Raw data from extant specimens appears as an Excel file in Figure 2-source data 1.

References

    1. Day MH
    (1973)
    Locomotor features of the lower limb in hominids
    Symp Zool Soc Lond 33:29–51.
    1. Day MH
    2. Thornton CMB
    (1986)
    The extremity bones of Paranthropus robustus from Kromdraai B, East formation member 3, Republic of South Africa: a reappraisal
    Anthropos 23:91–99.
  1. Book
    1. DeGusta DA
    (2004)
    Pliocene Hominid Postcranial Fossils from the Middle Awash, Ethiopia
    Berkeley: University of California.
  2. Book
    1. Gordon AD
    2. Green DJ
    3. Jungers WL
    4. Richmond BG
    (2020) Limb proportions and positional behavior: Revisiting the theoretical and empirical underpinnings for locomotor reconstruction in Australopithecus africanus
    In: Zipfel B, Richmond BG, Ward CV, editors. Hominid Postcranial Remains from Sterkfontein, South Africa, 1936-1995. Advances in Human Evolution Series. Oxford, United Kingdom: Oxford University Press. pp. 321–334.
    https://doi.org/10.1093/oso/9780197507667.001.0001
  3. Book
    1. Grausz HM
    2. Leakey REF
    3. Walker AC
    4. Ward CV
    (1988)
    Associated cranial and postcranial bones of Australopithecus boisei
    In: Grine FE, editors. Evolutionary History of the ‘“Robust”’ Australopithecines. Aldine de Gruyter. pp. 127–132.
  4. Book
    1. Leakey MG
    2. Leakey REF
    (1978)
    The Fossil Hominids and an Introduction to Their Context. 1968-1974
    Oxford: Clarendon.
  5. Book
    1. Leakey REF
    2. Walker A
    3. Ward CV
    4. Grausz HM
    (1987)
    A partial skeleton of a gracile hominid from the upper Burgi member of the Koobi Fora formation, East Lake Turkana, Kenya
    In: Giacobini G, editors. Hominidae: Proceedings of the 2nd International Congress of Human Paleontology Turin. Congresso Internazionale. pp. 167–174.
  6. Book
    1. Robinson JT
    (1972)
    Early Hominid Posture and Locomotion
    University of Chicago Press.
  7. Book
    1. Strait DS
    2. Grine FE
    3. Fleagle JG
    (2015) Analyzing hominin phylogeny: Cladistic approach
    In: Henke W, Tattersall I, editors. Handbook of Paleoanthropology. Springer-Verlag. pp. 1989–2014.
    https://doi.org/10.1007/978-3-642-39979-4_58
  8. Book
    1. Walker A
    2. Leakey REF
    (1993)
    The Nariokotome Homo erectus Skeleton
    Berlin, Heidelberg: Harvard University Press.
    1. Walker C
    2. Cofran ZD
    3. Grabowski M
    4. Marchi D
    5. Cook RW
    6. Churchill SE
    7. Tommy KA
    8. Throckmorton Z
    9. Ross AH
    10. Hawks J
    11. Yapunich GS
    12. Van Arsdale AP
    13. Rentzeperis F
    14. Berger LR
    15. DeSilva JM
    (2019)
    Morphology and evolution of the Homo naledi femora from Lesedi
    American Journal of Physical Anthropology 170:5–23.
  9. Book
    1. Wood B
    2. Schroer K
    (2017) Paranthropus: Where do things stand?
    In: Marom A, Hovers E, editors. Human Paleontology and Prehistory. Vertebrate Paleobiology and Paleoanthropology. Springer. pp. 95–107.
    https://doi.org/10.1007/978-3-319-46646-0

Decision letter

  1. George H Perry
    Senior and Reviewing Editor; Pennsylvania State University, United States
  2. Adam Gordon
    Reviewer; University at Albany-SUNY, United States
  3. Kevin Hatala
    Reviewer; Chatham University, United States

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Acceptance summary:

This paper will be useful for scholars interested in the evolution of bipedalism and the diversity of positional behavior in the hominin lineage and the pervasiveness of homoplasy/parallel evolution across various traits in human evolution. The authors expand the use of a proxy measure for arboreality/terrestriality to a much broader sample of fossil hominins than has previously been considered in a single analysis. Their results provide compelling support for the independent evolution of a high degree of terrestriality in one hominin species before it evolved in the modern human lineage.

Decision letter after peer review:

Thank you for submitting your article "Homoplasy in the evolution of modern human-like joint proportions in Australopithecus afarensis" for consideration by eLife. Your article has been reviewed by 2 peer reviewers, and the evaluation has been overseen by George Perry as the Senior and Reviewing Editor. The following individuals involved in review of your submission have agreed to reveal their identity: Adam Gordon (Reviewer #2); Kevin Hatala (Reviewer #3).

The reviewers have discussed their reviews with one another, and the Reviewing Editor has drafted this to help you prepare a revised submission.

The reviewers were both broadly positive and also raised similar core points requiring your attention, for us to consider the manuscript for publication – please see the individual reviews below.

With respect to the overlapping point raised by both reviewers (concerning ancestral reconstruction and phylogenetic uncertainty) but with slightly different proposed solutions (removal of the ancestral reconstruction analysis in favor of a parsimony-based approach, versus incorporating a sensitivity analysis for the ancestral reconstruction): we offer you flexibility in ultimately determining how best to proceed, with our slight consensus preference for removing the ancestral reconstruction (as the results would still stand on their own without this).

Reviewer #2:

In this manuscript, the authors expand on earlier work examining articular and diaphyseal proportions in the fore- and hindlimb of extant hominoids and extinct hominins to address the question of homology or homoplasy for the previously-documented similarity of proportions in modern humans and Australopithecus afarensis in contrast to the more ape-like proportions found in other fossil hominins. The authors use logged ratios of measurements related to forelimb and hindlimb size following Green et al. (2007), including associated measurements from the glenoid, scapula, and sacrum. Overall the methods and sample are appropriate to the question at hand, the paper is well-organized and clearly written, and the conclusions follow logically from their results: A. afarensis appears to have evolved relatively large hindlimb articular and diaphyseal dimensions independently from later Homo, and in both cases these dimensions are likely to relate to a greater reliance on terrestrial bipedalism than in other hominins. That said, there are a couple of areas that would benefit from a bit more attention. In particular, ratio values for hylobatids that are unexpectedly intermediate between modern humans and other extant apes in some analyses need explaining. Because this pattern appears to be driven by relatively large sacral bodies in hylobatids, the authors should include an additional analysis to compare logged ratios among taxa when sacrum size is excluded from ratios. In addition, while their overall conclusion of homoplasy in A. afarensis and later Homo is well supported, it's important to note that the results of the ancestral reconstruction analyses are highly dependent on the selection of phylogeny and evolutionary model.

As one of the authors of Green et al. (2007) and a proponent of using these types of logged ratios to make inferences about positional behavior in extinct hominins, I'm pleased to see that the authors are using these logged ratios to examine a broader set of fossil taxa than they've been applied to before, and I think that the authors do a great job here. However, I also think that it's important to address the limitations of what can be inferred from these ratios and to highlight unusual patterns in the data.

For example, the rank-ordering of mean RLSI among extant ape taxa differs among the various ratios shown in Figure 2. I'd encourage the authors to take a look at our recent piece (Gordon et al., 2020) in which we consider how different logged limb proportions relate to various ways "degree of arboreality" can be defined. I'd like to see the authors discuss in a bit more detail just what can and cannot be inferred from comparisons of ratio values between fossils and extant taxa for these different ratios.

In particular, I think it's critically important that the authors spend more time discussing the unexpected pattern where Hylobates has the lowest (and thus more human-like) forelimb:hindlimb ratio of all of the extant apes in some cases, despite them using the most forelimb-dependent locomotor behavior of all of the taxa in this study. I think that the information highlighted in Figure 4 is really useful in partitioning out the relative contribution of different variables to denominator of these ratios, and that figure shows that the hylobatids have relatively large sacral bodies compared to other taxa, particularly in relation to femoral head size. And it's clear from Figure 2 that it's the ratios that include sacrum size in the denominator that produce lower values for the hylobatids than the other apes.

I'd argue that the authors need to point this pattern out, briefly comment on why hylobatids have relatively large sacral bodies compared to the other extant apes, and explain why this isn't (or is) a concern for interpreting fossil RLSI values that include sacrum size. Given that (1) inclusion of sacrum size shifts gibbon ratios towards modern humans and (2) sacrum size appears in the denominator of RLSI calculated for four of the five fossil specimens in the three taxa argued to be most human-like (i.e., A.L.288-1 and KSD-VP-1/1 in A. afarensis, KNM-WT 15000 in H. erectus, and LES 1 in H. naledi), it would be informative to recalculate RLSI without sacrum size in those specimens and the comparative sample to see if those fossil specimens still fall inside the human range and outside of the ape range. That's shown for a subset of Lucy's measurements in Figure 2B and D, but it should be shown for the most complete set of measurements (excluding sacrum size) for each of these four fossils. I expect the overall pattern will remain unchanged; I've calculated RLSI for these fossils (A.L.288-1, KSD-VP-1/1, KNM-WT 15000, and LES 1) and the extant sample means using the data reported in Table 1 and the RLSI equations reported in Table S1 (adjusted to exclude sacral size). In all cases the patterns appear to be consistent with the patterns when sacrum size is included, although without the data for individual comparative specimens I can't comment on the degree of overlap in the species ranges of the logged ratios. I'd suggest (1) adding these comparisons to the supplementary material as figures in the same style as the other supplementary figures, and (2) mentioning in the text that the exclusion of sacrum size tends to shift hylobatid logged ratios in line with the other apes but doesn't remove those fossil hominins from sitting in the modern human range.

Considering all of the above as well as the authors' discussion of the impact of taxonomic uncertainty in some cases, I believe that the authors present a compelling argument that joint proportions similar to those of modern humans are only found in Homo erectus, H. naledi, and A. afarensis, and notably are not found in other australopiths, H. habilis, or H. erectus. I also agree that even if one were to consider a number of different possible cladograms, the most parsimonious interpretation of these results is that A. afarensis developed low forelimb:hindlimb joint ratios independently from later Homo taxa. However, I don't think that the ancestral state reconstruction analysis adds much, if anything, to the argument. It's clear that ancestral reconstructions are highly dependent on branching topology, branch lengths, and assumptions regarding the underlying evolutionary model, and that errors in specifying any of these components can produce highly divergent results (e.g., see Ponti et al., 2020). Figure 3 essentially shows just what a parsimony-based analysis of the same data would show, with the addition of what I would argue are unrealistically precise estimates of ancestral values without any depiction of model-derived ranges of uncertainty around the ancestral estimates – which themselves are likely to underestimate the actual uncertainty due to probable misspecification of the phylogeny and/or evolutionary model. Personally, I am both a big proponent of phylogenetic methods for comparative analysis and pretty sour on ancestral reconstruction methods. So feel free to take my comments on this part of the manuscript with a grain of salt, but also with the understanding that I've been working with these methods and thinking about them for twenty years.

Comment on significance testing:

Some reviewers might take issue with the lack of p-values and/or confidence intervals for comparison of RLSI values in this manuscript, but I'm not one of them. The comparative samples for extant taxa are reasonably large, and the consistent differentiation of modern human ratios from extant apes combined with the placement of fossil hominin ratios within the range of one of those two groups but not the other is a more compelling argument to me than any p-values or confidence intervals would be. If the authors chose to calculate p-values or confidence intervals, they would undoubtedly support the arguments the authors make here, but I really don't think they're necessary.

References:

Gordon AD, Green DJ, Jungers WL, Richmond BG. 2020. Limb proportions and positional behavior: revisiting the theoretical and empirical underpinnings for locomotor reconstruction in Australopithecus africanus. In Zipfel B, Richmond BG, and Ward CV, eds.: Hominid Postcranial Remains from Sterkfontein, South Africa, 1936-1995. Advances in Human Evolution Series. Oxford University Press. pp. 321-334.

Ponti R, Arcones A, Vieetes DR. 2020. Challenges in estimating ancestral state reconstructions: the evolution of migration in Sylvia warblers as a study case. Integrative Zoology. 15:161-173.

Reviewer #3:

In this manuscript, Prabhat et al. evaluate limb joint proportions in modern humans and extant non-human apes, and through comparisons with these morphological patterns they infer locomotor behaviors from partial skeletons of various fossil hominins. By evaluating their results in the context of prior cladistic analyses, they draw further inferences about the evolutionary patterns that may be evident from fossil hominin limb joint proportions.

Studies of hominin limb joint size proportions have been conducted previously, but none have included such a broad sample of fossil taxa and none have focused to a similar degree on inferring evolutionary patterns from this aspect of postcranial morphology. The authors' analyses of joint proportions and their inferences of locomotor patterns are interesting, and they largely support results from other functional analyses of postcranial morphology. They find that skeletons of Australopithecus afarensis, Homo erectus, and H. naledi have joint size proportions that are similar to humans, and they infer that these taxa used a manner and/or degree of bipedalism that closely represented those of modern humans. Meanwhile, A. africanus, A. sediba, Paranthropus robustus, P. boisei, H. habilis, and H. floresiensis all show ape-like patterns of limb joint proportions, and the authors infer that their locomotor patterns would have included greater degrees of arboreality. More interesting, however, is that the authors proceed to interpret their results in the context of past cladistic analyses, and they find that the morphological pattern shared by A. afarensis and modern H. sapiens is likely the product of homoplasy rather than shared ancestry. This result suggests that the manner of bipedal locomotion used by A. afarensis was not a linear precursor to the form of bipedalism practiced by later species of the genus Homo and by modern humans today. I appreciate the authors' discussion of how this interpretative framework would change the ways in which we understand functional morphological patterns in A. afarensis, as they relate to bipedalism. For decades, paleoanthropologists have debated the "humanness" of their locomotor style. However, if we view their bipedalism as a reflection of homoplasy then it need not be connected in any way to the locomotion of modern humans and can instead be viewed simply as a different evolutionary solution. Overall, the analyses and results support the conclusions of these authors, and the results of this study are likely to have significant impacts on the field of paleoanthropology.

Perhaps the most important potential weaknesses, from my perspective, are related to (1) the dichotomization of human-like and ape-like morphologies, and (2) the degree to which the inference of evolutionary patterns hinges upon the particular phylogenetic tree selected for this analysis. I do not believe that either of these problems are entirely avoidable, but merely suggest that these are areas where the analyses and interpretations can be clarified and/or extended.

With respect to the first point, the dichotomous analysis means that non-human apes with different locomotor strategies and differing degrees of arboreality are grouped together, while humans stand as a sole extant contrast. The variations in joint size proportions among the non-human apes are not directly explored, so it remains difficult to know just how sensitive these variables are to different loading regimes (e.g., I would expect gibbons to experience the loading regimes most different from those of humans, yet in Figure 2 they fall closest to human-like joint size ratios in two of four panels and they are never the most different). Exploring this in greater depth could be very informative.

With respect to the second point, the differences between Figures 3A and 3B reflects sensitivity of the ancestral state reconstruction to the placement of A. sediba. While homoplasy is the most likely pattern in both scenario, it does appear that relatively subtle differences in the placement of this taxon influences the likelihood of ancestral states at various nodes and along certain branches throughout the tree. We can only work with the fossil record that we have, but I would be curious to see just how much difference it takes in order for a pattern other than homoplasy to emerge as most likely. Evaluating the robustness of this signal could even further strengthen the paper.

This manuscript was a pleasure to review. I found the questions interesting, the analyses appropriate, and the results and discussion compelling. It is also very well-written. I think this will make a significant contribution to the field, and it will prompt new lines of thought and inquiry as readers see these results and reflect on past analyses of joint size and limb length proportions in various fossil hominins. I really enjoyed thinking about what these results mean for how we interpret A. afarensis postcranial morphology and locomotion, which have certainly been subject to some of the most vigorous debates in paleoanthropology.

I believe there are just a few areas that could be clarified and/or explored to strengthen the manuscript even further. First, related to patterns among non-human apes, I found this fascinating and they seemed unexpected to me in some places (such as the gibbon example that I cited above). I wonder if you could explain in greater detail whether joint size proportions reflect the differences in locomotor behaviors between the non-human apes? I don't necessarily expect that a clear pattern will emerge, but it seems worth discussing since there is a lot of variation in the extent to which these various apes load their upper vs. lower limbs, not to mention other differences in body size, etc. If it is not purely loading regimes that affect these proportions (and I expect it is not that clear and straightforward) then it may be worth some page space (supplementary if there is no room in main text) to discuss what other factors may appear likely to influence these proportions.

Second, in my mind it could be useful to do some kind of sensitivity analysis related to the ancestral state reconstruction and subsequent evolutionary inferences. I know that deriving phylogenetic trees is not a goal of the current study, but it could be interesting and worthwhile to explore the robustness of the signal of homoplasy. Assuming that it is quite robust (which I expect to be the case), then it would further strengthen that component of the paper. If not, or if a certain taxon's place really disrupts the pattern, this would warrant some discussion.

Last, and this is not mentioned above, in the supplementary plots S2-S17 I think it would be very useful to display all possible fossil specimens in each analysis (e.g., use only the traits available from the specimen-of-focus, for as many specimens as possible). I'd be curious to know whether the ape-like or human-like signals hold up even when certain more complete specimens (e.g., A.L. 288-1, KNM-WT 15000), are "rarefied". On the other side of the equation, it may tell you whether signals from less complete specimens are likely to hold up if more of a given skeleton were to be found.

At a much more specific level, I thought the authors might consider re-phrasing lines 57-58, as the wording seems to imply that the loss of adaptations to arboreality and the transition to exclusive terrestrial bipedalism occurred in a linear pattern. In light of their past works, I don't think the authors would intend to make this implication. Similarly, line 316 discusses that A. afarensis and later Pleistocene Homo evolved similar bipedal adaptations "in parallel". To me, that seemed to imply parallel evolution, a term which is sometimes (though not always) defined as requiring overlap in time and/or place, which is clearly not the case here. I was certainly able to understand what you meant here, but I wonder if a term other than "parallel" may be better?

https://doi.org/10.7554/eLife.65897.sa1

Author response

Reviewer #2:

In this manuscript, the authors expand on earlier work examining articular and diaphyseal proportions in the fore- and hindlimb of extant hominoids and extinct hominins to address the question of homology or homoplasy for the previously-documented similarity of proportions in modern humans and Australopithecus afarensis in contrast to the more ape-like proportions found in other fossil hominins. The authors use logged ratios of measurements related to forelimb and hindlimb size following Green et al. (2007), including associated measurements from the glenoid, scapula, and sacrum. Overall the methods and sample are appropriate to the question at hand, the paper is well-organized and clearly written, and the conclusions follow logically from their results: A. afarensis appears to have evolved relatively large hindlimb articular and diaphyseal dimensions independently from later Homo, and in both cases these dimensions are likely to relate to a greater reliance on terrestrial bipedalism than in other hominins. That said, there are a couple of areas that would benefit from a bit more attention. In particular, ratio values for hylobatids that are unexpectedly intermediate between modern humans and other extant apes in some analyses need explaining. Because this pattern appears to be driven by relatively large sacral bodies in hylobatids, the authors should include an additional analysis to compare logged ratios among taxa when sacrum size is excluded from ratios.

Thank you for this suggestion. We agree that the placement of the hylobatids in many of these analyses makes it challenging to interpret our findings along a terrestrial—arboreal continuum.

As proposed, we have re-run all of our analyses without contributions from the sacrum. Interestingly, while there is some shifting of the hylobatids, it is not enough to explain their intermediate positioning between the great apes and humans. We hypothesize, in part, an allometric explanation, as a result of further analyses which reveal that the biepicondylar width and glenoid size scale isometrically against femoral head diameter in all of the hominoids, except the hylobatids, which scale with negative allometry (m=0.75). Why this is the case remains unclear.

In addition, while their overall conclusion of homoplasy in A. afarensis and later Homo is well supported, it's important to note that the results of the ancestral reconstruction analyses are highly dependent on the selection of phylogeny and evolutionary model.

Yes, we agree. To that end, we have created a new Figure 3 in which six possible phylogenies are illustrated instead of the original two. Not only does this cover the different hypotheses current circulating, but also presents a phylogenetic scenario that would avoid RLSI homoplasy between later Homo and A. afarensis.

As one of the authors of Green et al. (2007) and a proponent of using these types of logged ratios to make inferences about positional behavior in extinct hominins, I'm pleased to see that the authors are using these logged ratios to examine a broader set of fossil taxa than they've been applied to before, and I think that the authors do a great job here. However, I also think that it's important to address the limitations of what can be inferred from these ratios and to highlight unusual patterns in the data.

For example, the rank-ordering of mean RLSI among extant ape taxa differs among the various ratios shown in Figure 2. I'd encourage the authors to take a look at our recent piece (Gordon et al., 2020) in which we consider how different logged limb proportions relate to various ways "degree of arboreality" can be defined. I'd like to see the authors discuss in a bit more detail just what can and cannot be inferred from comparisons of ratio values between fossils and extant taxa for these different ratios.

Thank you for alerting us to Gordon et al., 2020. We have added additional text that explores the limits of using RLSI to infer “degree of arboreality.”

In particular, I think it's critically important that the authors spend more time discussing the unexpected pattern where Hylobates has the lowest (and thus more human-like) forelimb:hindlimb ratio of all of the extant apes in some cases, despite them using the most forelimb-dependent locomotor behavior of all of the taxa in this study. I think that the information highlighted in Figure 4 is really useful in partitioning out the relative contribution of different variables to denominator of these ratios, and that figure shows that the hylobatids have relatively large sacral bodies compared to other taxa, particularly in relation to femoral head size. And it's clear from Figure 2 that it's the ratios that include sacrum size in the denominator that produce lower values for the hylobatids than the other apes.

I'd argue that the authors need to point this pattern out, briefly comment on why hylobatids have relatively large sacral bodies compared to the other extant apes, and explain why this isn't (or is) a concern for interpreting fossil RLSI values that include sacrum size. Given that (1) inclusion of sacrum size shifts gibbon ratios towards modern humans and (2) sacrum size appears in the denominator of RLSI calculated for four of the five fossil specimens in the three taxa argued to be most human-like (i.e., A.L.288-1 and KSD-VP-1/1 in A. afarensis, KNM-WT 15000 in H. erectus, and LES 1 in H. naledi), it would be informative to recalculate RLSI without sacrum size in those specimens and the comparative sample to see if those fossil specimens still fall inside the human range and outside of the ape range. That's shown for a subset of Lucy's measurements in Figure 2B and D, but it should be shown for the most complete set of measurements (excluding sacrum size) for each of these four fossils. I expect the overall pattern will remain unchanged; I've calculated RLSI for these fossils (A.L.288-1, KSD-VP-1/1, KNM-WT 15000, and LES 1) and the extant sample means using the data reported in Table 1 and the RLSI equations reported in Table S1 (adjusted to exclude sacral size). In all cases the patterns appear to be consistent with the patterns when sacrum size is included, although without the data for individual comparative specimens I can't comment on the degree of overlap in the species ranges of the logged ratios. I'd suggest (1) adding these comparisons to the supplementary material as figures in the same style as the other supplementary figures, and (2) mentioning in the text that the exclusion of sacrum size tends to shift hylobatid logged ratios in line with the other apes but doesn't remove those fossil hominins from sitting in the modern human range.

Thank you for this important observation. As suggested, we have rerun all of the analyses without the sacrum and include those as figure supplements. Additionally, we have added text to the main paper noting that the relatively large sacrum of gibbons is, in part, what is making their RLSI more human-like than the other hominoids.

Considering all of the above as well as the authors' discussion of the impact of taxonomic uncertainty in some cases, I believe that the authors present a compelling argument that joint proportions similar to those of modern humans are only found in Homo erectus, H. naledi, and A. afarensis, and notably are not found in other australopiths, H. habilis, or H. erectus. I also agree that even if one were to consider a number of different possible cladograms, the most parsimonious interpretation of these results is that A. afarensis developed low forelimb:hindlimb joint ratios independently from later Homo taxa. However, I don't think that the ancestral state reconstruction analysis adds much, if anything, to the argument. It's clear that ancestral reconstructions are highly dependent on branching topology, branch lengths, and assumptions regarding the underlying evolutionary model, and that errors in specifying any of these components can produce highly divergent results (e.g., see Ponti et al., 2020). Figure 3 essentially shows just what a parsimony-based analysis of the same data would show, with the addition of what I would argue are unrealistically precise estimates of ancestral values without any depiction of model-derived ranges of uncertainty around the ancestral estimates – which themselves are likely to underestimate the actual uncertainty due to probable misspecification of the phylogeny and/or evolutionary model. Personally, I am both a big proponent of phylogenetic methods for comparative analysis and pretty sour on ancestral reconstruction methods. So feel free to take my comments on this part of the manuscript with a grain of salt, but also with the understanding that I've been working with these methods and thinking about them for twenty years.

We agree and have modified Figure 3 so that it is no longer depicting ancestral state reconstructions but instead maps dichotomized RLSI (human-like or ape-like) on six different possible phylogenies.

Comment on significance testing:

Some reviewers might take issue with the lack of p-values and/or confidence intervals for comparison of RLSI values in this manuscript, but I'm not one of them. The comparative samples for extant taxa are reasonably large, and the consistent differentiation of modern human ratios from extant apes combined with the placement of fossil hominin ratios within the range of one of those two groups but not the other is a more compelling argument to me than any p-values or confidence intervals would be. If the authors chose to calculate p-values or confidence intervals, they would undoubtedly support the arguments the authors make here, but I really don't think they're necessary.

We appreciate that the reviewer does not find additional statistical analysis necessary.

References:

Gordon AD, Green DJ, Jungers WL, Richmond BG. 2020. Limb proportions and positional behavior: revisiting the theoretical and empirical underpinnings for locomotor reconstruction in Australopithecus africanus. In Zipfel B, Richmond BG, and Ward CV, eds.: Hominid Postcranial Remains from Sterkfontein, South Africa, 1936-1995. Advances in Human Evolution Series. Oxford University Press. pp. 321-334.

Ponti R, Arcones A, Vieetes DR. 2020. Challenges in estimating ancestral state reconstructions: the evolution of migration in Sylvia warblers as a study case. Integrative Zoology. 15:161-173.

We now cite Gordon et al., 2020 in our paper. Thank you.

Reviewer #3:

In this manuscript, Prabhat et al. evaluate limb joint proportions in modern humans and extant non-human apes, and through comparisons with these morphological patterns they infer locomotor behaviors from partial skeletons of various fossil hominins. By evaluating their results in the context of prior cladistic analyses, they draw further inferences about the evolutionary patterns that may be evident from fossil hominin limb joint proportions.

Studies of hominin limb joint size proportions have been conducted previously, but none have included such a broad sample of fossil taxa and none have focused to a similar degree on inferring evolutionary patterns from this aspect of postcranial morphology. The authors' analyses of joint proportions and their inferences of locomotor patterns are interesting, and they largely support results from other functional analyses of postcranial morphology. They find that skeletons of Australopithecus afarensis, Homo erectus, and H. naledi have joint size proportions that are similar to humans, and they infer that these taxa used a manner and/or degree of bipedalism that closely represented those of modern humans. Meanwhile, A. africanus, A. sediba, Paranthropus robustus, P. boisei, H. habilis, and H. floresiensis all show ape-like patterns of limb joint proportions, and the authors infer that their locomotor patterns would have included greater degrees of arboreality. More interesting, however, is that the authors proceed to interpret their results in the context of past cladistic analyses, and they find that the morphological pattern shared by A. afarensis and modern H. sapiens is likely the product of homoplasy rather than shared ancestry. This result suggests that the manner of bipedal locomotion used by A. afarensis was not a linear precursor to the form of bipedalism practiced by later species of the genus Homo and by modern humans today. I appreciate the authors' discussion of how this interpretative framework would change the ways in which we understand functional morphological patterns in A. afarensis, as they relate to bipedalism. For decades, paleoanthropologists have debated the "humanness" of their locomotor style. However, if we view their bipedalism as a reflection of homoplasy then it need not be connected in any way to the locomotion of modern humans and can instead be viewed simply as a different evolutionary solution. Overall, the analyses and results support the conclusions of these authors, and the results of this study are likely to have significant impacts on the field of paleoanthropology.

Thank you. We agree and hope that this paper adds a fresh, new perspective on terrestrial bipedalism in A. afarensis.

Perhaps the most important potential weaknesses, from my perspective, are related to (1) the dichotomization of human-like and ape-like morphologies, and (2) the degree to which the inference of evolutionary patterns hinges upon the particular phylogenetic tree selected for this analysis. I do not believe that either of these problems are entirely avoidable, but merely suggest that these are areas where the analyses and interpretations can be clarified and/or extended.

With respect to the first point, the dichotomous analysis means that non-human apes with different locomotor strategies and differing degrees of arboreality are grouped together, while humans stand as a sole extant contrast. The variations in joint size proportions among the non-human apes are not directly explored, so it remains difficult to know just how sensitive these variables are to different loading regimes (e.g., I would expect gibbons to experience the loading regimes most different from those of humans, yet in Figure 2 they fall closest to human-like joint size ratios in two of four panels and they are never the most different). Exploring this in greater depth could be very informative.

We agree that there are some fascinating (albeit unexpected) patterns in the non-human ape data. As recommended by Reviewer #2, we have re-run the analysis with the sacrum removed and included these new figures in the supplementary material. We have also added brief text acknowledging that RLSI does not track the degree of arboreality in living apes and hypothesize that allometry may help explain some of these findings. A closer examination of the patterns in the ape data is currently underway and is intended to be the subject of a follow-up manuscript.

With respect to the second point, the differences between Figures 3A and 3B reflects sensitivity of the ancestral state reconstruction to the placement of A. sediba. While homoplasy is the most likely pattern in both scenario, it does appear that relatively subtle differences in the placement of this taxon influences the likelihood of ancestral states at various nodes and along certain branches throughout the tree. We can only work with the fossil record that we have, but I would be curious to see just how much difference it takes in order for a pattern other than homoplasy to emerge as most likely. Evaluating the robustness of this signal could even further strengthen the paper.

We agree and have modified Figure 3 so that there are now six different phylogenetic hypotheses considered. Additionally, we no longer include ancestral state reconstructions.

This manuscript was a pleasure to review. I found the questions interesting, the analyses appropriate, and the results and discussion compelling. It is also very well-written. I think this will make a significant contribution to the field, and it will prompt new lines of thought and inquiry as readers see these results and reflect on past analyses of joint size and limb length proportions in various fossil hominins. I really enjoyed thinking about what these results mean for how we interpret A. afarensis postcranial morphology and locomotion, which have certainly been subject to some of the most vigorous debates in paleoanthropology.

I believe there are just a few areas that could be clarified and/or explored to strengthen the manuscript even further. First, related to patterns among non-human apes, I found this fascinating and they seemed unexpected to me in some places (such as the gibbon example that I cited above). I wonder if you could explain in greater detail whether joint size proportions reflect the differences in locomotor behaviors between the non-human apes? I don't necessarily expect that a clear pattern will emerge, but it seems worth discussing since there is a lot of variation in the extent to which these various apes load their upper vs. lower limbs, not to mention other differences in body size, etc. If it is not purely loading regimes that affect these proportions (and I expect it is not that clear and straightforward) then it may be worth some page space (supplementary if there is no room in main text) to discuss what other factors may appear likely to influence these proportions.

We, too, are struggling to understand the pattern in the data and do not have a satisfactory explanation. We have added some text hypothesizing that perhaps body size is driving some of the higher RLSI values in the great apes. Additionally, the sacrum appears to be contributing somewhat to the low RLSI value in gibbons. Other than that, we are somewhat baffled. However, we plan to expand our dataset to include cercopithecoids and Miocene hominoids and anticipate having a better explanation for these data in the near future.

Second, in my mind it could be useful to do some kind of sensitivity analysis related to the ancestral state reconstruction and subsequent evolutionary inferences. I know that deriving phylogenetic trees is not a goal of the current study, but it could be interesting and worthwhile to explore the robustness of the signal of homoplasy. Assuming that it is quite robust (which I expect to be the case), then it would further strengthen that component of the paper. If not, or if a certain taxon's place really disrupts the pattern, this would warrant some discussion.

Thank you for this suggestion. After consulting this comment and those made by Reviewer #2 and the editor, we decided to redraw Figure 3 using six different possible phylogenies. We hope this satisfies the concerns of the reviewer.

Last, and this is not mentioned above, in the supplementary plots S2-S17 I think it would be very useful to display all possible fossil specimens in each analysis (e.g., use only the traits available from the specimen-of-focus, for as many specimens as possible). I'd be curious to know whether the ape-like or human-like signals hold up even when certain more complete specimens (e.g., A.L. 288-1, KNM-WT 15000), are "rarefied". On the other side of the equation, it may tell you whether signals from less complete specimens are likely to hold up if more of a given skeleton were to be found.

This is a great point and we have now added all of these graphs as figure supplements. We have maintained the individual supplementary plots (Figure 2—figure supplements 1-16) and now have added these others (Figure 2—figure supplements 17-26)

At a much more specific level, I thought the authors might consider re-phrasing lines 57-58, as the wording seems to imply that the loss of adaptations to arboreality and the transition to exclusive terrestrial bipedalism occurred in a linear pattern. In light of their past works, I don't think the authors would intend to make this implication. Similarly, line 316 discusses that A. afarensis and later Pleistocene Homo evolved similar bipedal adaptations "in parallel". To me, that seemed to imply parallel evolution, a term which is sometimes (though not always) defined as requiring overlap in time and/or place, which is clearly not the case here. I was certainly able to understand what you meant here, but I wonder if a term other than "parallel" may be better?

We have added words and/or changed the language in both of these sections to avoid any misunderstandings.

https://doi.org/10.7554/eLife.65897.sa2

Article and author information

Author details

  1. Anjali M Prabhat

    Anthropology, Dartmouth College, Hanover, United States
    Contribution
    Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Writing – original draft, Writing – review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-0654-3455
  2. Catherine K Miller

    1. Anthropology, Dartmouth College, Hanover, United States
    2. Ecology, Evolution, Ecosystems, and Society, Dartmouth College, Hanover, United States
    Contribution
    Data curation, Formal analysis, Methodology, Writing – original draft
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-0352-3777
  3. Thomas Cody Prang

    Department of Anthropology, Texas A&M University, College Station, United States
    Contribution
    Conceptualization, Data curation, Formal analysis, Methodology, Writing – review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-3032-8309
  4. Jeffrey Spear

    1. Center for the Study of Human Origins, Department of Anthropology, New York University, New York, United States
    2. New York Consortium in Evolutionary Primatology, New York, United States
    Contribution
    Data curation, Formal analysis, Methodology, Writing – review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-0290-7090
  5. Scott A Williams

    1. Center for the Study of Human Origins, Department of Anthropology, New York University, New York, United States
    2. New York Consortium in Evolutionary Primatology, New York, United States
    Contribution
    Conceptualization, Formal analysis, Methodology, Supervision, Writing – review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-7860-8962
  6. Jeremy M DeSilva

    1. Anthropology, Dartmouth College, Hanover, United States
    2. Ecology, Evolution, Ecosystems, and Society, Dartmouth College, Hanover, United States
    Contribution
    Conceptualization, Formal analysis, Investigation, Methodology, Supervision, Writing – original draft, Writing – review and editing
    For correspondence
    Jeremy.M.DeSilva@dartmouth.edu
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-7010-1155

Funding

Leakey Foundation

  • Scott A Williams

Dartmouth College

  • Jeremy M DeSilva

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

The authors are grateful to the many individuals who made our data collection possible: Yohannes Haile-Selassie and Lyman Jellema (CMNH); Mark Omura (MCZ), Olivia Herschensohn, Kora Welsh, and Michèle Morgan (Harvard Peabody Museum of Archaeology and Ethnology); Neil Duncan, Eleanor Hoeger, Sara Ketelsen, Aja Marcato, Brian O’Toole, Marisa Surovy, and Eileen Westwig (AMNH); Emmanuel Gilissen and Wim Wendelen (Royal Museum for Central Africa); Lee Berger, Sifelani Jirah, and Bernhard Zipfel (Evolutionary Studies Institute, University of the Witwatersrand); Lazarus Kgasi, Stephany Potze, and Mirriam Tawane (Ditsong Museum of Natural History); Jared Assefa, Tomas Getachew, Getachew Senishaw, and Yonas Yilma (National Museum of Ethiopia, Authority for Research and Conservation of Cultural Heritage, and the Ethiopian Ministry of Culture and Tourism); Emma Mbua, Fredrick Manthi, and Job Kibii (National Museums of Kenya). We are additionally grateful to Matt Tocheri and Manuel Domínguez-Rodrigo for providing casts of hominin material included in this study.

Senior and Reviewing Editor

  1. George H Perry, Pennsylvania State University, United States

Reviewers

  1. Adam Gordon, University at Albany-SUNY, United States
  2. Kevin Hatala, Chatham University, United States

Publication history

  1. Received: December 18, 2020
  2. Accepted: April 19, 2021
  3. Version of Record published: May 12, 2021 (version 1)

Copyright

© 2021, Prabhat et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 1,198
    Page views
  • 106
    Downloads
  • 0
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)

Further reading

    1. Evolutionary Biology
    2. Genetics and Genomics
    Stephanie M Yan et al.
    Research Article

    Large genomic insertions and deletions are a potent source of functional variation, but are challenging to resolve with short-read sequencing, limiting knowledge of the role of such structural variants (SVs) in human evolution. Here, we used a graph-based method to genotype long-read-discovered SVs in short-read data from diverse human genomes. We then applied an admixture-aware method to identify 220 SVs exhibiting extreme patterns of frequency differentiation—a signature of local adaptation. The top two variants traced to the immunoglobulin heavy chain locus, tagging a haplotype that swept to near fixation in certain Southeast Asian populations, but is rare in other global populations. Further investigation revealed evidence that the haplotype traces to gene flow from Neanderthals, corroborating the role of immune-related genes as prominent targets of adaptive introgression. Our study demonstrates how recent technical advances can help resolve signatures of key evolutionary events that remained obscured within technically challenging regions of the genome.

    1. Evolutionary Biology
    2. Genetics and Genomics
    Ursula Oggenfuss et al.
    Research Article

    Genome evolution is driven by the activity of transposable elements (TEs). The spread of TEs can have deleterious effects including the destabilization of genome integrity and expansions. However, the precise triggers of genome expansions remain poorly understood because genome size evolution is typically investigated only among deeply divergent lineages. Here, we use a large population genomics dataset of 284 individuals from populations across the globe of Zymoseptoria tritici, a major fungal wheat pathogen. We built a robust map of genome-wide TE insertions and deletions to track a total of 2456 polymorphic loci within the species. We show that purifying selection substantially depressed TE frequencies in most populations, but some rare TEs have recently risen in frequency and likely confer benefits. We found that specific TE families have undergone a substantial genome-wide expansion from the pathogen’s center of origin to more recently founded populations. The most dramatic increase in TE insertions occurred between a pair of North American populations collected in the same field at an interval of 25 years. We find that both genome-wide counts of TE insertions and genome size have increased with colonization bottlenecks. Hence, the demographic history likely played a major role in shaping genome evolution within the species. We show that both the activation of specific TEs and relaxed purifying selection underpin this incipient expansion of the genome. Our study establishes a model to recapitulate TE-driven genome evolution over deeper evolutionary timescales.