The Os Navicular of Humans, Great Apes, OH 8, Hadar, and Oreopithecus: Function, Phylogeny, and Multivariate Analyses

AMNH NOVITATES Tuesday Dec 11 2001 10:13 AM 2000 novi 99146 Mp_1 Allen Press • DTPro System File # 01cc

PUBLISHED BY THE AMERICAN MUSEUM OF NATURAL HISTORY CENTRAL PARK WEST AT 79TH STREET, NEW YORK, NY 10024 Number 3288, 38 pp., 18 ®gures, 12 tables February 2, 2000

ESTEBAN E. SARMIENTO1 AND LESLIE F. MARCUS2

CONTENTS Abstract ...... 2 Introduction ...... 2 Materials and Methods ...... 3 Results ...... 6 Canonical Analyses ...... 6 Discussion ...... 7 Relative Orientations of Navicular Facets ...... 7 Curvature and Relative Size of Navicular Facets ...... 12 Variation in Relative Size and Set of Navicular Facets ...... 17 Foot Use and Inherited vs. Epigenetic Characters ...... 19 Body Weight, Lower-limb size, and Navicular Facet Area ...... 21 Navicular Function and Implied Locomotor Behaviors ...... 25 Fossil Foot Use ...... 28 OH8 ...... 28 Hadar ...... 29 Oreopithecus ...... 32 Phylogeny Implicit in Navicular Measurements ...... 32 Conclusions ...... 33 Acknowledgments ...... 33 References ...... 34

1 Research Associate, Division of Vertebrate Zoology, American Museum of Natural History. 2 Research Associate, Division of Paleontology, American Museum of Natural History; Professor of Biology, Queens College, Flushing NY, 11367.

Copyright ᭧ American Museum of Natural History 2000 ISSN 0003-0082 / Price $4.40 AMNH NOVITATES Tuesday Dec 11 2001 10:13 AM 2000 novi 99146 Mp_2 Allen Press • DTPro System File # 01cc

2 AMERICAN MUSEUM NOVITATES NO. 3288

ABSTRACT

To clarify fossil hominid behavior and phylogeny, and to test the accuracy of basing these studies on single bones, navicular measurements of Olduvai and Hadar hominids, Oreopithe- cus, and a representative sample of humans and great apes were compared. The measurements chosen for comparison quantify the relative orientation, articular area, and curvature of the navicular facets. The measurements demonstrate that the OH 8 navicular belongs to a rigid foot with an adducted hallux and a strong commitment to terrestriality. The Hadar naviculars belong to a foot which lacked a ®xed longitudinal plantar arch and had at least a degree of hallucal opposability comparable to that of mountain gorillas. The Oreopithecus navicular belongs to a mobile foot with a widely divergent hallux committed to arboreal behaviors. Multiple discriminant and canonical variate analyses of navicular measurements emphasize the uniqueness of Oreopithecus and the similarities between OH 8 and humans, and between Hadar and African apes. The African apelike morphology of the Hadar naviculars contradicts the alleged humanlike morphology of the Hadar pelvis and knee joints. This contradiction underscores the fallacies inherent in constructing phylogenies on the basis of single bones and/or fragmentary remains, and of reconstructing locomotor behaviors on the basis of localized anatomy.

INTRODUCTION of those past behaviors which have been strongly selected for. The hands and feet are those parts of the Despite their usefulness, carpal and tarsal integumentary and musculoskeletal structure bones have been largely ignored by most pa- which come directly into contact and interact leoanthropologists when reconstructing early with the physical variables of the environ- hominid behaviors (Lovejoy, 1974, 1978, ment. As such, their morphology closely cor- 1988; Robinson, 1972; Zihlman and Bunker, responds to these variables, responding di- 1979; Stern and Susman, 1983; Latimer, rectly both ontogenetically and phylogeneti- 1991; McHenry, 1991; Susman and Stern, cally to changes in use (i.e., changes in their 1991; Fleagle, 1998) or phylogenies (Leakey interactions with the physical variables of the et al., 1964; Robinson, 1965; Wolpoff, 1974; environment) over time (Sarmiento, 1985, Skeleton et al., 1986; Olson, 1978, 1981, 1988, 1994). Because hand and foot anatomy 1985; White et al., 1981; Chamberlain and re¯ect the animal's environmental interactions, they are excellent indicators of behav- Wood 1987; Edelstein, 1987; Verhaegen, ior. Composed of a large number of anatom- 1990, 1994; Tobias, 1991a, b; McHenry, ical elements that combine to produce the ap- 1996; Strait et al., 1997). This is all the more propriate morphology, hands and feet may surprising considering that carpal and tarsal also provide insights into past behaviors and bones are often the best preserved and most are useful for reconstructing phylogenies complete skeletal elements found at early (Schaeffer,1947; Beigert, 1963; Lewis, 1969, hominid fossil sites, and are well represented 1974, 1980a, b, c; Szalay and Decker, 1974; in fossil collections. It is the object of this Sarmiento, 1983, 1985, 1988, 1994; Beard et study, therefore, to analyze the navicular of al., 1988; Gebo and Dagosto, 1988). This es- humans, great apes, and some early ``homi- pecially applies to the carpus and tarsus nid'' and hominoid fossils in both a func- where many anatomical elements interact to tional and phylogenetic context. Because the provide the appropriate movements and dis- great ape navicular articulates variably with tribute loads. The anatomical complexity of all of the tarsal bones (Lewis, 1980b, c; Sar- the carpus and tarsus dictates that changes in miento, 1994), it should re¯ect functional hand or foot use are achieved with relatively differences throughout the tarsus. The pres- minimal changes in structure, i.e., the organ- ence of complete naviculars at many of the ism makes the best use of its inherited anat- early hominid fossil sites (Day and Napier, omy (Sarmiento, 1985, 1988). Thus, the car- 1964; Latimer et al., 1982; Clarke and To- pus and tarsus provide an anatomical record bias, 1995) should provide a phylogenetic AMNH NOVITATES Tuesday Dec 11 2001 10:13 AM 2000 novi 99146 Mp_3 Allen Press • DTPro System File # 01cc

2000 SARMIENTO AND MARCUS: OS NAVICULAR OF HOMINIDS 3

perspective of changes in foot use over time. wires to the articular surfaces and using the In addition, analyses and comparisons of a carpenter angle to measure the angles made single bone should serve to test the accuracy by the wires. All measurements are of the of reconstructing behavior or phylogeny right navicular except for some (fossils) in based on localized anatomy. which only the left specimen was available. Comparisons to body size relied on navic- MATERIALS AND METHODS ulars of specimens with reported body weights. For those specimens without re- Naviculars of Olduvai (OH 8) and Hadar ported body weight, the sums of the midshaft (AL 333±47, 333±36) hominids, Oreopithe- cross-sectional area of the femur and tibia cus (Basel #39), and a sample of naviculars were used for comparisons. The length of the from each of the great ape species and of femur (mm) times its midshaft cross-section- humans were measured and compared for di- al area plus the length of the tibia (mm) times mensions, curvature and relative orientation its midshaft cross-sectional area were used to of facets. Differences in the frequency of ter- compare lower-limb volume to correspond- restrial behaviors in Virunga gorillas (Gorilla ing navicular measurements for each individ- b. beringei) relative to western gorillas (G. ual specimen. Measurements for lower-limb g. gorilla) justi®ed considering the two as long bone lengths are after Sarmiento (1985) separate, although a speci®c designation for and Sarmiento et al. (1996). Cross-sectional Virunga gorillas is arguable (Sarmiento, areas of lower-limb long bones were arrived 1994; Sarmiento and Butynski, 1996; Sar- at by squaring the midshaft circumference miento et al., 1996). Great ape naviculars are and dividing it by 4␲. Measurements and the from the skeletal collections at the following indices formulated from them were chosen to institutions: American Museum of Natural re¯ect mechanical concerns. Figure 1 sum- History, New York; Museum of Comparative marizes the length and angular measurements Zoology at Harvard University, Cambridge; taken on the navicular and the indices for- Field Museum of Natural History, Chicago; mulated from these measurements. Tables 1± National Museum of Natural History, Wash- 10 summarize comparisons of the linear and ington D.C.; Philadelphia Academy of Sci- angular measurements. Figures 2±11 are bi- ences, Philadelphia; Powell Cotton Museum, variate plots of the proportions considered. Birchington; Royal African Museum, Ter- Functional interpretation of the navicular vuren; and Swedish Museum of Natural His- tory, Stockholm. Human naviculars are from of fossil forms relied on nearly 20 years of the Dart collection in the Department of studying foot use in humans and in free rang- Anatomy, University of the Witwatersrand, ing and captive great apes (Sarmiento, 1983, and the Anthropology collections at the 1985, 1994; Sarmiento and Butynski, in American Museum of Natural History. Fossil prep.). Articulation of human, great ape, and material is housed in the collections of the fossil foot bones in close-packed positions National Museum of Tanzania, Dar es Sa- were used to gauge the orientation of the laam (OH 8) and National Museum of Ethi- pedal segments imparted by the set (i.e., rel- opia, Addis Ababa (Hadar). Oreopithecus ative orientation) of the navicular facets. Lig- fossil material was studied at the Museum of amentous preparations of the foot of humans, Natural History, Basel; Istituto di Geologia western gorillas, common chimpanzees, and Universita de Firenze; and Instituto de Pa- orangutans were used to further verify the leontologia, Sabadell, Spain. relative orientations of articulated tarsals and Lengths were measured to the nearest 0.1 metatarsals. Notes on OH 8, Hadar, and Or- mm using a digital caliper connected to a eopithecus foot bones were used to corrob- computer and collected in a Lotus spread- orate functional interpretations of fossil foot sheet. A carpenter's angle accurate to 0.5Њ use based on the navicular study. was applied to the articular surfaces to mea- To complement results from the functional sure the cuboectocuneiform angle and hori- analysis and provide insights into phylogeny, zontal mesoectocuneiform angle. The re- a multiple discriminant and canonical variate maining angles were arrived at by aligning analysis was run employing 13 linear mea- AMNH NOVITATES Tuesday Dec 11 2001 10:13 AM 2000 novi 99146 Mp_4 Allen Press • DTPro System File # 01cc

4 AMERICAN MUSEUM NOVITATES NO. 3288

Fig. 1. Proximal (A), distal (B), lateral (C) and dorsal (D) views of the OH 8 navicular showing the measured lengths, and proximal (E), distal (F), lateral (G) and dorsal (H) views of a left gorilla navicular showing the measured angles; a ϭ talar facet major axis (dorsoplantar) diameter, b ϭ talar facet minor axis (mediolateral) diameter, c ϭ ectocuneiform facet dorsoplantar diameter, d ϭ ectocuneiform facet mediolateral diameter, e ϭ mesocuneiform facet dorsoplantar diameter, f ϭ mesocuneiform facet mediolateral diameter, g ϭ entocuneiform facet mediolateral diameter, h ϭ entocuneiform facet dorsoplantar diameter, i ϭ navicular maximum length, j ϭ cuboid facet dorsoplantar diameter, k ϭ cuboid facet mediolateral diameter, l ϭ depth of talar facet along major axis, m ϭ depth of talar facet along minor axis; 1 ϭ frontal talocuboid angle, 90Њ -2ϭ navicular torsion, 3 ϭ frontal mesoectocuneiform angle, 4 ϭ entoectocuneiform angle 5 ϭ sagittal taloectocuneifrom angle, 6 ϭ transverse mesoectocuneiform angle, 7 ϭ transverse cuboectocuneiform angle. In all cases the lines chosen for angular measurements bisect facets into approximately equal halves. For comparative purposes the major bisecting axes of the talar head and cuneiform facets are referred to in the text as the dorsoplantar axes. In neither great apes nor humans do all these axes have a dorsoplantar orientation, but are held in varying inclination to a dorsoplantar axis according to talar head and navicular torison and the frontal mesoectocuneiform angle. The cross-sectional area of the talar, ectocuneiform, mesocuneiform entocuneiform and cuboid facets are given by the products of a and b, c and d, e and f, g and h, and j and k, respectively. Relative cross-sectional area for each facet is compared as a percentage of the sum of all of the navicular facets. The subtended angle of curvature and the radius of curvature of the talar facet along the dosoplantar (major) and mediolateral (minor) axes are given by 4 arctan(2l/a) and (l2 ϩ a2/4)/ 2l, and 4 arctan(2m/b) and (m2 ϩ b2/4)/2m, respectively.

surements and 6 angles using PC SAS 6.12. be accurately measured in any human spec- Only human specimens with a distinct cu- imen and was dropped from the analysis. Re- boid facet (N ϭ 14) were included in this sults from these analyses are summarized as analysis. Due to small and irregular cuboid Mahalanobis distances in table 11 and ®gures facets, the frontal talocuboid angle could not 12±14. AMNH NOVITATES Tuesday Dec 11 2001 10:13 AM 2000 novi 99146 Mp_5 Allen Press • DTPro System File # 01cc

2000 SARMIENTO AND MARCUS: OS NAVICULAR OF HOMINIDS 5 AMNH NOVITATES Tuesday Dec 11 2001 10:13 AM 2000 novi 99146 Mp_6 Allen Press • DTPro System File # 01cc

6 AMERICAN MUSEUM NOVITATES NO. 3288

RESULTS

CANONICAL ANALYSES Multiple discriminant analysis distin- guished humans and the living great apes, with genera (i.e., Homo, Gorilla, Pan, and Pongo) showing much larger differences than species of the same genus. Mahalanobis distance summarizes the difference between any two groups as a single statistic and can be used as a test of signi®cance between groups (table 11), and as a distance measure for clustering groups (®gs. 12±14). Repeated attempts to reduce the number of measured variables in the discriminant analysis, including stepwise discrimination always led to a less distinctive pattern than the one arrived at when all the variables were considered. When grouped separately, neither the angles nor the length measurements gave patterns comparable to the combined data. Although the number of measured variables is relatively high compared to the sample size for each group, this is the strongest case one of us (LM) has seen where all of the chosen variables have considerable im- portance in discriminating among groups. AMNH NOVITATES Tuesday Dec 11 2001 10:13 AM 2000 novi 99146 Mp_7 Allen Press • DTPro System File # 01cc

2000 SARMIENTO AND MARCUS: OS NAVICULAR OF HOMINIDS 7

It is very dif®cult to interpret the contribution of the variables a posteriori (Marcus, 1990). Six of the nine canonical variates lead to signi®cant differences. Not summarized in the ®gures, the third through sixth canonical variates separate the two gorillas subspecies, the two chimpanzees species, and the orangutan even more distinctly than shown in ®g- ure 12. All of the fossils are also more dis- tinctively separated along some of these canonical axes. This is consistent with the large distances between the fossils and the living taxa (®gs. 13, 14). The two Hadar naviculars whether considered separately or together, are not signi®cantly different from either chimpanzee species (table 11). OH 8 is most similar to, but differs signi®cantly from humans (p ϭ 0.014).

DISCUSSION

RELATIVE ORIENTATION OF NAVICULAR FACETS The angles formed by the articular planes of the navicular facets re¯ect the set of the AMNH NOVITATES Tuesday Dec 11 2001 10:13 AM 2000 novi 99146 Mp_8 Allen Press • DTPro System File # 01cc

8 AMERICAN MUSEUM NOVITATES NO. 3288

tarsal and metatarsal bones (®gs. 15±18). In turn, the set of these bones bears on the mobility, rigidity, and weight-bearing capabilities of the foot (Elftman and Manter, 1935; Manter, 1941; Elftman, 1960; Day and Wood, 1968; Lewis 1980a, b, c; Sarmiento, 1994). The large sagittal taloectocuneiform angle (®g. 1G #5) in great apes and Hadar australopithecines (table 1) emphasizes the transfer of talar head loads to the substrate through the navicular's plantar tubercle (or secondarily through the entocuneiform), but sacri®ces the percentage of load transferred to the ectocuneiform. The great ape angle is associated with a mobile midtarsal joint and dorsi¯exed set of the talonavicular joint (®g. 15). The small sagittal taloectocuneiform angle in humans, on the other hand, emphasizes talar head load transfer to the cuneiforms at the expense of its transfer to the substrate. The human orientation is associated with a longitudinal arch and a rigid midtarsal and tarsometatarsal joint, and preferential loading of the ball of the foot (as opposed to the dis- AMNH NOVITATES Tuesday Dec 11 2001 10:13 AM 2000 novi 99146 Mp_9 Allen Press • DTPro System File # 01cc

2000 SARMIENTO AND MARCUS: OS NAVICULAR OF HOMINIDS 9

tal tarsal row and metatarsal bases) during the small angle value (table 2) is in accord weight support (Sarmiento, 1994). with an abducted hallux, a ®xed longitudinal The sagittal entoectocuneiform angle (®g plantar arch, and an entocuneiform which 1G #4, table 2) imparts a plantar divergence does not transfer any appreciable weight to of the entocuneiform relative to the ectocu- the substrate. The larger angle in the Hadar neiform facet3 (®g. 15). The larger the angle, fossils probably re¯ects weight transfer from the higher the degree of divergence. With a the entocuneiform to the substrate, and some plantar orientation of the entocuneiform fac- degree of hallucal opposability. The very et, talar head load can be transferred to the large angle of Oreopithecus is in accord with substrate through the navicular and plantar its markedly abducted hallucal postures tubercles of the entocuneiform. In taxa with (Kohler and Moya-Sola, 1997). an opposable hallux, the angle may re¯ect The transverse mesoectocuneiform angle enhanced opposition depending on the set of (®g. 1H #6, table 1) re¯ects the degree of its conarticular and the ®nal set imparted to divergence of the 2nd and 3rd pedal rays (®g. the hallucal long axis. In humans and OH 8, 16). A large angle approximating 180Њ as seen in humans and OH 8 imparts a nearly 3 Contributions of the measured navicular angles to parallel orientation to the long axes of the the set of the foot bones consider talar torsion values second and third pedal rays. Conversely, the approximating 90Њ and a talar facet held with its major smaller angle values seen in great apes, Had- axis in the vertical. With increasing deviation of the major axis of the talar facet toward the horizontal and lower ar, and Oreopithecus impart a divergent ori- navicular torsion values, the sagittal entoectocuneiform entation to the long axes of the pedal rays. angle would increasingly contribute to medial diver- The transverse cuboectocuneiform angle gence of the hallux relative to the third metatarsal. Like- (®g. 1H #7, table 1) re¯ects the degree of wise with horizontal postures of the major axis of the talar facet a high sagittal taloectocuneiform angle results divergence of the two most lateral rays rel- in a more laterally divergent third metatarsal relative to ative to the third ray and ectocuneiform (®g. the hallux and the talar head. 16). The smaller the value of this angle, the AMNH NOVITATES Tuesday Dec 11 2001 10:13 AM 2000 novi 99146 Mp_10 Allen Press • DTPro System File # 01cc

10 AMERICAN MUSEUM NOVITATES NO. 3288

greater the degree of lateral ray divergence. approximates the dorsoplantar (major) axis In this regard, pygmy chimpanzees have the of the ectocuneiform facet (®g. 17). With most divergent set and humans the least di- vertical postures of the dorsoplantar axis of vergent set of the two most lateral rays. the ectocuneiform, high torsion imparts a Torsion of the navicular (®g. 1F #2) affects plantarly rotated set to the talar facet and re- the orientation of the talar head relative to sults in a high transverse plantar arch. In go- the dorsoplantar axis of the ectocuneiform. rillas at least, the high torsion values are as- Torsion values approximating 90Њ indicate sociated with a higher transverse arch than that the major bisecting axis of the talar facet that seen in either chimpanzees or orangutans AMNH NOVITATES Tuesday Dec 11 2001 10:13 AM 2000 novi 99146 Mp_11 Allen Press • DTPro System File # 01cc

2000 SARMIENTO AND MARCUS: OS NAVICULAR OF HOMINIDS 11

Fig. 2. Talar facet cross-sectional area (mm2) vs. total navicular facet cross-sectional area (mm2)in humans, great apes, and fossil hominoids. Arrows point to fossils.

(Sarmiento, 1994). Because the rotated set of indicate that the major bisecting axis of the the navicular is also affected by talar head talar head is plantarly rotated relative to the torsion, and the dorsoplantar axis of the ec- cuboid in close-packed positions of the tal- tocuneiform may be set oblique to the ver- ocuboid joint. A low talocuboid angle, there- tical (®g. 17; Sarmiento 1994), humans do fore, attests to the relatively high transverse not necessarily have navicular torsion values arch of mountain gorillas (table 2; Sarmiento, as great as gorillas, despite a high transverse 1994). Because in humans and great apes the plantar arch (table 2). Without evidence from cubonavicular facet does not necessarily lie other joint sets, torsion of the OH 8 and Had- in a sagittal or parasagittal plane, talocuboid ar naviculars indicates a transverse plantar angles do not directly re¯ect plantar arch arch height which is at least comparable to height. Nevertheless, the low angle values in that of gorillas. both OH 8 and the Hadar fossils when taken The frontal talocuboid angle (®g. 1E #1) together with high navicular torsion (table 2) in¯uences the set of the navicular relative to indicate that these fossils had a rather high the cuboid, and also re¯ects the height of the transverse plantar arch when the tarsals were transverse plantar arch (®g. 18). Small angles in the close-packed position. AMNH NOVITATES Tuesday Dec 11 2001 10:13 AM 2000 novi 99146 Mp_12 Allen Press • DTPro System File # 01cc

12 AMERICAN MUSEUM NOVITATES NO. 3288

Fig. 3. Ectocuneiform facet cross-sectional area (mm2) vs. total navicular facet cross-sectional area (mm2) in humans, great apes, and fossil hominoids. Arrows point to fossils.

The frontal mesoectocuneiform angle (®g. in the sagittal plane and is important in both 1F #3) imparts a relative rotational set to the arboreal and terrestrial behaviors (Sarmiento, mesocuneiform and ectocuneiform (®g. 17). 1994). Although a large angle may imply greater opposition between the second and the third CURVATURE AND RELATIVE SIZE OF digit, metatarsal and cuneiform torsion also NAVICULAR FACETS affects the ®nal opposition set relative to the hallux. Considering a nonopposable hallux, Differences in the relative cross-sectional the large angle in humans (table 2) is best area of the navicular's talar, cuboid, and cu- associated with a tight curvature of the trans- neiform facets re¯ect differences in mobility verse plantar arch. In orangutans, on the oth- and load-bearing capabilities at each joint. er hand, the large angle may be best associ- Because planar articulations commonly im- ated with some degree of second to third dig- ply relatively restricted joint motion (Sar- it opposition and a high but mobile trans- miento, 1988), the relative area of the cune- verse arch. The transverse arch enhances the iform facets (largely planar articulations in foot's ability to withstand bending moments great apes and humans) is in large part an AMNH NOVITATES Tuesday Dec 11 2001 10:13 AM 2000 novi 99146 Mp_13 Allen Press • DTPro System File # 01cc

2000 SARMIENTO AND MARCUS: OS NAVICULAR OF HOMINIDS 13

Fig. 4. Mesocuneiform facet cross-sectional area (mm2) vs. total navicular facet cross-sectional area (mm2) in humans, great apes, and fossil hominoids. Arrows point to fossils.

indication of the relative load transmitted sent cuboid facet and transmission of the tal- across each joint. The relatively large ecto- ar head load largely to the cuneiforms. cuneiform and small entocuneiform facet ar- Absence or presence of a relatively small eas in OH 8, and the large entocuneiform and cuboid facet in humans, (®g. 6; tables 6 and small ectocuneiform facet areas in common 8) re¯ects a foot loaded parallel to its long chimpanzees, and gorillas (table 5), re¯ect axis and rarely subjected to a large magni- contrastingly different navicular loading pat- tude of mediolateral forces. Conversely, the terns. In OH 8 loading of the third digital ray large cuboid facet in pygmy chimpanzees, was achieved at the expense of the entocu- and orangutans can be associated with load- neiform and hallux. In gorillas and common ing of the foot in supinated postures with the chimpanzees, loading of the entocuneiform foot's mediolateral axis held approximately and hallux is achieved at the expense of load- vertical and parallel to the weight vector. In ing the third digital ray. In humans, the rel- African apes, loads across the cubonavicular atively large combined cuneiform facet area joint may also result from the force of the (®g. 8, table 9) re¯ects a small or nearly ab- peroneus longus tendon balancing a hallucal AMNH NOVITATES Tuesday Dec 11 2001 10:13 AM 2000 novi 99146 Mp_14 Allen Press • DTPro System File # 01cc

14 AMERICAN MUSEUM NOVITATES NO. 3288

Fig. 5. Entocuneiform facet cross-sectional area (mm2) vs. total navicular facet cross-sectional area (mm2) in humans, great apes, and fossil hominoids. Arrows point to fossils.

grasp. Similar foot postures and/or hallucal (table 3) implies that loads born by the talar opposability are implied by the relatively head can be transmitted through the navicu- large cuboid facets of the Hadar naviculars. lar to either of the cuneiforms, to the cuboid, The biconvex talar facet implies different to the substrate, or to any combination of loading capabilities from those of the more these three. planar cuboid or cuneiform facets. In contrast Because curvature compromises the total to planar articulations, curved articulations available articular surface area perpendicular present a perpendicular surface to force vec- to weight-bearing loads (Sarmiento, 1988), tors of varying orientation (Sarmiento 1988). curved articular surfaces must have more They, therefore, have an advantage over pla- area than planar articular surfaces to with- nar articulations in that they enable loading stand comparable loads (Sarmiento, 1988). of a segment in varying orientation relative Increasing surface area without increasing to the body weight and/or support, and are the degree of dorsoplantar curvature, the rel- less susceptible to shear forces (Sarmiento, atively wide mediolateral talar facet diameter 1988). The large degree of dorsoplantar cur- in humans (®g. 7) re¯ects a need to increase vature of the great ape and Hadar naviculars total perpendicular area available to weight- AMNH NOVITATES Tuesday Dec 11 2001 10:13 AM 2000 novi 99146 Mp_15 Allen Press • DTPro System File # 01cc

2000 SARMIENTO AND MARCUS: OS NAVICULAR OF HOMINIDS 15

Fig. 6. Cuboid facet cross-sectional area (mm2) vs. total navicular facet cross-sectional area (mm2) in humans, great apes, and fossil hominoids. Arrows point to fossils.

bearing loads. With a limited degree of dor- maximizes the total load it can transmit per soplantar curvature and a relative surface available articular area in comparison to Af- area comparable to that of gorillas and com- rican apes. A similar trade-off is seen in the mon chimpanzees4 (tables 3 and 9, ®g. 10), navicular of Oreopithecus and OH 8. The the human talar facet sacri®ces its capacity nearly parallel set of the OH 8 and human to transmit loads of varying orientation, but cuneiform and talar articular surfaces (tables 1 and 2), moreover, imply load transmission 4 Cross-sectional surface area is a more accurate es- from the talus through the navicular is large- timate of the maximum surface area available perpen- ly to the cuneiforms. In humans, this implied dicular to load than of true articular area. Articular cross- sectional area tends to increasingly underestimate true weight transmission is also indicated by the articular area with an increase in articular curvature, relatively large combined surface area of the where arc length (curvature (Њ) cord2 ϩ 2 arc height) ␲/ cuneiform facets (®g. 8). 720Њ and cross-sectional diameter is equal to cord length. Curved articular surfaces, when associated Because great apes have a more pronounced dorsoplantar curvature of the talar facet than do humans, cross- with differences in the dimensions and cur- sectional area underestimates their true talar facet area vatures of conarticulars, also imply joint mo- more than in humans. bility. To maintain a comparable surface area AMNH NOVITATES Tuesday Dec 11 2001 10:13 AM 2000 novi 99146 Mp_16 Allen Press • DTPro System File # 01cc

16 AMERICAN MUSEUM NOVITATES NO. 3288

Fig. 7. Dorsoplantar (major axis) diameter (mm) vs. mediolateral (minor axis) diameter (mm) of the talar facet (i.e., talar facet length vs. talar facet width) in humans, great apes, and fossil hominoids. Arrows point to fossils.

in contact throughout a range of joint motion, In light of its marked dorsoplantar and me- joints providing movement must have a larg- diolateral curvatures (table 3), the relatively er articular area (at least of one of their con- small cross-sectional area of the orangutan articulars) than those serving only to bear talar facet (table 6) re¯ects the large size of loads (Sarmiento, 1988). In orangutans, the the navicular's other facets (®g. 2). This size large ectocuneiform facet with marked bi- is most likely a correlate of a mobile foot concave curvatures (®g. 3, tables 5 and 7) which is loaded in a variety of postures (i.e., re¯ects considerable mobility at the navicu- the talar head load is variably transferred loectocuneiform joint (Rose, 1988). Marked among the three cuneiforms and cuboid). In mobility probably also explains the relatively contrast, a small degree of mediolateral cur- large mediolateral degree of curvature of the vature of the talar facet (table 3) with marked orangutan talonavicular joint (table 3), which differences in its dorsoplantar vs. its medio- out of all the great apes most closely ap- lateral radius of curvature in gorillas and proximates a ball and socket analog (Rose, common chimpanzees (table 4) indicates lim- 1988). ited rotatory (circumductory) motion at the AMNH NOVITATES Tuesday Dec 11 2001 10:13 AM 2000 novi 99146 Mp_17 Allen Press • DTPro System File # 01cc

2000 SARMIENTO AND MARCUS: OS NAVICULAR OF HOMINIDS 17

Fig. 8. Total cuneiform facet cross-sectional area (mm2) vs. total navicular facet cross-sectional (mm2) area in humans, great apes, and fossil hominoids. Arrows point to fossils.

talonavicular joint, at least during loading. joint compromised load bearing for mobility The same is probably true of OH 8 which in to the same degree seen in orangutans. spite of an absolutely small talar facet has a very large mediolateral radius curvature that VARIATION IN RELATIVE SIZE is much greater than its dorsoplantar radius AND SET OF NAVICULAR FACETS of curvature. With large degrees of dorsoplantar and mediolateral curvature, and no Large variation and overlap in navicular appreciable difference in the two radii of cur- angles and relative size of navicular facets vature (both of which are small), the rela- within and among human and great ape sam- tively small talar facet of Hadar ®nds a closer ples correspond, in part, to variation in over- mechanical analog to that of pygmy chim- all foot structure, i.e., corresponding varia- panzees and orangutans, and suggests con- tion in the abducted-adducted set of the hal- siderable rotatory mobility during loading. lux, the divergent set of the rays, and the However, without matching conarticulars to height of the plantar longitudinal and trans- gauge mobility at the joint it is not possible verse arches (®gs. 15±18). Large within-sam- to know whether the Hadar talonavicular ple variation in the set of the navicular facets, AMNH NOVITATES Tuesday Dec 11 2001 10:13 AM 2000 novi 99146 Mp_18 Allen Press • DTPro System File # 01cc

18 AMERICAN MUSEUM NOVITATES NO. 3288

Fig. 9. Bivariate plot of cube root of body weight (kg⅓) vs. square root of total navicular facet cross-sectional area (mm) in great apes. N ϭ 62, Slope ϭ 6.58, y intercept ϭ -143.77. At 95% con®dence limits OH 8, the two Hadar naviculars, and Oreopithecus were calculated to have body weights of 12.0± 100.3 kg, 17.6±143.6 kg (AL 333-47), 18.6±155.1 kg (AL 333-36), and 4.4±39.5 kg respectively.

however, need not all be re¯ected in corre- contours of the navicular's articular facets. sponding variation in overall foot structure. Other segments in the link may redirect load Because the foot is composed of a number and/or correct for variation in navicular joint of linked bony segments, the set of each facet mobility. For any one taxon, the overall foot on any one segment is not necessarily trans- structure need not demonstrate as much with- lated to all the linked segments, but may be in-sample variation, as is re¯ected in the di- corrected for by the joint sets of other seg- mensions and contours of any one navicular ments in the link. Day and Wood (1968) not- facet. Despite overlap in the range of values ed such a correction of set in the talonavic- for the set, dimensions, or contours of indi- ular articulation of OH 8 which provides for vidual facets, humans and great apes each an adducted hallux despite a divergent talar have a characteristic and distinct foot struc- neck. The same probably applies to the with- ture (Tuttle, 1970; Sarmiento, 1983, 1985, in-sample variation in the dimension and 1994; Rose, 1988). AMNH NOVITATES Tuesday Dec 11 2001 10:13 AM 2000 novi 99146 Mp_19 Allen Press • DTPro System File # 01cc

2000 SARMIENTO AND MARCUS: OS NAVICULAR OF HOMINIDS 19

Fig. 10. The sum of the femoral and tibial cross-sectional areas (mm2) vs. talar facet cross-sectional area (mm2) in humans, great apes, and fossil hominoids.

Because each characteristic great ape and facet measurements (tables 1±10), there is lit- human foot structure places constraints on tle or no overlap when all navicular mea- the range of variation in the relative size and surements are taken together, even the two set of each navicular facet, and the navicular most similar samples are separated by at least has at least four facets, it is highly unlikely 3.8 Mahalanobis distance units (table 11). (unless the sets, curvatures, and dimensions When taken together, therefore, the navicular for different facets on the same navicular are measurements do re¯ect the characteristic all positively correlated) that all four facets foot structure of each sample. on anyone navicular when compared individ- ually will each have sets, dimensions, and/or FOOT USE AND INHERITED VS.EPIGENETIC contours that deviate markedly (more than 1 CHARACTERS standard deviation) from their respective means. Despite a large overlap in the range Foot use (i.e., the series of external forces of variation between human and great ape applied to the foot through a range of foot samples for each of the individual navicular joint movements), through ontogenetic plas- AMNH NOVITATES Tuesday Dec 11 2001 10:13 AM 2000 novi 99146 Mp_20 Allen Press • DTPro System File # 01cc

20 AMERICAN MUSEUM NOVITATES NO. 3288

Fig. 11. Cube root of lower limb-volume (mm) vs. square root of total navicular cross-sectional area (mm) in humans, great apes, and fossil hominoids.

ticity in the dimension and relative orienta- Localization of forces and differences in tion of facets, modi®es the inherited mor- the duration of growth among the segments phology of the various segments to achieve comprising a structure are probably important the adult foot structure (Sarmiento, 1985). factors in determining which segments are With shifts in behavior, ontogenetic plasticity most strongly modi®ed. The latest segments allows the foot to adapt to a new use per- to complete epiphyseal fusion and fully de- mitting selection for the inherited morphol- velop have a longer time to respond to use, ogy which best suits this new use in descen- and thus may be more likely to be altered dant populations. Although both OH 8 and ontogenetically in response to use. Late de- humans have an adducted hallux (Day and veloping segments may very well be a focus Napier, 1964; Day and Wood, 1968), differ- of ontogenetic plasticity, since their epiphyses ences in the set of the talonavicular joint be- fuse at a time when the animal is closest to tween the two may in part represent differ- its adult weight. Because at this time the forc- ences in the number of generations hallucal es on the foot may be expected to most close- adduction has been selected for. ly approximate those of the adult, late devel- AMNH NOVITATES Tuesday Dec 11 2001 10:13 AM 2000 novi 99146 Mp_21 Allen Press • DTPro System File # 01cc

2000 SARMIENTO AND MARCUS: OS NAVICULAR OF HOMINIDS 21

oping segments are able to ®ne tune an adult total navicular or talar facet cross-sectional foot to its newly acquired use. area (®gs. 9±11; tables 9±10) reveal trade- Segments in direct interaction with envi- offs between body size, lower-limb loading, ronmental forces (i.e., those of the hands and and joint mobility. Although large facet sur- feet) are more likely to encounter greater var- face areas relative to joint load are charac- iation in forces both in direction and mag- teristic of mobile joints (Sarmiento, 1988), nitude, and may be expected to exhibit more differences in joint mobility are not neces- plasticity and be less conservative than more sarily re¯ected by differences in ratios of fac- proximal segments (i.e., those closer to the et cross-sectional area to body weight. In this animal's center of gravity). In the case of regard, comparatively large navicular facet proximal segments, intervening segments re- areas relative to body weight in female go- direct and buffer environmental forces, so rillas and common chimpanzees (®g. 9, table that changes in foot and hand use may cause 10) are more a factor of the lower limb bear- only negligible changes in the resultant forc- ing a proportionately greater percentage of es proximal segments are subjected to. In this the weight than an indicator of navicular regard, the morphology of proximal seg- joint mobility. Because midshaft cross-sec- ments is shaped by forces channeled through tional area of the lower-limb long bones is a inherited structures (i.e., muscles, connective factor of the load borne by the lower limbs, tissue, and adjacent segments), and thus ratios of talar or total navicular facet area to more likely to be conservative. midshaft cross-sectional area more accurately re¯ect facet size relative to facet load, and BODY WEIGHT,LOWER LIMB SIZE, AND hence are better indicators of navicular joint NAVICULAR FACET AREA mobility. In pygmy chimpanzees and orang- Relationships of body weight, lower-limb utans, therefore, the comparatively higher cross-sectional area, or lower-limb volume to values for these ratios (®g. 10, tables 9 and AMNH NOVITATES Tuesday Dec 11 2001 10:13 AM 2000 novi 99146 Mp_22 Allen Press • DTPro System File # 01cc

22 AMERICAN MUSEUM NOVITATES NO. 3288

Fig. 12. A plot of the ®rst two canonical variates with vectors representing the contribution of each of the measured variables to the scatter within and among measured taxa. Arrows point to fossils. Note the distinctiveness of H. sapiens, the uniqueness of Oreopithecus and the similarities of Hadar and African apes, of OH 8 and Homo, and of great ape species or subspecies within genera. The vectors representing the frontal mesoectocuneiform angle (EctMsFn) and the mediolateral diameter of the entocuneiform facet (EntFml) are nearly overlapping. Vector lengths are exagerated by a factor of ten, and owing to a two-dimensional projection, are not proportional to their actual length. Eighty percent of the variance among means relative to the within-group variance is summarized by the ®rst two canonical variates (see ®gure 14 for plotted means of the ®rst two canonical variates). Program written in Matlab version 5.1. This ``biplot'' is after Rohlf (1997); see Marcus (1993) for a discussion. The program and navicular data are available from one of us (LM). TalFlng ϭ talar facet dorsoplantar (major axis) diameter, TalFwd ϭ talar facet mediolateral (minor axis) diameter, EctFpdϭ ectocuneiform facet dorsoplantar diameter, EctFmlϭ ectocuneiform facet mediolateral diameter, MesFdpϭ mesocuneiform facet dorsoplantar diameter, MesFmlϭ mesocuneiform facet mediolateral diameter, EntFmlϭ entocuneiform facet mediolateral diameter, EntFdpϭ entocuneiform facet dorsoplantar diameter, MaxLngϭ navicular maximum length, CuFdpϭ cuboid facet dorsoplantar diameter, CuFmlϭ cuboid facet mediolateral diameter, TalFlDpϭ depth of talar facet along major axis, TalTrDpϭ depth of talar facet along minor axis; CubEcto ϭ transverse cuboectocuneiform angle, EctMstrϭ transverse mesoectocuneiform angle, Torϭ navicular torsion, EctMsFnϭ frontal mesoectocuneiform angle, TalEctϭ sagittal taloectocuneifrom angle, EctEmtϭ entoectocuneiform angle.

10) more than likely re¯ect comparatively behaviors, which are considerably limited greater joint mobility. The relatively larger among mountain gorillas (Sarmiento, 1994; total navicular facet cross-sectional areas of Sarmiento et al., 1996). western gorillas when compared to those of Relative reduction in navicular facet area mountain gorillas (®g. 9, table 9) likewise with an increase in body size among humans re¯ect more mobile navicular joints in west- and great apes (®gs. 9 and 10; tables 9 and ern gorillas, and correlate well with arboreal 10) is in part a factor of relative decrease in AMNH NOVITATES Tuesday Dec 11 2001 10:13 AM 2000 novi 99146 Mp_23 Allen Press • DTPro System File # 01cc

2000 SARMIENTO AND MARCUS: OS NAVICULAR OF HOMINIDS 23

Fig. 13. Dendrograms of the studied taxa constructed using unweighted pair group method (Rohlf, 1997). Inset shows portion of dendrogram which differs when both Hadar naviculars are considered as a single sample.

strength (proportional to muscle cross-sec- use among females. Additionally, orangutan tional area) with increase in body size (a cor- and gorilla females may be expected to show ollary of volume), as predicted by the square quicker foot segment and body movements, cube law. With proportionately weaker mus- given proportionately larger joint surface ar- cles, large-sized animals are unable to sta- eas to dissipate kinetic energy than their cor- bilize rotational forces of the same relative responding adult males. The relatively small magnitude as small-sized animals, and must navicular joint surface areas of humans de- limit their range of joint motion (around a spite a considerably smaller body size than point of load equilibrium) to limit the mag- gorillas (tables 9 and 10) is the result of a nitude of rotational forces. The relatively marked commitment to bipedality and the larger facet areas of orangutan and gorilla fe- additional loading on the feet associated with males when compared to their male counter- bipedal behaviors. parts re¯ect such a trade-off between body Despite marked sexual dimorphism in size and joint mobility (®g. 9, table 9). In body weight, orangutans and gorillas of cor- terms of behavior, these differences are re- responding sexes have similar-sized navicu- vealed in a greater commitment to arboreal lar facet areas (®g. 9). This phenomenon is behaviors (Sarmiento, 1985, 1994; Sarmiento a correlate of sexual dimorphism and a pu- et al., 1996) and probably more variable foot bescent growth spurt which doubles body AMNH NOVITATES Tuesday Dec 11 2001 10:13 AM 2000 novi 99146 Mp_24 Allen Press • DTPro System File # 01cc

24 AMERICAN MUSEUM NOVITATES NO. 3288

Fig. 14. Plot of mean canonical variate scores for studied taxa. All fossils are based on single samples. The actual Mahalanobis D for all of the canonical variates separating taxa is given as the value above each connecting line (table 11). Connecting lines represent a Minimum Spanning Tree (after Rohlf, 1997). Owing to a two-dimensional projection, the actual lengths of the connecting lines on the plot represent only a fraction of the D values.

size without a corresponding increase in foot Likewise navicular facet size does not size. With nearly adult foot size and navic- seem to have a constant relationship to low- ular facet area to body weight ratios closer er-limb volume (®g. 11, table 10). With var- to their female counterparts, subadult male ied foot use and navicular mobility and load- orangutans and gorillas are also more arbo- ing, including direct transfer of weight by the real, and show more variable foot use than navicular to the substrate, African apes have when fully adult (Sarmiento, 1985, 1994; the relatively largest navicular facet areas (ta- personal obs.). ble 10). A much larger size ratio in African Considering that (1) facet cross-sectional apes than in humans is expected given our area is largely a factor of joint load and mo- hypertrophied lower limbs, limited navicular bility, and (2) neither of these variables in joint mobility, and a longitudinal arch that humans and great apes have a constant re- prevents navicular to substrate contact. How- lationship with body weight, predictions of ever, a comparable or larger ratio in African body weight in fossils from interspecies re- apes than in orangutans despite the orangu- gressions of navicular facet area have a large tan's small lower limbs underscores the ef- degree of error (®g. 9). fect of terrestrial loading and generalized AMNH NOVITATES Tuesday Dec 11 2001 10:13 AM 2000 novi 99146 Mp_25 Allen Press • DTPro System File # 01cc

2000 SARMIENTO AND MARCUS: OS NAVICULAR OF HOMINIDS 25

Fig. 15. Contributions of the entoectocuneiform and sagittal taloectocuneiform angles (w and v respectively) to the relative set of the ®rst through third metatarsal in close-packed position as seen in line drawings of the exploded left foot of a human (A) and of a gorilla (B) in medial view. The large sagittal taloectocuneiform angle in gorillas imparts a dorsi¯exed set to the third metatarsal and is associated with a dorsi¯exed talar head, i.e. small angle of talar neck inclination (Day and Wood 1968). The gorilla entoectocuneiform angle imparts a plantar set to the entocuneiform relative to the ectocuneiform and is associated with an abducted hallux, i.e. plantar divergence of the hallux relative to second (x) and third metatarsals (y). The human taloectocuneiform and entoectocuneiform angles are associated with a plantar ¯exed talar head (i.e., large angle of talar neck inclination), nearly aligned ®rst to third metatarsals, and a longitudinal plantar arch. Due to a ®xed transverse arch in humans, however, the long axis of the second and third metatarsals must have a more plantar inclination than the hallux, and the value of x and y are negative. Because the major axis of the navicular's talar facet is not necessarily held vertically, the sagittal taloectocuneiform and entoectocuneiform angles may also impart some degree of medial divergence to the hallux.

foot use on African ape total navicular facet NAVICULAR FUNCTION area. Although knowledge as to behavior and AND IMPLIED LOCOMOTOR BEHAVIORS lower-limb use could possibly be used to im- prove predictions of body weight based on The relatively small Mahalanobis distanc- facet size, such information is not always es between Hadar and African ape naviculars present in isolated fossil naviculars. (table 11) attest largely to a similar mor- AMNH NOVITATES Tuesday Dec 11 2001 10:13 AM 2000 novi 99146 Mp_26 Allen Press • DTPro System File # 01cc

26 AMERICAN MUSEUM NOVITATES NO. 3288

Fig. 16. Dorsal view of the exploded left tarsus and metatarsus of a pygmy chimpanzee in the close- packed position showing the contribution of the transverse mesoectocuneiform angle (w) and the transverse cuboectocuneiform angle (v) to the divergence of the second through fourth metatarsals. Correction of the talocuboid angle by the facet sets on the cuboid and ectocuneiform results in third and fourth metatarsals that are nearly alinged (y). A relatively low transverse mesoectocuneiform angle results in a second metatarsal that is divergent from the most lateral three (x) despite a partial correction of this set by the mesocuneiform.

phology and imply similarities in function. be used by some to argue that the Hadar fos- The exclusive association of a localized mor- sils represent knuckle-walkers. This interpre- phology and its implied function to a speci®c tation is at odds with evidence from the pel- locomotor behavior, as is routine practice vis and knee joint, which although poorly among paleoanthropologists (Napier, 1962; quanti®ed and never adequately tested, is al- Day and Napier, 1964; Robinson, 1972; Con- most unanimously taken as indicative of bi- roy and Fleagle, 1972; Stern and Susman, pedality (Lovejoy, 1974; 1978; Stern and 1983; Latimer et al., 1987; Latimer and Susman, 1983; Susman et al., 1984; McHen- Lovejoy, 1989, 1990a, b; Lovejoy, 1978, ry, 1991; Susman and Stern, 1991; Ohman 1988; McHenry, 1991; Leakey et al., 1995; et al., 1977; Fleagle, 1998). The Hadar na- Ohman et al., 1997; Ward et al., 1999), may viculars, however, come from a different ho- AMNH NOVITATES Tuesday Dec 11 2001 10:13 AM 2000 novi 99146 Mp_27 Allen Press • DTPro System File # 01cc

2000 SARMIENTO AND MARCUS: OS NAVICULAR OF HOMINIDS 27

Fig. 17. Contribution of navicular torsion (v) and the frontal mesoectocuneiform angle (w) to the relative set of the ectocuneiform, mesocuneiform, and the talar facet in the close-packed position as seen in line drawings of an exploded right foot of a human (A) and a gorilla (B) from a dorsodistal view. High values of the frontal mesoectocuneiform angle in humans (w) do not result in marked opposition of the second and third metatarsals given metatarsal, ectocuneiform, and mesocuneiform torsion values which correct for the imparted set. Despite similar torsion values in humans and gorillas, the metatarsal, ectocuneiform, and mesocuneiform torsion all contribute to causing more marked opposition of the second and third metatarsals in gorillas. Marked talar torsion or large frontal mesoectocuneiform angles, are also associated to a high transverse arch (see text).

rizon (AL 333w/333) and an earlier time pe- of the navicular articulation on the AL 288- riod than the Hadar pelvis (AL 288±1an,±ao) 1 talus suggest an African apelike navicular (Johanson et al., 1982), and the two remains similar to those from the AL 333/333w ho- may represent different taxa with different rizon. The contradictions in behavior implicit locomotor behaviors (Ferguson, 1984, 1986; in the navicular, pelvic, and knee joint mor- Senut and Tardieu, 1985; Olson, 1981, 1985; phology actually arise from the practice of Gommery, 1997). associating localized anatomy exclusively to At Hadar a contradiction in implied be- any one locomotor behavior and underscores havior, however, also arises when comparing the fallacy inherent in such a practice. Be- remains from the same horizon and may not cause behavioral changes during evolution necessarily be the result of sampling different (i.e., adaptive shifts) require localized struc- taxa. The alleged bipedal knee joint mor- tures to satisfy the mechanical requisites of phology is also seen in fossils from the same two or more behaviors (Darwin, 1859), ex- horizon as the naviculars (AL 333w-56, 333- clusive association of localized morphology 4, 333-111, 333x-26, 333-42), and the di- to any one locomotor behavior denies evo- mensions, curvature, set, and con®guration lutionary change. Arguments for bipedality AMNH NOVITATES Tuesday Dec 11 2001 10:13 AM 2000 novi 99146 Mp_28 Allen Press • DTPro System File # 01cc

28 AMERICAN MUSEUM NOVITATES NO. 3288

Fig. 18. Proximal views of the right cuboid and navicular of a human (A) and a common chimpanzee (B) in a close-packed position showing the contribution of the talocuboid angle (v) to the height of the transverse arch (h). In the close-packed position the small angle shown by chimpanzees results in a higher transverse arch than normally seen in humans. The human arch, however, unlike that of chimpanzees, exhibits a ®xed height. Owing to a small, irregular, and often absent cuboid facet, the human talocuboid angle could not be accurately measured.

based on localized morphology do not ex- morphology. No matter how bipedal, quadru- clude the likelihood that these morphologies pedal or climbing-like a localized morphol- and their implied localized functions also ogy may be, it is usually not possible to preserved other behaviors (table 12). The pres- dict on the basis of that morphology alone ence of many of the alleged bipedal charac- the overall morphology of an unknown fossil ters in cursorial quadrupeds and the mechan- taxon and the range of behaviors the un- ical advantages these characters impart to known morphology enabled. In this regard, quadrupeds, support the theory that many of functional analysis of the navicular alone is the allegedly bipedal human characters arose limited to the functional role the navicular as a result of selection pressures for cursorial may have had in the foot, and in a range of quadrupedality (Sarmiento, 1998). The ab- hypothesized behaviors usually based on sence of a humanlike lumbar lordosis, the those seen in living taxa. orientation of the acetabulum's lunate surface, and the relative size of the humeral and FOSSIL FOOT USE femoral midshaft circumferences and ®rst sa- OH 8 cral body cross-sectional area, all characters which are altered ontogenetically as a result In OH 8 the comparatively shallow talar of use, show that AL 288-1 had a body-to- facet, small sagittal taloectocuneiform and limb weight distribution and pelvic joint pos- entoectocuneiform angles and the large trans- tures similar to those of quadrupeds (Sar- verse mesoectocuneiform angle suggest the miento, 1998). presence of a ®xed longitudinal arch. The lat- Although it may be argued that the pelvis erally expanded tubercle midway on the lat- is phylogenetically more conservative than eral border of the navicular which serves as the navicular (see above) and thus the latter the attachment site for the calcaneonavicular is more likely to re¯ect use, while the former (spring) ligament, and the large articular area is more likely to re¯ect heritage, the Hadar for this ligament on the talar head indicate a pelvic, femoral, and navicular morphology if longitudinal plantar arch with energy-storing associated must have satis®ed the mechanical properties comparable to or approximating requisites of many of the same behaviors. In those of humans (Alexander, 1989). A dor- this regard, it is not possible to fully interpret si¯exed talar head and plantar processes on what the range of these behaviors may have the navicular and entocuneiform, however, been without analyzing all of the animal's suggest the longitudinal arch height was low- AMNH NOVITATES Tuesday Dec 11 2001 10:13 AM 2000 novi 99146 Mp_29 Allen Press • DTPro System File # 01cc

2000 SARMIENTO AND MARCUS: OS NAVICULAR OF HOMINIDS 29

er than is characteristic of humans, and the tionship between body size and tree use distal tarsal row contacted the substrate (Day (Cartmill and Milton, 1977; Sarmiento, 1983, and Wood, 1968). Given a ®xed longitudinal 1985, 1988, 1994, 1998; Cartmill, 1985), an arch, however, only a small percentage of the animal corresponding in size to the OH 8 weight borne by the foot could have been foot, regardless of its terrestrial specializa- transferred to the substrate by the navicular tions, was no doubt better suited to move ar- and entocuneiform, hence the relatively borealy than humans and or male gorillas small plantar tubercles. (Sarmiento, 1985, 1994, 1998). Although the small frontal talocuboid angle (table 2) indicates a high transverse pedal Hadar arch, the development of a hamulus and groove for the peroneus longus tendon on the When considered in light of the other Al plantar surface of the ectocuneiform indi- 333/333x pedal remains, the Hadar navicu- cates this arch was not ®xed. Considering lars do provide considerable insight into foot very little rotatory mobility at the talonavic- use and function. The large sagittal taloec- ular (as implied by the shallow talar facet tocuneiform angle, the large degree of dor- with different mediolateral and dorsoplantar soplantar talar facet curvature (table 1) and radii of curvature) and calcaneocuboid joints the large in¯ated navicular tuberosity indi- (Lewis 1980c), transverse arch mobility must cate that the Hadar foot lacked the longitu- have resulted mainly from movement at the dinal plantar arch characteristic of modern subtalar joint. humans. The navicular's large plantar tuber- The sagittal entoectocuneiform angle and osity and the overlying horizontal portion of the set of the talonavicular, naviculoentocu- the talar facet enabled talar head loads to be neiform, and the ®rst tarsometatarsal joints transmitted directly to the substrate through all indicate an adducted hallux which was the navicular. The well-developed plantar achieved through a different combination of process on the entocuneiform (AL 333-28) joint sets than that seen in humans (Day and and the plantar set of the naviculoectocunei- Wood, 1968; Oxnard and Lisowski, 1980; form facet suggest the entocuneiform also Oxnard, 1984). Although some degree of op- participated in transferring weight to the sub- posability may have been possible, depend- strate. Although the high values of the frontal ing on the unknown metatarsophalangeal mesoectocuneiform angle and low values of joint set, this would have been very limited. the frontal talocuboid angles (table 2) re¯ect The adducted hallux and large transverse some degree of a transverse pedal arch, the mesoectocuneiform angle re¯ect a compact presence of a plantar process ¯anked distally foot that was strongly committed to terres- by the peroneus longus groove on the AL triality. The relatively small cuboid facet 333-79 ectocuneiform suggests the trans- suggests the foot was usually loaded in pro- verse arch was either lower than that of hu- nated postures (with the foot's mediolateral mans or was mobile. Given a high and ®xed axis approximating the horizontal and its transverse arch, the peroneus longus tendon long axis in the plane of forward movement). bowstrings and does not groove the plantar The relatively large ectocuneiform facet (®g. surface of the ectocuneiform (Sarmiento, 3, tables 5 and 7) and small mesocuneiform 1994). The implied mobility at the talonavic- and entocuneiform facets (®gs. 4 and 5, ta- ular joint also indicates the Hadar foot did bles 5 and 7) indicate a preference for load- not have a ®xed transverse pedal arch. ing the lateral side of the foot at the expense The large cuboid facet (®g. 6 and tables 6 of the hallux. Differences relative to humans and 8) and mobility at the talonavicular joint in the orientation of the subtalar and talocru- suggest the Hadar foot was loaded in supi- ral joints (Lewis, 1980c) probably re¯ect dif- nated postures as is customary among great ferences in terrestrial foot use in both qua- apes when climbing vertical supports of rel- drupedal and bipedal behaviors, and may atively large diameter (Sarmiento, 1985. have also been important for proper foot ori- 1994). Considering the large, in¯ated tuber entation in vertical climbing or walking calcaneus (Latimer and Lovejoy, 1989), along horizontal branches. Given the rela- prominent plantar processes on the anterior AMNH NOVITATES Tuesday Dec 11 2001 10:13 AM 2000 novi 99146 Mp_30 Allen Press • DTPro System File # 01cc

30 AMERICAN MUSEUM NOVITATES NO. 3288 AMNH NOVITATES Tuesday Dec 11 2001 10:13 AM 2000 novi 99146 Mp_31 Allen Press • DTPro System File # 01cc

2000 SARMIENTO AND MARCUS: OS NAVICULAR OF HOMINIDS 31

calcaneus, navicular, and entocuneiform (this of the hallux in Hadar than is characterstic study), the dorsi¯exed set at the metacarpo- of humans. Because the degree of hallucal phalangeal joints, and the short phalanges abduction is de®ned by the set of the talo- (Stern and Susman 1983; Latimer and Love- navicular, naviculoentocuneiform, ®rst tar- joy, 1990a,b), the Hadar foot was much bet- sometatarsal and metatarsophalangeal joints ter suited for terrestrial plantigrade postures (Day and Wood, 1968; Lewis, 1980b, c; Sar- than for arboreal grasps. Substrate contact by miento, 1994), and these are not all present the distal tarsus indicates a human bipedal among the Hadar AL 333/333w remains, es- stride involving heel-to-ball weight transfer timates as to degree of hallucal abduction are could not have been commonly employed. equivocal. Regardless, a degree of hallucal The more divergent set of the second and abductability comparable to or greater than third rays as implied from the transverse me- that seen in mountain gorillas is a reasonable soectocuneiform angle and the mobile trans- estimate. verse arch suggests a less compact foot than The Hadar navicular is best interpreted as in modern humans, one which was not as belonging to the foot of a generalized quad- strongly committed to terrestrial behaviors. ruped which probably employed plantigrade Mobility at the talonavicular joint and the bipedal postures and limited its arboreal be- relative plantar orientation of the entocunei- haviors to supports of relatively large diam- form facet indicates more abducted postures eter (Sarmiento, 1989, 1991, 1998). AMNH NOVITATES Tuesday Dec 11 2001 10:13 AM 2000 novi 99146 Mp_32 Allen Press • DTPro System File # 01cc

32 AMERICAN MUSEUM NOVITATES NO. 3288

Oreopithecus inciding phylogenies and phenetics may be unexpected, especially when epigenetic and The large sagittal taloectocuneiform and inherited characters are given equal weight. entoectocuneiform angles and the relatively Early anatomists and systematists, however, small transverse mesoectocuneiform angle did not usually test the inferred homologies (table 1 and 2) indicate the Oreopithecus foot used to construct phylogenies for the likeli- had markedly divergent rays. The low navic- hood of parallelisms (Sarmiento, 1998). ular torsion with the strong medial inclina- Thus, these phylogenies are in essence phe- tion of the talar neck speci®cally indicate a netic trees, i.e., they equate degree of overall markedly divergent hallux. Among the hom- morphological similarity with the degree of inoids such a marked divergence is unique to relationship. Without the morphological Oreopithecus and is corroborated when the complexity necessary to test for homologies, associated tarsal and metatarsal bones are ar- quanti®cation of overall similarities in frag- ticulated (Kohler and Moya-Sola, 1997). A mentary fossils may be the only analysis pos- shallow talar facet with small and approxi- sible. Results from such an analysis, how- mately equal radii of curvature (table 4) cor- ever, do not accurately re¯ect phylogenetic responds to a loose talonavicular joint with af®nities, since they con¯ate shared derived rotatory ability. Such joint laxity is also re- characters (synapomorphies) and parallel- ¯ected in the calcaneo-cuboid joint (Sar- isms (plesiomorphies) and do not distinguish miento, 1987; Szalay and Langdon, 1987) inherited characters from those originating and in the single continuous navicular facet ontogenetically with use. on the talar head. A talar facet which is large Through cladistic analysis and knowledge relative to the cuneiform facets (®g. 8) sug- as to what navicular characters are primitive gests an emphasis on mobility at the talona- vs. derived, testing for homologies may be vicular relative to the naviculocuneiform possible. For instance, a weight-bearing plan- joints. The absence of weight-bearing tuber- tar tubercle with a variable sustentacular ar- cles on the navicular and entocuneiform ticulation is a shared derived character of the (Sarmiento, 1987) suggest that the foot African ape navicular absent in other catar- lacked a commitment to terrestrial behaviors rhines and the earliest known fossil homi- and/or was not often used for walking along noids (i.e. Oreopithecus and Sivapithecus; large diameter horizontal supports. The large Sarmiento, 1994, this study). The presence of cuboid facet (®g. 6 and table 6) is in accord a large weight-bearing plantar tubercle and a with powerful hallucal opposability and a sustentacular facet on the navicular of Hadar foot loaded in supinated postures. forms may be used to argue that many of the The Oreopithecus navicular best corre- angles and metric characters shared by Hadar sponds to the foot of an arboreal vertical and African apes are also homologous. As climber. A relatively mobile foot with a wide such, the Hadar naviculars would seem to opposable grasp, this foot could be apposed share a special relationship with African apes against vertical trunks when climbing, or exclusive of humans and orangutans. As with used along horizontal supports of diameters all cladistic analyses this argument assumes permitting hallucal grasps. orthoselection with minimal reversals. In this case, the assumptions are that the large plan- PHYLOGENY IMPLICIT IN NAVICULAR tar tubercle and sustentacular facet was not MEASUREMENTS (1) independently acquired by chimpanzees, It is striking how close the phenetic tree gorillas and/or Hadar fossils, (2) a shared based on the Mahalanobis distances between trait of humans and African apes which was samples of human and great ape navicular later lost in humans, or (3) a shared hominoid measurements (®g. 13) coincide with homi- trait that was independently lost in humans, noid phylogenies as hypothesized by early orangutans, and Oreopithecus. With relative- anatomists (Keith, 1916, 1934, 1940; Morton ly few comparative taxa as outgroups and a 1927; Schultz, 1936; Le Gros Clark, 1971). limited number of shared derived characters Because navicular measurements were cho- for comparison, such an analysis may prove sen to re¯ect largely functional concerns, co- equivocal. In most cases more than a bit of AMNH NOVITATES Tuesday Dec 11 2001 10:13 AM 2000 novi 99146 Mp_33 Allen Press • DTPro System File # 01cc

2000 SARMIENTO AND MARCUS: OS NAVICULAR OF HOMINIDS 33

localized morphology is necessary for phy- a decreasing Mahalanobis distance to human logenetic resolution (Sarmiento, 1987). naviculars. Such a sequence of evolutionary Regardless of its inaccuracy, navicular changes in the foot are in accord with those phenetics and the cladistic analysis that can predicted in human phylogeny by current be done on the limited number of navicular models of hominoid and hominid divergence characters does re¯ect phylogeny to some (Sarmiento, 1995, 1998). degree. In this regard, there must be a limit Results from this study inspire caution as as to how much the naviculars of closely re- to how much of an animal's phylogeny or lated forms may differ from each other and/ overall behavior can be interpreted from a or how similar the navicular of more distally single bone. Phylogenies are unlikely to be related forms may be. Similarities in inher- accurate without the morphologic complexity ited anatomy predispose the types of prob- necessary to test for homologies, and without lems an organism encounters in its environ- at least a limited understanding of character ment and the behaviors used to solve these polarity. When interpreting either phylogeny problems. Differences in inherited anatomy, and/or function from fossil remains, incom- on the other hand, reduce the likelihood of plete remains are likely to lead to spurious encountering similar problems in the envi- conclusions. Interpretations of australopithe- ronment or arriving at similar solutions. Even cine systematics and behavior, therefore, can if the same problems are encountered and only be credible when all of the morpholog- similar solutions arrived at, inherited differ- ical evidence is accounted for. In retrospect, ences are more apt to result in greater dif- the notion that the australopithecine gait ferences in navicular morphology in more could have been predicted based on a single distally related taxa than in closely related os coxa (Sts 14) and some nonassociated ones. fragmentary femora (Lovejoy, 1974; see also Notably navicular phenetics also approxi- Johanson et al., 1976) is supercilious. mates the currently accepted hominoid phylogeny (Sarmiento, 1998). The only dis- agreement exists at the point of human di- ACKNOWLEDGMENTS vergence, and humans have a decidedly spe- cialized foot committed to terrestrial I thank the following for access to and as- bipedality (Weidenreich, 1922; Morton, sistance with comparative collections: W. 1924; Schultz 1963; Sarmiento, 1994, 1998). Van Neer, Royal African Museum Tervuren; The close similarities in the navicular of OH R. Thorington, National Museum of Natural 8 and humans, and of Hadar australopithe- History, Washington D.C.; O. GroÈndwall, cines and African apes, and the unique na- Swedish Museum of Natural History, Stock- vicular of Oreopithecus must re¯ect to some holm; Malcom Harmon, Powell Cotton Mu- degree phylogenetic af®nities. This applies seum, Birchington Kent; Maria Rutzmoser, whether or not the measured characters also Museum of Comparative Zoology, Cam- re¯ect functional concerns. As regards the bridge Massachusetts; and J. Kerbis-Peter- navicular, the AL 333/333w remains are Hans, Field Museum of Natural History, Chi- more likely to represent ancestral African cago. For access to fossil material I am grate- apes than ancestral humans. The measured ful to: Solomon Woerde Kal of the National differences between OH 8, Hadar, and Or- Museum of Ethiopia, Addis Ababa; M. L. eopithecus naviculars are consistent with ge- Mbago of the National Museums of Tanza- neric differences among humans and living nia, Dar es Salaam; M. Mazzinni, Istituto di great apes. Geologia, University of Florence; B. Enges- ser and the late J. Hurzeler, Natural History CONCLUSIONS Museum, Basel. The senior author would also like to extend a special thanks to M. The Oreopithecus, Hadar, and OH 8 na- Kohler and S. Moya-Sola for their gracious viculars, in order of decreasing geologic age, hopsitality during a visit to Sabadell to ex- show decreasing hallucal abduction, increas- amine Oreopithecus remains in their care. M. ing commitment to terrestrial behaviors, and D. Rose and F. J. Rohlf reviewed an earlier AMNH NOVITATES Tuesday Dec 11 2001 10:13 AM 2000 novi 99146 Mp_34 Allen Press • DTPro System File # 01cc

34 AMERICAN MUSEUM NOVITATES NO. 3288

version of this manuscript and provided help- ural selection or the preservation of fa- ful suggestions for its improvement. vored races in the struggle of life. New York: Modern Library. Day, M. H., and J. R. Napier REFERENCES 1964. Fossil foot bones. Nature 201: 969± Abitbol, M. M. 970. 1988. Evolution of the ischial spine and of the Day, M. H., and B. A. Wood pelvic ¯oor in the Hominoidea. Am. J. 1968. Functional af®nities of the Olduvai Phys. Antropol. 75: 53±67. hominid 8 talus. Man 3: 440±455. Alexander, R. McN. Edelstein, S. J. 1989. Elastic mechanisms in animal move- 1987. An alternative paradigm for hominoid ment. New York: Cambridge Univ. evolution. Hum. Evol. 2: 169±174. Press. Elftman, H. Beard, K. A., M. Dagosto, D. Gebo and M. God- 1960. The transverse tarsal joint and its con- inot trol. Clin. Orthop. 16: 41±46. 1988. Interrelationship among primate higher Elftman, H. and J. Manter taxa. Nature 331: 712±714. 1935. The evolution of the human foot, with Biegert, J. special reference to the joints. J. Anat. 1963. The evaluation of characteristics of the 70: 56±67. skull, hands and feet for primate tax- Ferguson, W. W. onomy. In S. L. Washburn (ed.), Clas- 1984. Revision of fossil hominid jaws from si®cation and human evolution: 116± the Plio/pleistocene of Hadar, in Ethi- 145. Chicago: Aldine. opia, including a new species of the ge- Biegert, J., and R. Mauer nus Homo (Hominoidea; Homininae). 1972. Rumpskelettlange, Allometrien und Primates 25: 519±529. Korperproportionen bie catarrhinen Pri- 1986. The taxonomic status of Praeanthropus maten. Folia Primatol. 17: 142±156. africanus (Primates: Pongidae) from Broom, R., and J. T. Robinson the late Pliocene of eastern Africa. Ibid. 1950. Notes on the pelves of the fossil ape- 27: 485±492. man, Pleisianthropus. Am. J. Phys. An- Fleagle, J. G. thropol. 8: 489±494. 1998. Primate adaptation and evolution. 2nd Carrano, M. T. ed. New York: Academic Press. 1997. Morphological indicators of foot pos- Gebo, D. L. ture in mammals, a statistical and bio- 1986. The anatomy of the prosimian foot and mechanical study. Zool. J. Linn. Soc. its application to the primate fossil re- 12: 77±104. cord. Ph.D. Diss. Duke Univ., Durham. Cartmill, M. Gebo, D. L., and M. Dagosto 1985. Climbing. In M. Hildebrand, D. M. 1988. Foot anatomy, climbing and the origin Bramble, K. F. Leim and D. B. Wake of the Indriidae. J. Human Evol. 17: (eds.), Functional vertebrate morphol- 135±154. ogy: 73±88. Cambridge: Belknap Press. Gommery, D. Cartmill, M., and K. Milton 1997. Le atlas et les axis des Hominides du 1977. The lorisform wrist joint and the evo- Plio-pleistocene: morphologie et syste- lution of brachiating adaptions in the matique. C.R. Acad. Sci. Paris 325: Hominoidae. Am. J. Phys. Anthropol. 639±642. 47: 249±272. Heiple, K. G., and C. O. Lovejoy Chamberlain, A. T. and B. A. Wood 1971. The distal femoral anatomy of Austral- 1987. Early Hominid phylogeny. J. Hum. opithecus. Am. J. Phys. Anthropol. 35: Evol. 16: 119±133. 75±84. Clarke, R. J., and P. V. Tobias Hill, W.C.O. 1995. Sterkfontein member 2 foot bones of 1960. Primates: comparative anatomy and the oldest South African hominid. Sci- taxonomy. IV. Cebidae, Pt. A. Edin- ence 269: 521±524. burgh: Edinburgh Univ. Press. Conroy, G. C. and J. G. Fleagle Johanson, D. C., C. O. Lovejoy, K. G. Heiple, and 1972. Locomotor behaviors in living and fos- A. H. Burstein sil pongids. Nature 237: 103±104. 1976. Biomechanical implications of the Afar Darwin, C. knee joint. Am. J. Phys. Anthropol. 44: 1859. The origin of species by means of nat- 188. AMNH NOVITATES Tuesday Dec 11 2001 10:13 AM 2000 novi 99146 Mp_35 Allen Press • DTPro System File # 01cc

2000 SARMIENTO AND MARCUS: OS NAVICULAR OF HOMINIDS 35

Johanson, D. C., M. Taieb, and Y. Coppens Leakey M. G., C. S. Feibel, I. McDougall, and A. 1982. Pliocene hominids from the Hadar for- Walker mation, Ethiopia (1973±1977): strati- 1995. New four-million-year-old hominid graphic, chronologic and paleo-envi- species from Kanapoi and Allia bay, ronmental contexts, with notes on hom- Kenya. Nature 376: 565±571. inid morphology and systematics. Am. Le Gros Clark, W. E. J. Phys. Anthropol. 57: 373±402. 1955. The os inominatum of the recent Pon- Jolly, C. J. gidae with special reference to that of 1965. The origins and specializations of the the Australopithecinae. Am. J. Phys. long-faced Cercopithecoidea. Ph.D. Anthropol. 13: 19±27. diss., Univ. of London. 1971. The antecedents of man. New York: 1970. The seed eaters: a new model of hom- Quadrangle Press. Lewis, O. J. inid differentiation based on baboon 1969. The hominoid wrist joint. Am. J. Phys. analogy. Man 5: 5±26. Anthropol. 30: 251±267. Keith, A. 1974. The wrist articulations of the Anthro- 1916. Lo schema dell'origine umana. Revista poidea. In F. A. Jenkins (ed.), Primate di Antropologia 20: 3±20. locomotion: 143±169. New York: Ac- 1934. The construction of man's family tree. ademic Press. London: Watts. 1980a. The joints of the evolving foot. Pt. I. 1940. Fifty years ago. Am. J. Phys. Anthro- The ankle joint. J. Anat. 130: 527±543. pol. 26: 251±267. 1980b. The joints of the evolving foot. Pt II. Kohler, M., and S. J. Moya-Sola The intrinsic joints. Ibid.: 833±857. 1997. Ape-like or hominid-like? The posi- 1980c. The joints of the evolving foot. Pt. III. tional behavior of Oreopithecus bam- The fossil evidence. Ibid. 131: 275± bolii reconsidered. Proc. Natl. Acad. 298. Sci. 94: 11747±11750. Lovejoy, C. O. Latimer, B. 1974. The gait of australopithecines. Yearb. 1991. Locomotor adaptations in Australopi- Phys. Anthropol. 17: 147±161. thecus afarensis: the issue of arboreal- 1978. A biomechanical review of the loco- ity. In Y. Coppens and B. Senut (eds.), motor diversity of early hominids. In C. Origine(s) de la bipedie ches les hom- J. Jolly (ed.), Early Hominids of Africa, inides: 169±176. Paris: CNRS. 403±429. New York: St. Martins Press. Latimer, B., and C. O. Lovejoy 1979. A reconstruction of the pelvis of AL- 1989. The calcaneus of Australopithecus 288 (Hadar Formation, Ethiopia). Am. afarensis and its implication for the J. Phys. Anthropol. 50: 460. evolution of bipedality. Am. J. Phys. 1988. Evolution of human walking. Sci. Am. Anthropol. 78: 369±386. 259: 118±125. 1990a. Hallucal tarsometatarsal joint in Aus- Lovejoy, C.O., K.G. Heiple, and A.H. Burstein tralopithecus afarensis. Ibid. 82: 125± 1973. The gait of Australopithecus. Am. J. 133. Phys. Anthropol. 32: 33±40. Lovejoy, C.O., T. D. White, M. J. Cohn, and J. C. 1990b. Metatarsophalangeal joints of Austral- Ohman opithecus afarensis. Ibid. 83: 13±23. 1996. The Maka femur: A case study in the Latimer, B., C. O. Lovejoy, D. C. Johanson, and application of analytical paradigms dic- Y. Coppens tated by new developments in morpho- 1982. Hominid tarsal, metatarsal and phalan- logical ontogeny. Am. J. Phys. Anthro- geal bones recovered from the Hadar pol., Suppl. 22: 152. formation: 1974±1977 collections. Am. MacLatchy, L. M. J. Phys. Anthropol. 57: 701±719. 1996. Another look at the Australopithecine Latimer, B., J. C. Ohman, and C. O. Lovejoy hip. J. Hum. Evol. 31: 455±476. 1987. Talocrural joint in African hominoids: Manter, J. T. implications for Australopithecus afar- 1941. Movements of the subtalar and trans- ensis. Am. J. Phys. Anthropol. 74: 155± verse tarsal joints. Anat. Rec. 80: 397± 175. 410. Leakey, L., P. V. Tobias, and J. Napier Marcus, L. F. 1964. A new species of the genus Homo from 1990. Traditional morphometrics. In F. J . Olduvai Gorge. Nature 202: 7±10. Rohlf and F. L. Bookstein (eds.). Pro- AMNH NOVITATES Tuesday Dec 11 2001 10:13 AM 2000 novi 99146 Mp_36 Allen Press • DTPro System File # 01cc

36 AMERICAN MUSEUM NOVITATES NO. 3288

ceedings of the Michigan morphomet- In C. B. Stringer (ed.), Aspects of hu- rics workshop: 77±112. Univ. Michigan man evolution: 99±128. London: Tay- Mus. Zool. Spec. Publ. lor and Francis. 1993. Some aspects of multivariate statistics 1985. Cranial morphology and systematics of for morphometrics. In L. F. Marcus, E. the Hadar fossil hominids and A. afri- Bello, and A. G. Valdecasas, (eds.), canus. In E. Delson and I. Tattersall Contributions to morphometrics: 99± (eds.), Ancestors: the hard evidence: 130. Madrid: Monogra®as Museo Nac- 102±119. New York: Alan R. Liss. cional Ciencias Naturales. Oxnard, C. E. McHenry, H. M. 1984. The Order of Man. New Haven, CT: 1991. First steps? Analyses of the postcran- Yale Univ. Press. ium of early hominids. In Y. Coppens Oxnard, C. E., and F. P. Lisowski and B. Senut (eds.), Origine(s) de la bi- 1980. Functional articulation of some homi- pedie chez les hominides: 133±141. noid foot bones: implications for the Paris: CNRS. Olduvai (Hominid 8) foot. Am. J. Phys. 1996. Homoplasy, clades and hominid phy- Anthropol. 52: 107±117. logeny. In W. E. Meikle, F. C. Howell, Rafferty, K. A. and N. G. Jablonksi (eds.), Contempo- 1998. Structural design of the femoral neck in rary issues in human evolution: 77±91. primates. J. Hum. Evol. 34: 361±383. San Francisco: California Acad. Sci. Reynolds, E. McHenry, H. M., R. S. Corruccinni, and F. C. 1931. The evolution of the human pelvis in Howell relation to the mechanics of erect pos- 1975. Analysis of an early hominid ulna from ture. Pap. Peabody Mus. Archaeol. Eth- the Omo basin, Ethiopia. Am. J. Phys. nol. Harvard Univ. 11: 249±334. Anthropol. 44: 295±304. Robinson, J. T. Mednick, L. W. 1965. Homo habilis and the australopithe- 1955. The evolution of the human ilium. Am. cines. Nature 205: 121±124. J. Phys. Anthropol. 13: 203±216. 1972. Early hominid posture and locomotion. Meldrum, D. J., M. Dagosto, and J. White Chicago: Univ. Chicago Press. 1997. Hindlimb suspension and hind foot re- Rohlf, F. J. versal in Varecia variegata and other 1997. NTSYSpc version 2.02h. Setauket, NY: arboreal mammals. Am. J. Phys. An- Exeter Software. thropol. 103: 85±102. Rose, M. D. Morton, D. J. 1988. Functional anatomy of the cheiridia. In 1924. Evolution of the human foot. II. Am. J. J. H. Schwartz (ed.), Orangutan biolo- Phys. Anthropol. 7: 1±52. gy: 509±524. London: Plenum Press. 1927. Human origin. Am. J. Phys. Anthropol. Sarmiento, E. E. 10: 173±203. 1983. The signi®cance of the heel process in Napier, J. H. anthropoids. Int. J. Primatol. 4(2): 127± 1962. Fossil hand bones from Olduvai Gorge. 152. Nature 196: 409. 1985. Functional differences in the skeleton Ohman, J. C., T. J. Krochta, C. O. Lovejoy, R. P. of wild and captive orang-utans and Mensforth, and B. Latimer their adaptive signi®cance. Ph.D. diss., 1997. Cortical bone distribution in the femo- New York Univ., New York. ral neck of hominoids: implications for 1987. The phylogenetic position of Oreopi- the locomotion of Australopithecus thecus and its signi®cance in the origin afarensis. Am. J. Phys. Anthropol. 104: of the Hominoidea. Am. Mus. Novita- 117±131. tes 2881: 44 pp. Olson, T. R. 1988. Anatomy of the hominoid wrist joints, 1978. Hominid phylogenetics and the exis- its evolutionary and functional impli- tence of Homo in Member I of the cations. Int. J. Primatol. 9: 281±345. Swartkrans formation, South Africa. J. 1991. Functional and phylogenetic implica- Hum. Evol. 7: 159±178. tions of the differences in the pedal 1981. Basicranial morphology of the extant skeleton of Australopithecines. Am. J. hominoids and Pliocene hominids: the Phys. Anthropol., Suppl. 12: 232. new material from the Hadar forma- 1994. Terrestrial traits in the hands and feet tion, Ethiopia, and its signi®cance in of gorillas. Am. Mus. Novitates 3091: early human evolution and taxonomy. 56 pp. AMNH NOVITATES Tuesday Dec 11 2001 10:13 AM 2000 novi 99146 Mp_37 Allen Press • DTPro System File # 01cc

2000 SARMIENTO AND MARCUS: OS NAVICULAR OF HOMINIDS 37

1995. Cautious climbing and folivory: a mod- Coppens and B. Senut (eds.), Origi- el of hominoid differentiation. Hum. ne(s) de la bipedie chez les hominides: Evol. 10(4): 289±321. 121±131. Paris: CNRS. 1998. Generalized quadrupeds, committed bi- Susman, R. L., J. T. Stern, Jr., and W. L. Jungers peds and the shift to open habitats: an 1984. Arboreality and bipedality in the Hadar evolutionary model of hominid diver- hominids. Folia Primatol. 43: 113±156. gence. Am. Mus. Novitates 3250: 78 Szalay, F. S., and R. L. Decker pp. 1974. Origin, evolution and function of the Sarmiento, E. E., and T. M. Butynski tarsus in late Cretaceous eutherians and 1996. Present problems in gorilla taxonomy. Paleocene primates. In F. Jenkins (ed.), Gorilla J. 12: 5±7. Primate locomotion: 223±259. New In prep. The foot musculature of Bwindi go- York: Academic Press. rillas: its functional and phylogenetic Szalay, F. S., and J. H. Langdon signi®cance. 1987. The foot of Oreopithecus: an evolution- Sarmiento, E. E., T. M. Butynski, and J. Kalina ary assessment. J. Hum. Evol. 15: 585± 1996. Gorillas of Bwindi-Impenetrable Forest 621. and the Virunga Volcanoes: taxonomic Tardieu, C., and H. Preuschoft implications of morphological and eco- 1996. Ontogeny of the knee joint in humans, logical differences. Am. J. Primatol. great apes and fossil hominids: Pelvi- 40: 1±21. femoral relationships during postnatal Schaeffer, B. growth in humans. Folia Primatol. 66: 1947. Notes on the origins and function of the 68±81. artiodactyl tarsus. Am. Mus. Novitates Tobias, P. V. 1356: 24 pp. 1991a. Olduvai Gorge, Vol. 4: the skulls en- Schultz, A. H. docasts and teeth of Homo habilis. 1930. The skeleton of the trunks and limbs of Cambridge: Cambridge Univ. Press. higher primates. Hum. Biol. 2: 303± 438. 1991b. Relationships between apes and hu- 1936. Characters common to higher primates mans. In A. Sahni and R. Gaur (eds.), and characters speci®c to man. Q. Rev. Perspectives in human evolution: 1±19. Biol. 11: 259±283; 425±455. Delhi: Renaissance Publishing House. 1963. Relations between the lengths of the Tuttle, R. H. main part of the foot skeleton in pri- 1970. Postural propulsive and prehensile ca- mates. Folia Primatol. 1: 150±171. pabilities in the cheiridia of chimpan- Senut, B., and C. Tardieu zees and other great apes. In G. H. 1985. Functional aspects of hominid limb Bourne (ed.), The chimpanzee. 2: 167± bones: implication for taxonomy and 253. Basel: Karger. phylogeny. In E. Delson and I. Tatter- Verhaegen, M. sall (eds.), Ancestors: the Hard Evi- 1990. African ape ancestry. Hum. Evol. 5: dence: 175±203. New York: Alan R. 295±297. Liss. 1994. Australopithecines: ancestors of the Af- Skelton, R. R., H. M. McHenry, and G. M. Draw- rican apes? Ibid. 7: 63±64. horn Walker, A.C. 1986. Phylogenetic analysis of early homi- 1973. New Australopithecus femora from nids. Curr. Anthropol. 27: 21±43. East Rudolf Kenya. J. Hum. Evol. 2: Stern, J. T., and R. L. Susman 545±555. 1983. The locomotor anatomy of Australopi- Ward, C.V., M.G. Leakey, B. Brown, F. Brown, J. thecus africanus. Am. J. Phys. Anthro- Harris and A. Walker pol. 60: 279±316. 1999. South Turkwel: A new Pliocene homi- Strait, D.S., F. E. Grine, and M. A. Moniz nid site in Kenya. J. Hum. Evol. 36: 1997. A reappraisal of early hominid phylog- 69±95. eny. J. Hum. Evol. 37: 17±82. Waterman, H. C. Straus, W. L., Jr. 1927. Studies on the evolution of the pelves 1929. Studies on the primate ilia. Am. J. of man and other primates. Bull. Am. Anat. 43: 403±460. Mus. Nat. Hist. 11: 585±641. Susman, R. L., and J. T. Stern Weidenreich, F. 1991. Locomotor behavior of early hominids: 1922. Der Menschenfuss. Z. Morphol. An- epistomology and fossil evidence. In Y. thropol. 22: 51±282. AMNH NOVITATES Tuesday Dec 11 2001 10:13 AM 2000 novi 99146 Mp_38 Allen Press • DTPro System File # 01cc

38 AMERICAN MUSEUM NOVITATES NO. 3288

White, T. D., D. C. Johanson, and W. H. Kimbel Zihlman, A. L. 1981. Australopithecus africanus: its phylo- 1969. Human locomotion: a reappraisal of the genetic position reconsidered. S. Afr. J. functional and anatomical evidence. Sci. 77: 445±470. Ph.D. diss., Univ. California at Berke- Wolpoff, M. H. ley. 1974. The evidence of two australopithecine 1978. Interpretations of early hominid loco- lineages in South Africa. Yearb. motion. In C. J. Jolly (ed.), Early Hom- Phys.Anthropol. 17: 113±139. 1999. Paleoanthropology. New York: Mc- inids of Africa: 361±377. New York: St Graw Hill. Martins Press. Wood, B. A. Zihlman, A. L., and L. Bunker 1974. Olduvai Bed I postcranial fossils; a re- 1979. Hominid bipedalism: then and now. assessment. J. Hum. Evol. 3: 373±378. Yearb. Phys. Anthropol. 22: 132±162. AMNH NOVITATES No RH this page!! Tuesday Dec 11 2001 10:13 AM 2000 novi 99146 Mp_40 Allen Press • DTPro System File # 01cc

Recent issues of the Novitates may be purchased from the Museum. Lists of back issues of the Novitates and Bulletin published during the last ®ve years are available at World Wide Web site http://nimidi.amnh.org. Or address mail orders to: American Museum of Natural History Library, Central Park West at 79th St., New York, NY 10024. TEL: (212) 769-5545. FAX: (212) 769- 5009. E-MAIL: [email protected]

a This paper meets the requirements of ANSI/NISO Z39.48-1992 (Permanence of Paper).