Foot strength and stiffness are related to footwear use in a comparison of minimally- vs. conventionally-shod populations

Abstract

The longitudinal arch (LA) helps stiffen the foot during walking, but many people in developed countries suffer from flat foot, a condition characterized by reduced LA stiffness that can impair gait. Studies have found this condition is rare in people who are habitually barefoot or wear minimal shoes compared to people who wear conventional modern shoes, but the basis for this difference remains unknown. Here we test the hypothesis that the use of shoes with features that restrict foot motion (e.g. arch supports, toe boxes) is associated with weaker foot muscles and reduced foot stiffness. We collected data from minimally-shod men from northwestern Mexico and men from urban/suburban areas in the United States who wear 'conventional' shoes. We measured dynamic LA stiffness during walking using kinematic and kinetic data, and the cross-sectional areas of three intrinsic foot muscles using ultrasound. Compared to conventionally-shod individuals, minimally-shod individuals had higher and stiffer LAs, and larger abductor hallucis and abductor digiti minimi muscles. Additionally, abductor hallucis size was positively associated with LA stiffness during walking. Our results suggest that use of conventional modern shoes is associated with weaker intrinsic foot muscles that may predispose individuals to reduced foot stiffness and potentially flat foot.

Full text

1
Scientific REpoRts | (2018) 8:3679 | DOI:10.1038/s41598-018-21916-7
www.nature.com/scientificreports
Foot strength and stiness are
related to footwear use in a
comparison of minimally- vs.
conventionally-shod populations
Nicholas B. Holowka, Ian J. Wallace & Daniel E. Lieberman
The longitudinal arch (LA) helps stien the foot during walking, but many people in developed countries
suer from at foot, a condition characterized by reduced LA stiness that can impair gait. Studies
have found this condition is rare in people who are habitually barefoot or wear minimal shoes compared
to people who wear conventional modern shoes, but the basis for this dierence remains unknown.
Here we test the hypothesis that the use of shoes with features that restrict foot motion (e.g. arch
supports, toe boxes) is associated with weaker foot muscles and reduced foot stiness. We collected
data from minimally-shod men from northwestern Mexico and men from urban/suburban areas in
the United States who wear ‘conventional’ shoes. We measured dynamic LA stiness during walking
using kinematic and kinetic data, and the cross-sectional areas of three intrinsic foot muscles using
ultrasound. Compared to conventionally-shod individuals, minimally-shod individuals had higher and
stier LAs, and larger abductor hallucis and abductor digiti minimi muscles. Additionally, abductor
hallucis size was positively associated with LA stiness during walking. Our results suggest that use
of conventional modern shoes is associated with weaker intrinsic foot muscles that may predispose
individuals to reduced foot stiness and potentially at foot.
As bipeds, humans have evolved dramatically dierent feet from other primates1. One of the most distinctive fea-
tures of the human foot is the longitudinal arch (LA), whose anatomical scaold is created by the conformation of
the tarsal and metatarsal bones, and which is reinforced by numerous so tissue structures that span the plantar
surface of the foot. e LA stiens the foot under loading, enabling it to function as a propulsive lever during
walking and running2. LA stiness partly derives from ligamentous structures, including the long and short
plantar ligaments, the spring ligament and the plantar aponeurosis, that traverse the plantar surface of the foot
longitudinally and act as trusses to resist compressive forces on the LA3. e intrinsic foot muscles also contribute
to LA stiness by contracting to help control LA deformation during walking and running4,5, thereby relieving an
unknown proportion of the stress borne by the plantar ligaments.
e standing height of the LA on the medial side of the foot is the most commonly used indicator of relative
arch height6. Individuals with exceptionally low LAs while standing are characterized as having at foot (pes
planus). All humans are born with a low arch, and most develop a fully adult conguration of the LA by 10–12
years of life7. However, roughly 20–25% of adults in the United States and Canada are diagnosed as having at
feet8–11, either because they fail to develop a normal height arch or because the arch collapses. Most individuals
diagnosed with at foot possess a so-called ‘exible’ at foot, characterized by substantial eversion of the rear foot
during weight-bearing, resulting in a marked drop in LA height12, and reduced LA stiness during walking13,14.
Although this condition is oen asymptomatic12, in some individuals it causes foot pain and fatigue aer long
durations standing and/or walking15. Reduced LA stiness is also a risk factor for numerous lower extremity
musculoskeletal disorders including plantar fasciitis, knee osteoarthritis, tibialis posterior tendinopathy, and met-
atarsal stress fracture11,16–19. us, developing strategies to prevent and treat this condition is an important health
objective.
Despite the high incidence at feet in the US and other developed nations, many studies report lower rates of
at foot in habitually barefoot or minimally-shod populations20–28. In one of the largest of these studies, which
Department of Human Evolutionary Biology, Harvard University, Cambridge, MA, 02138, USA. Correspondence and
requests for materials should be addressed to N.B.H. (email: nick_holowka@fas.harvard.edu)
Received: 5 September 2017
Accepted: 9 February 2018
Published: xx xx xxxx
OPEN
Content courtesy of Springer Nature, terms of use apply. Rights reserved

www.nature.com/scientificreports/
2
Scientific REpoRts | (2018) 8:3679 | DOI:10.1038/s41598-018-21916-7
included 1,846 adults from southern India, Sachithanandam and Joseph28 found that individuals who never wore
shoes before age 16 had roughly half the rate of at foot of those who grew up wearing shoes. More recently, in
a study of 810 school children between 6 and 18 years old, Hollander et al.25 found signicantly higher LAs in
children who were habitually barefoot compared to those who were habitually shod. ese ndings are poten-
tially signicant given that, until relatively recently, all humans were either barefoot or wore minimal footwear
lacking the cushioning, arch supports, restrictive toe boxes and other features of conventional shoes. It has also
been shown that the use of minimal shoes by adults who grew up in conventional modern shoes is associated with
increases in intrinsic foot muscle size, as well as LA height and stiness29–31. It is therefore reasonable to hypoth-
esize that the reduction in LA stiness that characterizes at foot is a mismatch condition caused by the human
foot being inadequately adapted to the novel environmental condition of wearing shoes that provide comfort and
protection at the expense of weaker foot muscles32. However, no study has investigated whether people who grew
up habitually barefoot or wearing minimal shoes have stronger foot muscles than those who grow up wearing
conventional modern shoes. us, the relationship between foot muscle strength, footwear use, and LA stiness
needs to be tested.
is study uses retrospective data as an initial test of the hypothesis that individuals who are habitually
minimally-shod throughout life have stronger foot muscles and stier feet than those who habitually wear ‘con-
ventional shoes’, which we dene here as shoes with some combination of features that aect natural foot motion,
including restrictive toe boxes, heel counters, arch supports and toe springs. Since few individuals in the United
States or other developed countries grow up barefoot or minimally-shod, we compared LA stiness and intrin-
sic foot muscle strength in conventionally-shod adults from the United States with age-matched adults from a
minimally-shod population of Tarahumara (Rarámuri) Native Americans from the Sierra Tarahumara, a moun-
tainous region of northwestern Mexico. Although, they have gained renown for ultra-long distance running33,
most Tarahumara run infrequently, with most of their physical activity consisting of farming and walking long
distances. During all activities, including running, they typically wear minimal sandals (huaraches) consisting of
soles made from car tire rubber axed to the foot and ankle by leather thongs (Fig.1). In recent times, conven-
tional modern footwear has become increasingly common among the younger Tarahumara, many of whom are
moving from isolated farms to urban environments. A recent study found that conventionally-shod Tarahumara
have signicantly less sti and lower LAs than minimally-shod Tarahumara33. For the present study, we predicted
Figure 1. A Tarahumara man wearing a typical sandal with a sole made from car tire rubber. Photo copyright©
2018 by David Ramos and used here with permission. All rights reserved.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

www.nature.com/scientificreports/
3
Scientific REpoRts | (2018) 8:3679 | DOI:10.1038/s41598-018-21916-7
that minimally-shod Tarahumara men have both larger intrinsic foot muscles and stier LAs than men of similar
ages and body sizes who have been habitually conventionally-shod most of their lives. We also predicted that
intrinsic foot muscle size is positively correlated with LA stiness across groups.
An additional hypothesis this study tested is whether static measures of LA stiness accurately reect the
dynamic function of the foot during walking. Most previous studies of barefoot and minimally-shod populations
measured only static LA height and stiness, but any inuence of LA deformation on musculoskeletal disorders
occurs primarily during dynamic loading via mechanisms such as increased stress on plantar so tissue struc-
tures and higher bending forces on the metatarsals16,17. Hollander et al.25 measured static arch height based on
palpable anatomical landmarks and found that these measurements do not always correspond to dynamic arch
indices measured during walking using a pedography platform. However, this result is perhaps not surprising, as
McPoil and Cornwall34 have demonstrated that pedography-based estimates of arch height are poor predictors
of LA height measurements that are based on anatomical landmarks, and thus may not accurately reect changes
in LA height in response to loading. is nding indicates the necessity of measuring of LA height and stiness
dynamically using kinematic data rather than just pedography measurements. We predicted that minimally-shod
individuals have dynamically stier LAs during walking than conventionally-shod individuals, and that dynamic
LA stiness is positively correlated with both static LA stiness and intrinsic foot muscle size.
Methods
Sample. For the minimally-shod population, we collected data from 75 Tarahumara men (mean ± SD: age,
64 ± 10 yrs; body mass, 64 ± 10 kg; height, 1.58 ± 0.06 m) from the area around the Barranca de Sinforosa in the
southwestern part of the state of Chihuahua, Mexico in May and June of 2016. Participants were recruited by
word of mouth with the help of members of the community. To exclude individuals who do not primarily wear
sandals, we limited our recruitment to individuals 50 years or older, as many younger Tarahumara have grown up
wearing shoes, and we excluded individuals who reported that they wore sandals less than ve days/week in warm
seasons. For the habitually conventionally-shod population, we recruited by word of mouth an age-matched
sample of 26 men (mean ± SD: age, 57 ± 11 yrs; body mass, 82 ± 12 kg; heig ht, 1.79 ± 0.08 m) from urban and
suburban areas in the United States (Boston, MA, Ithaca, NY, and Dayton, OH), all of whom habitually wear
conventional modern shoes. For all minimally- and conventionally-shod participants, exclusion criteria included
recent foot pain, previous injury to the foot and any overt gait abnormality. Participants also completed a survey
in which they estimated the average number of hours walked and run per day over the previous ve years.
All U. S. participants gave their written informed consent, and all Tarahumara participants provided verbal
informed consent, which was administered by translators who spoke Spanish and Rarámuri (the native language
of the Tarahumara). All procedures involving U. S. and Tarahumara participants were approved by Harvard
University’s Institutional Review Board, and all research was carried out in accordance with the approved guide-
lines and regulations.
Anthropometrics. We measured participant height and body mass and used these to calculate Body Mass
Index (BMI) as body mass/height2. We also measured lower limb length (distance from greater trochanter to the
ground), and used a custom-machined device to measure total foot length, truncated foot length (from the heel to
the rst metatarsophalangeal joint) and the dorsum height at 50% of foot length in both seated and standing con-
ditions. Arch Height Index (AHI) was calculated as the ratio of the foot’s dorsum height at 50% of foot length to
the total length of the foot excluding the toes when participants were standing. is has been shown to be a robust
and repeatable measure of LA height6. Following previous studies35,36, we classied participants as having ‘low’
LAs if their AHI values were below 0.297, which is 1.5 standard deviations below the average AHI reported in a
large sample of adult males from the U.S.6. Arch Stiness Index (ASI), which is a static measure of LA stiness,
was calculated using seated and standing AHI values: ASI = ( body mass*0.4)/(AHIseated − AHIstanding)37.
Ultrasound. Cross sectional areas of the intrinsic foot muscles in the right foot were captured using a Philips
L12-4 B-Mode Ultrasound Transducer (Philips Ultrasound, Inc., Bothell, WA), which has a 4–12 MHz frequency
range and a 41 mm linear array. To avoid possible inter-investigator error, all ultrasound images were captured
by a single, trained investigator (N.B.H.). To standardize image capture across participants, the navicular tuber-
osity was identied using surface palpation, and a line was drawn across the plantar surface of the participant’s
foot using an ink marker to indicate the frontal plane bisecting the navicular tuberosity (Fig.2). e ultrasound
transducer was moved along this line to take frontal plane images of the foot, which were instantly recorded on
a Samsung Galaxy Tablet S2 (Samsung Electronics America, Inc., Ridgeeld Park, NJ) using Lumify soware
(Philips Ultrasound, Inc., Bothell, WA).
Ultrasound images were used to quantify the cross-sectional areas of three intrinsic foot muscles: abductor
hallucis (AH; Fig.2a), exor digitorum brevis (FDB; Fig.2b), and abductor digiti minimi (ADM; Fig.2c). We
measured AH and FDB because both muscles are thought to help stien the LA during locomotion5,38,39. e role
of ADM in LA support is unclear4, but was included because Miller et al.29 found that this muscle increased size in
runners who switched to training in minimal shoes. e cross-sectional areas of these muscles have been shown
to be measureable using ultrasound with high intra-investigator reliability31,40,41. One investigator (N.B.H.) meas-
ured muscle cross-sectional area in ultrasound images using the draw tool in ImageJ42. To avoid potential bias in
measurement, all ultrasound images were assigned random ID numbers and measured without the investigator
knowing which individual or population the image came from. For some muscles in some participants, muscle
boundaries were not clear due to factors such as thick callused skin that impeded ultrasound waves, and in these
cases these muscles were not measured.
For statistical analyses, cross-sectional area values were scaled by dividing by (body mass)0.67, under the
assumption of geometric similarity, where cross-sectional area ∝ (body mass)0.67. is relationship assumes that
Content courtesy of Springer Nature, terms of use apply. Rights reserved

www.nature.com/scientificreports/
4
Scientific REpoRts | (2018) 8:3679 | DOI:10.1038/s41598-018-21916-7
anatomical structures scale isometrically with respect to body size, and may be appropriate for intra-species
comparisons where dierent individuals are expected to be geometrically similar to one another43. Alexander
et al.44 and Myatt et al.45 have found that hind limb muscle physiological cross-sectional area tends to scale with a
slightly higher coecient of allometry ~ (body mass)0.75 in bovids and non-human great apes than predicted by
geometric similarity, making this a conservative scaling metric. Physiological cross-sectional area diers from the
cross-sectional area measurements used in this study because it is also proportional to the cosine of ber penna-
tion angle. However, because the muscles measured in this study all have low ber pennation angles in humans
(<20°)46, their physiological cross-sectional areas will be only minimally aected by pennation.
Kinematic and kinetic data. We collected kinematic and kinetic data from all conventionally-shod par-
ticipants and from a similar-sized random subsample of minimally-shod participants (N = 30; mean ± SD: age,
60 ± 8 yrs; body mass, 66 ± 12 kg; height, 1.58 ± 0.06 m). Participants walked barefoot over an Emed q-100 pedog-
raphy platform (Novel GmbH, Munich, Germany) while they were being video recorded by two GoPro Hero
4 cameras (GoPro, Inc., San Mateo, CA, USA), with 7.5 mm 3MP M12 lenses (Back-Bone, Inc., Kanata, ON,
Canada). One camera was positioned 0.5 m from the pedography platform to record a medial view of the par-
ticipant’s right foot at a frame capture rate of 240 Hz. e second camera was positioned 2 m from the pressure
platform to record a lateral view of the participant’s full body at a frame capture rate of 120 Hz. e pedography
platform recorded vertical ground reaction forces at a rate of 100 Hz. Camera and platform recordings were syn-
chronized using a light on the platform that illuminated at the instant of foot contact (‘touchdown’). e end of
stance, ‘lio’, was determined from the medial camera video as the frame when the toes lost contact with the
platform.
Prior to recording, small circular white tape markers were placed on the right lower limbs of participants. To
measure LA angle, we placed markers on the medial aspect of the rst metatarsal head, the navicular tuberosity,
and the medial aspect of the posterior calcaneus (Fig.3), and to measure walking speed we placed a marker on
the greater trochanter. At the start of recording sessions, participants were instructed to practice walking barefoot
across the pedography platform until they could touchdown near the center of the platform with the right foot
Figure 2. Examples of muscle cross-sectional area images taken using ultasound. (a) Abductor hallucis, (b)
exor digitorum brevis, and (c) abductor digiti minimi. Dashed line in foot skeleton illustrations indicates plane
in which images were taken, based on palpation of the navicular tuberosity.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

www.nature.com/scientificreports/
5
Scientific REpoRts | (2018) 8:3679 | DOI:10.1038/s41598-018-21916-7
while maintaining a normal gait at a constant, comfortable speed. We then recorded subjects walking for a mini-
mum of three trials at self-selected speeds.
We digitized marker position in videos of the medial foot in MATLAB (MathWorks, Inc., Natick, MA) using
the DigitizingTools_20160818 package47. To reduce signal artifacts caused by digitizing error, we ltered the raw
marker coordinate data using a fourth order low-pass Butterworth lter with a 20 Hz cuto frequency. We used a
custom-written MATLAB routine to calculate two LA motion-related variables from the ltered data: maximum
LA angle during stance (θmax), and mid-stance LA stiness (kmid). θmax was calculated as the maximum angle
formed by the three markers on the medial foot during stance phase, with LA angle at touchdown set to 0°, and
LA angle increasing with greater midfoot deformation (Fig.3a). kMid was calculated using the following formula:
=Δ
k
F
LA height (1)
midmid
Fmid is the vertical ground reaction force at 50% of stance phase, which wemeasured with the pedography
platform. To determine ΔLA height, we calculated LA height as the perpendicular distance between the navicular
tuberosity and a line bisecting the rst metatarsal head and medial calcaneus markers (Fig.3b). We calculated
ΔLA height as the dierence between LA height at touchdown and LA height at 50% of stance phase. We calcu-
lated this LA stiness value at 50% of stancebecause this is when fore-a ground reaction forces are near 0 N,
making the three-dimensional ground reaction force vector nearly perpendicular to the ground. Because the
whole foot is in contact with the ground at this point in stance, the linear dimension used to measure LA height
is also roughly perpendicular to the ground, and therefore parallel to the ground reaction forcevector. us, we
expect the change in LA height at mid-stance to be caused by vertical ground reaction forces, and therefore kmid
should reect relative LA stiness at this point in stance. kmid was standardized by dividing by (body mass)0.67,
under the assumption that kmid should scale geometrically, as it does for limb stiness48.
For each stride analyzed we measured walking speed in ImageJ42 using the lateral camera videos. To do so we
calculated the distance travelled by the greater trochanter marker during the full stride cycle in which the foot
contacted the pedography platform divided by the stride duration. To analyze walking speed as a dimensionless
variable49 and thus facilitate comparisons among individuals with dierent leg lengths, we calculated Froude
number (Fr) as
=Fr v
gL (2)
2
where v is walking speed, g is the gravitational constant (9.81 m/s2), and L is greater trochanter height during
standing.
Statistical Analysis. All statistical analyses were performed in R50. All variables were inspected for nor-
mality and for homogeneity of variance between the two groups. ADM, ASI and kmid were log-transformed
Figure 3. Measurements of LA kinematics during walking. (a) LA angle, where Maxθ was measured at the
maximum value across stance as indicated by the dashed line in lower graph. (b) ΔLA height, where ΔLA
height was measured at mid-stance, as indicated by the dashed line in the lower graph. 0° and 0 cm height are
dened by marker positions at foot strike.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

www.nature.com/scientificreports/
6
Scientific REpoRts | (2018) 8:3679 | DOI:10.1038/s41598-018-21916-7
to achieve normality. To test for dierences between groups in each anthropometric (AHI, ASI, BMI), muscle
cross-sectional area (AH, FDB, ADM), and kinematic (kmid and θmax) response variable we created general linear
models in which group identity was included as a xed factor, and dierent predictor variables were included in
the model as follows: age for all response variables, BMI for response variables that were not scaled by body mass
(AHI and θmax), and Fr for kinematic variables (kmid and θmax). We performed ANOVAs on model variance to test
for dierences between groups. To test for relationships between muscle cross-sectional area and anthropometric/
kinematic variables, as well as ASI and kmid, we pooled data between groups and checked that variables were lin-
early related and t bivariate normal distributions. In cases where these assumptions were met, we used Pearson’s
product moment correlations to test for association, and otherwise we used Spearman’s rank correlation. Alpha
levels for all statistical tests were set at 0.05.
To assess the potential eects of physical activity on the variables measured in this study, we summed the
self-reported hours walked and hours run per day to create a physical activity (PA) variable. PA was poorly
matched between groups, and not normally distributed within groups. us, we conducted a matched sample
analysis by comparing the subset of minimally and conventionally-shod participants who had overlapping PAs
(1–3.9 hours walked/run per day), and used Wilcoxon Rank-Sum tests to test for dierences between groups in
each of the response variables.
Data Availability. All processed data analyzed for this study are available as supplementary data. Raw data
are available from the corresponding author on reasonable request.
Results
Anthropometrics and Muscle Cross-Sectional Area. Standing arch height as measured by AHI
and arch stiness as measured by ASI were 9% (GLM [General Linear Model]: P < 0.0001, df [degrees of free-
dom] = 97) and 27% (GLM: P = 0.009, df = 97) higher, respectively, in minimally-shod participants than in con-
ventionally-shod participants (Fig.4a; Table1; see Table2 for GLM results). Based on the AHI cut-o value of
0.297, 31% of conventionally-shod participants (8/26) had low arches, whereas only one of the 75 minimally-shod
participants had a low arch (1%). BMI also covaried signicantly with AHI (GLM: P = 0.01, df = 97), but BMI
did not dier between minimally- and conventionally-shod samples (GLM: P = 0.64, df = 97). Since clear images
were not obtainable for all muscles for all participants, we had smaller sample sizes for muscle cross-sectional
area comparisons (Table1). Cross-sectional areas of the AH and ADM in minimally-shod participants were
0.2 cm2 and 0.1 cm2 larger on average than in conventionally-shod participants, respectively, and these dierences
were signicant aer scaling by body size (AH – GLM: P < 0.0001, df = 69; ADM – GLM: P = 0.001, df = 42)
(Fig.4b; Tables1 and 2). Cross-sectional areas for FDB in conventionally-shod participants were 0.2 cm2 larger
on average than those in minimally-shod participants, but aer scaling for body size FDB was slightly but not
signicantly larger in minimally-shod participants (GLM: P = 0.2, df = 75) (Fig.4b; Tables1 and 2). Of the foot
muscle cross-sectional areas measured, only AH was signicantly associated with AHI (Pearson’s product-mo-
ment correlation [PPC]: P = 0.03, r = 0.26, df = 70). None of the scaled muscle cross-sectional areas were signi-
cantly associated with ASI (P > 0.05).
Kinematics and kinetics. Conventionally-shod participants had 12% longer legs on average and 7% faster
average walking speeds (1.04 ± 0.13 m/s) than minimally-shod participants (0.97 ± 0.18 m/s), but participants
from both groups walked with identical average Froude numbers (0.12 ± 0.03), indicating dynamic similarity.
For both minimally and conventionally-shod participants, LA height dropped gradually following touchdown
until reaching its lowest point around 75% of stance (Fig.5b). ereaer, LA height rapidly increased, reaching
approximately the same height at li o as at touchdown in minimally-shod participants, and a slightly greater
height than at touchdown in conventionally-shod participants. LA angle changed inversely with LA height, but
otherwise followed a nearly identical pattern during stance (Fig.5a).
Average θmax was 27% higher in conventionally-shod than minimally-shod participants (GLM: P = 0.002,
df = 50) (Fig.4c; Tables1 and 2). Midstance arch stiness, kmid, was 470 N/cm higher in the minimally versus
conventionally-shod participants, a dierence that remained signicant aer scaling by body mass and logging
(GLM: P = 0.002, df = 51). Froude did not covary signicantly with either variable. kmid was not signicantly
associated with participant ASI (PPC: P = 0.6, r = −0.07, df = 54). Scaled AH was negatively associated with θmax
(PPC: P = 0.001, r = −0.46, df = 44), and positively associated with scaled and logged kmid (PPC: P = 0.03, r = 0.31,
df = 44). Scaled FDB and ADM were not signicantly associated with either kinematic variable (P > 0.05).
Effect of Physical Activity. For the PA-matched comparisons, 10 conventionally-shod and 16
minimally-shod participants had PAs between 1 and 3.9 hours/day (Table3). Within these subgroups,
minimally-shod participants were 14 years older on average than conventionally-shod participants (68.8 ± 12.1
vs. 54.8 ± 9.9 years; WRS [Wilcoxon Rank-Sum]: P = 0.01, df = 27), and had slightly but not signicantly higher
BMI (WRS: P = 0.08, df = 27). As in the full sample comparison, minimally-shod participants had signicantly
higher AHI (WRS: P = 0.005, df = 27), larger scaled ADM (WRS: P = 0.02, df = 12), and higher scaled kmid (WRS:
P = 0.007, df = 16) than PA-matched conventionally-shod participants. ey also had larger scaled AH and FDB,
and lower θmax, but this dierence was not signicant (P > 0.05; Table3). Unlike in the full sample comparison,
conventionally-shod participants had slightly higher ASI than minimally-shod participants, but this dierence
was not signicant (WRS: P = 0.9, df = 27).
Discussion
is study compared intrinsic foot muscle size and foot biomechanics in two groups of adult men while con-
trolling for covariates such as age, body size and to some extent physical activity to test the hypothesis that people
who are habitually minimally-shod have bigger foot muscles and stier feet than people who habitually wear
Content courtesy of Springer Nature, terms of use apply. Rights reserved

www.nature.com/scientificreports/
7
Scientific REpoRts | (2018) 8:3679 | DOI:10.1038/s41598-018-21916-7
0.0 0.2 0.4
0 2000 4000
AHI
ASI
Arch Height Index Arch Stiffness Index
*
*
0.0 0.1 0.2
CSA/BM2/3 (cm2/kg2/3)
AH FD
BA
DM
*
*
0510 15
0100 200
a
b
θmax ( o )
kmid/BM2/3 (N/mkg2/3)
θmax kmid
**
c
Minimally-Shod Conventionally-Shod
Figure 4. Results of comparisons between minimally- and conventionally-shod individuals. (a) Static
measurements of arch height index (AHI) and arch stiness index (ASI). (b) Muscle cross-sectional area (CSA)
measurements. Note that values are scaled by (body mass[BM])2/3. (c) Dynamic measurements of maximum
arch deformation angle (θmax) and arch stiness (kmid). Note that kmid is scaled by (body mass)2/3. * denotes
statistically signicant dierence between groups.
Var i able Minimally Shod Conventionally Shod F df P-value
Static Nmean ± s.d. Nmean ± s.d.
AHI 75 0.35 ± 0.03 26 0.32 ± 0.03 28.6 97 <0.0001
ASI†75 2570 ± 1170 26 2020 ± 1070 7.1 97 0.009
BMI 75 25.7 ± 3.7 26 25.6 ± 3.3 0.2 97 0.64
Muscle CSA
AH*47 2.96 ± 0.51 25 2.77 ± 0.47 20 69 <0.0001
FDB*54 2.45 ± 0.38 24 2.66 ± 0.53 1.7 75 0.2
ADM*’† 25 1.24 ± 0.26 20 1.14 ± 0.38 12.7 42 0.0009
Dynamic
θmax 30 7.7 ± 2.1 25 9.8 ± 2.4 11.2 50 0.002
kmid*,†30 1774 ± 925 25 1304 ± 477 10.4 51 0.002
Table 1. Means and standard deviations ofstudy variables, and results of ANOVA tests for dierences between
minimally- and conventionally-shod participants. Tests are carried out on variance from general linear models.
*For general linear models, these variables were scaled by dividing by (body mass)0.67. †For general linear
models, these variables were logged.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

www.nature.com/scientificreports/
8
Scientific REpoRts | (2018) 8:3679 | DOI:10.1038/s41598-018-21916-7
conventional modern shoes. is hypothesis was supported for two of the three intrinsic foot muscles measured:
AH and ADM had signicantly larger cross-sectional areas in the minimally-shod than conventionally-shod
participants, but FDB did not dier signicantly between groups. Our prediction that minimally-shod individ-
uals would have stier LAs than conventionally-shod individuals was also supported. We measured LA stiness
both statically as the arch stiness index (ASI) as well as dynamically at midstance during walking (kmid), and for
both variables found signicantly higher values for the minimally-shod participants. Maximum LA angles (θmax)
were also signicantly higher in conventionally-shod participants, indicating that their arches deformed more
during walking than did those of the minimally-shod participants. We predicted that these static and dynamic
indicators of LA stiness would be correlated with intrinsic foot muscle cross-sectional area. While AH was sig-
nicantly associated with both dynamic LA measurements (θmax and kmid), it was not associated with ASI, and the
cross-sectional areas of the other two intrinsic foot muscles were not associated with any of the stiness variables
or AHI.
e positive correlation between AH cross-sectional area and foot stiness is likely related to the muscle’s
purported role in medial LA stabilization. In static loading experiments, Kelly et al.51 found that the AH raises the
medial LA when electrically stimulated, and several electromyography studies have also found that AH is active
when the LA is loaded during the stance phases of walking and running4,5,38. All else being equal, greater muscle
cross-sectional area should be directly related to greater force production, and therefore it follows that relatively
larger AH muscles should increase foot stiness more under loading. e fact that AH explained only a relatively
small percentage of the variance in θmax (24%) and kmid (10%) is not surprising given that numerous aspects of
Response Predictor Co ecient ± s.e. F P-value
AHI
Group −0.035 ± 0.007 28.6 <0.0001
Age −0.0001 ± 0.0003 0.2 0.63
BMI 0.002 ± 0.001 6.7 0.01
ASI†Group −0.11 ± 0.04 7.1 0.009
Age −0.002 ± 0.002 0.9 0.34
BMI Group −0.4 ± 0.85 0.2 0.64
Age −0.05 ± 0.04 1.7 0.2
AH*Group −0.03 ± 0.01 20 <0.0001
Age 0.0003 ± 0.0004 0.7 0.41
FDB*Group −0.008 ± 0.006 1.7 0.2
Age 0.0003 ± 0.0002 1.3 0.27
ADM*,†Group −0.12 ± 0.03 12.7 0.0009
Age 0.0001 ± 0.002 0.01 0.93
θmax
Group 2.09 ± 0.63 11.2 0.002
Age 0.005 ± 0.04 0.01 0.89
BMI 0.06 ± 0.08 0.6 0.44
Froude −7.91 ± 11.33 0.5 0.49
kmid*,†
Group −0.15 ± 0.05 10.4 0.002
Age 0.0005 ± 0.003 0.04 0.84
Froude −0.49 ± 0.87 0.32 0.58
Table 2. General linear model coecients, and results of ANOVA tests on model variance. Reference level for
‘Group’ is minimally-shod participants. *For general linear models, these variables were scaled by dividing by
(body mass)0.67. †For general linear models, these variables were logged.
Figure 5. Average foot kinematics during stance phase. (a) LA angle and (b) Δ LA height during in minimally-
shod (M-S; solid line) and conventionally-shod (C-S; dashed line) participants. Shaded regions represent ±one
standard deviation. 0° and 0 cm height are dened by marker positions at foot strike.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

www.nature.com/scientificreports/
9
Scientific REpoRts | (2018) 8:3679 | DOI:10.1038/s41598-018-21916-7
foot anatomy likely contribute to foot stiness, including bony geometry, ligamentous structures, and extrinsic
foot muscles (e.g. tibialis posterior). Although it is unclear why AH cross-sectional area explains a higher pro-
portion of variance in θmax than kmid, it is possible that AH activity is more important for stabilizing the LA when
it is maximally deformed, which generally occurs in the second half of stance following heel li (Fig.4). e
lack of association between AH and ASI may be related to the muscle’s near absence of activation during normal
standing4,52, suggesting that foot stiness during this behavior is mostly controlled by passive mechanisms such
as ligaments and bony geometry.
Unlike AH, neither FDB nor ADM was signicantly associated with dynamically stier feet during walking.
e result for FDB was surprising considering that Kelly et al.51 found that stimulating this muscle in statically
loaded individuals increased LA height, and other EMG studies have implicated the muscle in dynamically sti-
ening the foot during walking and running4,5,38. However, closer inspection of Kelly et al.’s51 results reveals that
back stimulation of FDB causes adduction and inversion between the calcaneus and metatarsals but not signif-
icant sagittal plane exion, suggesting that the muscle does not actively resist vertical compression of the LA.
Additionally, the muscle bers and tendons of FDB run longitudinally from the inferior calcaneus to the lateral
digits of the foot, making the muscle less favorably situated to stien the medial side of the LA than AH. We also
did not nd a signicant dierence in FDB size between minimally and conventionally-shod individuals, despite
nding signicant dierences between groups in static and dynamic measures of LA height and stiness. us,
the results of this study throw doubt on a potential role for FDB in stiening the medial LA.
e lack of association between ADM and medial LA stiness is not surprising given the muscle’s location on
the lateral side of the foot, which makes it poorly situated to serve as a stiening truss for the LA. However, the
larger size of ADM in minimally- versus conventionally-shod participants suggests an important, but currently
unknown role for the muscle in foot biomechanics. is muscle could participate in helping to stien the lateral
midfoot, which has been shown to be relatively compliant during walking in some individuals with low arches14.
Additionally, unlike most conventional shoes, the sandals worn by the Tarahumara lack restrictive toe boxes, ena-
bling the h digit greater mobility and likely necessitating more frequent use of ADM. However, the functional
importance of a more mobile h digit in locomotion remains to be investigated.
Our ndings for foot muscle cross-sectional area complement previous prospective studies on the eects of
minimal running shoes (lacking arch support, cushioning, and heel elevation) on intrinsic foot muscle size. Miller
et al.29 and Johnson et al.31 randomly assigned individuals to wear either minimal or conventional running shoes
for multi-week running regimes, and both found that those in the minimal shoe groups developed signicantly
larger AH muscles than those in the conventional shoe groups, but not larger FDB muscles. Miller et al. also
found signicantly larger ADM muscles and higher LAs in the minimal shoe group post-treatment, whereas
Johnson et al. did not measure either ADM size or LA height. Using a similar design to these studies, Chen
et al.30 found a signicant increase in overall foot muscle volume in their minimal shoe group but not in their
conventional shoe group, although they did not measure the sizes of individual muscles. Whereas these prospec-
tive studies suggest that adopting minimal footwear can help increase foot strength and LA height, the present
study indicates that habitual use of minimal footwear throughout life is related to bigger foot muscles and stier
LAs. Taken altogether, these studies suggest foot strength is related to the low incidence of at foot in habitually
barefoot and minimally-shod populations20–24,28,33, and also suggest the possibilitythat at foot can be treated by
switching to minimal footwear that does not restrict the natural motion of the foot.
An additional hypothesis this study tested was that static measurements of LA stiness would be correlated
with dynamic LA stiness during walking. Surprisingly, we did not nd a signicant association between our
static metric, ASI, and our dynamic metric, kmid, even though both were signicantly greater in minimally-shod
versus conventionally-shod participants. e likely explanation is that a single lower extremity experiences
roughly twice the ground reaction force during walking compared to normal standing, likely equating to a
two-fold increase in compressive forces experienced by the LA. To resist these higher forces, activity of the intrin-
sic and extrinsic foot muscles is elicited4,5,38,39, whereas these muscles are negligibly active during normal stand-
ing4,52. us, we argue that foot stiness is dependent on muscle activity during walking but not standing, and that
disparities in foot stiness between walking and standing will be related to the strength of muscles such as AH,
Var i able Minimally Shod Conventionally Shod W df P-value
Static Nmean ± s.d. Nmean ± s.d.
AHI 18 0.36 ± 0.03 10 0.31 ± 0.03 150 27 0.005
ASI 18 2470 ± 1097 10 2608 ± 1473 86 27 0.87
BMI 18 27.4 ± 4.3 10 24.5 ± 2.4 127 27 0.08
Muscle CSA
AH*12 2.91 ± 0.45 9 2.86 ± 0.50 78 21 0.10
FDB*15 2.40 ± 0.30 10 2.78 ± 0.46 70 24 0.81
ADM*4 1.08 ± 0.20 9 0.96 ± 0.16 33 12 0.02
Dynamic
θmax 7 8.2 ± 1.4 10 9.6 ± 2.5 25 16 0.36
kmid*7 1897 ± 823 10 1178 ± 364 78 16 0.007
Table 3. Means and standard deviations for study variables in physical activity-matched sub-samples, and
results of Wilcoxon rank sum tests between minimally and conventionally shod participants. *For statistical
tests, these variables were scaled by dividing by (body mass)0.67.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

www.nature.com/scientificreports/
10
Scientific REpoRts | (2018) 8:3679 | DOI:10.1038/s41598-018-21916-7
as well as other intrinsic and extrinsic foot muscles not investigated in this study (e.g. exor hallucis brevis and
longus, quadratus plantae, tibialis posterior, etc.). Because the forces experienced by the foot during walking are
greater than those experienced during standing, dynamic arch stiness is likely to have much greater implications
for the prevention of the common disorders associated with at foot, including plantar fasciitis, knee osteoarthri-
tis, and metatarsal stress fracture11,16–19. Dynamic assessments of arch stiness are thus probably more useful in
clinical examinations of lower extremity musculoskeletal disorders.
Following previous studies, we classied participants with AHI scores 1.5 standard deviations below the mean
from a large sample of adult males as having a ‘low arch’6,35,36, and found that 31% of our conventionally-shod
sample had low arches, whereas only one of the 75 minimally-shod participants that we measured had a low
arch. ere is not a single universally agreed upon denition of ‘at foot’12, and techniques for diagnosing the
condition are disputed15, and thus we do not assert that all low arched individuals in this study possessed at foot.
Nevertheless, the near absence of minimally-shod individuals with low arches and the relatively high incidence
of low arches in conventionally-shod individuals strongly supports previous studies that have reported low rates
of at foot in barefoot and minimally-shod populations20–24,28. Body size is unlikely to have been a factor in the
dierences found in the present study, as our minimally- and conventionally-shod samples had nearly identical
average BMIs with similar variances. Furthermore, the positive association between AH cross-sectional area
and AHI suggests that small foot muscles may be related to the presence of low arches, and thus by extension the
development of at foot.
is study has several limitations. First, we included only male participants, as we were unable to collect data
from enough Tarahumara females. We do not expect that inclusion of females would change the outcomes of our
minimally-shod versus conventionally-shod comparisons, but we do recognize a possible dierence in foot sti-
ness between males and females37. us far, there have been no investigations of sex dierences in intrinsic foot
muscle cross-sectional area and dynamic foot stiness, making this an area requiring further research. Another
study limitation is that the minimally-shod participants were subsistence farmers who reported considerably
more physical activity (PA)hours per day on average than the conventionally-shod participants, all of whom
worked in professions that require little to no physical activity. Despite PA being poorly matched between groups,
we assessed its possible eects on foot anatomy and mechanics by comparing the study variables between groups
in subsamples matched for PA. e results of these comparisons were similar to the full sample comparisons,
with minimally-shod participants having signicantly larger ADM muscles, higher AHI, and higher kmid values
than conventionally-shod participants. Dierences between groups for AH and θmax were not signicant, but
showed patterns that were consistent with the full sample analyses, with AH being higher and θmax being lower
in minimally-shod participants. ese results suggest that individuals with minimal shoes have stronger, stier
arches even aer accounting for dierences in physical activity, corresponding to the results of a previous inves-
tigation of minimally- and conventionally-shod Tarahumara33. at said, self-reported physical activity levels
are subject to error and do not reect lifetime dierences. Prospective studies are needed to better investigate the
eects of physical activity on the variables measured in this study.
One nal limitation is that we were only able to estimate dynamic LA stiness at one moment in a walk-
ing stride, mid-stance. is is because the pedography platform we used measures only vertical forces, and
mid-stance is when the non-vertical components of ground reaction force are closest to zero, as well as when the
linear dimension we used to measure LA displacement is roughly parallel to the ground reaction force. However,
as Fig.4 shows, LA displacement peaks in the second half of stance, following heel li, suggesting that mid-stance
may not be the most functionally relevant moment to calculate LA stiness. Inverse dynamics calculations of
intrinsic foot kinetics have revealed that midfoot moments also peak in the second half of stance2,13, at a similar
time to the maximum LA deformation angles (θmax) that we measured in this study. We measured signicantly
higher θmax in conventionally-shod versus minimally-shod participants, and found a signicant negative correla-
tion between this value and AH cross-sectional area. us, we expect that foot stiness at maximum arch defor-
mation will also be correlated with AH size, but testing this idea requires further investigation with equipment
that measures three-dimensional ground reaction forces.
Conclusions
e results of this study support the hypothesis that individuals who habitually wear minimal footwear have
LAs that are stier both statically and dynamically than those who habitually where conventional modern shoes.
Although prospective studies are necessary to conrm hypotheses of causality, these results also lend support
to the hypothesis that certain footwear features aect the cross-sectional areas of foot muscles such as AH, and
thus aect LA function. e relatively smaller AH and ADM muscles in the conventionally-shod individuals
measured here could be related to features in their shoes that immobilize and protect the foot, such as restrictive
toe boxes and raised arch supports. Humans have almost certainly been barefoot for most of the species’ exist-
ence, and although humans have been wearing minimal footwear for at least a few thousand years32, most of the
aforementioned features of modern shoes are extremely recent. us, given that the human foot evolved to func-
tion unshod and more recently in minimal footwear, its biomechanics may not be entirely adapted for modern
“conventional” shoes. We hypothesize that modern shoes reduce the role of the foot muscles in maintaining arch
stiness, thereby leading to less growth and possibly even atrophy through disuse. Subsequently, in some indi-
viduals when the arch is unsupported, the foot muscles are not strong enough to prevent arch collapse, resulting
in at foot. If correct, this hypothesis explains the relatively low rates of at foot among barefoot/minimally-shod
populations20–24,28.
Beyond the need for prospective studies, future research should target other muscles thought to be involved in
maintaining arch stiness (e.g. exor hallucis brevis, tibialis posterior). ere is also a need for data on individuals
from populations in which it is possible to control for potential confounding factors such as genetic ancestry and
physical activity levels. Additional research is also needed to determine the eects of specic features of modern
Content courtesy of Springer Nature, terms of use apply. Rights reserved

www.nature.com/scientificreports/
11
Scientific REpoRts | (2018) 8:3679 | DOI:10.1038/s41598-018-21916-7
shoes (e.g. arch supports) on foot biomechanics to isolate those that may inuence foot muscle activity. Finally,
further work is needed to determine how the use of modern shoes aects foot function at dierent stages of devel-
opment. Previous studies have indicated that shoe use early in life while the LA is still forming may be related
to a greater risk of developing at foot22,28. However, there is evidence to suggest that strengthening the intrinsic
foot muscles can also lead to a higher, stier arch in adults29–31,53. us, future studies should seek to determine if
early development of at foot is related to under-use of foot muscles in modern shoes, and if the use of minimal
footwear later in life can help treat the symptoms of at foot via strengthening of the foot muscles.
References
1. Lewis, O. Functional Morphology of the Evolving Hand and Foot (Oxford University Press, 1989).
2. Scott, S. H. & Winter, D. A. Biomechanical model of the human foot: inematics and inetics during the stance phase of waling. J.
Biomech. 26, 1091–1104 (1993).
3. Lapidus, P. W. inesiology and mechanical anatomy of the tarsal joints. Clin. Orthop. elat. es. 30, 20–36 (1963).
4. eeser, L. A., Susman, . L. & Stern, J. T. Electromyographic studies of the human foot: experimental approaches to hominid
evolution. Foot Anle 3, 391–407 (1983).
5. elly, L. A., Lichtwar, G., Cresswell, A. G. & Cresswell, A. G. Active regulation of longitudinal arch compression and recoil during
waling and running. J. . Soc. Interface 12, 20141076 (2015).
6. Butler, . J., Hillstrom, H., Song, J., ichards, C. J. & Davis, I. S. Arch height index measurement system: establishment of reliability
and normative values. J. Am. Podiatr. Med. Assoc. 98, 102–106 (2008).
7. Staheli, L. T., Chew, D. E. & Corbett, M. e longitudinal arch. J. Bone Joint Surg. Am. 69, 426–428 (1987).
8. Harris, . I. & Beath, T. Hypermobile at-foot with short tendo achillis. J. Bone Jt. Surg. Am. 30, 116–150 (1948).
9. Subotnic, S. I. e biomechanics of running: implications for the prevention of foot injuries. Sport. Med. 2, 144–153 (1985).
10. Dunn, J. E. et al. Prevalence of foot and anle conditions in a multiethnic community sample of older adults. 159, 491–498 (2004).
11. Gross, . D. et al. Association of at feet with nee pain and cartilage damage in older adults. Arthrit. Care es. 63, 937–944 (2011).
12. Mosca, V. S. Flexible atfoot in children and adolescents. J. Child. Orthop. 4, 107–121 (2010).
13. Saraswat, P., MacWilliams, B. A., Davis, . B. & D’Astous, J. L. inematics and inetics of normal and planovalgus feet during
waling. Gait Posture 39, 339–345 (2014).
14. DeSilva, J. M. & Gill, S. V. Brief communication: a midtarsal (midfoot) brea in the human foot. Am. J. Phys. Anthropol. 151,
495–499 (2013).
15. Benedetti, M. G. et al. Diagnosis of exible atfoot in children: a systematic clinical approach. Orthopedics 34, 94 (2011).
16. Simin, A., Leichter, I., Giladi, M., Stein, M. & Milgrom, C. Combined eect of foot arch structure and an orthotic device on stress
fractures. Foot Anle 10, 25–29 (1989).
17. Chandler, T. J. & ibler, W. B. A biomechanical approach to the prevention, treatment and rehabilitation of plantar fasciitis. Sport.
Med. 15, 344–352 (1993).
18. aufman, . ., Brodine, S. ., Shaer, . A., Johnson, C. W. & Cullison, T. . e eect of foot structure and range of motion on
musculoseletal overuse injuries. Am. J. Sports Med. 27, 585–593 (1999).
19. Shibuya, N., Jupiter, D. C., Ciliberti, L. J., Vanburen, V. & La Fontaine, J. Characteristics of adult atfoot in the United States. J. Foot
Anle Surg. 49, 363–368 (2010).
20. Engle, E. T. & Morton, D. J. Notes on foot disorders among natives of the Belgian Congo. J. Bone Joint Surg. Br. 13, 311–318 (1931).
21. Sim-Foo, L. & Hodgson, A. . A comparison of foot forms among the non-shoe and shoe-wearing Chinese population. J. Bone Joint
Surg. Am. 40, 1058–1062 (1958).
22. ao, U. B. & Joseph, B. e inuence of footwear on the prevalence of at foot: a survery of 2300 children. J. Bone Joint Surg. Br. 74,
525–527 (1992).
23. D’Août, . D., Patay, T. C., De Clercq, D. & Aerts, P. e eects of habitual footwear use: foot shape and function in native barefoot
walers. Footwear Sci. 1, 81–94 (2009).
24. Homann, P. e feet of barefooted and shoe-wearing peoples. J. Bone Joint Surg. Am. 3, 105–136 (1905).
25. Hollander, . et al. Growing-up (habitually) barefoot inuences the development of foot and arch morphology in children and
adolescents. Sci. ep. 8079 (2017).
26 . Echarri, J. J. & Forriol, F. e development in footprint morphology in 1851 Congolese children from urban and rural areas, and the
relationship between this and wearing shoes. J. Pediatr. Orthop. B 12, 141–146 (2003).
27. James, C. S. Footprints and feet of natives of the Solomon Islands. Lancet 234, 1390–1394 (1938).
28. Sachithanandam, V. & Joseph, B. e inuence of footwear on the prevalence of at foot. J. Bone Joint Surg. Br. 77–B, 254–257
(1995).
29. Miller, E. E., Whitcome, . ., Lieberman, D. E., Norton, H. L. & Dyer, . E. e eect of minimal shoes on arch structure and
intrinsic foot muscle strength. J. Sport Heal. Sci. 3, 74–85 (2014).
30. Chen, T. L., Sze, L. . Y., Davis, I. S. & Cheung, . T. H. Eects of training in minimalist shoes on the intrinsic and extrinsic foot
muscle volume. Clin. Biomech. 36, 8–13 (2015).
31. Johnson, A. W., Myrer, J. W., Mitchell, U. H., Hunter, I. & idge, S. T. e eects of a transition to minimalist shoe running on
intrinsic foot muscle size. Int. J. Sports Med. 37, 154–158 (2016).
32. Lieberman, D. E. e Story of the Human Body (Pantheon Boos, 2013).
33. Lieberman, D. E. Strie type variation among Tarahumara Indians in minimal sandals versus conventional running shoes. J. Sport
Health Sci. 3, 86–94 (2014).
34. McPoil, T. G. & Cornwall, M. W. Use of plantar contact area to predict medial longitudinal arch height during waling. J. Am.
Podiatr. Med. Assoc. 96, 490–494 (2006).
35. Butler, . J., Davis, I. S. & Hamill, J. Interaction of arch type and footwear on running mechanics. Am. J. Sports Med. 34, 1998–2005
(2006).
36. Butler, . J., Hamill, J. & Davis, I. Eect of footwear on high and low arched runners’ mechanics during a prolonged run. Gait Posture
26, 219–225 (2007).
37. Zifchoc, . A., Davis, I. S., Hillstrom, H., Song, J. e eect of gender, age, and lateral dominance on arch height and arch stiness.
Foot Anle Int. 367–372 (2006).
38. Mann, . A. & Inman, V. T. Phasic activity of the intrinsic muscles of the foot. J. Bone Joint. Surg. Am. 46, 469–481 (1964).
39. Gray, E. G. & Basmajian, J. V. Electromyography and cinematography of leg and foot (‘normal’ and at) during waling. Anat. ec.
161, 1–15 (1968).
40. Micle, . J., Nester, C. J., Cros, G. & Steele, J. . eliability of ultrasound to measure morphology of the toe exor muscles. J. Foot
Anle es. 6, 12 (2012).
41. Cros, G., Angin, S., Micle, . J., Hill, S. & Nester, C. J. eliability of ultrasound for measurement of selected foot structures. Gait
Posture 39, 35–39 (2014).
42. Schneider, C. A., asband, W. S. & Eliceiri, . W. NIH Image to ImageJ: 25 years of image analysis. Nature Methods 9, 671–675
(2012).
Content courtesy of Springer Nature, terms of use apply. Rights reserved

www.nature.com/scientificreports/
12
Scientific REpoRts | (2018) 8:3679 | DOI:10.1038/s41598-018-21916-7
43. Schmidt-Nielsen, . Scaling: Why is Body Size So Important? (Cambridge University Press, 1984).
44. Alexander, . M., Jayes, A. S., Maloiy, G. M. O. & Wathuta, E. M. Allometry of the leg muscles of mammals. J. Zool. 194, 539–552
(1981).
45. Myatt, J. P., Crompton, . H. & orpe, S. . S. Hindlimb muscle architecture in non-human great apes and a comparison of
methods for analysing inter-species variation. J. Ana t. 219, 150–166 (2011).
46. Le doux, W. ., Hirsch, B. E., Church, T. & Caunin, M. Pennation angles of the intrinsic muscles of the foot. J. Biomech. 34, 399–403
(2001).
47. Hedric, T. L. Soware techniques for two- and three-dimensional inematic measurements of biological and biomimetic systems.
Bioinspir. Biomim. 3, 34001 (2008).
48. Farley, C. T., Glasheen, J. & McMahon, T. A. unning springs: speed and animal size. J. Exp. Biol. 185, 71–86 (1993).
4 9. Alexander, . M. & Jayes, A. S. A dynamic similarity hypothesis for the gaits of quadrupedal mammals. J. Zool. 201, 135–152 (1983).
50.  Development Core Team. : A language and environment for statistical computing.  Foundation for Statistical Computing,
Vienna, Austria (2014).
51. elly, L., Cresswell, A. G., acinais, S., Whiteley, . & Lichtwar, G. Intrinsic foot muscles have the capacity to control deformation
of the longitudinal arch. J. . Soc. Interface 11, 20131188 (2014).
52. Basmajian, J. V. & Bentzon, J. W. An electromyographic study of certain muscles of the leg and foot in the standing position. Surg.
Gynecol. Obstet. 98, 662–666 (1954).
53. Mceon, P. O., Hertel, J., Bramble, D. & Davis, I. e foot core system: a new paradigm for understanding intrinsic foot muscle
function. Br. J. Sports Med. 49, 290 (2014).
Acknowledgements
We are especially grateful to Mickey Mahaey and Silvino Cubesare for their assistance with conducting research
in the Baranca de Sinforosa, as well as Christine Aguilar, Humberto Ramos Fernández, David Ramos, Gabriela
Yañez and the Yañez familyfor additional eld assistance. We thank Daniel Baird, Barbara Baird, David Holowka,
and Laurel Yohe for help with participant recruitment in the United States, Elizabeth Koch and Michael Ruiz
for assistance with data processing, and Steven Worthington for statistical advice. Funding for this project was
provided by the American School of Prehistoric Research (Harvard University) and the Hintze Family Charitable
Foundation.
Author Contributions
N.B.H., I.J.W. and D.E.L. designed the study and collected the data. N.B.H. and I.J.W. processed the data, N.B.H.
analyzed the data, and N.B.H., I.J.W. and D.E.L. wrote the manuscript.
Additional Information
Supplementary information accompanies this paper at https://doi.org/10.1038/s41598-018-21916-7.
Competing Interests: e authors declare no competing interests.
Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and
institutional aliations.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International
License, which permits use, sharing, adaptation, distribution and reproduction in any medium or
format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Cre-
ative Commons license, and indicate if changes were made. e images or other third party material in this
article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the
material. If material is not included in the article’s Creative Commons license and your intended use is not per-
mitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the
copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.
© e Author(s) 2018
Content courtesy of Springer Nature, terms of use apply. Rights reserved

1.
2.
3.
4.
5.
6.
Terms and Conditions
Springer Nature journal content, brought to you courtesy of Springer Nature Customer Service Center GmbH (“Springer Nature”).
Springer Nature supports a reasonable amount of sharing of research papers by authors, subscribers and authorised users (“Users”), for small-
scale personal, non-commercial use provided that all copyright, trade and service marks and other proprietary notices are maintained. By
accessing, sharing, receiving or otherwise using the Springer Nature journal content you agree to these terms of use (“Terms”). For these
purposes, Springer Nature considers academic use (by researchers and students) to be non-commercial.
These Terms are supplementary and will apply in addition to any applicable website terms and conditions, a relevant site licence or a personal
subscription. These Terms will prevail over any conflict or ambiguity with regards to the relevant terms, a site licence or a personal subscription
(to the extent of the conflict or ambiguity only). For Creative Commons-licensed articles, the terms of the Creative Commons license used will
apply.
We collect and use personal data to provide access to the Springer Nature journal content. We may also use these personal data internally within
ResearchGate and Springer Nature and as agreed share it, in an anonymised way, for purposes of tracking, analysis and reporting. We will not
otherwise disclose your personal data outside the ResearchGate or the Springer Nature group of companies unless we have your permission as
detailed in the Privacy Policy.
While Users may use the Springer Nature journal content for small scale, personal non-commercial use, it is important to note that Users may
not:
use such content for the purpose of providing other users with access on a regular or large scale basis or as a means to circumvent access
control;
use such content where to do so would be considered a criminal or statutory offence in any jurisdiction, or gives rise to civil liability, or is
otherwise unlawful;
falsely or misleadingly imply or suggest endorsement, approval , sponsorship, or association unless explicitly agreed to by Springer Nature in
writing;
use bots or other automated methods to access the content or redirect messages
override any security feature or exclusionary protocol; or
share the content in order to create substitute for Springer Nature products or services or a systematic database of Springer Nature journal
content.
In line with the restriction against commercial use, Springer Nature does not permit the creation of a product or service that creates revenue,
royalties, rent or income from our content or its inclusion as part of a paid for service or for other commercial gain. Springer Nature journal
content cannot be used for inter-library loans and librarians may not upload Springer Nature journal content on a large scale into their, or any
other, institutional repository.
These terms of use are reviewed regularly and may be amended at any time. Springer Nature is not obligated to publish any information or
content on this website and may remove it or features or functionality at our sole discretion, at any time with or without notice. Springer Nature
may revoke this licence to you at any time and remove access to any copies of the Springer Nature journal content which have been saved.
To the fullest extent permitted by law, Springer Nature makes no warranties, representations or guarantees to Users, either express or implied
with respect to the Springer nature journal content and all parties disclaim and waive any implied warranties or warranties imposed by law,
including merchantability or fitness for any particular purpose.
Please note that these rights do not automatically extend to content, data or other material published by Springer Nature that may be licensed
from third parties.
If you would like to use or distribute our Springer Nature journal content to a wider audience or on a regular basis or in any other manner not
expressly permitted by these Terms, please contact Springer Nature at
onlineservice@springernature.com