The arboreal origins of human
bipedalism
Susannah K.S. Thorpe
1
, Juliet M. McClymont
2
&
Robin H. Crompton
2
Almost a century and a half ago, Charles Darwin in The Descent of Man (1871: 141)
highlighted the evolution of bipedalism as one of the key features of the human lineage, freeing the
hands for carrying and for using and making tools. But how did it arise? The famous footprints
from Laetoli in Tanzania show that hominin ancestors were walking upright by at least 3.65
million years ago. Recent work, however, suggests a much earlier origin for bipedalism, in a
Miocene primate ancestor that was still predominantly tree-dwelling. Here Susannah Thorpe,
Juliet McClymont and Robin Crompton set out the evidence for that hypothesis and reject the
notion that the common ancestor of great apes and humans was a knuckle-walking terrestrial
species, as are gorillas and chimpanzees today. The article is followed by a series of comments,
rounded off by a reply from the authors.
Theories regarding the origins of hominin bipedalism have spent some considerable time
‘on the ground’ as a result of the knuckle-walking hypothesis, which postulates that our
earliest bipedal ancestor evolved from an ape that knuckle-walked on the ground in a way
similar to modern chimpanzees or gorillas. By contrast, we argue that there is compelling
and unequivocal evidence that bipedalism has arboreal origins.
The concept of an arboreal origin for habitual human bipedalism was first proposed over a
century ago. The arboreal behaviour that was considered to be exaptive (i.e. to have ‘prepared’
the body) for bipedalism has, however, changed fundamentally with the gradual discoveries
of new fossil evidence, and with the development of new approaches to reconstructing
the ecology and locomotion of extinct species. In particular, study of the ecology and
biomechanics of living apes has transformed our understanding of how bipedalism could
have evolved. Living apes offer broad models for how the dynamic between habitat and
morphology may combine to influence locomotor behaviour. Sir Arthur Keith (1903) was
the first to suggest that the arboreal locomotion of apes was important in understanding
the process by which upright posture evolved in human ancestors. His studies of primate
anatomy and behaviour led to the paradigm that an ape that moved by brachiating (arm-
swinging) underneath branches (suspension) later evolved into a habitual biped (e.g. Morton
1922; Keith 1923). Morphological and locomotor observations continued to be proffered
in support of this hypothesis for many decades (see Tuttle 1974 for a review). However,
1
Locomotor Ecology and Biomechanics L aboratory, School of Biosciences, University of Birmingham, Edgbaston,
Birmingham B15 2TT, UK (Author for correspondence; Email: s.k.thorpe@bham.ac.uk)
2
Department of Musculoskeletal Biology, Institute of Aging and Chronic Disease, University of Liverpool, Ashton
Street, Liverpool L69 3GE, UK
C
Antiquity Publications Ltd.
ANTIQUITY 88 (2014): 906–926 http://antiquity.ac.uk/ant/088/ant0880906.htm
906
https://doi.org/10.1017/S0003598X00050778 Published online by Cambridge University Press
Debate
Susannah K.S. Thorpe, Juliet M. McClymont & Robin H. Crompton
one of the most important lines of evidence to emerge relatively recently from new fossil
discoveries is that adaptations to suspension and arm-swinging must have evolved not once
only but convergently, across several millions of years, in multiple fossil ape species (e.g.
Almecija et al. 2009).
During the 1970s and ‘80s, Russell H. Tuttle (1974, 1981) proposed that the arboreal
ancestor of modern hominins would have been small-bodied—around the size of living
gibbons (9–13.5kg)—and would have engaged extensively in vertical climbing (that is,
climbing up and down vertical tree trunks with the torso in an upright position), an activity
that he considered to be functionally associated with bipedalism. The biomechanical link
was defined by Prost (1980) from apparent similarities in the range of joint angles exhibited
in vertical climbing by chimpanzees and in human bipedalism, and by Fleagle et al. (1981)
from joint movements and muscle activity in these behaviours in New World monkeys. But
recent work has undermined this hypothesis by showing that gorillas and orangutans have
more extended hip joint angles when moving bipedally than when they are using vertical
climbing (Crompton et al. 2003; Watson et al. 2009).
By the early 2000s the fossil record of the Eurasian and East African Miocene (23–5 million
years ago (Ma)) was burgeoning and revealing the body form of early ‘crown’ hominoids
(‘crown’ hominoids being the direct ancestors of all living apes, including humans). These
included fossils of species such as Morotopithecus bishopi (from approximately 18–22 Ma),
Pierolapithecus catalaunicus (c. 12 Ma), Hispanopithecus (Dryopithecus) laietanus (c.10Ma)
and Orrorin tugenensis (6 Ma). These fossils suggested that, contrary to expectations and
fossil evidence from Proconsul hesoloni and associated species, the early crown hominoids
stood and moved with an orthograde (upright) posture. Thus features such as their broad,
shallow trunks; scapulae positioned on the back rather than side of their bodies, and lumbar
vertebral bodies that increased in size towards the lower end of the spine all indicated that
these species were frequently upright (MacLatchy 2004; Moy
`
a-Sol
`
a et al. 2004; Nakatsukasa
et al. 2007; reviewed in Crompton et al. 2008). In addition, since they are estimated
to have weighed between 30 and 50kg, they were also at least as large as adult female
great apes (MacLatchy 2004; Moy
`
a-Sol
`
a et al. 2004; Nakatsukasa et al. 2007), a finding
which casts doubt on the validity of Tuttle’s (1981) model of a small-bodied gibbon-like
ancestor.
The fact that orthograde (upright) body postures had been evolving and diversifying
in our hominoid ancestry for in excess of 15 million years pushed study of the origins of
bipedalism back from the Pliocene into the early Miocene. It also challenged the commonly
held concept that the acquisition of habitual bipedalism is an appropriate marker of the
separation of the hominins from the panins (bonobos and chimpanzees), a separation
that is estimated to have occurred only 5–8 million years ago. It pushed the context
of bipedal origins back into the forest canopy from the ground (Senut 2011) where it
had spent some considerable time as a result of the knuckle-walking hypothesis. This
latter paradigm, that has dominated our vision of the evolution of bipedalism since the
1960s, held that because chimpanzees and gorillas move on the ground by quadrupedal
horizontal-trunked knuckle-walking, the pre-bipedal ancestor of hominins must also
have passed through a terrestrial knuckle-walking phase (e.g. Gebo 1996; Richmond &
Strait 2000).
C
Antiquity Publications Ltd.
907
https://doi.org/10.1017/S0003598X00050778 Published online by Cambridge University Press
The arboreal origins of human bipedalism
In parallel to the burgeoning fossil record, significant progress was being made in quanti-
fying the locomotor ecology of modern wild apes (i.e. the relative proportions of bipedalism
and other forms of movement exhibited by a given species in a given setting). Hunt and
colleagues (1996) advocated much-needed uniformity in the language used to describe loco-
motion across primate clades. They wished primarily to avoid the ubiquitous term ‘climbing’
to describe a wide range of locomotor behaviours that conflated pronograde (horizontal)
and orthograde (upright) body postures, and travelling in vertical and horizontal directions.
In the event, this has been adhered to more closely by the literature on living primates
than that on fossil forms. The significance of the approach was that it allowed comparative
quantification of the ecological context of locomotion (how much time a particular species
spent in knuckle-walking, brachiating, vertical climbing, etc.; in what kinds of setting—
e.g. forest canopy, forest floor, open grassland—and an indication of the stresses different
behaviours placed on the body). Thus it made it possible to quantify the adaptive advantages
of arboreal behaviours, a factor that was lacking from many earlier studies of locomotion that
were restricted to studies of captive animals or qualitative observations of wild-living taxa.
The approach revealed that all great apes occasionally choose to engage in arboreal
bipedalism—walking along and between branches on two legs (e.g. Hunt 1992; Remis
1995; Thorpe & Crompton 2005, 2006). It was from this that Hunt (1996) and Stanford
(2006) developed the arboreal foraging hypothesis. They showed that in chimpanzees, hand-
assisted bipedal posture (as opposed to bipedal locomotion) was associated with arboreal
feeding on relatively stable branches >100mm in diameter, and suggested such behaviour
might have been exaptive for terrestrial bipedalism. Postures are, however, less energetically
demanding to maintain than locomotion, and standing on large-diameter branches does
not pose the safety risks that are associated with balancing on thin, flexible branches. In
contrast we studied Sumatran orangutans (Pongo abelii), as they exhibit strong similarity to
humans in the extended-leg bipedal kinematics (joint angles) and kinetics (forces exerted
on the ground during locomotion) (Crompton et al. 2003; Crompton & Thorpe 2007).
Furthermore, they are the only exclusively arboreal great ape. We found that Sumatran
orangutans use extended-leg bipedal locomotion on highly flexible branches, <40mm in
diameter (Thorpe et al. 2007a) (Figure 1). This result countered traditional hypotheses
that had suggested that movement along flexible branches should be either via orthograde
suspension in which the animal gains stability by hanging with its centre of mass directly
under the branch; or by ‘compliant’ quadrupedalism, in which stability is maximised in
part by bending the knees and elbows substantially to reduce the movements of the branch
caused by the animal’s weight.
We also found that in 75 per cent of our observations of orangutan bipedal locomotion
along branches, they used their hands for stabilisation, as do chimpanzees (Hunt 1996;
Stanford 2006). Hand assistance ensures maximum safety while the bipedalism enables a
free hand to reach out for feeding, weight transfer, or balance in the peripheral branches
of trees, where the majority of preferred foods are situated and where primates must cross
between tree crowns. Being able to access these peripheral branches effectively is highly
advantageous because it allows large-bodied apes to cross more gaps between trees. Crossing
rather than circumventing gaps in the canopy can dramatically reduce the energy costs of
travel, especially where a change of height would otherwise be required (Thorpe et al. 2007b).
C
Antiquity Publications Ltd.
908
https://doi.org/10.1017/S0003598X00050778 Published online by Cambridge University Press
Debate
Susannah K.S. Thorpe, Juliet M. McClymont & Robin H. Crompton
Figure 1. Standardised cell residuals (SCRs) to demonstrate the primary results in the Log linear model of Thorpe et al.
2007a. The left-hand diagram shows the relationship between locomotion and the number of supports used and the right
shows the relationship between locomotion and the diameters of the supports used. SCRs indicate by their sign whether an
interaction is more (positive values) or less (negative values) common than predicted by the model and, by their size, to what
degree. SCRs greater than
+
−
2.0 indicate a lack of fit. The graphs show that quadrupedalism is strongly associated with
locomotion on single, large, stable supports >200mm in diameter; orthograde suspension is mostly associated with locomotion
on supports between 40–100mm in diameter. In contrast bipedalism is strongly associated with locomotion on multiple
supports and those that are <40mm in diameter.
We concluded that hand-assisted arboreal bipedalism as part of a smooth continuum of
orthograde behaviours ranging from suspending underneath branches to standing on top of
them confers a major selective advantage on orangutans and argued that arboreal bipedalism
would have been equally advantageous for ancestral crown hominoids (Figure 2).
We are convinced that the accumulating evidence for the arboreal origins of human
bipedalism is strong. Inevitably, some do not share our conviction. As part of a more general
critique on the use of ‘living referential models’ to understand fossil taxa, Sayers and Lovejoy
(2008) argued that our use of orangutan data was based on false premises. First, they
suggested that we studied bipedal posture and not bipedal locomotion. This indicates that
they didn’t read our paper well; even the title alluded to locomotion rather than posture. They
also suggested that orangutans are an unsuitable model because they have feet that are highly
specialised for gripping, such as very long toes that cannot therefore have been exaptations
for bipedality; and that they are rarely terrestrial, and when they are terrestrial they use
knuckle- or fist-walking (citing Tuttle & Beck 1972). We agree that the feet of orangutans
are highly specialised—but even then our recent work (Bates et al. 2013) shows that foot
pressures in the bipedalism of orangutans (and bonobos) overlap substantially with those
of humans, particularly under the middle of the foot. Nevertheless, orangutan footprint
morphology does not need to be exaptive for bipedality for the purposes of our model.
We studied Sumatran orangutans because, unlike Bornean orangutans (Pongo pygmaeus sp.)
and other great apes, they very rarely descend to the ground and should therefore be a
good model for arboreal locomotor ecology. If Sumatran orangutans did (hypothetically)
descend to the forest floor they probably would move quadrupedally because palmigrade
quadrupedalism is strongly associated with travel on broad, stable tree boughs (Figure 1)
(Thorpe & Crompton 2006; Thorpe et al. 2007a). Furthermore, anecdotal evidence suggests
wild Bornean orangutans generally use quadrupedalism when terrestrial. To our knowledge,
C
Antiquity Publications Ltd.
909
https://doi.org/10.1017/S0003598X00050778 Published online by Cambridge University Press
The arboreal origins of human bipedalism
Figure 2. A reconstruction of the arboreal bipedalism hypothesis depicting the evolution of modern great apes including
humans from an orthograde ancestral ape, capable of hand-assisted, arboreal bipedalism with extended lower limbs (from
O’Higgins & Elton 2007). Orangutan ancestors became a rboreal specialists, whereas the ancestors of gorillas and chimpanzees,
in response to changing and variable habitats, climbed vertically in and out of trees, and independently acquired knuckle-
walking. Hominins retained existing adaptations for extended-limb bipedalism and eventually became committed terrestrial
bipeds. Reprinted with permission from AAAS.
however, orangutan hand postures in terrestrial locomotion in the wild have never been
quantified because hands are difficult to see in the clutter of the forest floor from the distance
required for following orangutans in the wild. The hand postures are likely, however, to be
something akin to fist-walking simply because of the length of the digits.
The Tuttle and Beck (1972) paper that Sayers and Lovejoy (2008) reference as evidence
for orangutans employing knuckle-walking when terrestrial is based primarily on the mostly
postural descriptions of the behaviour of a single, captive, very obese male orangutan called
Felix. Tuttle and Beck (1972: 33–34) conclude that “although Felix often places his hands
in knuckle-walking postures, he rarely supports a major portion of his body weight on
C
Antiquity Publications Ltd.
910
https://doi.org/10.1017/S0003598X00050778 Published online by Cambridge University Press
