The hominins: a very conservative tribe? Last common ancestors, plasticity and ecomorphology in Hominidae. Or, What's in a name?

Abstract

In the early 20th century the dominant paradigm for the ecological context of the origins of human bipedalism was arboreal suspension. In the 1960s, however, with recognition of the close genetic relationship of humans, chimpanzees and bonobos, and with the first field studies of mountain gorillas and common chimpanzees, it was assumed that locomotion similar to that of common chimpanzees and mountain gorillas, which appeared to be dominated by terrestrial knuckle‐walking, must have given rise to human bipedality. This paradigm has been popular, if not universally dominant, until very recently. However, evidence that neither the knuckle‐walking or vertical climbing of these apes is mechanically similar to human bipedalism, as well as the hand‐assisted bipedality and orthograde clambering of orang‐utans, has cast doubt on this paradigm. It now appears that the dominance of terrestrial knuckle‐walking in mountain gorillas is an artefact seen only in the extremes of their range, and that both mountain and lowland gorillas have a generalized orthogrady similar to that seen in orang‐utans. These data, together with evidence for continued arboreal competence in humans, mesh well with an increasing weight of fossil evidence suggesting that a mix of orang‐utan and gorilla‐like arboreal locomotion and upright terrestrial bipedalism characterized most australopiths. The late split date of the panins, corresponding to dates for separation of Homo and Australopithecus, leads to the speculation that competition with chimpanzees, as appears to exist today with gorillas, may have driven ecological changes in hominins and perhaps cladogenesis. However, selection for ecological plasticity and morphological conservatism is a core characteristic of Hominidae as a whole, including Hominini.

Introduction

Phylogenetic systematics, which identifies biological groups (clades) by shared‐derived (synapomorphic) characters, is now the accepted method for analysing mammalian relationships. It has replaced older approaches based on perceived shifts in ‘levels of organization’ (grades) which were dominant in the 1950s and 1960s. Nevertheless, the very Victorian philosophy of progress at the heart of the ‘scala naturae’ concept, which itself underlies grade‐based approaches, is still all too pervasive: most of us at one time or the other (e.g. Wang et al. 2014) have been guilty of contrasting ‘human’ with ‘ape’ anatomy, with the (however inadvertent) implication that human anatomy is so distinct from that of all other living apes that they can all be lumped together on a different, lower rung of the ‘scala naturae’. Since grade‐based systematics are now almost universally rejected, under phylogenetic systematics this would imply that humans and their fossil relatives are members of one daughter lineage of the last common ancestor (LCA) we share with other apes, and that all other apes belong to another. It would be next to impossible to find any scientist who is not a creationist who would regard that implication as correct. The living apes and their fossil relatives are all members of an informal group, the crown hominoids, conservative in craniodental features but readily defined by shared‐derived characters of the postcranium, e.g. a transversely broad, not dorsoventrally deep thorax. This means that several higher taxa of fossil apes, such as the Proconsulidae, are root‐hominoids, not closely related to the living apes or to fossil crown hominoids of similar age, such as Morotopithecus. Even so, relationships within the crown hominoids are not easy to establish unequivocally on shared‐derived morphological characters. But, genetic distances and branching sequences are usually regarded as equivalent, although few studies have sought to establish concordances between the two approaches (Lockwood et al. 2004 is a notable exception). On this basis, gibbons and siamangs are generally accepted to form a clade very distinct from all other living crown hominoids, whose members fall into the family‐level taxon Hylobatidae. Although there is no consensus on family taxonomy within superfamily Hominoidea, I follow that adopted by Wood & Lonergan (2008), in the second of three Journal of Anatomy special issues on human evolution, where living members of the matching daughter clade belong to taxon Hominidae, containing all great apes (including humans) and nothing else. This requires that orang‐utans, gorillas, common chimpanzees and bonobos, and humans, and the immediate fossil relatives of each, must then belong to taxa at the subfamily level or below. It is generally accepted that genetic, and the available fossil evidence [e.g. Sivapithecus, ca. 11 Ma; Ankarapithecus, 9.5 Ma; Lufengpithecus, 9 Ma; and possibly Khoratpithecus chaingmuanensis, 13–10.5 Ma (Chaimanee et al. 2003); and Hispanopithecus laeitanius, 9.5 Ma (Moyà‐Solà & Köhler, 1996)] indicate a deep/ancient division between orang‐utans and their fossil relatives and all other great apes, and then a somewhat younger one between gorillas and their fossil relatives (e.g. Chororapithecus abyssinicus, 10.5 Ma; Suwa et al. 2007), and chimpanzees, bonobos and humans and their fossil relatives, evidenced by dental and gnathic fossil evidence for Pan‐like apes from the Ngorora Formation in the Tugen Hills at 12.5 Ma and from Niger at between 6 and 11.5 Ma (Pickford & Senut, 2005; Pickford et al. 2009), which would strengthen the argument for a Late Miocene division of Hominini and Panini.

Rather illogically, since genetic evidence supports the existence of a valid clade (see e.g. Lockwood et al. 2004), no formal name has yet been given to the African apes (which in my view unquestionably includes humans), but orang‐utans and their fossil relatives must be subfamily Ponginae, leaving Homininae as the only valid name for the clade containing chimpanzees, bonobos, humans and their fossil relatives, and Gorillinae as the name for the gorilla clade. The Homininae would then divide into taxa at the tribe level, Hominini and Panini, the latter containing chimpanzees and bonobos, which genetic evidence suggests divided between 5 and 8 Mya. Fossil evidence does not yet distinguish between these two living species of Pan.

Thus, contrasting characters in humans with those in ‘apes’, bears an implicit implication that the characters referred to in non‐human great apes (NHGAs) are ‘primitive’ or conserved, whereas those in humans are derived. We have thus become all too used to models of human evolution which describe an adaptive scenario whereby a chimpanzee‐like [interestingly, rarely bonobo‐like, with the notable exception of Zihlman's work (e.g. Zihlman et al. 1978)] common hominine ancestor should evolve into a human. Rarely do we consider Darwin's (1871) warning that we should not expect common ancestors to closely resemble any of their descendants. Hardly ever do we consider the possibility that in some systems, the common panin‐hominin ancestor may have been more human‐like than chimpanzee‐like. The main reason for adopting a chimpanzee/bonobo‐like model for the hominin‐panin LCA, rather than a hominin‐like model, according to Wood & Lonergan (2008), is that gorillas, the next most closely related clade, share more morphology with chimpanzees and bonobos than they do with humans. As they state, this assumes that we can exclude homoplasy (‘parallel’ or ‘convergent’ evolution). It also assumes, I would suggest, that evolution in craniodental features, which have received much attention, and postcranial features, which have not, go hand in hand and that any lack of homoplasy in the former implies a lack of homoplasy in the latter. One caveat when comparing panin/hominin‐like fossils with a hypothetical LCA is, as Wood & Lonergan (2008) rightly point out, that we cannot be sure that they do not represent a sister clade of both which left no descendants: this will be particularly the case the closer we get to the time of the panin‐hominin split (4.2–6.5 Ma, Stone et al. 2010), when attributing a fossil to either clade, let alone to such hypothetical sister clades, will be increasingly difficult.

The development of ideas on hominization

It was first assumed that an increase of brain size was the Rubicon of hominization: Darwin (1871) stressed the ‘immense’ difference between the mind of humans and other animals, and set the stage for the Piltdown ‘discovery’ in 1912, which combined a 500‐year‐old orang‐utan jaw with some ground down and stained chimpanzee teeth and a fragmentary modern human cranium. Although some, such as Boule (1915) expressed early doubts, it was not until 1953 that Oakley used fluorine relative dating techniques to demonstrate that the components were not from the same age, hence exposing the forgery (Weiner & Oakley, 1954). Thus Dart's (1926) publication of the Australopithecus child from the Taung limeworks, with an endocast demonstrating a very small brain, did not gain complete acceptance as an early human ancestor until the 1950s, following publication of the Sts 5 skull (Broom, 1949). During the intervening time, Dart (1953) became convinced that early human ancestors were savannah hunters by the taphonomic association of further australopith fossils with those of game animals in cave fill in the (nowadays predominantly savannah) environments of the ancient Vredefort crater (‘The Cradle of Humankind’):

‘The ancestors of Australopithecus left their fellows in the trees of Central Africa through a spirit of adventure and the more attractive fleshy food that lay in the vast savannahs of the southern plains’. (Dart & Craig, 1959, p. 195; quoted by Sussman, 1999).

At about this time the uniquely close genetic relationships of hominins and panins was first being recognized (reviewed in e.g. Ruvolo, 1997) and the first systematic field studies of great apes, by Schaller (e.g. 1963) on mountain gorillas and Goodall (e.g. 1968) on chimpanzees were being carried out. Such early studies, just like the first field studies of Old World monkeys (e.g. Washburn & DeVore, 1961) were naturally enough performed on more readily observable species at more open field sites, whether montane woodland or woodland‐savannah environments. Thus, the behaviour of common chimpanzees, and savannah baboons, with a high level of inter‐individual aggression, exerted particularly by males, together with clear evidence of hunting, exerted a strong influence on ideas of the likely behaviour of the panin‐hominin LCA (reviewed in e.g. Susman, 1987, 1999; Cartmill, 1999; Stanford & Bunn, 2001). Here was Dart's (1953) small‐brained australopith ancestor: the ‘killer ape’. As long ago as 1859, Owen had differentiated ‘knuckle‐walking’ as a behaviour which supposedly distinguished chimpanzees, gorillas and orang‐utans from hylobatids. But for the first four decades of the 20th century, the dominant paradigm saw the origins of human bipedalism as arboreal, in the effect of suspensory locomotion on body posture (e.g. Keith, 1923; Morton, 1927; Gregory, 1934; Schultz, 1936). But now the high levels of knuckle‐walking seen in the first systematic field studies of the NHGAs, together with their occasional bipedalism and high degree of terrestriality, and the surprising preference for meat in chimpanzees, seemed to fit the idea that hominization was a grade‐shift in which forest‐dwelling, arboreal, fruit‐eating quadrupedal and suspensory apes gradually became transformed into carnivorous, terrestrial, savannah‐dwelling ‘obligate’ bipeds. As noted by Roberts & Thorpe (2014) this new model was epitomized by the Zallinger ‘The Road to Homo sapiens’ illustration in Howell (1965), which identified the LCA of chimpanzees and humans unequivocally as a terrestrial, knuckle‐walking quadruped. Whether or not it has had any influence on the scientific adoption of a chimpanzee model for human ancestry, there can be little doubt that the public perception of human origins has been influenced by the immediacy of this image and its synchronicity with Ardrey's (1961) sensationalist reworking, in African Genesis, of Dart & Craig's (1959) ‘killer ape’ model, further popularized by adoption in Kubrick's (1968) film, 2001: A Space Odyssey, and finally by the adoption of a similar and frankly androcentric model of human nature and evolution by the sociobiologists (see Wilson, 1975; Wrangham & Peterson, 1996). A recent and more subtle attempt to restore the scientific credibility of the killer‐ape hypothesis is Wilson et al. (2014). Thus, the chimpanzee model has both become the dominant scientific paradigm and one regretfully associated with androcentric, biological‐determinist views of ‘human nature’ which have even included ‘scientific’ rationalizations for human rape as an adaptive strategy.

The first full statement of a knuckle‐walking model for the origins of human bipedalism was Washburn (1967). It received functional‐anatomical and biomechanical elaboration by Tuttle & Basmajian (1974), and then by Gebo (1992, 1996). Gebo went as far as claiming that a biomechanical link between knuckle‐walking and ‘heel‐strike plantigrady’ constituted a major shared‐derived character of the African apes. The latter idea is ruled out by the presence of both plantigrady and a strong heel‐strike in bipedalism of the arboreal, non‐knuckle‐walking orang‐utans (Crompton et al. 2003, 2008). Nevertheless, although strong doubts were already being expressed by in the 1991 CRNS symposium by, for example, e.g. Susman & Stern, and by Senut; the knuckle‐walking model has long retained its popularity. This popularity appears to be based on the concept that the common panin‐hominin ancestor must have had chimpanzee‐like locomotion (Wood & Richmond, 2000; Richmond et al. 2001) and has undoubtedly been reinforced by the very high frequency of knuckle‐walking reported in the first quantitative assessments of both mountain gorilla and chimpanzee locomotion (95 and 86%, respectively, reviewed in e.g. Crompton et al. 2010 and see Hunt, this volume). If knuckle‐walking had indeed been a major element in the locomotor repertoire of the hominin‐panin LCA, it might have been expected to leave some evidence in the early hominin fossil record: supposed knuckle‐walking features (see e.g. Jenkins & Fleagle, 1975) include dorsal ridges on the distal metacarpals and os centrale‐scaphoid fusion; however, increasing numbers of discoveries of fossil hominins have failed to produce any such evidence in hominins. Stern and Susman (1983) found none in Australopithecus afarensis; Clarke (1999, 2002) none in the earlier and more complete Australopithecus prometheus; Ward et al. (1999, 2001) none in earlier again Australopithecus anamensis. In 2008, at the time of the second Journal of Anatomy special issue on human evolution, despite Dainton & Macho's (1999) suggestion that knuckle‐walking is not a homologous phenomenon in chimpanzees and gorillas, it was probably controversial (Wood & Elton, 2008) for Crompton et al. (2008) to reject the knuckle‐walking model. But 7 years later, the lack of any evidence of knuckle‐walking in Ardipithecus ramidus (Lovejoy et al. 2009a), Kivell & Begun's (2007) demonstration of no clear functional link between centrale‐scaphoid fusion and knuckle‐walking, and Kivell & Schmitt's (2009) argument that there are two distinct modes of knuckle‐walking, that in chimps with extended‐wrist postures in an arboreal environment and that in gorillas with neutral wrist postures in a terrestrial environment, suggests that our position, previously held by several others, may not now be considered so heretical.

If the locomotor apparatus of the LCA of panins and hominins was not chimpanzee‐like, at least not that of a knuckle‐walker, whether of terrestrial or arboreal type (as defined by Kivell & Schmitt, 2009), then what was it like? A recent attempt at a shotgun phylogenetic analysis of characters in Hominoidea (Duda & Zrzavý, 2013) failed to come up with a single locomotor feature for the LCAs of Hominidae, African apes or Homininae. We have two major more likely sources of information: first, quantitative data on the locomotor behaviour and musculoskeletal anatomy of living Hominidae; and secondly, the early hominin, protohominin, and Mid‐late Miocene hominoid fossil record, which is much better known now than it was in 2008, particularly with respect to the locomotor apparatus, and fortunately in several cases with associated craniodental remains.

As we have seen, genetic data suggests the lineages leading to Pan troglodytes (the common chimpanzee) and Pan paniscus, the bonobo, split between 2.5 and 1.78 Ma. Data suggest much higher levels of population‐genetic stability than in humans, with more continuity, Stone et al. (2002, 2010). However, 1.6% of the human genome is more closely related to the bonobo genome than to the chimpanzee genome, and 1.7% more closely related to the chimpanzee genome than to the bonobo genome (Prüfer et al. 2012). This fact echoes inconsistent patterns of phenotypic resemblances, e.g. in social and sexual behaviour, although the reality of these has been challenged and put down to unbalanced population sampling (Stanford, 1998). But rapid genetic/phenotypic differentiation of the two species seems well supported. The recent date of separation of the two lineages postdates clear evidence of multiple species of early hominins, which are apparent in both the Vredefort crater region of South Africa and in the East African Rift Valley from 2.5 to 3.5 Ma. It even postdates the earliest securely known date, 2.8 Ma, for differentiation of Australopithecus and Homo (Villmoare et al. 2015). It is certainly worth noting that bonobos are reported to display much lower percentages of knuckle‐walking and ‘knuckle‐running’ than chimpanzees (7.8 vs. 86.1% of locomotion) but much higher percentages of vertical climbing and descent (50.4 vs. 6.5%) (reviewed in Crompton et al. 2010). But as the genetic data suggest that much of locomotor behaviour is likely to be recently derived, not conserved (eg. Stone et al. 2002, 2010, and see similar conclusions on different grounds in Crompton et al. 2010), Pan may be a relatively uninformative genus.

The remaining African ape, the gorilla, which indeed Sir Grafton Elliot Smith (1927) suggested as a model for the early ancestor of humans, appears to have split from the Homininae over 10 Mya based on tooth morphology in the Ethiopian fossil Chororapithecus abyssinicus (Suwa et al. 2007). Gorillas fall into a single species, with two main populations: the Western ‘lowland’ gorillas and Eastern ‘mountain’ gorillas, which are accorded the subspecific status of Gorilla gorilla gorilla and Gorilla gorilla beringei. More detailed fieldwork has been carried out on G. g. beringei, but most of the data comes from one or two locations at high altitude with limited availability of fruiting trees, particularly Karisoke in the Virungas. Observability is easy here, but at lower altitudes, particularly in the canopy, it is likely to be a substantial problem, and this is also a major issue in studies of G. g. gorilla. Although comparability of method is often a problem, as discussed in our most recent review (Crompton et al. 2010), we gave estimated values of 95% knuckle‐walking/knuckle‐running locomotion (whether arboreal or terrestrial) for G. g. beringei vs. 38.5% for G. g. gorilla, less than 1% vertical climbing/descent for G. g. beringei vs. 19.7% in G. g. gorilla, and 0.8% bipedalism and orthograde clamber/transfer in G. g. beringei vs. 6.1 and 3.3% in G. g. gorilla. Studies of locomotion of populations between G. g. gorilla and G. g. berengei, sometimes assigned (See eg. Bergl et al. 2011) to G. g. dielhi (Cross River gorilla) and G. g. graueri (eastern lowland gorilla), have not yet been published, but available data on ecology suggests they are not dissimilar in dietary habits to G. g. beringei from lower altitudes (see below), with relatively high fruit consumption and hence likely relatively high arboreality (Yamagiwa & Basabose, 2006).

In a laudable foray into (albeit museum‐based) ecomorphology, Dunn et al. (2014, p. 526) attempted to ‘test the hypothesis that gorilla talus morphology falls along a morphocline that tracks locomotor function related to a more inverted or everted foot set’. Their study follows a long tradition in palaeoanthropology, beginning with Morton (1927), of regarding particularly the shape of the talar trochlea as bearing a ‘signal’ indicating foot function. However, Dunn et al.'s (2014) ecological perspective is based on the premise that (p. 526) ‘all gorillas are obligate terrestrial knuckle‐walking quadrupeds’ although ‘those that live in lowland habitats eat fruits and climb more often than do those living in highland habitats’. Thus, they argue, a more ‘terrestrial’ lifestyle in G. g. beringei produces a foot which is morphologically (Schultz, 1927) and, claim Dunn et al. (2014), therefore functionally, more like that supposed to exist in humans than that of G. g. gorilla, with more inverted foot posture. (Presumably, this would imply a more medial path of the centre of pressure.) One must wonder about the extent to which we can expect to understand morphological clines in ecological terms if the 38.5% arboreal and terrestrial knuckle‐walking/running of G. g. gorilla is regarded as constituting ‘obligate’ terrestrial knuckle‐walking. ‘Obligate’ in its biological usage is defined by the Compact Edition of the Oxford English Dictionary as ‘that is of necessity such’. How can lowland gorillas be ‘of necessity’ terrestrial knuckle‐walkers when, on the basis of existing reports, 60% of their locomotion at least is not terrestrial knuckle‐walking? On the basis of the studies reviewed by Crompton et al. (2010) it might be safer to state that G. g. beringei is an obligate terrestrial knuckle‐walker; but although studies, for example of Karisoke (see e.g. Schaller, 1963; Doran, 1996), do tend to give this impression, Remis (1995, 1999) has noted that habitat structure and resource availability exerts substantial influence on gorilla locomotion. Remis (1999) suggests that locomotion exhibited at sites like Karisoke represent ‘adaptation’ to a locally extreme high‐altitude dwarf forest environment. Indeed, while study sites in the Virungas lie between 2700 and 3400 m, at extreme altitudes for gorillas, the ‘mountain’ gorillas of Bwindi live in forest at altitudes of 1160–2607 m, whereas the Bai Hokou lowland gorillas live at around 463 m. At Bai Hokou, canopy cover is near 100%, twice the cover in the densest parts of Karisoke, with five times the tree density. Whereas mean canopy height is around 40 m at Bai Hokou, with abundant arboreal fruit resources, trees at Karisoke are less than 7 m high (Remis, 1998), and arboreal fruit is very rare or non‐existent, so that it forms almost no part of the diet; terrestrial herbaceous forage, however, is abundant (Robbins & McNeilage, 2003).

However, the ‘mountain’ gorillas of Bwindi, only 25 km from the Virunga Volcanoes Conservation Area, live in forest at 1160–2607 m, and arboreal fruit is both more abundant and more used (on 27% of observation days according to Robbins & McNeilage, 2003). Stanford and Nkurunungi (2003) found that 24.6% of Bwindi gorilla diet was fruit.

Comparing Karisoke and Bwindi, Ganas et al. (2004) found that with a higher mean annual temperature and greater plant diversity, and particularly much greater fruit availability, the dietary choices of Bwindi and Karisoke mountain gorillas were 71% attributable to availability and only 28.8% attributable to ‘food profitability or local traditions’.

Gorilla g. beringei in the Bwindi Impenetrable Forest are genetically indistinguishable from those in the Virungas (Garner & Ryder, 1996; Jensen‐Seaman & Kidd, 2001). However, Redmond (pers. comm. to Colleen Goh) has observed G. g. beringei climbing up to heights of 40 m, and film records made available by him to Goh show G. g. beringei using not only vertical descent but orthograde clamber/transfer and, in the latter, using the hallux to grasp small supports, just like human indigenous populations who engage in arboreal foraging, such as the Batek (see e.g. Kraft et al. 2014). Of course, a full quantification of such behaviour – in both low‐altitude G. g. beringei and human tree climbers – is required, but our (Crompton et al. 2010) zero estimate for orthograde clamber/transfer is clearly wrong, and our <1% vertical/climb/descent very likely a very substantial underestimate.

Intriguingly, both gorillas (Watts, 1984, 1990; Tuttle & Watts, 1985; Remis, 1994, 1995) and human indigenous hunter‐gatherers (Kraft et al. 2014) seem to concentrate activity either at the centre of trees, utilizing both tree‐trunks and vines to change height, or on main branches, and only smaller gorillas venture often out into the peripheral canopy. In both cases, high‐energy foodstuffs appear to be the attraction: in the case of gorillas, fruit, and in the case of humans, honey and grubs of colonial hymenopterans (Kraft et al. 2014). In the case of the Bwindi gorillas, there is some, if certainly inconclusive, evidence that interspecific aggression with common chimpanzees occurs and we may speculate that this might be in part responsible: certainly, lower selectivity in dietary fruit and greater intake of lower quality vegetation and bark is consistent with evidence that gorillas may come off worse in interspecific encounters (Stanford & Nkurunungi, 2003). Parallel, or retained shared‐derived (African ape LCA) or possibly shared‐ancestral (hominid LCA?) characters in gorilla and human ecology and climbing behaviour thus do seem to exist, as well as in the morphology of the cheiridia. The existence of orthograde clamber/transfer in G. g. beringei is certainly pertinent to the suggestion by Thorpe et al. (2007) that hand‐assisted bipedality may have arisen as an adaptation to access to the peripheral canopy by enabling use of multiple small supports. However, it raises the question of why humans and gorillas should then have short, adducted halluces, which would seem to favour the preference for the tree core that characterizes the arboreal activity of both. I will suggest below that the recently discovered fossil hominin record is pertinent to this question.

Whereas African apes as a whole, including humans, are clearly much more terrestrial than Asian apes, and seem to travel primarily on the ground, rather than at canopy level as do orang‐utans, gorillas all seem to make use of arboreal fruit resources when they are available and retain a locomotor repertoire not dissimilar to that of orang‐utans (Thorpe & Crompton, 2006; Crompton et al. 2010). Humans have unquestionably become more terrestrial again than other African apes. But unless it can be shown, rather than merely assumed, that human hallux function is primarily an adaptation to hallucal toe‐off, not to use of small (single) supports in arboreal climbing, we must consider that the particular similarities which unquestionably exist between the feet of gorillas in particular and those of humans may be either (i) homoplastic, because of similar tree‐climbing methods; (ii) shared‐ancestral; or perhaps (iii) shared‐derived for African apes – bonobos and chimpanzees having subsequently lost them. That does not imply that the characters of similarity between the feet of gorillas and humans are not valuable in activities which are often (but not always) seen in a terrestrial context. Wang & Crompton (2004) showed that the static loads experienced by the foot in human‐like bipedal standing are indeed more human‐like in gorillas than in other living great apes, and Wang et al. (2014) have now extended this to show that joint torque and work in the foot during bipedal walking (using whole body inertial models, but driven by human walking functions) are again more human‐like in gorillas than are those in other great apes. This suggests that upright bipedal standing and walking may favour foot proportions in gorillas and humans, with a long tarsus and short lateral phalanges (Schultz, 1963). However, the almost completely arboreal orang‐utan, because of its very extended hip and knee postures, can sustain the most efficient bipedal walking, and this despite the derived elongation and curvature of the cheiridia, which assist its orthograde clambering and unusual pronograde suspension (Crompton et al. 2007). Lowland gorillas at least, particularly females (Remis, 1995; Thorpe & Crompton, 2006), engage in arboreal orthograde standing and walking, and in this and other respects have generally rather similar locomotor repertoires to orang‐utans (Thorpe & Crompton, 2006). But on the basis of quantitative studies of NHGA hindlimb architecture by Payne et al. (2006a,b), Thorpe & Crompton (2006) suggest that bipedal walking of gorillas and orang‐utans is not dynamically similar. Although Myatt et al. (2012) found no differences in hindlimb muscle architecture in a larger sample of species and individuals, available evidence still suggests that muscle moment arms of orang‐utans and gorillas differ (Payne et al. 2006b) and, kinematically, gorillas appear to have a smaller abduction of the hip, but greater flexion of the swing‐leg, whereas orang‐utans keep an extended posture of the knee and have higher hip extension (see e.g. Crompton et al. 2003, 2008; Watson et al. 2009). As adult, especially male, gorillas tend to concentrate activity on large boughs and tree trunks (e.g. Tuttle & Watts, 1985; Remis, 1995, 1999) the advantage orang‐utans gain in hand‐assisted bipedality on smaller, more compliant branches from extended hip and knee postures (Thorpe et al. 2009) may not accrue, or may not accrue to the same extent.

Both orang‐utans and panins have relatively elongated lateral phalanges. The similarity of humans and gorillas suggests that panins have acquired these features, rather than conserved them from a common African ape ancestor. The most relevant fossil crown hominoid, from its completeness and because its date antecedes or at least is paenecontemporaneous with dates for separation of the orang‐utan clade from the African ape clade, is Pierolapithecus catalaunicus (12.5–13 Ma, Moyà‐Solà et al. 2004). From the evidence for its hands, it does not appear to have exhibited extensive suspensory locomotion (Moyà‐Solà et al. 2004, 2005; Almécija et al. 2009). This implies that suspensory adaptations were derived independently in African and Asian apes, and that orthogrady, first evidenced in Morotopithecus (MacLatchy et al. 2000), arose in a fully arboreal context but in a compressive, in other words, bipedal mode (see e.g. Thorpe et al. 2007; Crompton et al. 2010). However, Pierolapithecus does not preserve enough foot bones to help with the issue of the proportions of the pedal digits in the common African ape ancestor. The later Hispanopithecus laeitanus (9.5 Ma) appears to be a pongine (Almécija et al. 2007) with a suspensory habit, but Hispanopithecus lacks any evidence of the feet.

Gorillas have further similarities to humans in the shortness of the hand in relation to the rest of the upper limb, the joint shortest in living apes (Schultz, 1927, 1930). While first ray lengths are short relative to third, compared with the human case, and the pollical phalanges also short relative to ray length, proportions of 3rd ray phalanges to 3rd ray length are unremarkable (Schultz, 1930). However, in modern humans, the short hand is not accompanied by a long brachium and antebrachium, and vertical climbing on large trunks often makes use of technological alternatives, whether a vine‐loop (Kraft et al. 2014) or the equivalent climbing aids of western tree‐surgeons. It is difficult to see how such similarity in hand proportions could have come about because of high levels of terrestriality, given the absence of any evidence for a knuckle‐walking period in human ancestry. Either there is an as yet unrecognized similarity in hand function in climbing, or humans and gorillas share rather conservative hands, with the exception of the human thumb. The relatively limited craniad orientation of the glenoid fossa in gorillas (Tuttle & Watts, 1985) allows the speculation that the upper limb of both may be well equipped to provide supportive and balancing torques to counter knee flexion under gravity during bipedalism as well as light‐touch balance at shoulder height or below.

If knuckle‐walking is functionally different and independently derived in Pan and Gorilla, and unlikely to have formed a major element of the locomotor repertoire of the common hominine stock, what of the ‘vertical climbing’ hypothesis of Fleagle et al. (1981)? This is based on the claim that hip and thigh muscle activation patterns in ‘vertical climbing’ are more similar to those in bipedalism than are those in other behaviours. Vertical climbing is seen in all great apes and it is therefore not tied to a chimpanzee or bonobo model of the locomotor behaviour of the common hominin ancestor. An immediate issue with the argument, however, is that ‘other behaviours’ do not of course include bipedalism itself, and yet all living apes include bipedalism in their repertoire, whether they be primarily terrestrial or primarily arboreal (see e.g. Thorpe & Crompton, 2006; Crompton et al. 2008). In addition, hip extension is markedly greater in such bipedalism than in vertical climbing in all NHGAs (see e.g. Crompton et al. 2003).

Qualitatively, Preuschoft & Witte (1991) suggested that in NHGAs, the flexed hindlimbs allow the feet to exert balancing torques against the tree trunk, against the gravitational force tending to rotate the upper body away from it while the elongated arms pull the body towards the trunk and upwards. Preuschoft contrasts this with what he sees as the case in humans and lemurs, where hindlimbs remain much more extended.

Isler (2002, 2003) and Isler & Thorpe (2003) found that in vertical climbing orang‐utans do not differ extensively in hindlimb flexion from Pan and Gorilla. However, they use longer stride lengths and lower stride frequencies, particularly in the forelimb, with more extended shoulder and elbow at hand contact, and the foot more elevated relative to the hip. Lower stride frequencies are consistent with the finding of Myatt et al. (2011, 2012) that no consistent quantitative differences exist in either hindlimb or forelimb muscle architecture between the orang‐utans and the nonhuman African apes, which would imply that longer stride lengths in vertical climbing must be offset by lower stride frequencies. Myatt et al.'s (2011, 2012) broader conclusion is that selection has been, if anything, for plasticity in capability within a broad envelope of possible behaviours. Similarly, Isler et al. (2006) found that segment inertial properties show extensive overlap between NHGA genera. Segment mass distribution also overlaps between NHGAs and humans, with the exception of the shank. However, owing to a different mass distribution between the limb segments, the centre of mass of both arms and legs is more distal in NHGAs than in humans and, with the exception of common chimpanzees in the Isler et al. (2006) sample, where differences were very small, the natural pendular period of NHGA forelimbs is larger than that of the hindlimbs, in contrast to the limbs of both cursorial mammals and cercopithecoids (Isler et al. 2006). Thus, as mechanical principles show that a mismatch in pendular period, and hence disturbance of the coordination of limb swing frequencies between fore‐ and hindlimbs, must be inefficient in quadrupedalism, just as it is in bipedalism (Wang et al. 2003a,b), it appears that hominoid limbs are not optimized for efficiency in quadrupedal walking but rather for an ability to perform well in various locomotor modes. Only common chimpanzees differ, having proportions which favour a more efficient quadrupedal gait.

The narrower date range for the hominin–panin split which now prevails (4.2–6.5 Ma, Stone et al. 2010; vs. 4–8 Ma, Bradley 2008) raises some interesting questions. The protohominin Oreopithecus, at ca. 7–9 Ma, clearly antedates the split. It shows unequivocally that a mix of orthograde clambering/hand‐assisted bipedality can lead to hominin‐like features of pelvis, lumbar spine and femur, with retention of a markedly abducted hallux, but features of the pollex which foreshadow a precision grip (Harrison & Rook, 1997; Köhler & Moyà‐Solà, 1997, 2003; Moyà‐Solà et al. 1999). However, Sahelanthropus, at 6–7 Ma (Brunet et al. 1995, 2002; Brunet, 2002) remains more enigmatic: the face could be that of a female gorilla, and the teeth are like those of Ardipithecus kaddaba (Haile‐Selassie et al. 2004), rather similar to those of bonobos. The significance of the forwards position of the foramen magnum is disputed: Schaefer (1999) arguing for overlap in foramen magnum position between Homo and Pan, and Russo & Kirk (2013) arguing for the validity of foramen magnum position as an indicator of bipedalism on the basis of patterns in mammals as a whole. Nonetheless, habitual orthogrady – rather than, for example, Pan‐like vertical‐climbing habits – currently appears more likely. Thus, so near to the time of the panin–hominin split, the morphology of Sahelanthropus does not tend to endorse a vertical climbing model for the origins of bipedality.

Orrorin, at 5.7–6.1 Ma (Sawada et al. 2002), shows a range of features (Senut et al. 2001; Senut, 2003) that convince me at least that it is a good match for the arboreal orthograde clamberer/hand‐assisted biped envisioned by Thorpe et al. (2007). The presence of an obturator externus groove on the back of the femoral neck does not, in isolation, make it a biped: Stern & Susman (1983) argue that this feature also appears in some pitheciines, atelines and Pongo: but hip hyperextension in bipedalism certainly characterizes the latter, and all three use it in bridging (R. H. Crompton, pers. obs.) and so are unquestionably signalled by the strength of the groove in Orrorin. While one sympathizes with Stern & Susman's (1991) preference for establishment of a ‘total morphological pattern’ over the single ‘magic trait’, the obturator externus tendon groove in Orrorin does form part of a consistent total morphological pattern, that of an arboreal, and perhaps semi‐terrestrial, biped.

At the time of the 2008 Journal of Anatomy review issue, the single best‐known early hominin skeleton remained the AL‐288‐1 ‘Lucy’ skeleton of Australopithecus afarensis, discovered in 1974, and with reference to the interpretation of which the Stern & Susman (1991) paper concerned itself. Briefly, an intractable debate ensued over whether this long‐armed, short‐legged, 1.1‐m hominin, and other members of the hypodigm, such as AL‐333 were semi‐arboreal creatures whose apparent adaptations to arboreality would have ‘compromised’ its obvious terrestrial bipedality, so that it would have walked ‘bent‐hip, bent‐knee’ as Stern and Susman (e.g. 1983, 1991) and Susman and Stern (1984) held, or whether arboreal features were anachronisms and it was an effective, committed, fully erect biped as Latimer (1983, 1991) and Latimer & Lovejoy (1982, 1989, 1990a,b) held. The debate was not conceded on either side.

Inspired by a chance meeting with Stern in Paris at the time of the 1990 CRNS symposium ‘Origines de la Bipédie chez les Hominidés’, my own group undertook a series of studies (e.g. Crompton et al. 1998, 2012; Wang et al., 2004; Sellers et al., 2005), in which we implemented ‘total morphological pattern’ in silico by inverse and forwards dynamic modelling, and made use of topographical statistics to calculate the mean tendency and relationships of the Laetoli G‐1 footprint trails. These trails, attributed to Au. afarensis, as with the fossil bones themselves, have been subject to often almost diametrically opposed interpretations (reviewed in Crompton et al. 2008, 2010). Our modelling studies (along with ancillary experimental studies (e.g. Wang et al. 2003a,b; Carey & Crompton, 2005) showed for example that the energetic costs of transport would double, and core body temperature increase, if ‘Lucy’ had walked bipedally, but that this species could have been effective upright bipedal walkers over relatively short distances if unloaded (Crompton et al. 1998; Wang et al. 2004; Sellers et al. 2005 and see Ward, 2002). Further, they showed that stride lengths in the G‐1 trackway are equivalent to typical small‐town walking speeds of modern humans. Our studies of the central tendency of the Laetoli G‐1 trail (Crompton et al. 2012) showed that the maker of the G‐1 prints (whose taxonomic affiliation remains uncertain) certainly walked upright, not BHBK, had a clear medial arch and, despite Meldrum et al.'s (2011) claim from a mark on a single footprint, no mid‐tarsal break. They further had lateral to medial pressure transfer and a hallucal ‘toe‐off’ impression. Altogether, the prints have, at least qualitatively, the main functional features of prints created by modern human feet, 3.65 million years later.

Despite the concerns expressed by Stern & Susman (1991) about use of ‘magic traits’ palaeoanthropologists continue to make functional generalizations from single bones. Ward et al. (2011) argued that torsions in the shaft, and the shape of the base of a single complete fourth metatarsal of Au. afarensis show that this species had pedal arches and no midtarsal break. Considering also the morphology of the Ardipithecus ramidus foot, the AL‐333 foot, and the Laetoli footprints, Ward et al. (2011) conclude that the lateral midfoot of Au. afarensis was ‘rigid’ and that: ‘The evolutionary trajectory suggested by these fossil remains makes it unlikely that selection continued to favor substantial arboreal behaviors by the time of A. afarensis’ (p. 753). However, Crompton et al. (2008) had already figured a human foot pressure record with a mid‐tarsal break and in 2013 we (Bates et al. 2013) showed that 7% of 21 500 human foot pressure records show lateral mid‐foot pressures over 200 kpa, the clinical threshold for diagnosis of midfoot collapse. Our database is now well in excess of 30 000 records and continues to show that two‐thirds of us produce one or more such records with a mid‐tarsal break within 5 min of treadmill walking. Thus, it hardly matters if a single fourth metatarsal of Au. afarensis (Ward et al. 2011) or the whole Laetoli G‐1 trackway (Crompton et al. 2012) show no evidence of a mid‐tarsal break: most Homo sapiens produce steps with a mid‐tarsal break as a quite normal, if transient, phenomenon. Our work on bipedalism of orang‐utans (Thorpe et al. 2007) was criticized by Begun et al. (2007), p. 1066d), who claim that ‘Pongo does not load the lower limbs in a manner even approaching the human condition (with extended lower limbs positioned beneath the center of mass, knees in sagittal plane, feet fully plantigrade’ and that ‘chimpanzees and bonobos appear to be better bipeds’, ignoring previously published work (e.g. Crompton et al. 2003 and see Crompton et al. 2007), showing that the whole‐body kinematics of orang‐utan bipedalism gait more closely resembles those of humans than do those of chimpanzees. We have since shown (Crompton et al. 2008) that, indeed, the kinematics of orang‐utan bipedalism achieves energy transformation rates (stride‐to‐stride energy savings) of some 50%, much closer to the human 70% energy savings than to the 8% seen in chimpanzees. Further, despite differences in foot morphology, we then (Bates et al. 2013) showed that the distribution of midfoot pressure in humans, bonobos and even the long, curved feet of orang‐utans actually overlaps. Humans do have relatively stiff feet for a great ape, but they are far from rigid.

Thus, as even some modern human populations do exploit arboreal resources by unassisted tree climbing as part of their normal foraging (Kraft et al. 2014) Ward and colleagues’ (2011) claim that it is ‘unlikely that selection continued to favor substantial arboreal behaviors by the time of A. afarensis’ ceases to have much significance. Even Homo sapiens does indeed retain – we must presume, from its arboreal ancestry (Crompton et al. 2010) – a foot which, although relatively stiff, retains substantial midfoot compliance and frequently displays a mid‐tarsal break, the character absence of which is traditionally supposed (since Elftman & Manter, 1935a,b) to distinguish our feet from those of other great apes. The perhaps surprising overlap in foot pressure, that is, in external function, between human feet and the long‐toed, curved feet of even orang‐utans is explicable through the engineering concept of ‘functional redundancy’ (see e.g. Alexander, 2003). Generally, the more components there are in a system, the less a single optimal solution is likely to exist, so that multiple alternative solutions with broadly equivalent function (a solution‐space) will tend to exist. A system like the foot, with many bones, joints, muscles, ligaments and tendons, will be able to produce the same external function in different ways in different individuals. For example, some individuals produce more midfoot flexion at the calcaneocuboid joint, some at the cuboid‐fifth metatarsal joint (Lundgren et al. 2008).

With an increasing pace of fossil hominin discovery, there is now abundant evidence that substantial arboreality continued in some Australopithecus species at least, just as it seems to have done in Homo sapiens. The highly abducted hallux of the footbones of the 3.4 Ma Woranso‐Mille hominin suggested to Haile‐Selassie et al. (2012) that close to the time of Au. afarensis, other hominins may have had hallucal function more similar to that claimed by Lovejoy et al. (2009b) for the predominantly arboreal Ardipithecus: that is, ‘gripping rather than propulsive’. Haile‐Selassie et al. (2012) found ratios of the fourth metatarsal to fall with the gorilla cluster. Indeed, gorillas and modern humans overlapped in this parameter. An associated scapula from Woranso‐Mille bears particular morphological resemblances to that of Pongo (according to Churchill et al. 2013), suggesting a similarly labile locomotor repertoire. Further, the 3.3 Ma juvenile partial skeleton of Au. afarensis from Dikika was originally described as combining long and curved manual phalanges with a gorilla‐like scapula (Alemseged et al. 2006), but Green & Alemseged (2012) now stress affinities to Pongo.

The later (1.98 Ma) partial australopith skeletons from Malapa in South Africa referred to Au. sediba again share ‘with other australopiths forelimb features that have been interpreted as indicative of an arboreal habitus’ (Churchill et al. 2013, p. 4), Again, the scapula resembles that of Pongo and, to a lesser extent, that of Gorilla rather than those of other hominoids. The Au. sediba hand combines evidence of strong flexor muscles with a long thumb and short fingers, which suggest to Kivell et al. (2011) precision gripping and possible stone tool manufacture. This idea is very similar to that presented much earlier by Clarke (1999) in reference to the StW 573 skeleton of Au. prometheus. Some confusion exists concerning the foot of Au. sediba. Zipfel et al. (2011) note that the talocrural joint is rather human‐like, and there also may be evidence of a human‐like arch and even an Achilles’ tendon. The latter at least would argue for terrestrial adaptation, in contradiction to the evidence of the forelimb. DeSilva (2009) claimed that the human ankle joint is incompatible with use of the plantar surface of the feet ventral to the trunk, as in vertical climbing of chimpanzees. DeSilva et al. (2013) then elaborated an hypothesis that in terrestrial bipedalism Au. sediba must have been obligate hyperpronators, to counter a ‘required’ inverted set of the ankle to allow the feet to be used against the side of tree‐trunks in climbing. They bolstered this idea by details of, for example, the position of the tuber calcanei. However, Venkataraman et al. (2013a,b) have demonstrated that soft‐tissue plasticity in modern human arboreal foragers is quite adequate to counter any derived features of the ankle associated with bipedalism, so much that sufficient dorsiflexion exists to allow chimpanzee‐like vertical climbing with the plantar surface of the foot placed against the tree, directly ventral to the trunk. Venkataraman et al. (2013a,b) found that most traits of distal tibia morphology associated with differences in climbing frequencies and behaviour in living NHGAs are not distinguishable between human populations which do, and do not, engage in arboreal foraging. They conclude that the demands of terrestrial bipedalism swamp any morphological ‘signal’ for climbing. However, we might equally conclude that these morphological features make little difference to climbing performance. This conclusion bears on the significance of the almost right‐angled relationship between the trochlear facet and shaft of the Au. anamensis partial tibia from Kanapoi, KNM‐KP 29285. Ward et al. (2001) regarded it as indicating such a reduction of arboreal capability by 3.9–4.2 Ma suggesting habitual, even obligate, terrestrial bipedality. However, it would appear that human terrestrial bipedalism is by no means incompatible with maintenance of a high level of arboreal capability, at least in climbing in the core, rather than peripheral canopy, of trees (cf. Kraft et al. 2014).

The single most complete early hominin skeleton is StW 573 Au. prometheus from the Silverberg Grotto at Sterkfontein, South Africa, which has recently been securely dated at 3.67 Ma (Granger et al. 2015), following resolution of stratigraphic issues related to flowstone intrusion which lead to dates between 2.2 and 4 Ma being assigned. Although some parts of the postcranium are still in matrix, and thus awaiting full description, this specimen is so complete that limb proportions are unequivocal. This female was probably of similar stature to a relatively small modern human woman, and unlike the younger Lucy AL‐288‐1 had subequal arms and legs (Clarke, 1999, 2002; R. Clarke, pers. comm.; R. H. Crompton, pers. obs.). She combines human‐like knee joints with shoulder and elbow joints more resembling those of a gorilla or orang‐utan. The hallux is moderately divergent, perhaps similar to that of the Woranso‐Mille foot, and the uniquely complete hand has short fingers and a long, powerful thumb, rather resembling the much later Au. sediba (Clarke, 1999, 2002; R. Clarke, pers. comm.; R. H. Crompton, pers. obs.). It is unquestionably the skeleton of a substantially arboreal hominin which was also a capable terrestrial biped.

Lovejoy et al. (2009a,b,c) regard Ardipithecus ramidus (ca. 4.4 Ma) as close to the LCA of hominins with panins, although 4.4 Ma is quite late for that to be the case. Its upper limb shows no particular features of a vertical climber or of a knuckle‐walker, and its short hands are consistent with those of some, but not all, later hominins: thus use of ‘careful climbing’ (Lovejoy et al. 2009a) seems a reasonable conclusion. However, Lovejoy et al. (2009b) claim it combines rather stiff feet with a strongly abducted hallux. Given the substantial arboreality now attributed to other, later hominins with (from evidence of the knee joints) greater terrestrial capability but similarly abducted halluces (such as Au. prometheus; the Dikika Au. afarensis, and Au. sediba), Lovejoy and colleagues' (2009a,b,c) description of A. ramidus as bipedal on the ground but primarily quadrupedal in the trees seems untenable.

Concluding remarks

Thus, although Wood & Elton (2008) thought it controversial for us (Thorpe et al. 2007 and Crompton et al. 2008) to argue strongly for an arboreal origin for bipedalism in orang‐utan or gorilla‐like arboreal orthograde clambering and hand‐assisted bipedality, the weight of the fossil evidence now seems to endorse this proposal. Indeed, there was little new about the overall idea, as an ‘arboreal origins’ model of one kind or another goes back to Jones (1916), Keith (1923) and Smith (1927). We now know, as we did not in 2008, that AL‐288‐1 Lucy was very short for an australopith, and that long arms and short legs, whether or not they go with short stature (see e.g. Pontzer, 2012), are not likely to have been typical in australopiths, although limb segment length was evidently variable. We know also that in South (e.g. Clarke, 2002) and probably East Africa, multiple australopith species, quite likely occupying somewhat different (e.g. more or less mesic) environments, coexisted more or less sympatrically and synchronically. Indeed from 3.5 Ma onwards, hominins tend to be associated with ecologically diverse habitats (see e.g. Bedaso et al. 2013). This may well have facilitated niche differentiation and selected for retention of the remarkable phenotypic, ecological and locomotor plasticity which we have seen is typical of family Hominidae.

Genus Homo may be as old as 2.4–2.8 Ma (Schrenk et al. 1993; Villmoare et al. 2015) and by the younger date is known from Ethiopia to Malawi. If OH 62 is regarded as part of the hypodigm for Homo habilis, the type being OH 7, the postcranium is in no substantial way distinguishable from even the longest‐armed, shortest‐legged australopith. But Clarke (2013) proposes that OH 62 is not part of the Homo habilis hypodigm, which should rather include KNM‐ER 1470 and OH 65 with OH7. If Clarke is correct, then the KNM‐ER 1472 and KNM‐ER 1500 postcranials, with a morphology (and thus probable body stature) very similar to early African Homo erectus, would also be Homo habilis, and this species would be clearly distinct in ecology from Australopithecus spp. This raises an interesting question that we can at present only address speculatively.

As the date range at which Homo appears is more or less contemporaneous with the date of splitting of panins into their present two clades, was it perhaps aggressive interactions with chimpanzees which led to hominins becoming more restricted to the cores of forest and woodland trees and to the ground, just as may be the case with gorillas? Was the increase in stature seen from Australopithecus to Homo another response to competition with chimpanzees? Something, after all, must have lead to hominins becoming progressively more associated with the more open environments, which had been available since 7–8 Ma (see e.g. Elton, 2008). But just as Homo sapiens is typically hominid and typically conservative in retaining ecological plasticity and arboreal capabilities, it appears that australopiths, in the form of Paranthropus, may have retained substantial arboreality as late as 1.34 Ma (Dominguez‐Rodrigo et al. 2013). Of course it is quite possible that, as Oxnard (1973) argued long ago, australopiths are not directly ancestral to Homo, but the common ancestor would still most likely have been both arboreal and bipedal (see Senut, 2006).

Nevertheless, all the above tends to point to generalized arboreal clambering/orthograde scrambling and selection for ecological plasticity being key to African ape, including hominin, evolution, in the period of climate change which set in from the mid‐Miocene (see e.g. Elton, 2008).

For myself, having begun my career working on locomotor ecology of bushbabies and tarsiers, two clades in which it is relatively easy to discern homoplastic continua of ecomorphological adaptation (e.g. in calcaneus length) on the theme of a clade‐typical locomotor adaptation, whether this should be termed ‘vertical clinging and leaping’ (see e.g. Oxnard et al. 1990; Crompton & Sellers, 2007), the much broader range of capabilities which may be elicited in given environments (locomotor totipotentiality, sensu Prost, 1965) in Hominoidea and indeed Hominini is both salutary and refreshing!

I hope this review has highlighted several cases where qualitative descriptions have become enshrined as paradigms from which palaeoanthropology has found it difficult to escape, such as the idea that humans are somehow not ‘ape‐like’; that this ‘otherness’ includes and indeed is typified by a rigid lateral midfoot, and of course the misconception which has been brought about by the use of the name ‘mountain gorilla’. Paradigms should always be, but all too often are not, challenged by testing against quantitative data.