THE ANATOMICAL RECORD 300:932–945 (2017)

Hominin Hip Biomechanics:

Changing Perspectives
ANNA G. WARRENER1,2*
1
Department of Anthropology, University of Colorado Denver, Denver,
Colorado 80217-3364
2
Department of Human Evolutionary Biology, Harvard University, Cambridge,
Massachusetts

ABSTRACT
The shape of the human pelvis reflects the unique demands placed
on the hip abductor muscles (gluteus medius and gluteus minimus),
which stabilize the body in the frontal plane during bipedal locomotion.
This morphological shift occurred early in hominin evolution, yet impor-
tant shape differences between hominin species have led to significant
disagreement about abductor function and locomotor capability in these
extinct taxa. A static biomechanical model that relies on a close associa-
tion between skeletal measurements of the pelvis and femur has tradi-
tionally been used to reconstruct hip biomechanics in these species.
However, experimental biomechanical approaches have highlighted the
dynamic nature of mediolateral balance in walking and running, chal-
lenging the assumptions of the static hip model. This article reviews tra-
ditional approaches for understanding hip abductor function, shows how
they have been applied to the fossil hominin record, and discusses new
techniques that integrate the dynamic nature of mediolateral balance
during human locomotion. Anat Rec, 300:932–945, 2017. V C 2017 Wiley

Periodicals, Inc.

Key words: pelvis; hip; biomechanics; hominin

The adaptation of the human pelvis to the unique
demands of bipedal locomotion is a frequently repeated
story told by anthropologists and anatomists. The long,
anteroposteriorly narrow pelvic girdle characteristic of
quadrupeds, and presumably the last common ancestor
of humans and chimpanzees (Lovejoy, 2005), became
short and wide to facilitate mediolateral balance of the
trunk over a single supporting limb during locomotion.
The anterior rotation and broadening of the iliac blades
associated with this restructuring of the pelvis is partic-
ularly important for human striding bipedalism because
it alters the placement and function of the hip abductor
muscles (gluteus medius and gluteus minimus). In chim-
panzees, the minor gluteal muscles arise from the poste-
rior surface of the ilium. In such an orientation, their
primary action has often been cited as extensors of the
lower limb (Lovejoy, 1988), but electromyographic (EMG)
profiles of gluteal muscle activation in chimpanzees has
shown significant gluteus medius activity during bipedal
walking, thought to correspond to medial support at the
hip during single-leg stance (Stern and Susman, 1981).
Whatever the role of the minor gluteals in chimpanzees,
in humans, the biomechanical challenge of stabilizing
the body in the mediolateral plane during single-leg sup-
port phase of bipedal walking and running requires
active muscle management of the mediolateral compo-
nent of the ground reaction force (GRF) to maintain the
body center of mass (COM) over the standing limb. The
reorientation of the ilium and lateral placement of the
hip abductors on the human pelvis allow these muscles
to provide the single largest contribution to mediolateral
GRF during both walking and running gaits (Pandy and
*Correspondence to: Anna G. Warrener, University of
Colorado Denver, Department of Anthropology, Campus Box
103, PO Box 173364, Denver, CO 80217-3364.
E-mail: awarrener@fas.harvard.edu
Received 22 June 2016; Revised 8 November 2016; Accepted
26 November 2016.
DOI 10.1002/ar.23558
Published online in Wiley Online Library (wileyonlinelibrary.
com).

C 2017 WILEY PERIODICALS, INC.

V
 HOMININ HIP BIOMECHANICS
Andriacchi, 2010; John et al., 2012). The importance of
hip abductor force contribution to normal bipedal loco-
motion is evident from the impaired gait of those with
abductor weakness or paralysis, known as Trendelen-
burg gait (Gottschalk et al., 1989). When the hip abduc-
tors do not function properly, the pelvis rotates away
from the standing limb during single-leg support phase
of walking, requiring a compensatory trunk tilt toward
the supporting side to maintain the body COM over the
supporting foot.
Some of the earliest fossil hominins show morphologi-
cal changes in the pelvis to facilitate hip abduction, indi-
cating the early evolutionary origin of this adaptation
and its crucial role in the adoption of bipedalism in our
lineage (Lovejoy et al., 2009a). However, while the gen-
erally recognizable form of the hominin pelvis was well
developed by at least 3.2 million years ago, as demon-
strated by the relatively complete pelvis of A.L. 288-1
(known as “Lucy”), important differences in the width of
the pelvis, shape of the ilia, and morphology of the proxi-
mal femur have led to debate about just how the abduc-
tor mechanism functioned in early hominins (Lovejoy
et al., 1973; Lovejoy, 1988; Berge, 1994; Ruff, 1995,
1998; Arsuaga et al., 1999; Lovejoy, 2005; Richmond and
Jungers, 2008; Ruff and Higgins, 2013). Most analyses
have relied on a static model of hip abductor function
where the moment generated about the hip is estimated
from skeletal measurements of the pelvis and femur
(Lovejoy et al., 1973; Lovejoy, 1988; Berge, 1994; Ruff,
1995, 1998; Lovejoy, 2005; Richmond and Jungers, 2008;
Ruff and Higgins, 2013). The static biomechanical model
determines abductor muscle force, Fm, as the product of
the external force acting on the hip, generally measured
as body weight minus the weight of the supporting limb,
and the ratio of the moment arm of the external force of
body weight acting about the hip, R, to the hip abductor
muscles’ moment arm, r (Fig. 1a) (Inman, 1947; Osborne
and Fahrini, 1950; Denham, 1959; Hirsch and Frankel,
1960; Merchant, 1965; McLeish and Charnley, 1970;
Pauwels, 1976; Ruff, 1995; Warrener et al., 2015). Hip
abductor moment arm length is defined as the perpen-
dicular distance from the hip joint center of rotation to
the line of action of the hip abductor muscles, with femo-
ral neck length often being used as a proxy for this
dimension in skeletal samples (Lovejoy et al., 1973; Ruff,
1995, 1998; Ruff and Higgins, 2013; Warrener et al.,
2015). The static model assumes that during single-leg
stance, the external force vector passes nearly vertically
through the mid-sagittal line of the body (Inman, 1947;
Merchant, 1965). Therefore, one-half biacetabular width,
or a given proportion of this distance, has frequently
been used to estimate the length of R during locomotion
(Lovejoy et al., 1973; Lovejoy, 1988; Berge, 1994; Ruff,
1995, 1998; Lovejoy, 2005; Richmond and Jungers,
2008), leading to greater moment calculations about the
hip as biacetabular width increases. Interestingly, the
same static mechanical model has produced significantly
different interpretations of hip abductor force production
in Australopithecus and other early hominins (reviewed
below). This static model is also persistent in the medi-
cal literature and is often employed to understand
the effectiveness of orthopedic interventions at the hip
(Sariali et al., 2008; Traina et al., 2009).
A significant problem with the static biomechanical
model is that it ignores the dynamic nature of
933
mediolateral balance during locomotion. While the static
model considers only the pelvis segment in determining
the external moment about the hip joint, during locomo-
tion moments generated about the ankle and knee are
also important contributors to the stability of the body
COM over the standing limb (Pandy and Andriacchi,
2010; Pandy et al., 2010; John et al., 2012). Additionally,
by assuming the external force vector remains near the
body midline during single-leg support phase of gait, the
static model ignores the mediolateral component of the
GRF. While small in magnitude compared to the vertical
component, any mediolateral deviation of the GRF will
affect the total moment generated about each joint of
the lower limb, ultimately affecting the force required of
the hip abductors (Winter, 2009). A more appropriate
model is the link-segment model that includes the foot,
shank, and thigh segments and requires the external
moment about each joint to be simultaneously balanced
by internal moments generated by the joints’ muscles
and anatomical structures (Fig. 1b) (Zajac et al., 2002;
Winter, 2009). The link segment model is central to tech-
niques for analyzing muscle contributions to walking
and running including inverse and forward dynamics
solutions (Winter, 2009). These approaches are present-
ing challenges to our basic understanding of how pelvic
morphology relates to dynamic hip abductor function
during locomotion, requiring a re-evaluation of tradition-
al interpretations of pelvic adaptation in the hominin
lineage.
Early Medical Perspectives and Approaches
Early experimental work on hip abductor function
aimed to provide a sound theoretical basis for improving
impaired gait through surgical intervention at the hip.
Inman (1947) measured hip abductor muscle activation
in vivo to compare muscle force production to theoretical
calculations of the moment about the hip joint. Using
the static abductor muscle force model, the minimum
moment about the hip was calculated as the product of
half biacetabular width and body weight (including the
weight of the stance limb), while abductor moment arm
length was measured from X-rays to determine muscle
force (Inman, 1947). Experimental hip abductor force
was measured through EMG, with electrical impulse cal-
ibrated to moment measurements at various levels of
standing hip abduction. Inman (1947) concluded that hip
abductor force production was approximately 1.65 times
body weight. The position of the pelvis also played a cru-
cial role in determining the activation of the hip abduc-
tors. When the pelvis was elevated toward the
supporting limb, the experimentally measured moment
values agreed closely with those predicted by the static
theoretical calculations. However, when the pelvis was
allowed to tilt away from the supporting side, the action
potential of the muscles decreased significantly causing
Inman (1947) to conclude that the moment about the hip
was being resisted exclusively by the fasciae latae. While
Inman’s (1947) experimental approach was significant in
its effort to simulate abductor force production in vivo,
the determination of hip force production from EMG is
problematic. Action potential magnitude to muscle force
generation is poorly correlated and electormyographic
readings are difficult to obtain from deep muscles such
as the gluteus minimus (Pandy and Andriacchi, 2010).
934 WARRENER

Fig. 1. Static and link segment models of hip abductor force pro-
duction. (A) The standard static biomechanical model of hip abductor
force production determines hip abductor muscle force (Fm) as the
product of body weight, minus the weight of the standing limb (BW-
limb), and the ratio of the external force moment arm (R) to the moment
arm of the hip abductor muscles (r). (B) The link segment model of the
lower limb assumes that each limb segment has a fixed-point mass at
its center of mass (m1, m2, m3), as well as a fixed segment length and
moment of inertia (I1, I2, I3). The segments articulate at frictionless
hinge or ball and socket joints. Gravitational, ground reaction, and
muscle and ligament forces acting on each segment can then be cal-
culated through inverse dynamic or forward dynamic solutions.
 HOMININ HIP BIOMECHANICS
Cross-talk from electrical signals produced by muscles
near the location of surface electrodes can also produce
erroneous EMG readings, making interpretation of EMG
signals difficult (Disselhortst-Klug et al., 2009).
Studies by Osborne and Fahrini (1950) and Merchant
(1965) used skeletal models to estimate hip abductor
force production using a combination of pulls and load
cells to simulate and measure the force of body weight
and the counteracting force of the hip abductors. These
experiments also explored the role of pelvis abduction
and femoral rotation (Merchant, 1965) in abductor force
production. Both authors reported lower hip abductor
forces than those found by Inman (1947), approximately
1.3 times body weight. Contrary to Inman’s (1947) con-
clusions, both these studies indicated that hip abductor
force production was greatest with the pelvis in an
adducted position, as the body weight moment arm
moved farther from the hip joint center of rotation
(Osborne and Fahrini, 1950; Merchant, 1965). Inman
(1947) did not directly account for changes in the exter-
nal moment arm, R, of body weight in different pelvic
positions, but Osborn and Farhini (1950) directly mea-
sured this parameter using radiographs. One subject
stood on a single limb with the pelvis in neutral, maxi-
mum abduction and maximum adduction with a radio-
opaque plumb-line centered through the midpoint of the
ankle joint indicating the line of the body center of grav-
ity during single leg stance. The results showed that
during single-leg stance with the pelvis in neutral posi-
tion, the body center of weight did not pass directly
through the mid-sagittal line as previous research had
supposed (Inman, 1947), but aligned just lateral to the
mid-sagittal line toward the supporting hip (Osborne
and Fahrini, 1950). By incorporating direct measures of
R into their model, Osborn and Farhini (1950) recog-
nized an important dynamic variable in their model.
McLeish and Charnley (1970) also directly addressed
the position of the body center of mass and its relation-
ship to hip abductor activation, noting the asymmetrical
nature of single-leg stance. In their experiments, sub-
jects stood on a single limb while abducting the pelvis
through multiple positions while the force exerted by the
standing limb was measured relative to known marker
locations. Radiographs of the pelvis and lower limb were
taken simultaneously during limb abduction. From these
measurements and images, the static moment about the
hip joint and hip abductor force were calculated in each
position of the pelvis. The results from two subjects
clearly demonstrate the variability of hip moments, and
hence muscle force, relative to pelvic position. For sub-
ject A, hip moments decreased rapidly as the pelvis
abducted, from nearly 5 times body weight with the pel-
vis tilted away from the stance limb by 8.5 degrees, to
just 1 times body weight with the pelvis elevated toward
the standing limb by 13 degrees. In subject C, pelvic tilt
away from the standing limb of 17.8 degrees produced a
hip moment of 4 times body weight, while pelvic eleva-
tion of 17.2 degrees produced a moment nearly 3 times
body weight. These variations were in part attributed to
postural and contralateral limb placement during single
limb stance, but regardless the variability is striking.
The variation in R is particularly noteworthy. In subject
A, R ranged from 154% of half biacetabular width in the
most adducted pelvic position to just 30% when the pel-
vis was abducted. However, subject C maintained a
935
more symmetrical placement of the bodyline of action,
with R being 120% of half biacetabular width in the
most adducted position to just over 80% in the most
abducted orientation. McLeish and Charnley (1970, p.
208) concluded “all moments and forces are subject to
considerable variation due to spinal and limb move-
ments which are, to quite a large extent, under volun-
tary control.” This conclusion is particularly important
in the context of previous analyses, which, while
acknowledging the dynamic aspects of locomotion and
the effect on center of mass displacement (Saunders
et al., 1953; Denham, 1959), presumed that dynamic
shifts in R had to be proportional to pelvic width
(Inman, 1947; Osborne and Fahrini, 1950; Denham,
1959; Merchant, 1965). Therefore, hip abductor force
production could be easily extrapolated for any individu-
al based on known skeletal and weight measurements.
Despite McLeish and Charnley’s (1970) conclusions, this
static approach to modeling hip abductor force produc-
tion has been used extensively in the anthropological lit-
erature to try and understand gait mechanics in extinct
hominins.
Pelvic Shape and Hip Biomechanics in Extinct
Hominins
Biomechanical reconstructions of locomotor perfor-
mance in extinct hominins are foundational to under-
standing the totality of hominin behavior within a given
ecological context. Feeding behavior, travel distances,
and energy expenditure are all dependent on and inter-
woven with locomotor strategies. Therefore, the intensity
of study and debate on this topic within anthropology is
not surprising. However, our understanding is often
muddied by the nature of our evidence. While analysis
of fossilized skeletons has provided a wealth of informa-
tion, we must always be careful to not overestimate the
strict relationship between form and function (who
would predict that goats are such good tree climbers
from the shape of their hooves?).
In addition to the biomechanical demands placed on
the pelvis by bipedalism, hominin pelvic shape has also
been driven by adaptations for obstetrics and thermoreg-
ulation. The large size of the human infant head and
body has created a tight fit for the fetus through the
maternal pelvis during birth. While encephalization, and
therefore fetal head size, likely did not place significant
demands on the shape of the female pelvis until the
Middle Pleistocene (Ruff et al., 1997; Rightmire, 2004),
the selective force of obstetrics is clearly evident from
the sexually dimorphic nature of the human pelvis, with
females having larger pelvic diameters throughout the
true pelvis, as well as the unique rotational mechanics
that typify human childbirth (Rosenberg, 1992).
Hominin pelvic shape is also influenced by variation
in overall body shape associated with the demands of
thermoregulation. Body heat production is a function of
body mass while heat dissipation is a function of body
surface area. Because mass is the cube of linear dimen-
sions, while surface area is the square of linear dimen-
sions, the surface area-to-body mass ratio of an object
will always decrease as the object gets larger. If the
human body is modeled as a cylinder, increasing height
alone will not affect the surface area-to-mass ratio, but
an increase in breadth will cause a decrease in the ratio
936 WARRENER

H. sapiens
allowing more heat generation and less heat dissipation

26.9 6 1.8
11.1 6 0.8
13.4 6 0.8

0.8
0.7
0.4
0.9
0.9
Femaleg
for a given body height (Ruff, 1991). Increases in body

6
6
6
6
6
–
breadth are most profoundly expressed in the size of the

12.7
12.6
10.2

12.0
11.9
pelvis, particularly, the false pelvis at the level of the ili-
ac blades. Greater pelvic breadth has been well docu-
mented across living human populations in high-latitude
cold climates and narrow pelves are characteristic of

H. sapiens

27.7 6 1.5
10.2 6 0.9
13.0 6 0.8

0.7
0.7
0.6
0.8
1.0
populations in low-latitude equatorial areas (Ruff, 1991)

Malef

6
6
6
6
6
reflecting the different thermoregulatory demands of

11.9
11.8
8.6
11.1
10.2
these environments.

Ardipithecus and Australopithecus. The earli-

est good fossil evidence of hominin pelvic shape comes
from Ardipithecus ramidus (4.4mya), which appears to

neanderthalensis
show distinctive features in the false pelvis associated

Kebara
with bipedalism (Lovejoy et al., 2009a). Although the fos-

12.9h
2f H.

0.5

11.8

9.4

8.8
11.4
30.5

14.2
sil was distorted and required significant reconstruction,

–
TABLE 1. Pelvic dimensions of extinct hominins and modern humans
the ilia are derived in the direction of later hominins,
being laterally broad and more sagittally oriented than
in earlier hominoids. This change indicates that the
reorganization of the minor gluteal muscles may already
have been well underway in this early species despite
still being committed to extensive arboreal behaviors

‘archaic’
Pelvis 1e

Homo
(Lovejoy et al., 2009a, 2009b). However, the true pelvis

0.6

11.6
33.5
10.8
13.9

13.1
13.5

12.7
13.3
remained more primitive with long ape-like ischia and SH
no evidence of hamstring and gluteus maximus expan-
sion seen in later hominins (Lovejoy et al., 2009a).
The pelvis of Australopithecus is known primarily
through two relatively complete specimens: A.L. 288-1,
H. erectus
KNM-ER
15000d

“Lucy,” assigned to Australopithecus afarensis (Johanson

1.5

8.8
22.5

10.0

10.2
10.5

12.6
–

–
et al., 1982) from Hadar, Ethiopia, and Sts14, Australo-
pithecus africanus (Lovejoy et al., 1973; Berge and Gou-
laras, 2010), from Sterkfontein, South Africa. Both
individuals are thought to be females (Johanson et al.,
1982; Berge and Goularas, 2010). Additionally, pelvic
BSN49/P27c

fragments recovered from Sterkfontein, South Africa

H. erectus
0.9–1.4

have allowed the reconstruction of the previously known

Gona

9.8

11.1

11.4
28.8

12.4

13.1

13.3
Stw 431 A. africanus male pelvis (Kibii and Clarke,
2003) and a relatively complete innominate and second
sacral segment have recently been described for A. afar-
ensis from Woranso-Mille, Ethiopia (Lovejoy et al.,
2016). Other more fragmentary remains have been
A. africanus

found in South Africa from Makapansgat (MLD7 and

MLD8) and Swartkrans (SK50, SK3155). Although taxo-
Sts 14b

Schmid (1983) (cited in Berge and Goularas, 2010).

8.5g
2.4

8.3
11.6

7.3

9.3
25.6

10.7

10.5

nomic attribution is not clear for all these specimens,

Tague (1989) data for white males and females.

they share morphological characteristics that allow dis-

cussion of overall pelvic morphology in the genus (Kibii
and Clarke, 2003).
Despite their small stature, the australopith pelvis is
most notable for being exceptionally mediolaterally
A. afarensis

Simpson et al. supplemental (2008).

A.L. 288-1a

broad (Tague and Lovejoy, 1986; Kibii and Clarke, 2003;

26.8

12.4
3.2

7.7

8.5
11.8
9.4
8.8
8.7

Berge and Goularas, 2010), particularly in the false pel-

vis, but also at the mid and inferior pelvis (Table 1)
ausler and Schmid (1995).
Berge and Goularas (2010).

Rak and Arensburg (1987).

(Tague and Lovejoy, 1986; Berge and Goularas, 2010).

Walker and Ruff (1993).

Richmond and Jungers (2008) have argued that Orrorin

Bonmatı et al. (2010).

tugenensis femoral morphology indicates that the trans-

versely wide pelvic complex seen in Australopithecus is
False pelvis (cm)

True pelvis (cm)

the ancestral condition for the hominin clade. The recon-

Midplane AP
Biacetabular

structions of Sts14 by Berge and Goularas (2010) and

A.L. 288-1 by Schmid (1983) place these two specimens
Outlet ML
Bispinous
Outlet AP
Date (Ma)

Inlet ML

within one standard deviation of the human means in

Inlet AP
Bi-iliac

bi-iliac, biacetabular, bispinous, and bituberous diame-

ters (Berge and Goularas, 2010). The relative width and
H€

height proportions of the australopith pelvis indicate the

h
d
a
b

g
e
c

f
 HOMININ HIP BIOMECHANICS
overall shape was both broad and short when compared
to modern humans. Pelvic indices of bi-iliac and biace-
tabular diameters to pelvic length are 163%–173% and
69%–75%, respectively, across the australopith recon-
structions, while the human mean ratios are 128% and
55%, respectively (Berge and Goularas, 2010). In addi-
tion, australopiths have a narrow anteroposterior diame-
ter at the pelvic midplane (Tague and Lovejoy, 1986;
Berge and Goularas, 2010) compared to modern humans
where greater anteroposterior diameters reflect obstetric
demands (Tague and Lovejoy, 1986; Ruff, 1995).
The pelvis of Australopithecus sediba, the species dis-
covered in Malapa, South Africa in 2008 and dated to
1.9mya (Pickering et al., 2011), is a partial exception to
the general australopith pelvic type. The best preserved
specimen, MH2, has been identified as belonging to a
female and shows a mosaic of primitive hominin and
derived Homo-like traits. As in other australopiths, the
biacetabular distance is relatively wide, the sacral and
acetabular joint surfaces are small, and the pubis is rela-
tively long (Kibii et al., 2011). However, the iliac blades
are more sagittally oriented and less flaring than in oth-
er australopiths, their shape is sigmoid as in later
Homo, and the blades are more robust and show a dis-
tinctive iliac tuberosity (Kibii et al., 2011). Additionally,
the ischium is shorter than in earlier australopiths and
the ischial tuberosities are distinctively everted and the
pubis more superiorly oriented (Kibii et al., 2011). The
differences in pelvic shape seen in A. sediba have been
argued to reflect changes in locomotor biomechanics as
the species retained the small cranial capacity of other
australopiths and therefore the obstetric demands for
birthing encephalized infants were not yet acting on pel-
vic shape (Kibii et al., 2011). However, the specific rele-
vance of each of these morphological changes to the
biomechanics of bipedal locomotion are not well under-
stood, and Kibii et al. (2011) do not propose any mechan-
ical explanations to support their claims. DeSilva et al
(2013) have proposed that A. sediba had a novel form of
bipedalism characterized by a fully extended hind limb,
inverted foot during swing-phase and initial ground con-
tact, with a large pronatory moment through the ankle
during stance phase. In the pelvis, the presence of osteo-
phytic growths at the origin of the rectus femoris and
the insertion of the biceps femoris are argued to indicate
increased strain on the hip joint and knee during loco-
motion (DeSilva et al., 2013). While no other australo-
piths are thought to have walked with similar bipedal
kinematics, these features in A. sediba may reflect
greater reliance on arboreal movement requiring a com-
pensatory strategy while on the ground (DeSilva et al.,
2013).
Few australopith specimens preserve muscular and
ligamentous attachments with sufficient clarity to allow
definitive evaluation of muscle placement, but several
important characteristics can be assessed. The very
broad ilium present in the genera clearly indicates an
expanded area of attachment for the minor gluteals
when compared to chimpanzees (Lovejoy et al., 1973),
and the presence of a posterior gluteal line indicates
that posterior portions of the gluteus maximus arose
directly from the ilium (H€ aeusler and Berger, 2001). The
orientation of the A.L. 288-1 ilium is slightly different
from Sts 14 and Sts 431, with the external surface more
posteriorly oriented (H€ aeusler, 2002; Berge and
937
Goularas, 2010), but whether this affected the function
of the hip abductors is uncertain. Berge (1994) concluded
that an ape-like arrangement of the hip musculature in
A.L. 288-1 would have provided better mechanical
advantage for the hip extensors and facilitated medial
rotation and adduction of the thigh during bipedal loco-
motion. However, other analyses have proposed that a
human-like muscular arrangement was more likely in
australopiths (H€ aeusler, 2002). A.L. 288-1 differs from A.
africanus in other aspects of the pelvis as well, particu-
larly the lack of a well-developed attachment for the ilio-
femoral and sacroiliac ligaments, which stabilize the
femoral head and sacrum during locomotion (Stern and
Susman, 1983; H€ aeusler, 2002). “Lucy’s” pelvis also lacks
a defined iliopsoas groove inferior to the anterior inferior
iliac spine or iliopectineal eminence, which has been
interpreted as compromising bipedal locomotion (Stern
and Susman, 1983; H€ aeusler, 2002). However, the newly
described pelvis, KSD-VP 1/1 from Woranso-Mille, Ethio-
pia has a strong iliopsoas groove, a prominent anterior
inferior iliac spine, robust anteriorly placed iliac pillar,
rugose area of attachment for the sacroiliac and iliofe-
moral ligaments and a robust ischial tuberosity, a pat-
tern more typical of other australopith species (Lovejoy
et al., 2016).
Controversy over the “effectiveness” and “efficiency” of
australopith gait could fill an entire manuscript. The
problems in determining locomotor behavior related to
the function of the hip are numerous, including differing
reconstructions of the A.L. 288-1 (Tague and Lovejoy,
1986; H€ ausler and Schmid, 1995) and Sts14 pelves (Rob-
inson, 1972; Abitbol, 1995; H€ ausler and Schmid, 1995;
Berge and Goularas, 2010), and poorly defined function-
al parameters for efficient and “effective” bipedalism.
While the overwhelming majority of researchers agree
that Australopithecus was a habitual biped (see Ward,
2002 for a full review), the argument remains over the
specific mechanics of bipedal progression in this genus,
and whether a shift toward bipedalism reduced arboreal
practice and competency (Ward, 2002).
In accordance with the divergent views of locomotor
behavior, biomechanical analyses of the hip abductors
have produced two very different reconstructions of func-
tion and gait in Australopithecus. In a detailed biome-
chanical analysis of the A.L. 288-1 pelvic complex, Ruff
(1998) calculated potential hip abductor and joint reac-
tion forces as well as mediolateral bending forces of the
femur based on Charnley’s and McLeish’s (1970) model
of hip abductor function during single-leg stance.
Assuming similar stance kinematics as the subjects in
their study, Ruff (1998) determined that joint reaction
force was 12% greater, hip abductor muscle force 27%
higher, and mediolateral bending forces at upper and
mid-shaft of the femur 36% and 39% greater, respective-
ly, than in modern humans (Ruff, 1998). Based on these
measurements, if A.L. 288-1 walked with a normal
human gait, Ruff (1998) hypothesized that she would
have greater femoral head diameter, iliac buttressing,
and mediolateral reinforcement of the femur relative to
body mass, than seen in modern humans. Because these
features are absent in the skeleton, Ruff (1998) conclud-
ed that this individual walked with a slightly waddling
gait, which would have diminished hip joint forces but
been energetically costly. Other researchers have drawn
similar conclusions regarding the kinematics and
938 WARRENER
energetics of australopith locomotion (Napier, 1964;
Stern and Susman, 1983; Berge, 1994; Hunt 1998).
In contrast, Lovejoy et al. (1973), Lovejoy (1988, 2005)
and McHenry (1975) have argued that australopith pel-
vic morphology maintained proper mechanical advan-
tage of the hip abductor muscles, and differed from the
modern human form because of parturition require-
ments in Homo, not because of locomotor differences
between the taxa. Although A.L. 288-1 and Sts 14 have
wide biacetabular dimensions, the ratio of r/R may have
been similar to modern humans due to the very long
femoral necks of this genus (McHenry and Temerin,
1979; Lovejoy, 2005). Therefore, the transversely wide
pelvis seen in australopiths would not have increased
hip abductor muscle force production during locomotion.
Based on this analysis of hip mechanics, Lovejoy et al.
(1973) and Lovejoy (2005) argue a fully striding bipedal
gait pattern, with effective mediolateral stabilization of
the trunk, would have characterized australopith loco-
motion (see also Kimbel and Delezene, 2009 for a review
of the debate). In addition to clear changes in the pelvis
associated with habitual bipedalism (Lovejoy et al.,
1973; Lovejoy, 1988, 2005; Kimbel and Delezene, 2009;
Haile-Selassie et al., 2010), Australopithecus displays a
suite of features, including valgus knee (Ward, 2002;
Haile-Selassie et al., 2010), development of transverse
and longitudinal arches (Ward et al., 2011), and adduc-
tion of the great toe (Latimer and Lovejoy, 1990), which
all suggest a form of bipedal progression similar to later
Homo (Ward et al., 2011).
Homo erectus and Archaic Homo. Pelvic mor-
phology of Homo erectus has been primarily described
based on the reconstructed pelvis of KNM-WT 15000
(Walker and Ruff, 1993), the os coxae of KNM-ER 3228
(Rose, 1984) from Lake Turkana, Kenya, and the nearly
complete pelvis, BSN49/P27, from Gona, Ethiopia (Simp-
son et al., 2008). A partial os coxae, OH 28 is also known
from Olduvai, Tanzania (Day, 1971). The Homo erectus
pelvis is generally robust with a large iliac tubercle and
strong iliac pillar, and the lateral surface of the ilium
shows a deep fossa for the attachment of gluteus medius
and minimus muscles (Rose, 1984; Simpson et al., 2008).
The iliac blades also flare laterally but the auricular sur-
face is relatively small compared to modern humans
(Rose, 1984; Ruff, 1995; Simpson et al., 2008). The ischi-
al tuberosity is large and the anterior superior iliac
spine is described as being protuberant (Rose, 1984;
Ruff, 1995). KNM-ER 3228 has a clear attachment site
for the iliofemoral ligament on the superior aspect of the
acetabular margin and a shallow groove for the iliopsoas
tendon between the anterior inferior iliac spine and the
iliopubic eminence (Rose, 1984). The attachments for the
sacroiliac ligaments are well developed and can be seen
in the Gona pelvis (Simpson et al., 2008) OH 28 (Day,
1971) and KNM-ER 3228 (Rose, 1984). Some of these
features are missing in KNM-WT 15000, but this is
readily attributable to the young age of that individual
(Walker and Ruff, 1993).
There is a general agreement that gait and kinematics
in Homo erectus were very similar to those of modern
humans including the function of the mediolateral bal-
ance mechanisms of the pelvis (Day, 1971; Rose, 1984;
Ruff, 1995; Arsuaga et al., 1999). The laterally flaring
ilia in combination with a long femoral neck (Brown
et al., 1985; Ruff, 1995) have been interpreted as
increasing hip abductor mechanical advantage in Homo
erectus (Brown et al., 1985), even though the gluteal
musculature appears to be very well developed (Rose,
1984; Simpson et al., 2008) and hip abductor forces have
been estimated to be greater than those of modern
humans based on the traditional static model of abduc-
tor mechanics (discussed in detail below) (Ruff, 1995).
Further information about the shape and mechanical
properties of the pelvis in early Homo has been deduced
from femoral morphology, for which there is a signifi-
cantly larger fossil collection. Ruff (1995) analyzed fem-
oral cross-sectional geometry in early Homo and argued
that, based on the increase in mediolateral bending
strength of the proximal femur, the primitive wide pel-
vic complex seen in australopiths was maintained until
0.5 Ma, although changes in gait altered the biome-
chanical loading of the lower limbs. The mediolaterally
broad pelvic shape would have limited increases in cra-
nial capacity within the genus until after the advent of
rotational birth, which was facilitated by an increase in
anteroposterior dimensions of the mid-pelvis (Ruff,
1995). This allowed biacetabular width and femoral
neck length to decrease and changed the biomechanical
loading of the femur, reducing mediolateral stress (Ruff,
1995).
Ruff (1995) excluded KNM-WT 15000 from his analy-
sis of femoral cross-sectional geometry because the indi-
vidual’s femoral morphology likely would not follow
predicted adult patterns due to his young age. Yet, while
conclusions about its adult morphology are necessarily
tentative, the argument that biacetabular width
remained wide in early Homo is at odds with KNM-WT
15000 pelvic morphology as originally reconstructed.
Walker (1993) speculated that the long femoral neck and
very narrow biacetabular width of this individual indi-
cated increased hip abductor leverage in Homo erectus
compared to modern humans. Walker (1993) further
argued that with selection for larger brains in later
Homo, which necessitated increases in the size of the
bony birth canal, long femoral necks were maintained
only in populations where cold climates allowed for wid-
er body builds. In equatorial regions, femoral neck
length was reduced to minimize body breadth, thereby
sacrificing abductor function (Walker, 1993). These inter-
pretations may need to be revised in light of the recent
recognition that the original reconstruction of the KNM-
WT 15000 pelvis likely underestimated the true breadth
(Arsuaga et al., 1999).
This picture has been further complicated by the dis-
covery of the Gona pelvis, which is relatively complete,
though slightly distorted (Simpson et al., 2008).
Although the Gona pelvis shares many diagnostic fea-
tures with other Homo erectus specimens (Simpson
et al., 2008, 2014), her extremely small estimated femo-
ral head diameter, 33.4–36.8 mm, (Simpson et al., 2008)
and inferred body size, make her the smallest individual
of the known Early or Middle Pleistocene Homo speci-
mens (Ruff, 2010). The pelvis is also unique in being
extremely mediolaterally broad. In bi-iliac, bispinous,
biacetabular, and mediolateral outlet diameters, Gona is
larger than the modern female means from human com-
parative samples (Tague, 1989; Simpson et al., 2008)
(Table 1). The curious combination of very small stature
 HOMININ HIP BIOMECHANICS
and body mass with very great mediolateral dimen-
sions of the pelvis has prompted some to question the
assignment of the Gona pelvic specimen to H. erectus
(Ruff, 2010). However, Simpson et al. (2014) note mul-
tiple morphological features, such as the robusticity
and placement of the iliac pillars, the size of the glute-
al fossa and ischial tuberosity, orientation of the iliac
blades, and the derived condition of the pubic symphy-
seal face that distinguish the Gona pelvis from those
of australopiths. Additionally, the enigmatic nature of
the Homo erectus remains from Dmanisi, Georgia, and
the calvarium from Ileret, Kenya (KNM-ER 42700) are
expanding the known variability of this species both
cranially and postcranially (Gabunia et al., 2000;
Spoor et al., 2007). Homo naledi also demonstrates the
mosaic nature of hominin pelvic evolution. While no
full reconstructions are yet available, the pelvic frag-
ments so far described show a mix of australopith-like
features in the ilium, and Homo-like features of the
lower pelvis (Berger et al., 2015). Regardless of taxo-
nomic categorization, the known pelves from Early
and Middle Pleistocene Homo (except for KNM-WT
15000) share the ancestral mediolaterally wide pelvis
and long femoral neck of Australopithecus (Arsuaga
et al., 1999), which has implications for the interpreta-
tion of hip abductor function through the early evolu-
tion of our genus.
The most complete fossil pelves representing “archaic”
Homo are those from Jinniushan, China, dated to
260 ka (Rosenberg, 1998; Rosenberg et al., 2006) and SH
Pelvis 1 from Sima de los Huesos, Spain (Arsuaga et al.,
1999; Bonmatı et al., 2010). Originally thought to be
around 200 ka (Arsuaga et al., 1999), more recent mass-
spectrometer analysis has shown the Sima de los Huesos
hominin deposit to be approximately 600 ka (Bischoff
et al., 2007), placing these hominins at the very begin-
ning of the Neandertal ancestral lineage (Bischoff et al.,
2007). The pelves of Kebara 2 from Mt. Carmel, Israel
(65 ka) (Rak and Arensburg, 1987; Gr€ un and Stringer,
1991) and the recently virtually reconstructed pelvis of
Tabun C1 (110 ka) (Ponce de Leon et al., 2008; Weaver
and Hublin, 2009) provide important insights into
derived Neandertal pelvic morphology. Fragmentary
remains from Shanidar, Tabun, Krapina and La Ferras-
sie also highlight differences in Neandertal pubic mor-
phology compared to modern humans (Rosenberg, 1998).
Morphologically, these pelves share many similarities
with those of Early and Middle Pleistocene Homo,
including a mediolaterally broad false pelvis with flaring
ilia, a general platypelloid shape, developed iliac but-
tress and tubercle, and prominent anterior superior iliac
spines (Rosenberg, 1998; Arsuaga et al., 1999; Rosenberg
et al., 2006). The extremely long and thin pubic ramus
of the Neandertal specimens (Trinkaus, 1976; Rak, 1990)
has historically been interpreted as part of a suite of
obstetric and fetal developmental features thought to be
unique to this group (Rosenberg, 1998). However, Nean-
dertal males have longer pubic rami than females, which
is at odds with an obstetrical interpretation of this trait
(Rosenberg, 1998). Additionally, both Pelvis 1 and Jin-
niushan share this pubic elongation, which is probably
an ancestral characteristic that has been lost in modern
humans (Rosenberg, 1998; Arsuaga et al., 1999). In
cross-section, the pubic rami of these earlier specimens
is intermediate between the reduced cross-sectional area
939
seen in Neandertals and the broader area characteristic
of modern humans (Rosenberg 1998; Arsuaga et al.,
1999). The functional significance of this difference
between the earlier and later “archaic” groups is unclear
(Rosenberg, 1998).
While the long pubic rami of Neandertals was once
argued to be an obstetric adaptation (Trinkaus, 1984;
Rosenberg, et al. 1988; Weaver and Hublin, 2009), Rak and
Arensburg (1987) argue that the long pubic bones seen in
Kebara 2, place the acetabula more laterally, effectively
increasing the body weight moment arm about the hip and
requiring greater hip abduction forces. However, other loco-
motor biomechanical interpretations of Neandertal pelvic
morphology have noted that the combination of flared iliac
blades, large biacetabular diameter, and long femoral necks
would indicate similar hip abductor mechanics in “archaic”
Homo and earlier H. erectus specimens (Arsuaga et al.,
1999). While these biomechanical interpretations may be
correct, the extremely broad shape of the Neandertal pelvis
is either an ancestral morphology shared with early mem-
bers of the genus Homo (Bonmatı et al., 2010), or reflects
adaptation to thermoregulatory demands in this cold
adapted species (Ruff, 1994).
The fossil evidence of hominin pelvic shape shows
that a mediolaterally broad pelvis has been characteris-
tic of many of our extinct ancestors, but biomechanical
implications of this general trend are debated. All these
reconstructions rely on the static model of hip abductor
function, which is understandable given that kinematics
in extinct species is unknowable. However, new techni-
ques are demonstrating that mediolateral stability of the
body during locomotion is significantly more complex
than this model allows, bringing into question previous
conclusions about hominin pelvic biomechanics. While
we can never know how these species moved on the
ground with certainty, increasing our understanding of
the relationship between anatomical structure and func-
tion in species that we can test experimentally (humans
and some primates) may provide new insights into the
evolution of hominin locomotor behavior.
Alternative Approaches to Hip Biomechanics
Inverse dynamics. Analysis of dynamic body
movement is generally performed using either Newton–
Euler inverse dynamics or through model-based simula-
tions (Zajac et al., 2002, 2003; Winter, 2009). Inverse
dynamics can be characterized as a ground up approach
where the resolution of unknown forces acting on a
body can be resolved from the known resultant force
(GRF) and the kinematic and inertial properties of the
body itself. These data are now relatively easy to obtain
using force-platforms and high-speed motion capture
systems to monitor the movement of body segments
during locomotion. Inertial properties of the body seg-
ments are generally obtained by scaling subject body
dimensions to those from cadavers where segment cen-
ter of mass and inertial properties can be directly mea-
sured (Dempster, 1955; de Leva Paolo, 2004; Winter,
2009). Assuming the foot, shank, and thigh are rigid
bodies articulating at frictionless joints, the net joint
moments about the ankle, knee, and hip can be
resolved sequentially using the Newton–Euler equa-
tions of motion:
940 WARRENER
X introduced either through instrument error (poor camera
Fx 5max
X markers from anatomical landmarks, which can affect
Fz 5maz
X et al., 2002). Such errors can dramatically alter net
M5I /
where m is the mass of the segment, Fx is the force of
the segment COM in the plane of movement (horizontal
or mediolateral), ax is the linear acceleration of the seg-
ment COM in that plane, Fz is the vertical force of the
segment COM, az is the vertical linear accelertation of
the segment COM, I is the moment of inertia, and a is
the angular acceleration of the body segment about its
COM. The resulting net joint moments represent the
total moment about a joint generated by all the anatomi-
cal structures acting on the adjoining body segments.
Joint contact forces can also be calculated using inverse
dynamics. With additional anatomical data on muscle
moment arms and cross-sectional area obtained through
imaging, muscle force and effective mechanical advan-
tage (EMA) of individual muscles and muscle groups act-
ing to generate the measured net joint moments can also
be estimated (Zajac et al., 2002; Biewener, 2004; Winter,
2009).
One advantage of inverse dynamics analysis for deter-
mining lower limb mechanics is that upper body seg-
ment kinematics do not have to be included in the
biomechanical model because their contribution to body
kinetics is represented in the measured GRF (Zajac
et al., 2002). This allows resolution of ankle, knee, and
hip moments without the inclusion of a body segment
representing the pelvis or torso. Inverse dynamics has
been shown to provide accurate measurements of hip
joint contact forces measured through instrumented
prostheses (Brand et al., 1994; Zajac et al., 2002), dem-
onstrating the general reliability of the method. Addi-
tionally, as computational power has increased and
commercially available biomechanics analysis software
has become available in the past 15 years, inverse
dynamics data has become significantly easier to collect,
and computation time has decreased, allowing larger
studies to be conducted (Otten, 2003).
One limitation of inverse dynamics is that net joint
moments cannot be easily decomposed into individual
muscle or ligament forces at the joint (Zajac et al., 2002;
Robertson et al., 2004). While individual muscle contri-
butions can be estimated from static anatomical data
(see above), muscle EMA and mechanical properties
change with movement, influencing the muscle’s force
producing capacity and therefore its contribution and
coordination to the net joint moment (Anderson and
Pandy, 2001; Pandy, 2001). Nevertheless, Anderson and
Pandy (2001) report that muscle forces derived from
inverse dynamics were nearly identical to muscle forces
predicted from dynamic optimization models during
normal walking. However, the close match between
inverse dynamics and simulation derived muscle forces
may not hold for running or impaired gaits (Zajac et al.,
2002). In addition, the contribution of individual muscles
to segment accelerations and joint contact forces are not
deducible from the inverse dynamics equations (Zajac
et al., 2002). Another potential limitation of inverse
dynamics solutions for understanding gait mechanics is
the sensitivity of the calculations to kinematics noise
coverage of body markers) or the movement of body
body segment measurements during locomotion (Zajac
joint moments because acceleration calculations expo-
nentially amplify such errors. Care must be taken to
properly filter and gap fill kinematic data to prevent
erroneous calculations through later analysis (Winter,
2009; Christodoulakis et al., 2010).
Forward dynamics. Another powerful technique
for understanding human movement is forward dynam-
ics. Forward dynamics is a model-based approach where
the movement of a multisegment body is driven by simu-
lated forces. These models can be actuated through net
joint moments or based on individual muscular contribu-
tions to segmental accelerations (Zajac et al., 2003). For-
ward dynamics simulations must include a full body
link-segment model that incorporates all degrees of free-
dom and physical constraints at each joint (Winter,
2009). The initial positions and velocity of each segment
must be known, with the only inputs being the external-
ly applied and internally generated moments and forces
(Winter, 2009). As an internal check, the model must be
validated against kinematic and kinetic data collected in
the lab to demonstrate the reliability of the simulated
forces in producing the observed subject movement (Win-
ter, 2009).
Forward dynamics solutions are most useful for ask-
ing “what if questions” to understand how changes in
the action of a particular muscle or anatomical structure
may affect the entire system (Winter, 2009). This tech-
nique allows clinicians to test the effectiveness of tar-
geted anatomical interventions on subjects with
pathological gait (Zajac et al., 2003) and has been used
to model walking dynamics in extinct hominins (Nagano
et al., 2005; Sellers et al., 2005). Forward dynamics mod-
els have helped elucidate the contribution of individual
muscles to forward progression and vertical support of
the body during walking (Zajac et al., 2003; Liu et al.,
2006), as well as partitioning the contribution of elastic
energy storage to body accelerations (Zajac et al., 2003).
Forward dynamics has also significantly increased our
understanding of joint coupling, demonstrating that
muscles that do not cross a joint can contribute to
moments generated about that joint, thus redistributing
their mechanical energy across the body segments (Zajac
et al., 2003; Pandy and Andriacchi, 2010).
The most significant drawback of forward dynamics
solutions is that the resolved muscle forces cannot be
experimentally validated in living subjects by noninva-
sive means (Pandy and Andriacchi, 2010). The kinemat-
ics and GRF produced by the model simulation can be
compared to gait data obtained from subjects in the lab,
but because the musculoskeletal system is redundant,
with many muscles crossing each joint, the observed
motion can be the product of many different combina-
tions of muscle forces. Therefore, the forward dynamics
solution may or may not accurately replicate the specific
muscle activation responsible for the observed movement
(Pandy and Andriacchi, 2010), and sensitivity studies
must be used to judge the acceptable limits of a
 HOMININ HIP BIOMECHANICS
Fig. 2. Skeletal dimensions of the hip and pelvis versus external and
hip abductor moment arm lengths. (A) Biacetabular width, defined as
the distance between the centers of the femoral heads, is not signifi-
cantly correlated with R at mid-stance during walking (dashed line,
y 5 0.51x 2 3.7, R2 5 0.05, P 5 0.28, N 5 25) or running gaits (solid
line, y 5 0.65x 2 5.7, R2 5 0.06, P 5 0.28, N 5 21). (B) Hip abductor
moment arm length, r, is significantly correlated with femoral neck
particular forward dynamics solution to support its con-
clusions (Zajac et al., 2003).
Skeletal anatomy and hip abductor function.
Using these new techniques, our understanding of the
relationship between hip abductor mechanics and skele-
tal anatomy has been challenged in recent years. War-
rener et al. (2015) explicitly tested the relationship
between skeletal estimates of hip abductor EMA and
measures of EMA and locomotor cost determined during
walking and running. Kinematics, kinetics, and MRI
data were collected for 26 males and females, as well as
oxygen consumption data for this sample and an addi-
tional fifteen subjects (Warrener et al., 2015). An inverse
dynamics approach was used to determine joint
moments at the hip, knee, and ankle and muscle forces
and EMA were resolved using subject-specific anatomical
data from MRI as well as muscle fascicle lengths derived
from cadaver dissection (Warrener et al., 2015). Our
results show that biacetabular width is a poor predictor
of R during walking and running (Fig. 2a), but the hip
941
length (y 5 1.0x 2 0.62, R2 5 0.62, P < 0.0001). (C) Hip abductor
moment arm length, r, is also significantly correlated with bi-iliac width
(y 5 0.32x 2 2.9, R2 5 0.51, P < 0.001) in males (N 5 13) and females
(N 5 13). Lines indicate OLS regressions. Circles 5 walk, square 5 run,
open symbols 5 female, closed symbols 5 male. Data and methods
are available in Warrener et al. (2015).
abductor moment arm, r, is strongly correlated with
femoral neck length and bi-iliac width (Fig. 2b and c)
(Dunsworth et al., 2013; Warrener et al., 2015). Further-
more, the variability of dynamic measures of R during
walking and running across subjects demonstrates that
lower limb kinetics and kinematics unrelated to skeletal
measures influence forces experienced at the hip (War-
rener et al., 2015). Baggaley et al. (2015) also report no
significant correlation between measures of pelvic width
relative to lower limb length and hip kinematics during
walking.
The results of these studies have important implica-
tions for interpreting variation in pelvic shape both in
the hominin fossil record and also in living humans. Our
analysis of dynamic hip abductor function using inverse
dynamics demonstrates that greater biacetabular width
in early hominins cannot be assumed to decrease hip
abductor EMA (Berge, 1994; Hunt, 1998; Ruff, 1998).
Longer femoral neck lengths and greater bi-iliac width
in australopiths, early Homo, and later Neandertals like-
ly indicate a long hip abductor muscle moment arm, r,
942 WARRENER
which may have increased EMA of this muscle group.
However, without a clear understanding of what (if any)
skeletal anatomical traits influence the deviations of the
GRF from the body midline during locomotion, our under-
standing of mediolateral stabilization mechanisms at the
hip in extinct hominins remains frustratingly unclear.
The lack of correlation between pelvic width and hip
abductor function also suggests that traditional interpreta-
tions of the effect of sexual dimorphism on locomotor perfor-
mance in modern humans are incorrect. The increased
width of the human female pelvis to accommodate a neonate
has long been argued to reduce locomotor efficiency by
increasing the metabolic cost of hip abductor activation
because of reduced EMA in women (Zihlman and Brunker,
1979; Meindl et al., 1985; Rosenberg, 1992). Warrener et al.
(2015) show that while hip abductor EMA is slightly lower
in women during both walking and running, these differ-
ences are driven by shorter r in women compared to men
due to shorter femoral neck lengths. Because femoral neck
length is strongly correlated with overall femoral length,
and thus height, the reduction in EMA in women can be
seen primarily as a consequence of smaller body size. Fur-
thermore, any increase in hip abductor force production in
women does not translate into greater overall locomotor
cost, which is not significantly different between the sexes
(Bunc and Heller, 1989; Bourdin et al., 1993; Hall et al.,
2004; Warrener et al., 2015).
In recent years, forward dynamics simulations have
been used to address locomotor performance in A. afar-
ensis based on the relatively complete skeleton of “Lucy”
(Nagano et al., 2005; Sellers et al., 2005). Nagano et al.
(2005) generated a neuromusculo-activated model where
the muscle activation profiles were optimized to mini-
mize energy expenditure. Their results show that
smooth, striding gait kinematics can be generated with
“Lucy’s” skeletal anatomy, and muscle activation pat-
terns in the sagittal plane were generally comparable to
those from modern humans (Nagano et al., 2005). In the
mediolateral plane, Nagano et al. (2005) report higher
muscle activation in the hip abductors and a tendency
for the swing leg to move away from the midsagittal line
during locomotion, with the swing leg abducted to nearly
10 degrees throughout most of the step. They attribute
these distinctive kinematics and muscle activation pro-
files to “Lucy’s” greater pelvic width, concluding that
greater moments must be generated at the hip to main-
tain trunk stability. However, their model reconstruction
does not include a torso or upper extremities, so evaluat-
ing the effect of a wider pelvis on trunk stability is diffi-
cult. Additionally, the ankle joint was represented as a
monoaxial hinge joint prohibiting any mediolateral
moment contributions at the ankle to frontal plane sta-
bility. In modern humans, the ankle plantarflexors and
everters generate substantial lateral moments on the
body COM during stance phase, and the plantar inver-
ters assist the minor gluteal muscles in generating medi-
al moments (Pandy and Andriacchi, 2010). Ignoring
these moment contributions on body COM movement in
the mediolateral plane may cause erroneous kinematic
and kinetics in the simulated A. afarensis model.
Dynamic mechanics of mediolateral balance
during locomotion. Biomechanical analyses over
the last 10 years have demonstrated the complexity of
mediolateral balance during locomotion. While the hip
abductor muscles are the primary producers of the medi-
al component of the GRF during walking and running,
other muscles of the lower limb, and gravity provide
important contributions to both the medial and lateral
movement of the body COM, altering the force required
by the hip abductors (Pandy and Andriacchi, 2010; Pan-
dy et al., 2010). Peak hip abduction force during walking
has been measured between one and two times body
weight (Pandy and Andriacchi, 2010; Warrener et al.,
2015, unpublished data) and up to four times body
weight at moderate running speeds (Pandy and Andriac-
chi, 2010; Lenhart et al., 2014; Warrener et al., 2015,
unpublished data). During early stance phase of both
walking and running, the body center of mass moves lat-
erally through the action of gravitational forces and the
activity of the vasti muscles, which also decelerate the
body’s forward progression until midstance (Pandy and
Andriacchi, 2010; Pandy et al., 2010). During the second
half of stance, the soleus and gastrocnemius accelerate
the body laterally, with peak forces of roughly 5% and
10% of body weight (Pandy and Andriacchi, 2010).
Throughout stance phase, the primary contributors to
the medial component of the GRF are the minor gluteal
muscles, but the plantar inverters also contribute medi-
ally directed forces. However, the minor gluteals are also
important for vertical support of the body, particularly
during walking, and forward progression during running
(Pandy and Andriacchi, 2010; Pandy et al., 2010). The
contribution of so many muscles of the lower limb to
mediolateral body stability, and the coupling of joint
moments, helps to explain why hip abductor mechanics
cannot be simply explained through a static lever model
system at the hip.
Because of the multijoint effect of muscle force produc-
tion, the hip abductors have been a primary focus in
attempts to understand frontal plane knee moments.
Abnormal moments about the knee joint are thought to
be responsible for several chronic knee injuries including
medial compartment osteoarthritis (Valente et al., 2013;
Shull et al., 2015), patellofemoral pain syndrome, and
anteriocruciate ligament injuries (Powers, 2010). Howev-
er, the specific effects of body kinetics and force profiles
with injury occurrence have been difficult to determine.
During locomotion, the medially directed GRF vector
acts to adduct the tibia, thereby generating a large
external knee adduction moment and placing the medial
knee compartment into compression (Pandy and
Andriacchi, 2010; Shull et al., 2015). Because the hip
abductors are the primary muscles producing the medi-
ally directed GRF, hip abductor forces have the potential
to increase external knee adduction moments (Pandy
and Andriacchi, 2010). Conversely, individuals with
increased dynamic genu valgus, a complex alignment
issue that involves hip placement and tibial orientation,
may have reduced external knee adduction moments,
placing more stress on the lateral compartment of the
knee and increasing lateral displacement of the patella
during stance phase (Barrios et al., 2016). Large dynam-
ic knee adduction angles during locomotion are often
associated with hip abductor weakness (Hewett, 2006;
Powers, 2010). However, therapies targeted at increasing
hip abductor strength to prevent knee adduction during
stance have had mixed results (Kean et al., 2015; Shull
et al., 2015).
 HOMININ HIP BIOMECHANICS
The complicated nature of individual muscle and kine-
matic modifications during locomotion is highlighted
by two recent studies. Pohl et al. (2015) administered a
superior gluteal nerve block to eight participants,
recording isometric hip abductor strength and gait data
before and after the intervention. Although hip abductor
strength was significantly reduced in all participants
after receiving the nerve block, during walking no signif-
icant changes in hip or knee frontal plane moments or
angular kinematics could be detected when hip abductor
function was impaired. Shull et al (2015) hypothesized
that subjects with medial compartment knee osteoarthri-
tis who were asked to walk with a toe-in gait to mini-
mize knee adduction moments would show uniform
muscle recruitment strategies associated with the new
gait kinematics. However, no consistent muscle force
recruitment patterns emerged across the subjects even
though individuals showed large changes in force pro-
duction in many muscles of the lower limb between the
two conditions. These studies suggest that the contribu-
tion of individual muscles to frontal plane kinematics
during locomotion remains poorly understood and may
be strongly individualized.
The results of numerous inverse dynamics and for-
ward dynamics studies now indicate the static model of
hip abductor function, strongly influenced by pelvic
anatomy, is incorrect. Unfortunately, these studies have
not yet provided clarity on how skeletal anatomy of the
lower limb, coupled with muscular force production is
specifically responsible for observed variations in frontal
plane locomotor kinematics across subjects. For anthro-
pological study, the lack of a clear correlation between
skeletal anatomy and hip abductor function is particu-
larly frustrating given that our only means of assessing
locomotor performance in extinct hominins is through
fossilized bones. Clinicians also must be wary of the tra-
ditional hip abductor model, understanding that many
other anatomical and dynamic factors contribute to
mediolateral stability of the body during locomotion.
However, forward dynamics modeling and experimental
gait data are now allowing us to capture the complexity
of the this essential locomotor task, which will ultimate-
ly lead to a clearer model of how humans and our extinct
ancestors managed the unusual task of taking to the
ground on two feet.
ACKNOWLEDGMENTS
The author would like to thank Jeremy DeSilva and
Karen Rosenberg for the invitation to contribute to this
article and three anonymous reviewers for their helpful
comments, which improved the article. She would also
like to thank Daniel Lieberman for his review of an ear-
lier version of this work.
LITERATURE CITED
Abitbol M. 1995. Reconstruction of the STS 14 (Australopithecus
africanus) pelvis. Am J Phys Anth 96:143–158.
Anderson FC, Pandy MG. 2001. Static and dynamic optimization
solutions for gait are practically equivalent. J Biomech 34:153–
161.
Arsuaga JL, Lorenzo C, Carretero JM, Gracia A, Martınez I, Garcıa
N, Berm udez de Castro JM, Carbonell E. 1999. A complete
943
human pelvis from the Middle Pleistocene of Spain. Nature 399:
255–258.
Barrios JA, Heitkamp CA, Smith BP, Sturgeon MM, Suckow DW,
Sutton CR. 2016. Clinical Biomechanics. Clin Biomech 31:7–11.
Baggaley M, Noegren B, Clasey J, Shapiro R, Pohl M. Frontal plane
kinematics of the hip during running: are they related to hip
anatomy and strength? Gait Posture 42:505–510.
Berge C, Goularas D. 2010. A new reconstruction of Sts 14 pelvis
(Australopithecus africanus) from computed tomography and
three-dimensional modeling techniques. J Hum Evol 58:262–272.
Berge C. 1994. How did the australopiths walk? A biomechanical
study of the hip and thigh of Australopithecus afarensis. J Hum
Evol 26:259–273.
Berge L, et al. 2015. Homo naledi, a new species of the genus Homo
from the Dinaledi Chamber, South Africa. eLIFE 4:1–35. e09560:
Biewener AA. 2004. Muscle mechanical advantage of human walk-
ing and running: implications for energy cost. J Appl Physiol 97:
2266–2274.
Bischoff JL, Williams RW, Rosenbauer RJ, Aramburu A, Arsuaga
JL, Garcia N, Cuenca-Besc os G. 2007. High-resolution U-series
dates from the Sima de los Huesos hominids yields: implications
for the evolution of the early Neanderthal lineage. J Archaeol Sci
34:763–770.
Bonmatı A, G omex-Olivencia A, Arsuaga JL, Carretero JM, Gracia
A, Martınez I, Lorenzo C, B ermudez de Castro JM, Carbonell E.
Middle Pleistocene lower back and pelvis from an aged human
individual from the Sima de los Huesos site, Spain. PNAS 107:
18386–18391.
Bourdin M, Pastene J, Germain M, Lacour J. 1993. Influence of
training, sex, age and body mass on the energy cost of running.
Eur J Appl Physiol O 66:439–444.
Brand RA, Pedersen DR, Davy DT, Kotzar GM, Heiple KG,
Goldberg VM. 1994. Comparison of hip force calculations and
measurements in the same patient. J Arthroplasty 9:45–51.
Brown F, Harris J, Leakey R, Walker A. 1985. Early Homo erectus
skeleton from west Lake Turkana, Kenya. Nature 316:788–792.
Bunc V, Heller J. 1989. Energy cost of running in similarly trained
men and women. Eur J Appl Physiol O 59:178–183.
Christodoulakis G, Busawon K, Caplan N, Stewart S. 2010. On the
filtering and smoothing of biomechanical data. Communication
Systems Networks and Digital Signal Processing (CSNDSP), 2010
7th International Symposium: 512–516.
Day M. 1971. Postcranial remains of Homo erectus from bed IV,
Olduvai Gorge, Tanzania. Nature 232:382–387.
de Leva P. 2004. Adjustments to Zatsiorsky segment inertia param-
eters. J Biomech 29:1223–1230.
Dempster W. 1955. The anthropometry of body action. Ann NY
Acad Sci 63:559–585.
Denham RA. 1959. Hip mechanics. J Bone Joint 41b:550–557.
DeSilva JM, Holt KG, Churchill SE, Carlson KJ, Walker CS, Zipfel
B, Berger LR. 2013. The lower limb and mechanics of walking in
Australopithecus sediba. Science 340:1232999.
Disselhortst-Klug C, Schmitz-Rode T, Rau G. 2009. Surface electro-
myography and muscle force: Limits in sEMG-force relationship
and new approaches for applications. Clin Biomech 24:225–235.
Dunsworth HM, Warrener AG, Deacon T, Ellison PT, Pontzer H.
2013. Metabolic hypothesis for human altriciality. Proc Natl Acad
Sci USA 109:15212–15216.
Gabunia L, Vekua A, Lordkipanidze D, Swisher C, Ferring R,
Justus A, Nioradze M, Tvalchrelidze M, Ant on S, Bosinski G.
2000. Earliest Pleistocene hominid cranial remains from Dmanisi,
Republic of Georgia: taxonomy, geological setting, and age. Sci-
ence 288:1019.
Gottschalk F, Kourosh S, Leveau B. 1989. The functional anatomy
of the tensor fasciae latae and gluteus medius and minimus.
J Anat 166:179–189.
Gr€un R, Stringer C. 1991. Electron spin resonance dating and the
evolution of modern humans. Archaeometry 33:153–199.
Haile-Selassie Y, Latimer BM, Alene M, Deino AL, Gibert L, Melillo
SM, Saylor BZ, Scott GR, Lovejoy CO. 2010. An early Australopi-
thecus afarensis postcranium from Woranso-Mille, Ethiopia. Proc
Natl Acad Sci USA 107:12121–12126.
944 WARRENER
Hall P, Figueroa A, Fernhall B, Kanaley JA. 2004. Energy expendi-
ture of walking and running: comparison with prediction equa-
tions. Med Sci Sports Exerc 36:2128.
H€ausler M, Schmid P. 1995. Comparison of the pelves of Sts 14 and
AL288-1: implications for birth and sexual dimorphism in austral-
opiths. J Hum Evol 29:363–383.
H€ausler M, Berger L. 2001. Stw 441/465: a new fragmentary ilium
of a small-bodied Australopithecus africanus from Sterkfontein,
South Africa. J Hum Evol 40:411–417. 1
Haeusler M. 2002. New insights into the locomotion of Australopi-
thecus africanus based on the pelvis. Evol Anthropol 11:53–57.
Hewett TE. 2006. Anterior cruciate ligament injuries in female ath-
letes: Part 1, mechanisms and risk factors. Am J Sports Med 34:
299–311.
Hirsch C, Frankel H. 1960. Analysis of forces producing fractures of
the proximal end of the femur. J Bone Joint Surg 42:633–640.
Hunt KD. 1998. Ecological morphology of Australopithecus afaren-
sis: traveling terrestrially, eating arboreally. In: Strasser E, Flea-
gle J, Rosenberger L, McHenry M, editors. Primate locomotion.
New York: Plenum Press. p 397–418.
Inman VT. 1947. Functional aspects of the abductor muscles of the
hip. J Bone Joint Surg Am 29:607–619.
Johanson DC, Lovejoy CO, Kimbel WH, White TD, Ward C, Bush
ME, Latimer BM, Coppens Y. 1982. Morphology of the Pliocene
partial hominid skeleton (AL 288-1) from the Hadar Formation,
Ethiopia. Am J Phys Anthropol 57:403–451.
John CT, Seth A, Schwartz MH, Delp SL. 2012. Contributions of
muscles to mediolateral ground reaction force over a range of
walking speeds. J Biomech 45:2438–2443.
Kean CO, Bennell KL, Wrigley TV, Hinman RS. 2015. Clin Biomech
30:226–230.
Kibii JM, Clarke RJ. 2003. A reconstruction of the Stw 431 Austral-
opithecus pelvis based on newly discovered fragments: news and
views. S Afr J Sci 99:225–226.
Kibii JM, Churchill SE, Schmid P, Carlson KJ, Reed ND, de Ruiter
DJ, Berger LR. 2011. A partial pelvis of Australopithecus sediba.
Science 333:1407–1411.
Kimbel W, Delezene L. 2009. “Lucy” redux: a review of research on
Australopithecus afarensis. Am J Phys Anthropol Suppl 140:2–48.
Latimer B, Lovejoy O. 1990. Metatarsophalangeal joints of Austral-
opithcsu afarensis. Am J Phys Anthropol 81:13–23.
Lenhart R, Thelen D, Heiderscheit B. 2014. Hip muscle loads dur-
ing running at various step rates. J Orthop Sports Phys Ther 44:
766. A4.
Liu MQ, Anderson FC, Pandy MG, Delp SL. 2006. Muscles that
support the body also modulate forward progression during walk-
ing. J Biomech 39:2623–2630.
Lovejoy CO, Heiple KG, Burstein AH. 1973. The gait of Australopi-
thecus. Am J Phys Anthropol 38:757–779.
Lovejoy CO, Suwa G, Spurlock L, Asfaw B, White TD. 2009a. The
pelvis and femur of Ardipithecus ramidus: The emergence of
upright walking. Science 326:71–71.
Lovejoy CO, Latimer B, Suwa G, Asfaw B, White TD. 2009b. Com-
bining prehension and propulsion: The foot of Ardipithecus rami-
dus. Science 326:72–72.
Lovejoy CO, Latimer BM, Spurlock L, Haile-Selassie Y. 2016. The
pelvic girdle and limb bones of KSD-VP-1/1. In: Haile-Selassie Y,
Su DF, editors. The postcranial anatomy of Australopithecus afar-
ensis. Springer. p 155–178.
Lovejoy CO. 1988. Evolution of human walking. Sci Am 259:118–125.
Lovejoy CO. 2005. The natural history of human gait and posture.
Gait Posture 21:95–112.
McHenry H. 1975. Biomechanical interpretation of the early homi-
nid hip. J Hum Evol 4:343–355.
McHenry H, Temerin L. 1979. The evolution of hominid bipedalism:
evidence from the fossil record. Yearb Phys Anthropol 22:105–
131.
McLeish R, Charnley J. 1970. Abduction forces in the one-legged
stance. J Biomech 3:191–209.
Meindl RS, Lovejoy CO, Mensforth RP, Carlos LD. 1985. Accuracy
and direction of error in the sexing of the skeleton: implications
for paleodemography. Am J Phys Anthropol 68:79–85.
Merchant AC. 1965. Hip abductor muscle force. J Bone Joint Surg
Am 47:462–476.
Nagano A, Umberger BR, Marzke MW, Gerritsen KGM. 2005. Neu-
romusculoskeletal computer modeling and simulation of upright,
straight-legged, bipedal locomotion of Australopithecus afarensis
(A.L. 288-1). Am J Phys Anthropol 126:2–13.
Napier JR. 1964. The evolution of bipedal walking in the hominids.
Arch Biol 75:673–708.
Osborne GV, Fahrini WH. 1950. Oblique displacement osteotomy for
osteoarthritis of the hip joint. J Bone Joint Surg Br 32-B:148–
160.
Otten E. 2003. Inverse and forward dynamics: models of multi-body
systems. Philos T Roy Soc B Biol Sci 358:1493–1500.
Pandy MG. 2001. Computer modeling and simulation of human
movement. Annu Rev Biomed Eng 3:245–273.
Pandy MG, Andriacchi TP. 2010. Muscle and joint function in
human locomotion. Annu Rev Biomed Eng 12:401–433.
Pandy MG, Lin Y-C, Kim HJ. 2010. Muscle coordination of medio-
lateral balance in normal walking. J Biomech 43:2055–2064.
Pauwels F. 1976. Biomechanics of the normal and diseased hip. Ber-
lin: Springer-Verlag.
Pickering R, Dirks PHGM, Jinnah Z, de Ruiter DJ, Churchil SE,
Herries AIR, Woodhead JD, Hellstrom JC, Berger LR. 2011. Aus-
tralopithecus sediba at 1.977 Ma and implications for the origins
of the genus Homo. Science 333:1421–1423.
Pohl M, Kendall K, Patel C, Wiley J, Emery C, Ferber R. Experi-
mentally reduced hip-abductor muscle strength and fronta-plane
biomechanics during walking. J Athl Training 50:385–391.
Ponce de Le on MS, Golovanova L, Doronichev V, Romanova G,
Akazawa T, Kondo O, Ishida H, Zollikofer CPE. 2008. Neander-
thal brain size at birth provides insights into the evolution of
human life history. Proc Natl Acad Sci USA 105:13764–13768.
Powers CM. 2010. The influence of abnormal hip mechanics on
knee injury: A biomechanical perspective. J Orthop Sports Phys
Ther.
Rak Y, Arensburg B. 1987. Kebara 2 Neanderthal pelvis: First look
at a complete inlet. Am J Phys Anthropol 73:227–231.
Rak Y. 1990. On the differences between two pelvises of Mousterian
context from the Qafzeh and Kebara caves, Israel. Am J Phys
Anthropol 81:323–332.
Richmond B, Jungers W. 2008. Orrorin tugenensis femoral morphology
and the evolution of hominin bipedalism. Science 319:1662–1665.
Rightmire GP. 2004. Brain size and encephalization in early to Mid-
PleistoceneHomo. Am J Phys Anthropol 124:109–123.
Robertson G, Caldwell G, Hamill J, Kamen G. 2004. Research
methods in biomechanics. Champaign: Human Kinetics.
Robinson J. 1972. Early hominid posture and locomotion. Chicago:
University of Chicago Press.
Rose M. 1984. A hominine hip bone, KNM-ER 3228, from East
Lake Turkana, Kenya. Am J Phys Anthropol 63:371–378.
Rosenberg K, Brace D, Frayer E, Geise M, Greene E, Tague R,
Tappen N, Trevathan W, Trinkaus E. 1988. The functional signifi-
cance of neandertal public length [and comments and reply]. Curr
Anthropol 29:595–617.
Rosenberg KR. 1992. The evolution of modern human childbirth.
Am J Phys Anthropol 35:89–124.
Rosenberg K. 1998. Morphological variation in west Asian postcra-
nia. In Akazawa T, Aoki K, Bar-Yosef O, editors. Neandertals and
modern humans in Western Asia. New York: Kluwer Acedemic
Publishers. p 367–379.
Rosenberg KR, Zun e L, Ruff CB. 2006. Body size, body proportions,
and encephalization in a Middle Pleistocene archaic human from
northern China. Proc Natl Acad Sci 103:3552–3556.
Ruff C. 1991. Climate and body shape in hominid evolution. J Hum
Evol 21:81–105.
Ruff C. 1994. Morphological adaptation to climate in modern and
fossil hominids. Am J Phys Anthropol 37:65–107.
Ruff C. 1995. Biomechanics of the hip and birth in early Homo. Am
J Phys Anthropol 98:527–574.
Ruff C. 1998. Evolution of the hominid hip. In: Strasser E, Fleagle
J, Rosenberger A, McHenry H, editors. Primate locomotion:
Recent advances. New York: Plenum Press. p 449–469.
 HOMININ HIP BIOMECHANICS
Ruff C. 2010. Body size and body shape in early hominins - Implica-
tions of the Gona Pelvis. J Hum Evol 58:166–178.
Ruff C, Higgins R. 2013. Femoral neck structure and function in
early hominins. Am J Phys Anthropol 150:512–525.
Ruff C, Trinkaus E, Holliday TW. 1997. Body mass and encephaliza-
tion in Pleistocene Homo. Nature 387:173–176.
Sariali E, Veysi V, Stewart T. 2008. Biomechanics of the human hip –
Consequences for total hip replacement. Curr Ortho 22:371–375.
Saunders JB, Inman VT, Eberhart HD. 1953. The major determi-
nants in normal and pathological gait. J Bone Joint Surg Am 35-
A:543–558.
Schmid. 1983. Eine Rekonstruktion des Skelettes von A.L. 288-1
(Hadar) und deren Konsequezen. Folia Primatol 40:283–306.
Sellers WI, Cain GM, Wang W, Crompton RH. 2005. Stride lengths,
speed and energy costs in walking of Australopithecus afarensis:
using evolutionary robotics to predict locomotion of early human
ancestors. J Roy Soc Inter 2:431–441.
Shull PB, Huang Y, Schlotman T, Reinbolt JA. 2015. Muscle force
modification strategies are not consistent for gait retraining to
reduce the knee adduction moment in individuals with knee oste-
oarthritis. J Biomech 48:3163–3169.
Simpson SW, Quade J, Levin NE, Butler R, Dupont-Nivet G,
Everett M, Semaw S. 2008. A female Homo erectus pelvis from
Gona, Ethiopia. Science 322:1089–1092.
Simpson SW, Quade J, Levin NE, Semaw S. 2014. The female Homo
pelvis from Gona: response to Ruff (2010). J Hum Evol 68:32–35.
Spoor F, Leakey M, Gathogo P, Brown F, Ant on S, McDougall I,
Kiarie C, Manthi F, Leakey L. 2007. Implications of new early
Homo fossils from Ileret, east of Lake Turkana, Kenya. Nature
448:688–691.
Stern J, Susman R. 1981. Electromyography of the gluteal muscles
in Hylobates, Pongo, and Pan: implications for the evolution of
hominid bipedality. Am J Phys Anthropol 55:153–166.
Stern J, Susman R. 1983. The locomotor anatomy of Australopithe-
cus afarensis. Am J Phys Anthropol 60:279–317.
Tague RG. 1989. Variation in pelvic size between males and
females. Am J Phys Anthropol 80:59–71.
Tague RG, Lovejoy CO. 1986. The obstetric pelvis of AL 288-1
(Lucy). J Hum Evol 15:237–255.
945
Traina F, De Clerico M, Biondi F, Pilla F, Tassinari E, Toni A. 2009.
Sex differences in hip morphology: Is stem modularity effective
for total hip replacement?. J Bone Joint Surg 91:121–128.
Trinkaus E. 1976. The morphology of European and Southwest
Asian Neandertal pubic bones. Am J Phys Anthropol 44:95–103.
Trinkaus E. 1984. Neandertal pubic morphology and gestation
length. Curr Anthropol 25:509–514.
Valente G, Taddei F, Jonkers I. 2013. Influence of weak hip abduc-
tor muscles on joint contact forces during normal walking: proba-
bilistic modeling analysis. J Biomech 46:2186–2193.
Walker A. 1993. Perspectives on the Nariokotome discovery. In
Walker A, Leakey R, editors. Nariokotome Homo erectus skeleton.
Cambridge: Harvard University Press. p 1–6.
Walker A, Ruff C. 1993. The reconstruction of the pelvis. In Walker
A, Leakey R, editors. Nariokotome Homo erectus skeleton. Cam-
bridge: Harvard University Press. p 221–233.
Ward C. 2002. Interpreting the posture and locomotion of Australo-
pithecus afarensis: where do we stand?. Am J Phys Anthropol
Suppl 35:185–215.
Ward C, Kimbel W, Johanson D. 2011. Complete fourth metatarsal
and arches in the foot of Australopithecus afarensis. Science 331:
750–753.
Warrener AG, Lewton KL, Pontzer H, Lieberman DE. 2015. A wider
pelvis does not increase locomotor cost in humans, with implica-
tions for the evolution of childbirth. PLoS ONE 10:e0118903.
Weaver TD, Hublin J-J. 2009. Neandertal birth canal shape and the
evolution of human childbirth. Proc Natl Acad Sci USA 106:8151–
8156.
Winter DA. 2009. Biomechanics and motor control of human move-
ment. 4th ed. Hoboken: John Wiley & Sons, Inc.
Zajac FE, Neptune RR, Kautz SA. 2002. Biomechanics and muscle
coordination of human walking. Part I: introduction to concepts,
power transfer, dynamics and simulations. Gait Posture 16:215–
232.
Zajac FE, Neptune RR, Kautz SA. 2003. Biomechanics and muscle
coordination of human walking: part II: lessons from dynamical
simulations and clinical implications. Gait Posture 17:1–17.
Zihlman A, Brunker L. 1979. Hominid bipedalism: Then and now.
Yearb Phys Anthropol 22:132–162.