the anomalous archaichomo femur from berg aukas, namibia: a biomechanical assessment

The Anomalous Archaic Homo Femur From Berg Aukas,Namibia: A Biomechanical Assessment

ERIK TRINKAUS,1,2* CHRISTOPHER B. RUFF,3AND GLENN C. CONROY1,4

1Department of Anthropology, Washington University, St. Louis,Missouri 631302U.M.R. 5809 du C.N.R.S., Laboratoire d’Anthropologie,Universite de Bordeaux I, 33405 Talence, France3Department of Cell Biology and Anatomy, Johns Hopkins UniversitySchool of Medicine, Baltimore, Maryland 212054Department of Anatomy and Neurobiology, Washington University Schoolof Medicine, St. Louis, Missouri 63110

KEY WORDS hominid; lower limb; biomechanics; Pleistocene

ABSTRACT The probably Middle Pleistocene human femur from BergAukas, Namibia, when oriented anatomically and analyzed biomechanically,presents an unusual combination of morphological features compared to otherPleistocene Homo femora. Its midshaft diaphyseal shape is similar to mostother archaic Homo, but its subtrochanteric shape aligns it most closely withearlier equatorial Homo femora. It has an unusually low neck shaft angle. Itsrelative femoral head size is matched only by Neandertals with stockyhyperarctic body proportions. Its diaphyseal robusticity is modest for aNeandertal, but reasonable compared to equatorial archaic Homo femora. Itsgluteal tuberosity is relatively small. Given its derivation from a warmclimatic region, it is best interpreted as having had relatively linear bodyproportions (affecting proximal diaphyseal proportions, shaft robusticity, andgluteal tuberosity size) combined with an elevated level of lower limb loadingduring development (affecting femoral head size and neck shaft angle). Am JPhys Anthropol 110:379–391, 1999. r 1999 Wiley-Liss, Inc.

In 1995, Grine, Jungers, Tobias and Pear-son described the proximal half of a heavilymineralized human femur, which had beenrecovered 30 years earlier during vanadiummining operations at the Berg Aukas minein northern Namibia (Conroy et al., 1993;Grine et al., 1995). The specimen is undatedgiven its lack of stratigraphic associationwith any chronological indicators (biologicalor geological), but an assessment of its mor-phology relative to that of other hominidfemora led Grine and colleagues to concludethat it is best attributed to Middle or LatePleistocene archaic Homo, a conclusion withwhich we concur. In particular, the morphol-ogy of the proximal epiphyseal region distin-guishes it from Australopithecus and some

early Homo femora and aligns it with mem-bers of H. erectus and later PleistoceneHomo. The absence of a pilaster distin-guishes it from early and recent modernhumans, whereas the strong development ofa medial buttress places it closest to Middleand Late Pleistocene archaic Homo femora(Fig. 1).

In their assessment of the femur, Grine etal. (1995) emphasized the large size of thefemoral head, its diaphyseal cross-sectional

Grant sponsors: National Science Foundation; C.N.R.S.; theWenner-Gren Foundation; the L.S.B. Leakey Foundation; theNational Geographic Foundation.

*Correspondence to: Erik Trinkaus, Department of Anthropol-ogy, Campus Box 1114, Washington University, St. Louis, MO63130, U.S.A. E-mail: [email protected]

Received 25 November 1998; accepted 23 June 1999.

AMERICAN JOURNAL OF PHYSICAL ANTHROPOLOGY 110:379–391 (1999)

r 1999 WILEY-LISS, INC.

properties, its low neck shaft angle, andseveral details of the diaphyseal shape. How-ever, their assessment did not include abiomechanical evaluation of the specimen.On-going assessments of Pleistocene andrecent Homo femoral and tibial cross-sec-tional geometry (e.g., Ruff et al., 1993; Ruff,1995; Trinkaus, 1997; Trinkaus et al., 1999)and proximal femoral proportions (Ruff etal., 1991; Trinkaus, 1993) have developed aframework within which we can now assessthe biomechanical implications of the BergAukas femur. Given the dearth of MiddlePleistocene human postcrania, any insightwhich this, albeit undated, specimen mightprovide would add to our knowledge of ge-nus Homo locomotor evolution.

MATERIALS AND METHODSComparative samples

The biomechanically relevant parametersof the Berg Aukas femur are compared pri-marily to those of three chronological (andby default partially geographical) samples ofarchaic Homo (see Tables 2 and 3 for speci-mens).

The first sample, designated ‘‘Early Ar-chaic Homo,’’ consists of mature femorawhich are distinct from those of Australopi-thecus, assigned to early Homo and H. erec-tus, and date from the terminal Pliocene (ca.2.0 ma B.P.) to the early Middle Pleistocene(ca. 600 ka B.P.). Additional data are in-cluded for the Early Pleistocene early adoles-cent KNM-WT 15000 specimen. This sampleand period represent a long stage of stasis inrelative endocranial capacity and pelvic pro-portions (Ruff, 1995; Ruff et al., 1997); it alsoincludes only specimens from warm climatezones. Since proximal femoral proportionsare influenced by pelvic configurations,which in turn are affected by trends inencephalization and ecogeographical pat-terns in body shape (Ruff, 1995), this sampleshould encompass specimens whose femoralvariation reflects primarily locomotor param-eters and not separate effects from encepha-lization, pelvic aperture shape, and relativebody breadth.

The second sample, the one to which theBerg Aukas femur may have close affinities,is designated ‘‘Middle Archaic Homo.’’ Itconsists of specimens from the middle por-tion of the Middle Pleistocene, ca. 500 kaB.P. to ca. 200 ka B.P. Femoral variability inthis group reflects both locomotor variationand the marked encephalization that oc-curred during the Middle Pleistocene (Ruffet al., 1997). It also includes specimens fromacross the Old World (Zambia to Germany toChina), and some of its variation may reflecthuman geographical and ecogeographicvariation.

The third sample, ‘‘Late Archaic Homo,’’includes the Neandertals sensu lato fromthe terminal Middle Pleistocene to themiddle of oxygen isotope stage 3 (ca. 150 kaB.P. to ca. 35 ka B.P.). This sample, the onlyone of late archaic humans which preserves

Fig. 1. Anterior view of the Berg Aukas hominidfemur, taken relative to the anteversion plane of theproximal femur. Scale in centimeters.

380 E. TRINKAUS ET AL.

femora, is notable for both its encephaliza-tion approaching that of recent humans(Ruff et al., 1997) and its hyperarctic bodyproportions (Holliday, 1997b).

In addition, a Eurasian sample of earlymodern humans ($ 18 ka B.P.) is used toprovide a robust modern human baseline forthe assessments of archaic Homo femora.These specimens derive from the sites ofArene Candide, Barma Grande, Caviglione,Cro-Magnon, Dolnı Vestonice I and II,Grotte-des-Enfants, Minatogawa, Mladec,Nahal Ein Gev, Ohalo II, Paglicci, Parabita,Paviland, Pavlov I, Predmostı, Qafzeh, LaRochette, Skhul, and Willendorf (N 5 16 to34 for various comparisons).

Methods

Six biomechanical parameters of interestare preserved on the Berg Aukas femur:head size reflecting hip joint reaction forcelevels and hence body mass (Jungers, 1988;McHenry, 1988; Ruff et al., 1991), femoralneck length influencing gluteal abductormuscle moment arms and load distributionsthrough the proximal femur (Lovejoy et al.,1973; Ruff, 1995), gluteal tuberosity sizerelated to hip muscle hypertrophy (Trinkaus,1976), diaphyseal strength reflecting overalllocomotor levels (Ruff et al., 1993), subtro-chanteric diaphyseal shape reflecting hipregion proportions and the resultant rela-tive anteroposterior versus mediolateral loadlevels (Ruff, 1995), and midshaft diaphysealshape reflecting in part hip proportions butmainly locomotor levels (Ruff, 1987, 1995,1999; Trinkaus et al., 1998). The first fourrequire the scaling of the relevant femoralmeasure to body mass or body mass timesbeam length. The second two involve com-parisons of relevant perpendicular secondmoments of area.

The scaling of head size, neck length,gluteal tuberosity size, and midshaft diaphy-seal rigidity require the estimation of femo-ral length, since it approximates the beamlength for the femur and is the best estima-tor of stature (Feldesman et al., 1990). Bodymass is dependent upon stature and relativebody breadth (Ruff, 1994; Ruff et al., 1997),and both recent and Pleistocene Homo areknown to have varied ecogeographically with

respect to their body proportions (Ruff, 1991,1994; Holliday, 1997a,b). It is therefore nec-essary to know the body breadth of theindividuals in question or to estimate itbased on ecogeographical patterns, or totake known variation in relative bodybreadth into consideration in assessments ofproximal and diaphyseal dimensions to femo-ral length. Given the incomplete state of theBerg Aukas femur, the latter approach istaken here.

The assessment of subtrochanteric (80%)diaphyseal shape was done using maximumto minimum second moments of area (Imaxand Imin). Given variation in the orientationof the major axis at midshaft, from mediolat-eral in Early Archaic Homo to anteroposte-rior in early modern humans, the anatomi-cally oriented Ix and Iy values are employedat that level. For Ix and Iy, the sagittal planeof the femur is defined as passing throughthe linea aspera and the mediolateral mid-point of the diaphysis near midshaft.

The cross-sectional geometric parameters(see Table 1 for a list) for the comparativesamples were computed using versions ofSLICE (Nagurka and Hayes, 1980). Most ofthe cross sections were obtained noninva-sively by molding the subperiosteal contour,determining cortical thicknesses using multi-or biplanar radiography, and interpolatingthe endosteal contour within the boundariesset by the cortical thicknesses. A few werefrom scaled photographs of appropriate fos-silization breaks, and several were takenfrom published cross-sections (e.g., Weiden-reich, 1941; Mallegni et al., 1983).

Anteroposterior head diameter was mea-sured directly or, for OH-28, Arago 44 andAmud 1, was estimated from acetabularheight using a regression based on recenthumans (r2 5 0.896). Neck length was quan-tified using the biomechanical neck length ofLovejoy et al. (1973), the distance perpen-dicular to the diaphyseal axis from the mostlateral point of the greater trochanter to itstangent to the proximal femoral head takenin the coronal plane of the femoral head andneck. Gluteal tuberosity size was quantifiedas the maximum breadth of the rugose areaof the tuberosity proper (Trinkaus, 1976).

381BERG AUKAS HOMO FEMUR

Assessments of the biomechanical affini-ties of the Berg Aukas femur relative tothose of other archaic Homo specimens weredone graphically for all but the neck lengthto bone length comparison (Figs. 3–6). Sinceall of these variables are, or may well be,related to each other in nonlinear ways (Ruffet al., 1993), the data were logged prior tographing and analysis. In addition, a z-scorefrom the early modern human reduced ma-jor axis regression line is provided for eacharchaic Homo femur.

Berg Aukas femur

This analysis is based on data derived forthe Berg Aukas femur, combining the datapresented by Grine et al. (1995) (head diam-eter, neck length, and neck shaft angle),data derived from computed tomographic(CT) scans of the specimen by GCC (takenusing a Siemens Somatom DR3 scanner at125 kV with a window width of 2050 andslice thickness of 1mm) (for cross-sections)and a high quality resin cast of the specimen(for 50% cross-sections and gluteal tuberos-ity breadth).

Grine et al. (1995, pp.159–160) estimatedlength using a multiple regression based ona recent human (African American) sampleand six measurements of the head and neck.This provided an ‘‘interarticular’’ length of518 6 16 mm. To avoid any circularity inassessing proximal epiphyseal proportionsin the present analysis, the midshaft of theBerg Aukas femur was determined by therelative position of the medial buttress, aswelling of bone along the medial proximaldiaphysis which begins at the level of mid-

gluteal tuberosity on the medial diaphysisand gradually rotates posteriorly as it goesdistally (Trinkaus, 1976, 1984). On femorawhere it is well developed, the medial but-tress reaches the posteromedial aspect ofthe diaphysis near midshaft. On the BergAukas femur, this buttress is directly pos-teromedial at the distal break of the shaft. Itwas therefore assumed that the distal breakis close to midshaft (see Grine et al., 1995).

The distance parallel to the diaphysealaxis, from the proximal neck to the distalbreak is ca. 235 mm, providing a biomechan-ical length (Ruff and Hayes, 1983) of ca. 470mm. Given that the medial buttress hadalready reached the posteromedial aspect ofthe shaft 10 to 15 mm proximal of the distalbreak, this estimate is more likely to overes-timate than underestimate the bone’s origi-nal length. The addition of the 10 mm (thedistance from the proximal neck to the proxi-mal head measured parallel to the diaphy-seal axis) provides a maximum length ca.480mm, 38 mm below the previous mean predic-tion of 518 mm.

To obtain Ix and Iy as well as the othercross-sectional parameters, the bone wasoriented using the linea aspera, the distalbreak of the cast was photographed (Fig. 2),the image was projected onto a Summa-graphics digitizing tablet enlarged linearly5.95 times, and the subperiosteal and endos-teal contours were digitized (not includingthe trabeculae in the posterior medullarycavity) (Fig. 2). The same procedure wasfollowed for the hardcopy print-out of our CTimage taken just proximal of the distalbreak, linearly enlarged 3.84 times. Cross-

TABLE 1. Cross-sectional geometric properties of the midshaft (50% level)and subtrochanteric level (80%) of the Berg Aukas femur1

50%Grine et al. (1995)

50%CT scan

50% cast(adjusted)

80% Grine et al.(1995)

80%adjusted

TA 853 800 805 898 842CA 750 708 697 794 744Ix — 51,599 50,365 — —Iy — 51,050 52,720 — —Imax 67,149 61,019 59,889 84,677 75,604Imin 49,112 41,631 43,917 47,362 42,288J 116,261 102,649 103,085 132,038 117,892Theta — 134° 139° — —1 Areas in mm2 and second moments of area in mm4. TA, total area; CA, cortical area; Ix, anteroposterior second moment of area; Iy,mediolateral second moment of area; Imax, maximum second moment of area; Imin, second moment area perpendicular to Imax; J, polarmoment of area; theta: orientation of Imax.


sectional parameters were computed using aPC-DOS version (Eschman, 1992) of SLICE(Nagurka and Hayes, 1980). Each imagewas digitized twice and the results aver-aged. The resultant values were then ad-justed for linear shrinkage (ca.1.6%) of thecast (Table. 1). The final values are less thanthose calculated by Grine et al. (1995), butthey remain consistent with the externaldiameters of the diaphysis.1

For comparable 80% measures, the valuesprovided by Grine et al. (1995) were ad-justed by the percentage differences foundfor the midshaft cross sections (6.7% and12.0%) (Table 1). Since these are linearcorrections and the only subtrochanteric val-ues of interest here are Imax and Imin, this hasa trivial effect on the comparisons.

RESULTSOverall size

The Berg Aukas femur is very large. Itsfemoral head diameter of 57.6mm is thelargest known, although femoral head diam-eters between 50 and 55 mm are known forseveral Middle and Late Pleistocene hu-mans, including Arago 44, Broken Hill E689and E907, Krapina 213, La Chapelle-aux-Saints 1, La Ferrassie 1, Neandertal 1, andSpy 2 (only the largest male early modernhumans have femoral head diameters ex-ceeding 50 mm). Its midshaft cortical area of697 mm2 is the largest known for an archaicHomo femur, even though it is approachedby those of KNM-ER 736 (659 mm2),Ehringsdorf 5 (666 mm2), and Amud 1 (659mm2). Its midshaft polar moment of area of103,085 mm4 is similarly among the largestknown, being surpassed only by KNM-ER736 (116,628 mm4) and matched by those ofEhringsdorf 5 (102,449 mm4), and Qafzeh 8(102,428 mm4).

These large femoral dimensions and thebody mass estimate based on femoral headdiameter of ca. 93 kg (Grine et al., 1995) areimpressive. They also emphasize the need toscale these parameters to appropriate mea-sures of body size.

Hip proportions

There are few archaic Homo specimensthat provide biomechanical neck length andfemoral length (or reasonable estimates ofthem), and it is inappropriate to scale femo-ral neck length against other proximal femo-ral dimensions since they are frequentlyhighly correlated (Wolpoff, 1978; Ruff, 1995).All of the other archaic Homo femora fallabove the regression line for the early mod-ern human sample (Table 2), with KNM-ER1481a and Spy 2 having relatively highvalues. In addition, the early H. erectusKNM-WT 15000 early adolescent provides az-score of 2.93, in agreement with previousassessment of its femoral neck length (Ruff,1995). Archaic Homo tend to have relativelylong femoral necks (see also Wolpoff, 1978).

The Berg Aukas femur is in the middle ofthe early modern human sample and has therelatively shortest neck of the archaic Homo

1To assess which values are more reasonable, total subperios-teal areas were regressed on the product of the midshaft diam-eters (‘‘area’’) for archaic Homo femora (50% TA 5 0.705 3‘‘area’’ 1 8.33, r2 5 0.953, N 5 31; Berg Aukas AP: 34.0 mm, ML:33.5 mm), and z-scores for the different Berg Aukas total arearesiduals were computed. The Berg Aukas z-scores are 1.34 forthe total area from Grine et al. (1995), 20.45 for the CT scandigitized here, and 20.27 for the shrinkage-adjusted cast values.

Fig. 2. Distal view of a cast of the Berg Aukas righthominid femur, showing the cross-sectional distributionof bone revealed by the transverse break of the diaphy-sis. Anterior is above and medial is to the right. Thepencil mark indicates the position of the linea aspera.Scale in millimeters.


femora. As with the comparison of anatomi-cal neck length to its vertical diameter (Grineet al., 1995), the Berg Aukas femoral necklength is unexceptional from a modern hu-man perspective, although relatively shortfor an archaic Homo.

This moderate neck length (for an archaicHomo) is associated with the lowest known(presumably) nonpathological human femo-ral neck shaft angle. Its value of 106° (Grineet al., 1995, p.163) is below all other knownvalues for archaic Homo (being approachedby KNM-WT 15000 at 110° and Amud 1 at113°) (Walker and Leakey, 1993; Trinkaus,1993) and 3.4 and 4.1 standard deviationsbelow the means of two Holocene humangroups with the lowest mean neck shaftangles [Khoisan males (Grine et al., 1995)and Japanese Jomon foragers (Ishisawa,1931), respectively] and below a minimumvalue of 110° for a sample of 1376 recenthumans from 15 populations (Anderson andTrinkaus, 1998).

Femoral diaphyseal shape

The distribution of 80% Iman versus Imin(Fig. 3) reflects the pattern previously noted(Ruff, 1995; Trinkaus et al., 1998; Trinkausand Ruff, 1999). In this, Early Archaic Homo

have platymeric proximal femoral diaphy-ses, reflecting the biomechanical loads im-posed on their femora from long femoralnecks and platypelloid pelves. The EurasianLate Archaic Homo sample has relativelyround proximal femoral diaphyses, with thetwo exceptions being the early (terminalMiddle Pleistocene) Krapina specimens.Most of the early modern humans exhibitplatymeric femora, with the exception of thevery linear Qafzeh-Skhul sample. This isreflected in their z-scores, which are positiveto minimally negative for the early sampleand negative (except for the Krapina femora)for the late archaic sample (Table 3).

The Middle Pleistocene remains exhibitconsiderable variability. Two Broken Hillfemora are quite round, one is relativelyplatymeric, and the Tabun E1 femur is closeto the early modern human line. The BergAukas femur is in the middle of the earlymodern human and overall distributions(Table 3). It is therefore relatively platy-meric, more similar to the Early ArchaicHomo sample than it is to the Late ArchaicHomo group, falling very close to specimenssuch as KNM-ER 736 and 803a and GesherBenot Ya’aqov 1. However, it fits well within

TABLE 2. z-Scores for comparisons of femoral neck length, head diameter, and gluteal tuberosity breadth

Neck length/length

Head Diameter/length

Gluteal tuberosity/length

Gluteal tuberosity/head

Berg Aukas 0.06 2.48 21.02 24.88Early Archaic Homo

KNM-ER 737 — — 2.51 —KNM-ER 1472 1.27 20.36 — —KNM-ER 1481a 2.00 1.65 4.44 2.17Gesher Benot Ya’aqov 1 — — 4.27 —OH 28 — 0.09 1.63 1.21

Middle Archaic HomoBroken Hill E689 — — — 21.01Broken Hill E690 — — 0.82 —Broken Hill E907 — — — 21.85

Late Archaic HomoAmud 1 — 20.26 0.79 0.89La Chapelle 1 0.60 3.09 2.30 22.02La Ferrassie 1 1.09 2.34 2.79 20.07La Ferrassie 2 0.40 1.70 4.92 2.86Krapina 213 — — — 0.30Krapina 214 — — — 4.01Neandertal 1 0.95 2.65 2.90 20.60St.-Cesaire 1 — — 3.02 —Shanidar 1 — — 1.62 —Shanidar 4 — 2.11 4.09 1.46Shanidar 5 — 0.52 — —Shanidar 6 — — 5.48 —Spy 2 2.58 3.59 3.09 21.68Tabun 1 0.40 0.92 3.91 2.67


the variable distribution of the Middle Pleis-tocene femora.

In the midshaft, Berg Aukas falls with themajority of the archaic Homo specimens. Itlacks any evidence of a pilaster, since it onlyhas a slight projection of the linea asperadorsally (Fig. 2). There is a marked medialbuttress, one of the largest known for anarchaic Homo femur, being approached bythose of Amud 1, La Ferrassie 1 and Saint-Cesaire 1 (Trinkaus et al., 1991, 1998;Trinkaus and Ruff, 1999). Therefore, eventhough the major axis is oriented anterolat-

eral to posteromedial, as is reflected in itstheta value of 139°, the anteroposterior andmediolateral second moments of area (Ixversus Iy) are very similar (Table 1).

With the exceptions of the Broken HillE690 and Castel di Guido 1 femora, all of thearchaic Homo femora fall on the roundedside of the early modern human distribution(Fig. 3). The mean z-scores for the Early,Middle and Late Archaic Homo samples are23.31, 22.44 (22.96 without Broken HillE690 and Castel di Guido 1), and 22.85,respectively. The Berg Aukas femur with az-score of 23.51 falls well within the rangesof archaic Homo femora and outside of therange of the largely pilastric early modernhuman femora.

Femoral diaphyseal robusticity

Even though the Berg Aukas femur islarge, its level of robusticity (strength scaledto appropriate measures of body size andlimb length) is less apparent relative toother Pleistocene Homo femora. This ambi-guity derives from the combined contribu-tions of body proportions, body mass andactivity level to the hypertrophy of theweight-bearing locomotor skeleton (Ruff etal., 1993).

In comparisons of both midshaft corticalareas and polar moments of area to femorallengths, there is a consistent pattern. LateArchaic Homo specimens are generally onthe high side of the early modern humandistribution, with mean z-scores of 1.49 and1.53 (Fig. 4 and Table 3). The Early ArchaicHomo femora plus the Middle PleistoceneBroken Hill E690 femur, all of which derivefrom Africa or the neighboring Near East,have more modest levels of relative corticalarea and polar moments of area, beingslightly higher on average than the earlymodern human sample (mean z-scores of0.52 and 0.36, respectively).

Berg Aukas has a z-score of 1.39 for rela-tive cortical area, well within the Late Ar-chaic Homo plus Zhoukoudian range (mean:1.57) but overlapping those of Aın Maarouf 1and KNM-ER 803a. It has a z-score of 1.00for the relative polar moment of area, be-tween the mean values for the Early (plusBroken Hill; 0.36) and Late (plus Zhoukou-dian; 1.59) Archaic Homo samples. Is Berg

Fig. 3. Bivariate distributions of perpendicular sec-ond moments of area for the subtrochanteric (80%) level(above) and the midshaft (50%) level (below). The subtro-chanteric distribution employs maximum and perpen-dicular to maximum second moments of area (Imaxversus Imin), whereas the midshaft one plots anteroposte-rior versus mediolateral ones (Ix versus Iy). Symbols:solid hexagon, Berg Aukas femur; gray diamonds, earlyarchaic Homo; dark gray pentagons, middle archaicHomo; open squares, late archaic Homo; open triangles,early modern humans. The reduced major axis line forthe early modern human sample is provided.


Aukas one of the more robust earlier Pleis-tocene archaic Homo femora, similar to amodestly built Late Pleistocene archaicHomo, or can its relative diaphyseal hyper-trophy be otherwise explained?

Relative femoral head size

To assess the relative size of the BergAukas femoral head, it needs to be scaled toa measure of body mass. However, the onlypossible indications of body mass are eitherthe femoral head itself or calculations basedon estimated stature (from its reconstructedlength) and body breadth (from ecogeo-graphical patterning given its inferred paleo-climatological context) (Ruff et al., 1997).Alternatively, it can be compared to femorallength, bearing in mind that such a compari-son assumes similar body proportions across

the comparative samples, an assumptionknown to be false.

Such a comparison for Pleistocene Homoplaces Berg Aukas at the top of the distribu-tion, exceeded in relative femoral head sizeonly by three European Neandertals (Fig. 5and Table 2). Its z-score of 2.48 is close to themean (2.67) of the European Late ArchaicHomo sample and well above the more mod-est values for most of the late Near Easternspecimens and the three EarlyArchaic Homospecimens. The Berg Aukas femur has alarge femoral head, but not one which isexceptional relative to at least one Pleis-tocene hominid sample.

Relative gluteal tuberosity size

The gluteal tuberosities of many Pleis-tocene archaic and early modern humans

TABLE 3. z-Scores for comparisons of femoral cross-sectional properties

80% Imax/Imin 50% Ix/Iy 50% CA/Len 50% J/Len

Berg Aukas 20.10 23.51 1.39 1.00Early Archaic Homo

Aın Maarouf 1 — 21.75 1.73 1.42KNM-ER 736 20.26 23.81 (20.78) (21.11)KNM-ER 737 1.06 24.54 0.34 0.57KNM-ER 803a 20.02 23.71 1.60 1.72KNM-ER 1472 0.96 21.91 1.06 0.51KNM-ER 1481a 0.42 22.74 0.19 20.44KNM-ER 1808mn 2.17 23.46 (0.53) (1.44)Gesher Benot Ya’aqov 1 20.03 22.87 0.14 20.47OH 28 1.30 25.02 20.39 20.01

Middle Archaic HomoBroken Hill E689 22.04 — — —Broken Hill E690 1.23 20.23 (0.76) (20.08)Broken Hill E793 — 21.81 — —Broken Hill E907 21.73 — — —Castel di Guido 1 — 20.53 — —Ehringsdorf 5 — 22.91 — —Tabun E1 20.26 23.02 — —Zhoukoudian F1 — 23.57 (2.50) (2.25)Zhoukoudian F2 — 23.35 — —Zhoukoudian F4 — 23.39 — —Zhoukoudian F5 — 23.41 — —Zhoukoudian F6 — 22.20 — —

Late Archaic HomoAmud 1 20.55 23.29 1.40 0.93La Chapelle 1 22.57 23.45 2.20 2.08La Ferrassie 1 24.11 23.89 0.60 0.56La Ferrassie 2 21.41 23.35 1.35 1.80Fond-de-Foret 1 — 22.91 1.01 1.13Krapina 213 0.31 — — —Krapina 214 0.48 — — —Neandertal 1 22.04 21.99 0.95 1.08St.-Cesaire 1 21.74 21.67 2.33 2.29Shanidar 4 — 22.87 2.51 2.87Shanidar 5 — 23.10 1.89 2.32Shanidar 6 — 22.39 1.61 1.95Spy 2 21.95 22.89 1.09 1.01Tabun 1 21.00 23.58 0.93 0.39Tabun 3 — 21.62 — —


are large and rugose (Trinkaus, 1976). TheBerg Aukas tuberosity is similarly marked,descending from a third trochanter just be-low the level of the lesser trochanter andbecoming increasingly rugose and concaveas it descends distally onto the dorsal proxi-mal diaphysis. Its maximum breadth of 9.0mm is located proximally, at the distal end ofthe third trochanter, and the tuberosity nar-rows to ca. 7.5 mm distally.

Even though it is difficult to assess actualmuscle size from osteological attachmentdimensions and only a portion of M. gluteusmaximus inserts into the gluteal tuberosity(the remainder blending into the iliotibialtract), the size of the gluteal tuberosityappears to provide a reasonable indication

of M. gluteus maximus hypertrophy (and byextension that of the hip musculature gener-ally, given the need for balance betweensynergists and antagonists).

Gluteal tuberosity breadth relative tofemoral length largely separates archaicHomo and early modern human femora; theonly overlap involves the modest tuberositydimensions for Broken Hill E690 (possiblyunderestimated given surface abrasion tothe specimen) and La Chapelle-aux-Saints 1and the pronounced one for DolnıVestonice13 (Fig. 6). It places Berg Aukas among themore gracile of the specimens, with a z-scoreof 20.02. An alternative approach is to com-pare gluteal tuberosity breadth to femoralhead diameter, assuming that the latter isbetter reflecting body mass than is femorallength across these ecogeographically vari-able samples (Fig. 6). The result is a generaluniformity of early modern human and ar-chaic Homo femora, with only three smallLate Archaic females (La Ferrassie 2,Krapina 214 and Tabun 1) plus the earlyKNM-ER 1481a being robust outliers (Fig. 6and Table 2). Berg Aukas, with its modestgluteal tuberosity breadth and large femoralhead, is an unusually low outlier.

DISCUSSIONPaleoenvironmental context

The Pleistocene paleoenvironments ofsouthwestern Africa are poorly known, butit appears from geological evidence (Ward et

Fig. 4. Bivariate distributions of midshaft (50%)cortical area (above) and polar moment of area (below)versus femoral biomechanical length. Symbols as in Fig.3. The reduced major axis line for the early modernhuman sample is provided.

Fig. 5. Bivariate distribution of femoral head antero-posterior diameter versus femoral length. Symbols as inFig. 3. The reduced major axis line for the early modernhuman sample is provided.


al., 1983) that it remained relatively warmin the Pleistocene and that Namibia wassemi-arid to arid throughout this period. Aserious limitation to Pleistocene paleocli-matic reconstructions of Namibia is the pau-city of time successive mid-to-late tertiaryvertebrate fossil localities (Shackley, 1980;Senut et al., 1992; Pickford et al., 1994).This problem has been alleviated partiallyby the discovery of a dozen fossiliferouslocalities in northern Namibia, of which therichest, both in terms of taxonomic diversityand specimen abundance, is the Berg Aukasmine (Senut et al., 1992; Conroy, 1996). Ingeneral, the biostratigraphic evidence atBerg Aukas largely supports other evidence(Siesser, 1980) suggesting that Namibia’s

current arid to extreme-arid environmentprogressively developed in conjunction withthe late Miocene origin of the Benguelacurrent.

The scattered Middle Pleistocene archeo-logical evidence and geological evidence forat least ephemeral lakes (e.g., Shackley,1980; Sandelowsky, 1983), as well as thepresence of the Berg Aukas hominid femur,indicate that there were periods duringwhich rainfall was sufficient to support asavanna grassland and associated fauna.

We can only assume that the Berg Aukashominid was associated with the post-Miocene (probably Pleistocene) micro-mam-malian fauna represented in one of theclusters of breccia blocks from the BergAukas mine dump (Conroy, 1996). Ten ex-tant mammalian genera are exclusive tothis cluster, the majority of which are associ-ated with semi-arid to desert conditions(Skinner and Smithers, 1990).

It is therefore likely that the paleoenviron-mental context of the Berg Aukas hominidwas similar to that of modern Namibia,warm and semi-arid to arid. A linear humanbody form, such as is associated with suchenvironments (Ruff, 1994), appears mostlikely for this individual based on its paleo-environmental context.

Diaphyseal shape

The subtrochanteric and midshaft diaphy-seal cross sections of the Berg Aukas femurfit best within the variable Middle ArchaicHomo sample. The specimen is fully non-pilastric, and its midshaft structural distri-bution of bone places it close to most of theother archaic Homo femora (all except twounusual Middle Pleistocene specimens).Proximally it is less rounded than the Nean-dertals and closer to the earlier (and lowerlatitude) specimens. However, it is unclearwhether the trend toward rounder proximalfemoral diaphyses through later archaicHomo was the result of changing load pat-terns in this region from trends in pelvicaperture shape (Ruff, 1995) or was a combi-nation of such aperture changes combinedwith ecogeographically influenced pattern-ing in pelvic breadth (Ruff, 1991).

Fig. 6. Bivariate distributions of gluteal tuberositybreadth versus femoral biomechanical length (above)and femoral head diameter (below). Symbols as in Fig. 3.The reduced major axis line for the early modern humansample is provided.


Femoral robusticity and body proportions

Analyses of Pleistocene Homo femoral andtibial diaphyseal robusticity, which take intoaccount body proportions, estimated bodymass, and inferred activity levels (e.g., Ruffet al., 1993; Trinkaus, 1997; Trinkaus et al.,1998, 1999, Trinkaus and Ruff, 1999), haveshown that there is a trend (with individualvariation) through Pleistocene Homo inwhich average femoral and tibial robusticitydeclines very slowly through time acrossthese groups. The apparent gracility of manyof the Early Archaic Homo and warm cli-mate MiddleArchaic Homo femora and tibiaeis the product of their ecogeographicallyinfluenced linear body proportions. The ap-parent hyper-robusticity of the Late Pleis-tocene Neandertals disappears once theirhyperarctic body proportions are taken intoaccount, and the moderately to stronglylinear body proportions of northwestern OldWorld early modern humans (Holliday,1997a) account for most of their apparentgracility relative to at least that of theNeandertals.

Assessment of the robusticity of the BergAukas femur relative to these archaic Homofemora must therefore take into accountinferred ecogeographically determined bodyproportions. Given a warm paleoclimate forPleistocene Namibia, the Berg Aukas indi-vidual would be expected to have had linearbody proportions, although not necessarilyas linear as those seen among Pleistoceneand recent equatorial Africans (Ruff, 1994).Assuming this, its level of femoral corticalbone hypertrophy would indicate a level ofrobusticity similar to those of African andNear Eastern Early/Middle Archaic Homo.However, if one infers less linear body propor-tions for this individual, the indicated levelof robusticity would decrease accordingly.The modest size of the Berg Aukas glutealtuberosity relative to femoral length (andmore so relative to femoral head size) eitherindicates little development of its hip muscu-lature or, more likely, a linear body shape.

However, since femoral head size is propor-tional to body mass (Jungers, 1988; McHenry,1988; Ruff et al., 1991), scales relativelysimilarly across the genus Homo once varia-tion in body proportions are taken into ac-

count (Ruff et al., 1993), and remains stableonce growth is completed (Ruff et al., 1991),the relatively large size of the Berg Aukasfemoral head indicates cold adapted bodyproportions, somewhere between the aver-age for the hyperarctic European Neander-tals and temperate populations.

An inference of a hyperarctic body formdoes not fit appropriately with the knownnature of Pleistocene Namibian environ-ments. It would also imply that the BergAukas femur had one of the more gracile ofthe known Pleistocene human femoral di-aphyses. In addition, its relatively platy-meric proximal femur contrasts with thoseof the cold-adapted Neandertals with theirbroad pelves, suggesting that it did notpossess the very broad bi-iliac breadth ofthose late archaic humans.

It remains possible that the femoral lengthestimate employed here is in error. A shorterone would increase the degree of capitularhypertrophy (similar to the more extremeEuropean Neandertals), and the resultantdiaphyseal robusticity would fall among themore robust of Early and Middle ArchaicHomo (if given warm climate proportions) orin the middle of the Neandertals (if givencold adapted body proportions). A longerfemoral length would make the relative headsize less pronounced and indicate less of acold-adapted body mass to stature propor-tion, but it would make it gracile for anarchaic Homo femur in terms of diaphysealhypertrophy and especially hip musculaturedevelopment.

Given both the anatomical basis for thelength estimate employed here and the rea-sonable level of diaphyseal and gluteal hyper-trophy for the individual if provided withrelatively warm climate (but not equatorial)body proportions, it is the exceptionally largesize of the femoral head that is anomalous.Femoral head dimensions respond primarilyto levels of hip joint reaction force duringdevelopment, which implies an elevated levelof biomechanical loading during develop-ment if a more moderate level (for a Pleis-tocene hominid) during mature life.

This correlates with its exceptionally lowneck shaft angle. Neck shaft angles are highin infancy and then decrease steadily duringthe first decade and a half of life (Humphry,


1889; Billing, 1954; Houston and Zaleski,1967; Yamaguchi, 1993) [a pattern also evi-dent in juvenile Late Pleistocene Homofemora (Trinkaus and Ruff, 1996)]. The de-gree of decrease in the angle, and its even-tual level in adulthood, is largely deter-mined by the level of loading of the lowerlimb from locomotor activity during develop-ment. This is evident from both clinicalstudies (e.g., Houston and Zaleski, 1976;Laplaza et al., 1993; Yamaguchi, 1993; Sajiet al., 1995) and patterns across recenthumans in which the mean neck shaft angleis positively correlated with the level ofsedentism and mechanization of the society(Anderson and Trinkaus, 1998).

The implication of the anomalously largefemoral head, low neck shaft angle, andmoderate diaphyseal robusticity (for a warmclimate archaic Homo specimen) is that thisindividual experienced unusually high lev-els of lower limb loading during develop-ment, combined with average locomotor load-ing levels for a Pleistocene foragingpopulation during mature life. What is un-known, given the absence of any other bonesof this individual, is whether this unusualcombination was associated with any local-ized or systemic pathological condition, ormerely represents an extreme of individualvariation with respect to lower limb loadingduring development.

ACKNOWLEDGMENTS

We thank P.V. Tobias for access to the BergAukas human femur and R. van der Riet forhelp with CT scanning the specimen. Collec-tion of the comparative Pleistocene hominidfossil data has been possible through thecourtesy of T. Akazawa, B. Arensburg, L.Beck, Y. Coppens, B. Endo, D. Geraads, I.Hershkovitz, J.J. Hublin, J. Jelınek, H.Joachim, W.J. Kennedy, T. Kimura, A. Lan-ganey, A. Leguebe, H. de Lumley, MuayedSa’id al-Damirji, M. Oliva, R. Orban, D.Pilbeam, H.P. Powell, J. Radovcic, Y. Rak,C.B. Stringer, J. Svoboda, M. Teschler-Nicola, B. Vandermeersch, A. Walker, J.Zias, I. Zohar, the Government of Kenya andthe Governors of the National Museums ofKenya. B. Holt and T. Kimura providedcomparative data on early modern humanremains.

LITERATURE CITED

Anderson JY, Trinkaus E. 1998. Patterns of sexual,bilateral and inter-populational variation in humanfemoral neck-shaft angles. J Anat (Lond) 192:279–285.

Billing L. 1954. Roentgen examination of the proximalfemur end in children and adolescents. Acta RadiolSuppl. 110:1–80.

Conroy GC. 1996. The cave breccias of Berg Aukas,Namibia: a clustering approach to mine dump paleon-tology. J Hum Evol 30:349–355.

Conroy GC, Pickford M, Senut B, Mein P. 1993. Dia-monds in the desert: the discovery of Otavipithecusnamibiensis. Evol Anthropol 2:46–52.

Eschman PN. 1992. SLCOMM Version 1.6. Albuquer-que: Eschman Archeological Services.

Feldesman MR, Kleckner JG, Lundy JK. 1990. Thefemur/stature ratio and estimates of stature in mid-and late-Pleistocene fossil hominids. Am J Phys An-thropol 83:359–372.

Grine FE, Jungers WL, Tobias PV, Pearson OM. 1995.Fossil Homo femur from Berg Aukas, northernNamibia. Am J Phys Anthropol 97:151–185.

Holliday TW. 1997a. Body proportions in Late Pleis-tocene Europe and modern human origins. J HumEvol 32:423–447.

Holliday TW. 1997b. Postcranial evidence of cold adapta-tion in European Neandertals. Am J Phys Anthropol104:245–258.

Houston CS, Zaleski WA. 1967. The shape of vertebralbodies and femoral necks in relation to activity. Radi-ology. 89:59–66.

Humphry G. 1889. The angle of the neck with the shaftof the femur at different periods of life and underdifferent circumstances. J Anat Physiol 23:273–282

Ishisawa M. 1931. Anthropologische Untersuchungenuber das Skelett des Yoshiko-Steinzeitmenschen. IIITeil. Die unteren Extremitaten (in Japanese). J An-thropol Soc Nippon 46:1–192.

Jungers WL. 1988. New estimates of body size inaustralopithecines. In: Grine FE, editor. Evolutionaryhistory of the ‘‘robust’’ Australopithecines. New York:Aldine de Gruyter. p 115–125.

Laplaza FJ, Root L, Tassanawipas A, Glasser DB. 1993.Femoral torsion and neck-shaft angles in cerebralpalsy. J Ped Orthop 13:192–199

Lovejoy CO, Heiple KG, Burstein AH (1973) The gait ofAustralopithecus. Am J Phys Anthropol 38:757–780.

Mallegni F, Mariani-Costantini R, Fornaciari G, LongoET, Giacobini G, Radmill AM. 1983. New Europeanfossil hominid material from an Acheulian site nearRome (Castel di Guido). Am J Phys Anthropol 62:263–274.

McHenry HM .1988. New estimates of body weight inearly hominids and their significance to encephaliza-tion and megadontia in ‘‘robust’’ australopithecines.In: Grine FE, editor. Evolutionary history of the‘‘robust’’Australopithecines. New York:Aldine de Gruy-ter. p 133–148.

Nagurka ML, Hayes WC. 1980. An interactive graphicspackage for calculating cross-sectional properties ofcomplex shapes. J Biomech 13:59–64.

Pickford M, Mein P, Senut B. 1994. Fossiliferous Neo-gene karst fillings in Angola, Botswana and Namibia.S Afr J Sci 90:227–230.

Ruff CB. 1987. Sexual dimorphism in human lower limbbone structure: Relationship to subsistence strategyand sexual division of labor. J Hum Evol 16:147–151.

Ruff CB. 1991. Climate, body size and body shape inhominid evolution. J Hum Evol 21:81–105.

Ruff CB. 1994. Morphological adaptation to climate in


modern and fossil hominids. Yrbk Phys Anthropol37:65–107.

Ruff CB. 1995. Biomechanics of the hip and birth inearly Homo. Am J Phys Anthropol 98:527–574.

Ruff CB. 1999. Skeletal structure and behavioral pat-terns of prehistoric Great Basin populations. In: Hemp-hill BE, Larsen CS, editors. Understanding prehis-toric lifeways in the Great Basin Wetlands:Bioarchaeological reconstruction and interpretation.Salt Lake City: University of Utah Press (in press).

Ruff CB, Hayes WC. 1983. Cross-sectional geometry ofPecos Pueblo femora and tibiae — A biomechanicalinvestigation: I. Method and general patterns of varia-tion. Am J Phys Anthropol 60:359–381.

Ruff CB, Scott WW, Liu AYC (1991) Articular anddiaphyseal remodeling of the proximal femur withchanges in body mass in adults. Am J Phys Anthropol86:397–413.

Ruff CB, Trinkaus E, Holliday TW (1997) Body mass andencephalization in Pleistocene Homo. Nature 387:173–176.

Ruff CB, Trinkaus E, Walker A, Larsen CS. 1993.Postcranial robusticity in Homo, I: Temporal trendsand mechanical interpretations. Am J Phys Anthropol91:21–53.

Saji MJ, Upadhyay SS, Leong JCY. 1995. Increasedfemoral neck-shaft angles in adolescent idiopathicscoliosis. Spine 20:303–311.

Sandelowsky B. 1983. Archaeology in Namibia. Am Sci7:606–615.

Senut B, Pickford M, Mein P, Conroy GC, Van CouveringJA. 1992. Discovery of 12 new Neogene fossiliferoussites in paleokarsts in the Otavi Mountains, Namibia.Compt Rend Acad Sci Paris 314:727–733.

Shackley M. 1980. An Acheulian industry with Elephasrecki fauna from Namib IV, South West Africa(Namibia). Nature 284:340–341.

Siesser WG. 1980. Late Miocene origin of the Benguelaupwelling system off northern Namibia. Science 208:283–285.

Skinner J, Smithers R. 1990. The mammals of thesouthern African subregion. Pretoria: University ofPretoria Press.

Trinkaus E. 1976. The evolution of the hominid femoraldiaphysis during the Upper Pleistocene in Europe andthe Near East. Z Morphol Anthropol 67:291–319.

Trinkaus E. 1984. Does KNM-ER 1481A establish Homoerectus at 2.0 myr BP? Am J Phys Anthropol 64:137–139.

Trinkaus E. 1993. Femoral neck-shaft angles of theQafzeh-Skhul early modern humans, and activitylevels among immature Near Eastern Middle Paleo-lithic hominids. J Hum Evol 25:393–416.

Trinkaus E. 1997. Appendicular robusticity and thepaleobiology of modern human emergence. Proc NatAcad Sci USA 94:13367–13373.

Trinkaus E. 1999. The ‘‘robusticity transition’’ revisited.In: Stringer CB, Finlayson C, editors. Gibraltar andthe Neanderthals. Oxford: Oxbow Press (in press).

Trinkaus E, Ruff CB. 1996. Early modern human re-mains from eastern Asia: the Yamashita-cho 1 imma-ture postcrania. J Hum Evol 30:299–314.

Trinkaus E, Ruff CB. 1999. Diaphyseal cross-sectionalgeometry of Near Eastern Middle Paleolithic humans:The femur. J Archaeol Sci 26:409–424.

Trinkaus E, Churchill SE, Villemeur I, Riley KG, HellerJA, Ruff CB. 1991. Robusticity versus shape: thefunctional interpretation of Neandertal appendicularmorphology. J Anthropol Soc Nippon 99:257–278.

Trinkaus E, Ruff CB, Churchill SE, Vandermeersch B.1998. Locomotion and body proportions of the Saint-Cesaire 1 Chatelperronian Neandertal. Proc Nat AcadSci USA 95:5836–5840.

Trinkaus E, Stringer CB, Ruff CB, Hennessy RJ, Rob-erts MB, Parfitt SA. 1999. Diaphyseal cross-sectionalgeometry of the Boxgrove 1 Middle Pleistocene humantibia. J Hum Evol 37:1–25.

Walker A, Leakey RE. 1993. Postcranial skeleton. In:Leakey RE, Walker A, editors. The Nariokotome Homoerectus skeleton. Cambridge: Harvard UniversityPress. p 221–233.

Ward JD, Seely MK, Lancaster N. 1983. On the antiq-uity of the Namib. S Afr J Sci 79:175–183.

Weidenreich F. 1941. The extremity bones of Sinanthro-pus pekinensis. Palaeontol Sinica 5D:1–150.

Wolpoff MH. 1978. Some implications of relative biome-chanical neck length in hominid femora. Am J PhysAnthropol 48:143–147.

Yamaguchi O. 1993. A radiologic study of the hip joint incerebral palsy. J Japan Orthop Assoc 67:1–11


the anomalous archaichomo femur from berg aukas, namibia: a biomechanical assessment

Documents