Professional Documents
Culture Documents
...........................................................................................................................................................................................................................
Striding bipedalism is a key derived behaviour of hominids that possibly originated soon after the divergence of the chimpanzee
and human lineages. Although bipedal gaits include walking and running, running is generally considered to have played no major
role in human evolution because humans, like apes, are poor sprinters compared to most quadrupeds. Here we assess how well
humans perform at sustained long-distance running, and review the physiological and anatomical bases of endurance running
capabilities in humans and other mammals. Judged by several criteria, humans perform remarkably well at endurance running,
thanks to a diverse array of features, many of which leave traces in the skeleton. The fossil evidence of these features suggests that
endurance running is a derived capability of the genus Homo, originating about 2 million years ago, and may have been
instrumental in the evolution of the human body form.
M
ost research on the evolution of human locomotion How well do humans run long distances?
has focused on walking. There are a few indications In considering human running, it helps to start from the perspective
that the earliest-known hominids were bipeds1,2, of the basic biomechanical differences that distinguish running and
and there is abundant fossil evidence that australo- walking gaits in all mammals, including human bipeds. These
pithecines habitually walked by at least 4.4 million differences are well characterized. Walking uses an ‘inverted pen-
years (Myr) ago3,4. Many researchers interpret the evolution of an dulum’ in which the centre of mass vaults over a relatively extended
essentially modern human-like body shape, first apparent in early leg during the stance phase, efficiently exchanging potential and
Homo erectus, as evidence for improved walking performance in kinetic energy out-of-phase with every step (Fig. 1a, b). The meta-
more open habitats that came at the expense of retained adaptations bolic cost of transport (COT) for human walking, like that of other
in the australopithecine postcranium for arboreal locomotion (for mammals, is a ‘U’-shaped curve, in which optimal speed, approxi-
example, refs 5–8). Although the biomechanics of running, the mately 1.3 m s21, is largely a function of leg length15. Most humans
other human gait, is well studied, only a few researchers (see refs 9, voluntarily switch to running at approximately 2.3–2.5 m s21,
which corresponds closely to the intersection of the COT curves
10 for example) have considered whether running was a mode of
for walking and running in humans (Fig. 2b)16,17. At these higher
locomotion that influenced human evolution. This lack of attention
speeds running becomes less costly than walking by exploiting a
is largely because humans are mediocre runners in several respects. mass-spring mechanism that exchanges kinetic and potential energy
Even elite human sprinters are comparatively slow, capable of very differently (Fig. 1b). Collagen-rich tendons and ligaments in
sustaining maximum speeds of only 10.2 m s21 for less than 15 s. the leg store elastic strain energy during the initial, braking part of
In contrast, mammalian cursorial specialists such as horses, grey- the support phase, and then release the energy through recoil during
hounds and pronghorn antelopes can maintain maximum gallop- the subsequent propulsive phase18,19. To use these springs effectively,
ing speeds of 15–20 m s21 for several minutes11. Moreover, running the legs flex more in running than in walking: flexing and then
is more costly for humans than for most mammals, demanding extending at the knee and ankle during the support phase (Fig. 1a).
roughly twice as much metabolic energy per distance travelled than Limb stiffness relative to body mass in running humans is similar to
is typical for a mammal of equal body mass12. Finally, human that of other mammalian cursors20.
runners are less manoeuvrable and lack many structural modifi- Although extensive data on endurance capabilities are not avail-
cations characteristic of most quadrupedal cursors such as elongate able for most quadrupedal mammals, several lines of evidence
digitigrade feet and short proximal limb segments. indicate that humans, using criteria such as speed and sustainable
However, although humans are comparatively poor sprinters, distance, are much better endurance runners than has generally
they also engage in a different type of running, endurance running been appreciated. Human ER speeds range from approximately 2.3
(ER), defined as running many kilometres over extended time to as much as 6.5 m s21 in elite athletes. Average ER speeds for
periods using aerobic metabolism. Although not extensively studied recreational joggers range between 3.2–4.2 m s21 (ref. 21). From an
in non-humans, ER is unique to humans among primates, and evolutionary perspective, it is important to note that human ER
uncommon among quadrupedal mammals other than social carni- speeds are exceptional compared to non-human primates. Apes
vores (such as dogs and hyenas) and migratory ungulates (such as such as chimpanzees, and other primates, such as patas monkeys,
can sprint rapidly, but they do so rarely and only for short
wildebeest and horses)13,14. Here, we review the evidence for and
distances22,23. No primates other than humans are capable of ER.
impact of ER in human evolution. We begin with a discussion of the
Quadrupedal cursors easily sprint faster than humans over short
mechanical differences between walking and running, and how well distances, but sustainable ER speeds of humans are surprisingly
humans perform at ER compared to other mammals. We then comparable to specialized mammalian cursors such as dogs and
review what is known about the key structural specializations horses in two respects. The first comparison to make is with
thought to underlie human ER capabilities, the extent to which trotting, because bipeds are incapable of galloping, but also because
they may be features that evolved originally for bipedal walking, and human bipedal running and quadrupedal trotting are biomechani-
the evidence for their appearance in the hominid fossil record. We cally most comparable. Both gaits synchronize contralateral fore-
conclude by outlining some hypotheses for why ER capabilities and hindlimbs, effectively restricting each stride cycle to just two
initially arose in the genus Homo, and the significance of this steps, and both are inherently ‘bouncy’ gaits with substantial
behaviour for human evolution. vertical displacements of the centre of mass18,24. When compared
NATURE | VOL 432 | 18 NOVEMBER 2004 | www.nature.com/nature
©2004 Nature Publishing Group 345
review article
to quadrupedal trotting, human ER speeds are relatively high when outdistance specialized quadrupeds. Some horse and dog breeds, for
adjusted for body mass (Fig. 2a). The predicted preferred trotting example, can be made to run more than 100 km day21 while
speed for a human-sized (65 kg) quadruped is approximately carrying or pulling a human. Such extreme and human-induced
2.8 m s21, and the trot–gallop transition is 3.8 m s21 (ref. 25). A feats, however, should not detract from the fact that humans can
more extreme comparison of performance that is not adjusted for and do run long distances well, despite a primate ancestry.
body size is between humans and large mammals such as ponies and The one category in which humans perform poorly compared to
horses (Fig. 2a). Human ER speeds exceed the preferred trotting many quadrupeds is the energetic cost of running. The mass-
(3.1 m s21) and the trot–gallop transition (4.4 m s21) speeds of adjusted COTof human running is about 50% higher than a typical
ponies (110–170 kg)26, and even the preferred trotting speed mammal, including other primates12. Compared to the only value
predicted for a 500-kg quadruped25. measured for a chimpanzee (a 17.5-kg juvenile), human running is
Most cursorial quadrupeds such as zebra, antelopes, and African 25% less costly in absolute terms, but about 10% more costly when
hunting dogs trot when running long distances14, but a few such as adjusted for body mass29. Interestingly, other endurance cursors
hyenas and wildebeest are known to run long distances using a low- such as wolves and African hunting dogs also have high mass-
speed gallop (typically a canter)13. When galloping, species with adjusted COT relative to the average mammal12. One important
high sustainable speeds such as dogs or horses can usually outrun characteristic of human ER may be its range of accessible economi-
humans. The maximum sustainable (,10–15 min) galloping speed cal speeds. Horses have U-shaped COT curves with narrow ranges of
predicted for a 65-kg quadruped is 7.7 m s21, and elite racing horses preferred speeds for trotting and galloping and gait transitions that
can gallop 10 km at 8.9 m s21 (refs 25, 27). However, human ER minimize cost, thereby achieving an effectively flat COT curve that
speeds are quite comparable to the preferred galloping speeds that excludes many speeds within the aerobic range (Fig. 2b)26. It is
cursors use over longer distances and times. Minetti27 has shown not known whether other quadrupedal cursors such as dogs have
that sustainable galloping speeds in horses decline considerably for U-shaped COT curves, but human runners differ from horses in
runs longer than 10–15 min, accounting for the average daytime employing a single gait, with a flat COT curve at all but the fastest
speed of 5.8 m s21 at which long-distance postal horses were endurance speeds9,16. Like another group of cursorial bipeds,
consistently run for millennia. Wildebeests (,100 kg) prefer to kangaroos and wallabies, humans are thus able to adjust running
canter at 5.1 m s21 (ref. 13). Well-conditioned human runners speed continuously without change of gait or metabolic penalty
exceed the predicted preferred galloping speed for a 65-kg quad- over a wide range of speeds. Further research is necessary to
ruped25 and can occasionally outrun horses over the extremely long determine whether other cursors are capable of such a broad
distances that constrain these animals to optimal galloping speeds, range of economic speeds.
typically a canter (Fig. 2a)9,10.
Humans also perform well at ER by another criterion, sustainable Structural bases and fossil evidence for endurance running
distance. Approximately 10% of Americans habitually jog or run The human capacity for ER raises several questions. What features
several kilometres a day (the percentage is higher if one includes make ER possible? When do these features first appear in the fossil
treadmill exercise and related sports)28. Fit human amateurs can record? How might such features relate to adaptations for bipedal
regularly run 10 km, and longer distances such as marathons walking? Many of the anatomical and physiological features
(42.2 km) are achieved by tens of thousands of people each year. involved in running are well studied in mammals, including
Such distances are unknown if not impossible for any other primate, humans, but most have not been explicitly evaluated in the
but are comparable to those observed in specialized mammalian human fossil record. A useful approach is to consider separately
cursors in open habitats. African hunting dogs travel an average of the evidence for structural features relevant to four types of
10 km per day, and wolves and hyenas travel on average 14 and demands posed by ER: energetics, strength, stabilization and
19 km day21, respectively14. This is not to say that humans can thermoregulation. The skeletal traces of these features, and the
346 ©2004 Nature Publishing Group NATURE | VOL 432 | 18 NOVEMBER 2004 | www.nature.com/nature
review article
evidence for their first presence in the fossil record, are summarized However, the transverse groove into which the Achilles tendon
in Table 1 and illustrated in Fig. 3. Several issues need to be kept in inserts on the posterior surface of the calcaneus is chimpanzee-like
mind when evaluating these features. in size in three early australopithecine Hadar specimens (AL 333-8,
First, it is useful to distinguish between structures that benefit 333-37 and 333-55)32,33, and contrasts with the substantially wider
both walking and running from those that are specific to the unique and taller attachment area characteristic of H. sapiens. We hypoth-
biomechanics of running and are functionally unrelated to walking. esize that, as in modern apes, a developed Achilles tendon was
Second, the limitations of the fossil record complicate our ability to absent in Australopithecus and originated at some point after 3 Myr
test evolutionary hypotheses concerning many structural modifi- ago, probably in the genus Homo.
cations that are derived in humans relative to chimpanzees. Some, Another well-developed set of springs important to human
such as Achilles tendon length, leave no clear skeletal evidence— running is the longitudinal arch of the foot. During walking, the
rendering uncertain their first appearance. Others, particularly in plantar arch helps to maintain mid-tarsal rigidity for powered
the foot, are not yet adequately sampled in the fossil record to make plantar flexion during toe-off, and absorbs some impact force
it possible to identify their origins. (but only after heel strike); during running, the elastic structures of
the plantar arch function as a spring, returning approximately 17% of
Energetics the energy generated during each stance phase19. Several features in
Humans exhibit many musculoskeletal specializations for bipedal- australopithecine foot bones from Hadar and Sterkfontein (STW
ism. Given the fundamental biomechanical contrasts between 573) suggest that some sort of plantar arch was present, including an
walking and running, which features are specifically relevant to elongated lateral cuneiform and insertions for the plantar liga-
the energetic cost of running? As noted above, the mass-spring ments4,34,35. But analyses of the Hadar and Sterkfontein specimens
mechanics of running differ from the pendular mechanics of suggest that they may have had a partial arch only, as indicated by
walking: running uses a compliant limb in which muscles and the enlarged medial tuberosity of the navicular, which is also
tendons in the legs sequentially store and then release strain energy enlarged and weight-bearing in chimpanzees, but is diminutive
during the stance phase of the stride cycle. In contrast to apes, and not weight-bearing in Homo36. In addition, for the plantar arch
human legs have many long spring-like tendons connected to short to be an effective spring during running, the transverse tarsal joint
muscle fascicles that can generate force economically30. These must restrict rotation between the hind foot and the anterior tarsals,
springs (see Fig. 3) can have comparatively little effect on energy allowing passive stretching of the plantar ligaments during a mid-
savings during an inverted pendulum-like walk, particularly at heel foot strike. In humans, this rotation is restricted primarily by a
strike when the limb is not compliant, but are estimated to save projecting medial flange on the proximal cuboid, which causes the
approximately 50% of the metabolic cost of running17,19. The most calcaneocuboid joint to form a close-packed position following
important of these springs is the Achilles tendon, which connects several degrees of rotation37. There are no preserved early H. erectus
the heel with the major plantar flexors of the foot; other elongated feet, but this feature—together with a fully adducted big toe—is
tendons that are derived features of the human leg include the first apparent in the OH 8 foot4,36,37, which is generally ascribed to
iliotibial tract and m. (muscle) peroneus longus31. Unfortunately, H. habilis.
there are no preserved early Homo calcanei, and leg tendon length An additional energetic factor to consider is stride length. Unlike
probably cannot be estimated reliably from attachment sites. most quadrupeds25, humans increase speed during ER mostly by
Figure 2 Comparative ER performance in humans and quadrupeds. a, Range of have U-shaped COT curves for walking, and trotting has a similar-shaped curve in
speeds for human ER and sprinting, and minimum trot (Tm), preferred trot (Tp), trot– the horse, but the human COT is essentially flat at ER speeds. Preferred speeds
gallop transition (T–G), preferred gallop (Gp), and maximum sustained gallop (Gms) (dotted rectangles) correspond to the most energy-efficient speeds in horses and
for ponies (ref. 26), and predicted for quadrupeds of 65 and 500 kg (ref. 25). Also walking humans, but speed selection is unrestricted in human ER. Note also that
indicated is Gld, the optimal long distance (,20 km), daytime galloping speed for human running, like quadrupedal trotting, involves synchronized movements of
horses (ref. 27). Note that quadrupeds sprint at speeds above Gms. b, Comparison diagonally opposite appendages (dots).
of the metabolic cost of transport (COT) in humans and ponies9,16,17. Both species
348 ©2004 Nature Publishing Group NATURE | VOL 432 | 18 NOVEMBER 2004 | www.nature.com/nature
review article
of speed48. Like Pan and early Homo, australopithecines have robust running to help ensure stabilization and balance. Most obviously,
femoral shafts relative to body mass, but they are less wide the trunk and neck of human runners are more forwardly inclined
transversely than in early Homo7. Although the distinctly shorter during running than walking (Fig. 1a), resulting in a greater
femoral neck of humans compared to Pan or Australopithecus tendency to pitch forward, especially at heel strike. Homo has a
decreases the mechanical advantage of the hip abductors, it might number of derived features that enhance trunk stabilization, includ-
also facilitate running by reducing bending moments in the femoral ing expanded areas on the sacrum and the posterior iliac spine for
neck. The reduction in interacetabular hip breadth in Homo also the attachment of the large erector spinae muscles, and a greatly
reduces lateral bending moments on the pelvis and lower back enlarged m. gluteus maximus4,46. The latter muscle, whose increased
generated at footstrike, and likewise helps minimize the angular size is among the most distinctive of all human features, is strongly
momentum in the trunk caused by rapid oscillation of long, heavy recruited in running at all speeds but not in walking on level
legs49. surfaces50. In addition, the transverse processes of the sacrum are
also relatively larger in Homo than Australopithecus, suggesting a
Stabilization more mechanically stable sacroiliac joint34.
Bipedal gaits are inherently unsteady, but several differences Independent rotations within the trunk play a crucial role in
between running and walking call for special mechanisms during dynamic stabilization during human running and may help to
Figure 3 Anatomical comparisons of human, chimpanzee, H. erectus and are reduced or absent in humans. e, Reconstruction of H. erectus based primarily on
A. afarensis. a, c, Anterior and posterior views of human, enumerating features KNM-WT 15000 (from refs 4, 65); f, reconstruction of A. afarensis based primarily
related to endurance running listed in Table 1. b, d, Anterior and posterior views of on AL-288 (from refs 4, 66).
chimpanzee. Labelled muscles connect the head and neck to the pectoral girdle and
example, sweating, hairlessness, cranial cooling systems), would be 1. Haile-Selassie, Y. Late Miocene hominids from the Middle Awash, Ethiopia. Nature 412, 178–181
(2001).
useful for prolonged walking in hot environments, but they would 2. Galik, Y. et al. External and internal morphology of the BAR 1002 0 00 Orrorin tugenensis femur. Science
have been essential to tolerate the considerably higher impulsive 305, 1450–1453 (2004).
loads and endogenous heat produced by distance running. Second, 3. Ward, C. V. Interpreting the posture and locomotion of Australopithecus afarensis: where do we stand?
Yb. Physical Anthropol. 35, 185–215 (2002).
a few derived features of Homo that improve ER capabilities
4. Aiello, L. & Dean, M. C. An Introduction to Human Evolutionary Anatomy (Academic, London, 1990).
(notably forearm shortening and decoupling of the head and 5. Rose, M. D. in Origine(s) de la Bipédie chez les Hominides (eds Coppens, Y. & Senut, B.) 37–49 (CNRS,
pectoral girdle) are unrelated to walking, but would have hindered Paris, 1991).
arboreal locomotor capabilities. Thus some of the differences 6. Jungers, W. L. Relative joint size and hominid locomotor adaptations with implications for the
evolution of hominid bipedalism. J. Hum. Evol. 17, 247–265 (1988).
between Homo and Australopithecus that have been attributed to
7. Ruff, C. B. et al. in Primate Locomotion: Recent Advances (ed. Strasser, E.) 449–469 (Plenum, New York,
selection for more efficient long-distance walking may instead have 1998).
evolved for ER, thereby helping to make Homo the first fully 8. Wood, B. & Collard, M. The human genus. Science 284, 65–71 (1999).
terrestrial hominoid. 9. Carrier, D. R. The energetic paradox of human running and hominid evolution. Curr. Anthropol. 25,
483–495 (1984).
Considering all the evidence together, it is reasonable to hypoth- 10. Heinrich, B. Why We Run: A Natural History (Harper Collins, New York, 2002).
esize that Homo evolved to travel long distances by both walking and 11. Garland, T. Jr. The relation between maximal running speed and body-mass in terrestrial mammals.
running. New fossils and more detailed analyses of the existing fossil J. Zool. 199, 157–170 (1983).
record are needed to test whether these two locomotor capabilities 12. Taylor, C. R., Heglund, N. C. & Maloiy, G. M. Energetics and mechanics of terrestrial locomotion.
I. Metabolic energy consumption as a function of speed and body size in birds and mammals. J. Exp.
emerged concurrently or whether ER evolved after selection for Biol. 97, 1–21 (1982).
long-distance walking. An even more difficult task is to determine 13. Pennycuick, C. J. in Serengeti: Dynamics of an Ecosystem (eds Sinclair, A. R. E. & Norton-Griffiths, M.)
what behaviours selected for ER in the first place. Why would early 164–184 (Univ. Chicago Press, Chicago, 1979).
352 ©2004 Nature Publishing Group NATURE | VOL 432 | 18 NOVEMBER 2004 | www.nature.com/nature