Foot strike patterns and hind limb joint angles during running in Hadza hunter-gatherers

Hrman Pontzr,Klly Sucman,Davi A.Raicln,Brian M.Woo, Auax Z.P.Mabulla,Frank W.Marlow

aDepartmentof Anthropology,Hunter College,New York,NY 10065,USA

bCity University of New York,New York,NY 10016,USA

cNew York Consortium for Evolutionary Primatology,USA

dEvolutionary Anthropology,Duke University,Durham,NC 27708,USA

eSchool of Anthropology,University of Arizona,Tucson,AZ 85721,USA

fDepartment of Anthropology,Yale University,New Haven,CT 06511,USA

gDepartmentof Archeology,University of Dar es Salaam,Dar es Salaam,Tanzania

hDepartment of Archaeology and Anthropology,University of Cambridge,Cambridge CB2 3DZ,UK

Foot strike patterns and hind limb joint angles during running in Hadza hunter-gatherers

Herman Pontzera,b,c,*,Kelly Suchmand,David A.Raichlene,Brian M.Woodf, Audax Z.P.Mabullag,Frank W.Marloweh

aDepartmentof Anthropology,Hunter College,New York,NY 10065,USA

bCity University of New York,New York,NY 10016,USA

cNew York Consortium for Evolutionary Primatology,USA

dEvolutionary Anthropology,Duke University,Durham,NC 27708,USA

eSchool of Anthropology,University of Arizona,Tucson,AZ 85721,USA

fDepartment of Anthropology,Yale University,New Haven,CT 06511,USA

gDepartmentof Archeology,University of Dar es Salaam,Dar es Salaam,Tanzania

hDepartment of Archaeology and Anthropology,University of Cambridge,Cambridge CB2 3DZ,UK

Background:Investigations of running gait among barefoot and populations have revealed a diversity of foot strike behaviors,w ith some preferentially employing a rearfootstrike(RFS)as the foot touches down while others employ a m idfootstrike(MFS)or forefootstrike(FFS). Here,we report foot strike behavior and joint angles among traditional Hadza hunter-gatherers living in Northern Tanzania.

Methods:Hadza adults(n=26)and juveniles(n=14)ran ata range of speeds(adults:mean 3.4±0.7 m/s,juveniles:mean 3.2±0.5 m/s)over an outdoor trackway while being recorded via high-speed digital video.Foot strike type(RFS,MFS,or FFS)and hind limb segment angles at foot strike were recorded.

Results:Hadza men preferentially employed MFS(86.7%of men),while Hadza women and juveniles preferentially employed RFS(90.9%and 85.7%of women and juveniles,respectively).No FFS was recorded.Speed,the presence of footwear(sandals vs.barefoot),and trial duration had no effect on foot strike type.

Conclusion:Unlike other habitually barefoot populations which prefer FFS while running,Hadza men preferred MFS,and Hadza women and juveniles preferred RFS.Sex and age differences in footstrike behavior among Hadza adults may reflectdifferences in running experience,w ith men learning to prefer MFS as they accumulate more running experience.

CopyrightⒸ2014,Shanghai University of Sport.Production and hosting by Elsevier B.V.All rights reserved.

Barefoot;Biomechanics;Foot strike;Foraging;Running

1.Introduction

Like many other animals,humans employ a bouncing, mass-spring gait when running,w ith the hind limb storing and releasing elastic strain energy each step.1,2This springlike behavior of the lower limb reduces the amount of muscle work required and improves running efficiency.1—3From a mechanical perspective,the two most important anatom ical springs in the human leg are the Achilles tendon and theplantar arch;together,these structures store and return roughly half of the potentialand kinetic energy losteach step during running.1These anatom ical springs are most effective when runners land on the m iddle or frontof the foot,allow ing the Achilles tendon and plantar arch to stretch as the foot is loaded during early stance phase.4Landing on the midfootor forefoot may also reduce the incidence of running-related injuries.5Nonetheless, many runners habitually heel strike,6—8landing on the rear portion of their foot,and the effects of training,footwear,and speed on footstrike patterns remain unclear.Here,we exam ine running mechanics among Hadza hunter-gatherers to assess foot strike patterns in an untrained,physically active,traditional population w ith m inimal footwear.

Foot strike patterns have recently emerged in debates over the role of endurance running in human evolution.Endurance running has been cited by several researchers as a critical adaptation in the hom inin lineage,marking a departure away from an ape-like,plant-based foraging ecology and toward a more active,omnivorous ecological strategy that included scavenging and hunting.9—11Bramble and Lieberman11noted thatmany of the anatom ical features associated w ith effective endurance running in modern humans fi rst appear inHomo erectusand proposed that key evolutionary changes seen in our genus followed the evolution of endurance running.Selection for endurance may have even played a critical role in the evolution of increased brain size.12

Subsequent work by Lieberman and colleagues6has suggested that the anatom ical adaptations in the human foot are particularly advantageous during unshod running w ith a forefootormidfootstrike(FFS,MFS).In a study of habitually barefoot Kenyan runners from the Kalenjin population,Lieberman and colleagues6noted that these renowned endurance runners tend to land on the front or m iddle of their foot while running.In contrast,habitually shod American runners tend to rearfoot strike(RFS).Lieberman and colleagues6hypothesized that the population difference in footstrike behaviorwas influenced by differences in footwear:barefoot running, common among Kalenjin individuals,allows runners to experience the high impact forces imparted by RFS and leads to the adoption of MFS or FFS.In contrast,conventional running shoes absorb much of the impactassociated w ith RFS, and their elevated heel increases the likelihood and incidence of RFS.This hypothesis suggests that RFS has become more common w ith the development and popularity of modern athletic footwear,and thatRFS should be rare orabsentamong unshod or m inimally shod populations.

More recently,Hatala and colleagues8studied foot strike and impact forces at different running speeds in 38 habitually unshod adults from the Daasanach population of Northern Kenya.The Daasanach are traditional pastoralists;they typically walk long distances to tend herds,gather water,and in other daily tasks,but run much less than the Kalenjin.In contrast to the Kalenjin,Hatala and colleagues8found that the Daasanach often RFS,and that running speed affects foot strike behavior.At speeds less than 5.01 m/s,the Daasanach used RFS at a higher frequency than MFS or FFS.Between 5.01 and 6 m/s,frequencies of MFS and FFS were sim ilar, while MFS was the predom inant pattern at speeds greater than 6.01 m/s.These results indicate that not all unshod populations prefer to MFS or FFS while running,and that training,experience,and speed may affect foot strike patterns.

Yet another pattern of foot strike use is reported for the Tarahumara,a m inimally shod population of traditional farmers living in the Sierra Madre Occidentalof Northwestern Mexico.13The Tarahumara are renowned endurance runners, running 75 km ormore in traditionalballgames and,in recent years,competing in ultramarathons.14Tarahumara traditionally wore simple rawhide sandals(huaraches),and many continue to do so today,although some have adopted conventional running shoes.13Lieberman13reported that 89%of Tarahumara who wear conventional shoes habitually RFS, while Tarahumara who wear traditionalhuarachestend to MFS or FFS.

In this study we exam ined foot strike patterns and running kinematics among traditional Hadza hunter-gatherers in Northern Tanzania.As described in detail elsewhere,15traditional Hadza subsist almost entirely on w ild foods,hunting and gathering each day on foot and w ith simple hand tools. Traditional Hadza have no agriculture,livestock,or machinery.Women typically walk 6 km each day gathering w ild berries,tubers,and other plant foods,while men walk an average of 11 km per day,hunting small and large game w ith bow and arrow,and gathering honey.15,16The landscape they inhabit is sem iarid savannah w ith a patchy m ix of forest and grassland cover;the ground is often rocky,and low craggy hills are common.While the Hadza are highly active,they rarely run.15

Musiba and colleagues17conducted a study of walking gait and footdimensions among traditional Hadza.As discussed in that report,Hadza adults typically wear simple sandals made from repurposed tire rubber,common throughout East Africa. These sandals have relatively thin(～1 cm)soles that offer protection from sharp rocks and thorny plants but do not provide any cushioning or elevate the heel.Traditional Hadza can therefore be categorized accurately as“m inimally shod”, and their feetdisplay many of the same features(e.g.,splayed toes,greater foot w idth)evident in habitually unshod populations.17,18While Musiba and colleagues17did notexamine running,self-selected speeds during walking trials reported for Hadza adults(1.15 m/s)were relatively fastcompared to other traditional,unshod populations,and the Hadza also used greater stride frequencies and stride lengths.

We used high-speed digital video to analyze foot strike patterns and limb-segment angles of Hadza adults and juveniles running at a range of speeds.Our objectives were to determ ine the frequency of RFS,MFS,and FFS among the Hadza,to investigate the effects of speed and age on foot strike patterns,and to compare these data to published values for the Kalenjin and Daasanach.We predicted that the Hadza, who lack the training and experience in endurance running common among the Kalenjin,would exhibit foot strike patterns more sim ilar to the Daasanach.

2.M aterials and m ethods

2.1.Subjects

In May and June 2010,we recruited 26 Hadza adults (n= 15 males,body mass:50.4 ± 4.2 kg,height: 158.0±6.4 cm,hip height:84.8±4.0 cm;n=11 females, body mass:45.9±6.0 kg,height:150.0±7.0 cm,hip height: 83.5±4.0 cm)and 14 Hadza juveniles(9 males and 5 females,mean age:8.6 years,range 5—14,body mass: 20.3 ± 5.9 kg,height:111.1 ± 21.4 cm,hip height: 58.2±9.0 cm)in two camps(Setako and Sengeli)to participate in walking and running trials,as partof a larger study on Hadza energy expenditure.16Body mass and height were measured using a digital scale and stadiometer,respectively. Hip height was measured as the distance from the greater trochantor to the ground while standing unshod.Prior to the study,human research perm issions were obtained from all legally cognizant institutional and governmental agencies, including the Tanzanian Council for Science and Technology and National Medical Research Institute.Verbal informed consent and,for juveniles,verbal parental consent,was obtained prior to participation.Communication was conducted in Swahili,in which the Hadza are generally fluent.

2.2.Experimental setup

In each Hadza camp we established a clear,level pathway for walking and running trials.Other than clearing small shrubs and loose rocks,no alteration was made to the trackway;its surface generally consisted of hard and dry soil of mixed sand and silt common to that region.A high-speed digital camera(Exilim F1;300 fps;Casio America,Dover, NJ,USA)was placed 7.5 m from the track and oriented perpendicular to the direction of travel in order to capture kinematics in the sagittal plane.For 11 adultsubjects,running trials were recorded during respirometry trials designed to measure energy expenditure(oxygen consumption and carbon dioxide production;see Pontzer et al.16which reports energetics data for walking trials).Subjects in these trials wore a Cosmed k4b2(COSMED USA Inc.,Chicago,IL,USA) respirometry unit in a light chest harness,as well as a light plastic mask,to collectand measure expired air.During these respirometry trials,subjects were asked to run at slow(“pole pole”in Swahili),medium(“kati kati”),and fast(“haraka”) speeds for 2—3 m in each,completing 2—3 laps of the 200-m trackway at each speed.Speeds were calculated by tim ing these laps w ith a stopwatch,and a researcher(HP)paced each subject to maintain a constant speed.Most subjects(9/11) chose to wear their sandals during respirometry trials;the other two ran barefoot.Respirometry results were reported previously.16

Allother running trials were recorded during short～7—10-m bouts along a portion of the trackway,w ithout respirometry equipment.No direction was given regarding running speed; subjects chose theirown speed.These“shortbouts”were begun several meters out of frame so that the subject was at a steady speed during video capture.Subjectswere barefootduring these short-bout trials.Four adults completed both respirometry and short-bout trials.Two adults(men)performed additional shortbout trials shod.Combining respirometry and short-bout trials, we collected a total of 66 foot strike recordings.

All video analysis was performed using Kinovea software version 0.8.15(http://www.kinovea.org/).Running speeds were calculated using the autotrack feature,calibrated using a 1-or 2-m scale bar placed along the trackway for all bouts. Because the scale bar was placed along the side of the trackway farthest from the camera,this method overestimates true running speed:the subject,running in front of the scale bar(i.e.,between the scale barand camera,Fig.1),willappear to run faster than she is actually traveling.To account for this difference,we compared speeds calculated from video to those calculated using a stopwatch for a setof 13 respirometry trials. As expected,we found that video-based estimates of running speed were 13.4%±8.1%faster than speeds calculated using a stopwatch.This comports w ith the camera’s angle of view(～26°)and distance from the trackway:a subject running 100 cm in front of the scale bar should appear to be moving～13%faster than she actually was.Therefore,speeds calculated from digital video were decreased by 13.4%for subsequent analysis and comparisons with other studies.

Fig.1.(A)Schematic aerialview of the camera and trackway setup.Given the distance between the camera and the scale bar(7.5 m),the camera’s angle of view (26°),and the positioning of the subjectbetween the scale bar and camera,the sighted distancedis approximately 13%shorter than the scale bar distance of 2 m. As a result,speeds calculated using the scale bar were～13%faster than the actual speed measured using a stopwatch.Speeds were corrected for subsequent analyses.(B)A Hadza woman using a rearfootstrike.Footangle convention isshown;dorsiflexion relative to the ground plane is negative.(C)A Hadza man using a m idfoot strike.Knee angle and ankle angle conventions are shown.

2.3.Ankle,knee,and foot angles

Ankle,knee,and plantar foot angles at foot strike were calculated using Kinovea,follow ing angle conventions used by Lieberman and colleagues6(Fig.1).Footstrike was defined as the fi rstvideo frame in which the foot is in contactw ith the ground.The locations of anatom ical landmarks were estimated;markers were not placed on the foot or leg.Ankle angle was defined as the angle connecting the head of fi fth metatarsal,the lateral malleolus,and the knee.A negative ankle angle corresponds to dorsiflexion,while a positive angle indicates plantarflexion.Knee angle was defined as the angle connecting the lateral malleolus,the center of the knee,and major axis of thigh.The plantar foot angle was measured as the angle between the ground plane and the line connecting the posterior calcaneal tuber and distal fi fth metatarsal.The lack of anatom ical markers limited the resolution w ith which angles could be determ ined.Additionally,for plantar angles±1°at footstrike,the angle between the plantar surface and ground plane was somewhatobscured by the shadow of the footon the ground.As a result,formany MFS,where plantar angles were ±1°,plantar angles were recorded as 0°as itwas notpossible to reliably distinguish the angle between the plantar surface and the ground plane w ith greater precision.

2.4.Foot strike categorization and statistical analyses

All foot strikes were classified as RFS,MFS,or FFS follow ing criteria reported by A ltman and Davis.19Strike type was defined by the plantar angle and by the portion of the foot contacting the ground at foot strike.Strikes w ith a negative plantar angle less than-5°,in which the heel contacted the ground fi rst,were classified as RFS.Strikes w ith a plantar angle between-5°and+1°,in which the m iddle portion of the foot contacted the ground fi rst,were classified as MFS.We did not record any FFS,defined as a positive plantar angle greater than 1°and the frontportion(i.e.,the distal portion of the metatarsals)striking the ground fi rst.To assess reliability of footstrike determination,two authors(HP and KS)assessed strike type for all trials independently.Their categorization agreed in all but one trial(65/66 trials,or 98.5%agreement).

Foot strike behavior(RFS,MFS,or FFS)was exam ined in relation to age class(adultor juvenile),sex,footwear(barefoot or shod),and trial type(respirometryvs.short-bout).Because subjects varied in the number of trials collected,foot strike was compared among individuals rather than among trials.For comparisons among age-class and sex,subjects were counted only once in each comparison(e.g.,each adult male was counted once in the comparison of adultmen and women).For comparisons across footwear and trial type,subjects that completed both conditions were counted once in each condition(e.g.,a subjectwho completed 2 respirometry trials and 2 short-bout trials would be counted once in each condition).To account for the multiple comparisons among adults (sex,footwear,and trial type)and the inclusion of some subjects in both conditions,we used Bonferroni correction to adjustour significance criterion fromp=0.05 top=0.01 for analyses of adults.Comparisons of foot strike usage for each condition were done using chi-squared tests in ExcelⓇ(M icrosoft,Redmond,WA,USA).Mulitvariate comparisons were performed in JMPⓇ10.0.0(SAS,New York,NY,USA) using nominal logistic regression.

3.Results

3.1.Foot strike

A totalof 66 running trials were recorded.Across all trials, 30(45.4%)were RFS and 36(54.6%)were MFS;no FFS was recorded.When data from adults and juveniles were combined,60%(24/40)of subjects used RFS and 40%(16/40) used MFS.A substantialdifference in footstrike behavior was evident across age-classes.Adults used MFS more often (53.8%,14/26 subjects)than did juveniles(14.3%,2/14),p=0.015.Due to this difference adults were analyzed separately for subsequent analyses.

Among adults,more men used MFS(86.7%,13/15)than women(9.1%,1/11),p＜0.001.In contrast,there was no significantdifference between adults in respirometry trials(54.5% MFS,6/11)versus short-bout trials(61.9%MFS,13/21),p=0.469,norbetween adultswearing sandals(66.7%MFS,6/9) versus running barefoot(52.4%MFS,11/21),p=0.687.Four adults(3 males,1 female)completed trials in four conditions (barefootand shod;respirometry and short-bout);none of these fourchanged their footstrike behavioracross conditions.

Median speed for all adult trials was 3.4 m/s.Below this speed more adults used RFS(57.9%RFS,11/19),while above the median speed more subjects used MFS(71.4%MFS,10/ 14),but this trend did not achieve significance(p=0.095). Further exam ination suggests that this trend derives from differences among men and women in their chosen running speed rather than an effect of speedper se.Running speeds (mean±SD)for women and men were 2.98±0.44 and 3.74±0.59 m/s,respectively,and the difference was significant(p=0.001,ttest).As noted above,all but one woman used RFS while allbut two men used MFS.Further,of the six adults with trials atboth slow(＜3.4 m/s)and fast(＞3.4 m/s) speeds,none changed their foot strike usage at faster speeds. In fact,in all subjects w ith multiple recorded trials,none changed foot strike usage between trials.Thus,women were more likely to use RFS and to use a slower running speed than men.There is no evidence thatsubjects changed from RFS to MFS as speed increased.

Results from bivariate comparisons were consistent w ith those of a multivariate nom inal logistic regression.When speed,sex,and footwear(shod,barefoot)were used as independent variables predicting foot strike,only sex was a significant factor(p=0.001).When adult and juvenile trialswere pooled,both sex(p=0.001)and age-class(p＜0.001) were significant predictors of foot strike usage,while speed (p=0.157)and footwear(p=0.101)were not.

Fig.2.Foot,ankle,and knee angles at footstrike for Hadza adults.Each point is one trial;some subjects are represented in multiple trials.Gray triangles:rearfoot strike trials;black circles:midfoot strike trials.See Fig.1 for angle conventions.

3.2.Joint angles

Foot,ankle,and knee anglesatfootstrike forHadza adultsare plotted againstspeed in Fig.2.The effectsof footwear,speed,and footstrike usage were entered into a multivariate nom inal logistic regression to exam ine their effect on these angles.Not surprisingly,footstrike usage(RFSvs.MFS)wasa significantpredictor of footangle at impact(p＜0.001),butspeed(p=0.54)and footwear(shodvs.unshod,p=0.37)had no effect.Similarly,foot strike usage significantly predicted ankle angle at foot strike (p＜ 0.001),while neither speed(p=0.21)nor footwear (p=0.74)were significant factors.For knee angle,both foot strike(p=0.006)and speed(p=0.011)weresignificantfactors, w ith moreacuteknee flexion atfasterspeeds,butfootwearhad no effect(p=0.54).When juvenile trials are added to these comparisons,age-class does not significantly affect foot,ankle,or knee angles(p＞0.05 all comparisons).

4.Discussion

4.1.Comparisons with other studies

Foot strike usage among Hadza adults was intermediate between that reported among the Kalenjin and Daasanach populations(Table 1),and sim ilar in some ways to the pattern reported for Tarahumara adults.When Hadza juveniles,adult men,and adult women are exam ined separately,some sim ilarities w ith other populations emerge.Hadza men rarely use RFS(13.3%of subjects),similar to foot strike patterns of barefoot Kalenjin adolescents and Kalenjin adults who grew up barefoot,and to m inimally-shod Tarahumara.6,8,13In contrast,Hadza women and juveniles often used RFS(90.9% and 85.7%of subjects,respectively),similar to Daasanach adults,habitually shod Kalenjin adolescents,and Tarahumara wearing conventional running shoes.The high rate of RFS among Hadza women and juveniles was also sim ilar to that reported for habitually shod adults.7Unlike Kalenjin adults that grew up barefoot,habitually barefoot Kalenjin adolescents,and habitually barefoot U.S.adults,Hadza runners never used FFS in trials recorded for this study.

Due to the m ix of MFS and RFS among the Hadza,mean plantar foot strike angle among adults was intermediate between habitually shod U.S.adults and Kalenjins.U.S.adults ran w ith a high frequency of RFS,thus leading to the large dorsiflexion upon plantar foot strike,causing smaller(negative)footangles.Kalenjins had a high frequency of FFS,thus show ing the large plantarflexion upon foot strike and larger angles.Ankle angles among Hadza adults were sim ilar to those of habitually barefoot U.S.adults,barefoot Kalenjin adolescents,and Kalenjin adults who grew up barefoot.Knee angles at foot strike were consistently greater(i.e.,more flexed)among Hadza adults than for Kalenjin or U.S.groups.The difference in knee angle is more substantial when differences in running speed are considered;Hadza speeds were lower,on average,than those of the Kalenjin or U.S.groups reported by Lieberman and colleagues,6yet knee flexion generally increased w ith speed in our sample.

Table 1 Foot strike,limb angles at foot strike,and speed for Hadza,Kalenjin,Daasanach,and U.S.populations.

Direct comparison of joint angles among studies is hampered by the different methods used to measure them. Unlike Lieberman and colleagues,6we did not place visual markers on anatom ical landmarks.Instead,the knee angle in our study was calculated using the major axes of the thigh and shank,which may have resulted in systematic differences in knee angle calculation relative to the analysis of Lieberman and colleagues.The image resolution and lack of visual markers probably also decreased the precision of our angle measurements,an effect that was most evident in our inability to distinguish plantarangles±1°for MFS trials(Fig.2).Thus, while we took care to calculate angles in a manner thatwould maxim ize comparability to other studies(Fig.1),it is possible that some differences between studies arise from methodological differences.

4.2.Factors influencing foot strike type

Foot strike behavior among traditional Hadza huntergatherers was m ixed,w ith consistent differences between men and women and between juveniles and adults.Women and juveniles used RFS more often than MFS,while men used MFS almost exclusively.There was no difference between shod versus barefoot conditions,noramong respirometry trials (which lasted for severalm inutes)and short-bout trials(which lasted a few seconds).The lack of difference between shortbout and respirometry trials lends confidence that the duration of the trial did notaffect footstrike choice.Further,there is no evidence thatHadza adults sw itched from RFS to MFS as speed increased.

While the Hadza used MFS ratherthan FFS,comparisonsw ith other populations suggest that Hadza men are sim ilar to experienced barefoot runners such as the Kalenjin in avoiding RFS.In addition,Hadza men achieve MFS patterns using generally comparable jointkinematicsto othergroupsthathabitually MFS. Hadza women and juveniles are sim ilar to shod U.S.runners and inexperienced runners such as the Daasanach in preferring RFS, and use comparable jointkinematicsto achieve these footstrikes. Thispattern of footstrikeusagesuggestsrunning experiencemay be important in developing footstrike preferences.As children learn to walk and theirgaitmatures,RFS develops as a normal partof the walking gaitcycle;20thus RFS is the behavior learned fi rst.As the musculoskeletal system and motor control develop further during adolescence,experience running barefoot or m inimally shod may lead to a preference forMFS or FFS during running,perhaps in response to the high impact forces21experienced when running w ith RFS.Individuals who rarely run m ightnothave the same accumulated experience ofhigh impact forces due to RFS,and thus never sw itch from RFS to MFS or FFS for running.

Our data are cross-sectional and do not provide the ontogenetic data or other measures of personal history and experience that longitudinal studies m ightafford.Nonetheless, the pattern of foot strike use among the Hadza are consistent w ith the hypothesis that running experience and skill play a role in shaping foot strike behavior.Hadza adolescents used RFS almostexclusively.Indeed,the only two adolescents that used MFS were also the oldest(13-and 14-year-old boys). Hadza women apparently maintain this preference for RFS into adulthood,while Hadza men come to prefer MFS.We suggest that the change in foot strike behavior by Hadza men may develop as they learn to huntand track w ild game.While Hadza men do not typically engage in endurance running,it is likely that they run more often as they learn to hunt than their female counterparts do in learning to gather plant foods. Indeed,our measurements of travel speeds used while out of camp on forays,taken using wearable GPS devices,16indicate that men use running speeds approximately tw ice as often as women(Fig.3).Perhaps men’s running experience,and the greater impact force experienced during RFS,lead Hadza men to prefer running w ith MFS as their foraging efforts and experience grow.

Fig.3.Percentage of travelduring forays(i.e.,outside of theircamps)in which travel speed exceeded froude=0.5 for 13 adult Hadza women and 10 adult Hadza men.Travel was defined as a GPS epoch in which speed exceeded 0.5 m/s.Speed was calculated from GPS units worn during daylight hours. Froude=0.5 was calculated from hip height for each subject using the formula:Froude=speed2/(hip height×g).Epoch duration was variable,w ith a median of 10 s.See Pontzer and colleagues16for GPS methods and protocol.

An alternative explanation for the observed differences in foot strike usage between Hadza men and women,and between Hadza children and adultmen,is that adult men experience larger ground reaction forces due to their greater body mass and running speed,leading to proprioceptive responses in foot strike preference.Hadza men in this sample were 10.0%heavier than women(p=0.04,ttest)and 5.4%taller (p=0.01,ttest)and,as noted above,used faster running speeds than women.While we did not measure ground forcesin this study,the difference in mass and speed suggests men would have experienced correspondingly larger ground forces. Nonetheless,the observed variation in foot strike preference indicates that this mechanical hypothesis does not capture all of the inter-individualvariance in behavior.The two men who employed RFS were among the largest(54.6 and 58.2 kg),and their mean running speed(3.55 m/s)was near the mean of the men’s sample.Notably,there was no difference in hip height between men and women in the Hadza sample(p=0.44,ttest)indicating that sex differences in footstrike usage were nota resultof differences in hind limb length.

Whatever the reason for their foot strike preference,it is notable that MFS is common among Hadza men even though they rarely run.This finding holds implications for the evolution of human running gait.In populations w ith even m inimal experience running,we can expect thatmany individuals would prefer MFS(or perhaps FFS)rather than RFS on occasions when they do run.Some threshold levelof exposure to running may be necessary to promote MFS or FFS,but extensive running experience is not needed.Thus MFS(and perhaps FFS)may have been common among hunter-gatherer groups in the past,even those thatdid notengage in endurance running or employ exhaustion hunting techniques regularly.

Including our data from this study,foot strike behavior during running has been described for only three habitually barefootor m inimally shod populations.The variability in foot strike preference both w ithin and between these groups is notable,and suggests caution is warranted when draw ing conclusions about“average”or“typical”gait in unshod populations.For example,it is possible that groups such as the Daasanach run w ith RFS due simply to a lack of endurance running experience.Documenting foot strike behavior and otheraspects ofwalking and running gaitin otherbarefootand m inimally shod populations w ill improve ourunderstanding of culturaland ecological factors influencing locomotor behavior and anatomy in humans.

Acknow ledgments

We thank Fides Kirei for assistance w ith data collection, and Lauren Christopher,Annie Qiu,and Khalifa Stafford for assistance w ith video analysis.Daniel Lieberman and two anonymous reviewers provided comments that improved this manuscript.Funding was provided by the National Science Foundation(BCS-0850815)and Hunter College.

1.A lexander RM.Energy saving mechanisms in walking and running.J Exp Biol1991;160:55—69.

2.Ker RF,Bennett MB,Bibby SR,Kester RC,A lexander RM.The spring in the arch of the human foot.Nature1987;325:147—9.

3.Roberts TJ,Marsh RL,Weyand PG,Taylor CR.Muscular force in running turkeys:the economy of minimizing work.Science1997;275:1113—5.

4.Perl DP,Daoud AI,Lieberman DE.Effects of footwear and strike type on running economy.Med Sci Sports Exerc2012;44:1335—43.

5.Daoud AI,Geissler GJ,Wang F,Saretsky J,Daoud YA,Lieberman DE. Foot strike and injury rates in endurance runners:a retrospective study.Med Sci Sports Exerc2012;44:1325—34.

6.Lieberman DE,Venkadesan M,Werbel WA,Daoud AI,D’Andrea S, Davis IS,et al.Foot strike patterns and collision forces in habitually barefoot versus shod runners.Nature2010;463:531—5.

7.Larson P,Higgins E,Kam inski J,Decker T,Preble J,Lyons D,etal.Foot strike patterns of recreationaland sub-elite runners in a long-distance road race.J Sports Sci2011;29:1665—73.

8.Hatala KG,DingwallHL,Wunderlich RE,Richmond BG.Variation in foot strike patternsduring running among habitually barefootpopulations.PLoS One2013;8:e52548.http://dx.doi.org/10.1371/journal.pone.0052548.

9.Carrier DR.The energetic paradox of human running and hominid evolution.Curr Anth1984;25:483—95.

10.Shipman P,Walker AC.The costs of becom ing a predator.J Hum Evol1989;18:373—92.

11.Bramble DM,Lieberman DE.Endurance running and the evolution of Homo.Nature2004;432:345—52.

12.Raichlen DA,Polk JD.Linking brains and brawn:exercise and the evolution of human neurobiology.Proc Biol Sci2013;280:20122250.http:// dx.doi.org/10.1098/rspb.2012.2250.

13.Lieberman DE.Strike type variation among Tarahumara Indians in m inimal sandals versus conventional running shoes.J Sport Health Sci2014;3:86—94.

14.McDougall C.Born to run.New York:Knopf;2009.

15.Marlowe FW.The Hadza:hunter-gatherers of Tanzania.Berkeley:University of California Press;2010.p.336.

16.Pontzer H,Raichlen DA,Wood BM,Mabulla AZ,Racette SB, Marlowe FW.Hunter-gatherer energetics and human obesity.PLoS One2012;7:e40503.http://dx.doi.org/10.1371/journal.pone.0040503.

17.Musiba CM,Tuttle RH,Hallgrimsson B,Webb DM.Sw iftand sure-footed on the savanna:a study of Hadzabe gaits and feet in Northern Tanzania.Am J Hum Biol1997;9:303—21.

18.D’Aouˆt K,Pataky TC,De Clercq D,Aerts P.The effects of habitual footwear use:foot shape and function in native barefoot walkers.Footwear Sci2009;1:81—94.

19.Altman AR,Davis IS.A kinematic method for footstrike pattern detection in barefoot and shod runners.Gait Posture2012;35:298—300.

20.Saunders JB,Dec M,Inman UT,Eberhart HD.The major determ inants in normal and pathological gait.J Bone Jt Surg Am1953;35-A:543—58.

21.Lieberman DE.What can we learn about running from barefoot running? An evolutionary medical perspective.ExercSportSciRev2012;40:63—72.

Received 10 September 2013;revised 3 February 2014;accepted 10 March 2014

*Corresponding author.Department of Anthropology,Hunter College,New York,NY 10065,USA.

E-mail address:herman.pontzer@hunter.cuny.edu(H.Pontzer)

Peer review under responsibility of Shanghai University of Sport

2095-2546/$-see front matter CopyrightⒸ2014,Shanghai University of Sport.Production and hosting by Elsevier B.V.A ll rights reserved. http://dx.doi.org/10.1016/j.jshs.2014.03.010