humans. Richmond and Strait1 discount thepossibility that knuckle-walking evolvedindependently in gorillas and in a commonhuman–chimpanzee ancestor, an idea sup-ported by the different ontogenetic develop-ment of gorilla and chimpanzee carpusload-bearing morphology2 and by terrestrialorang-utans who normally walk on the dor-sum of their fingers, albeit in a less efficient3

fist-walking posture4,5, and occasionallyknuckle-walk6,7. The entire great-ape lineagemay thus be predisposed towards knuckle-walking when terrestrial quadrupedalismbecomes a major selective factor6,7, under-mining the claim that the four cited traitscan resolve the phylogenetic relationships ofchimpanzees, humans and gorillas1.

Richmond and Strait1 assume thatbecause some features of the present-daydistal radius fulfil one of the mechanicalrequisites of African-ape knuckle-walking(limiting extension at the radiocarpal joint),then they must have arisen in response tothe requirements of knuckle-walking. But itis debatable whether form, function andadaptive significance should be so linkedwithin a modern evolutionary frame-work8,9. Claims for knuckle-walking adapta-tions in the wrist need careful evaluation asthis behaviour describes a novel posture ofthe metacarpals and phalanges in relationto a substrate4,5. Digitigrade, semidigiti-grade or fist-walking postures may makesimilar demands on the wrist joints toknuckle-walking4 and so could also havegiven rise to stabilizing and extension-limit-ing mechanisms at the radiocarpal joint.

Extension-limiting structures allow con-tact between the hand and substrate to bemaintained more efficiently when climbinglarge vertical supports4 and so may havearisen in a climbing ancestor of humans andgreat apes10. From this perspective, Rich-mond and Strait1 only demonstrate thatbehaviours requiring limited extension at theradiocarpal joint were important in hominidancestry. Methodological considerations,however, suggest that the proposed exten-sion-limiting mechanism of the distal radiusmay not have been suitably quantified.

A clearly delimited scaphoid notch is notalways evident in the hominoid speciesincluded in the analysis1, so the measure-ments may not be replicable. It is unclearwhy the angle between the scaphoid andlunate facets of the radius may be signifi-cant in quantifying the extension-limitingmechanisms at this joint. This dimensionshould discriminate between African andAsian apes, with a narrower angle in the lat-ter. The greater radio-ulnar curvature of thedistal radial articulation in orang-utans andgibbons is atypical for anthropoid primates,and probably facilitates the rotary move-ment of the proximal carpal joint neededfor forearm suspensory locomotion4,11. Butit is not explained why increased rotary

movement should preclude an extension-limiting mechanism at the radiocarpaljoint1. Part of the discrimination betweenAfrican apes, Asian apes and hominids(their Fig. 2) may therefore reflect differ-ences in suspensory behaviour and havenothing to do with knuckle-walking.

The first two canonical axes of Fig. 2b ofref. 1 do not separate the A. afarensis radiifrom orang-utans any more than fromchimpanzees. The variables that stronglycontribute to the remaining axis are notidentified, so it is unclear why theMahalonobis D2 distance is greater betweenorang-utans and A. afarensis than betweenchimpanzees and A. afarensis (their Fig. 2c,based on all canonical axes), a differencethat bears on the conclusion that A. afaren-sis had a knuckle-walking ancestry.

Although hominids may have evolvedfrom a knuckle-walker, I do not believe thatRichmond and Strait’s analysis1 has provedthis. I cannot, therefore, support relatedinterpretations regarding the phylogeneticaffinity1 and locomotor mode12 of earlyhominids, or of the chimpanzee-like char-acteristics of their immediate ancestor1.Mike DaintonDepartment of Human Anatomy and Cell Biology,New Medical School, University of Liverpool,Ashton Street, Liverpool L69 3GE, UKe-mail: mdainton@liverpool.ac.uk1. Richmond, B. G. & Strait, D. S. Nature 404, 382–385 (2000).

2. Dainton, M. & Macho, G. A. J. Hum. Evol. 36, 171–194 (1999).

3. Tuttle, R. H. J. Morphol. 128, 309–364 (1969).

4. Sarmiento, E. E. Int. J. Primatol. 9, 281–345 (1988).

5. Tuttle, R. H. Am. J. Phys. Anthropol. 26, 171–206 (1967).

6. Tuttle, R. H. (ed.) in Primate Functional Morphology and

Evolution 203–212 (Mouton, The Hague, 1975).

7. Tuttle, R. H. & Beck, B. B. Nature 236, 33–44 (1972).

8. Gould, S. J. & Vrba, E. S. Paleobiology 8, 4–15 (1982).

9. Lauder, G. V. in Functional Morphology in Vertebrate

Paleontology (ed. Thomason, J. J.) 1–18 (Cambridge Univ.

Press, Cambridge, 1995).

10.Sarmiento, E. E. Hum. Evol. 10, 289–321 (1995).

11. Jenkins, F. A. Jr & Fleagle, J. G. in Primate Functional

Morphology and Evolution (ed. Tuttle, R. H.) 213–221 (Mouton,

The Hague, 1975).

12.Collard, M. & Aiello, L. C. Nature 404, 339–340 (2000).

Richmond and Strait argue that evidence ofancestral knuckle-walking is retained in thedistal radius of Australopithecus1, basing theirconclusions on a canonical variate analysisthat uses four measurements of the distalradial joint surface in higher primates. Wedispute their claim that early hominidsretained a specialized wrist morphologyassociated with knuckle-walking and ques-tion the biological and mechanical interpre-tations they use to attribute knuckle-walkingfeatures to the radius of Australopithecus.

The authors believe that in African apesthere is a mechanism that ‘locks’ the wristinto place, which they base on their infer-ence that the scaphoid conforms with a dis-tally facing notch in the dorsal joint marginof the radius in full extension. They suggestthat this ‘locking mechanism’ acts by limit-ing wrist extension and argue that it exists

brief communications

NATURE | VOL 410 | 15 MARCH 2001 | www.nature.com 325

because chimpanzees do not consistentlyrecruit the major flexor muscles of the digitsduring knuckle-walking.

We question these conclusions on sever-al grounds. First, very little motion occursat the radioscaphoid joint — Richmondand Strait’s X-ray evidence shows that ittakes place almost entirely at the midcarpaljoint. All synovial joints conform through-out their range of motion, and would beexpected to do so in full flexion. Mostimportant, synovial joint stability derivesfrom its muscles and ligaments2, not fromits essentially frictionless joint surfaces: theprimary role of joint-surface geometry is toassure that the joint reaction force remainsorthogonal to rigid-body contact surfaces,and that velocity vectors remain tangentialthroughout the range of motion3.

Neither do normal joints lock. Lockingwould prevent motion and thus the perfor-mance of cartilage-sparing, negative workby the joint’s soft-tissue envelope. Cartilagecongruity maintains joint reaction forcenormality throughout the flexion–exten-sion cycle, especially under high externalloads. Neither peripheral nor edge loadingoccurs in normal, stable joints3. Further,neither the location nor direction of thejoint reaction force can be determined froman X-ray image — comprehensive free-body analysis is required.

Finally, tendons and fascial sheaths arecrucial to joint stability, particularly those ofthe long digital flexors4. An absence of elec-tromyographic activity suggests that passiverecruitment of the joint capsule and themuscle’s connective-tissue envelope are allthat is required for joint stability duringlight-to-moderate loading5, a possibilityconsidered as likely in a study6 cited by Rich-mond and Strait. It is unclear how otherjoints of the wrist and hand might be sus-tained (see the authors’ Fig. 1b) during sig-nificant loading if no tendons are recruited.

This confusion extends to ontogeny.Details of surface conformity, as in Rich-mond and Strait’s ‘scaphoid notch’, resultfrom cartilage modelling (they cannot bedetermined by hereditary ‘descriptive’ spec-ification7). Chondral modelling modulatesthe curvature, shape, congruence andsmoothness of a joint’s articular surface8

and postnatal joint motion and joint reac-tion forces determine the final adult jointcontours9,10. It is therefore unlikely that Aus-tralopithecus retained details of joint con-tours specific to knuckle-walking withoutperforming that behaviour.

Richmond and Strait used four variablesfor their canonical variate analysis, of whichtwo were the breadth and orientation oftheir ‘scaphoid notch’ — defined as thelength of the portion along the dorsal mar-gin of the distal radial articular facet that isbevelled. This ‘notch’ must be either a spe-cial morphological substructure of the joint

© 2001 Macmillan Magazines Ltd

margin, or simply some fraction of it. Itcannot be the former, as the authors’ dataare neither dichotomous nor bimodal (orall primates would have scaphoid notchesand would, by induction, be adapted toknuckle-walking). Their notch thereforeappears to be an undefined (neither arc norchord location is specified) estimate of thecurvature of some fraction of this alreadydiminutive joint margin whose form is dic-tated by chondral modelling. Without thefirst axis, Homo and Pongo collapse into asingle distribution, as do terrestrial mon-keys and the African apes, so this analysiscannot distinguish their highly distinctivelocomotor patterns, and all but onehominid (the extensively damaged KNM-ER-20419) would distribute with Homo.

Human and chimpanzee distal radii dodiffer. However, their dissimilarity derivesprimarily from a difference in angulationbetween the epiphysis and diaphysis (seewww.kent.edu/anthropology/naturefigure).The greater angulation in African apes is thesource of another character that Richmondand Strait refer to as distal projection. As theentire epiphysis is so angled, such differencesare more likely to be due to morphogen orparacrine signalling during pattern forma-tion, for example by modulating the para-thyroid-hormone-related protein–IndianHedgehog loop11,12, and to asymmetricphysis compression induced by knuckle-walking during ontogeny (the same process-es that produce human knee valgus13). C. Owen Lovejoy*†, Kingsbury G. Heiple†,Richard S. Meindl**Division of Biomedical Sciences, Department ofAnthropology, Kent State University, Kent, Ohio 44242, USA†Division of Surgery, Department of Orthopaedics,School of Medicine, Case Western ReserveUniversity, Cleveland, Ohio 44206, USAe-mail: olovejoy@aol.com1. Richmond, B. G. & Strait, D. S. Nature 404, 382–385 (2000).

2. Martin, R. B., Burr, D. B. & Sharkey, N. A. Skeletal Tissue

Mechanics (Springer, New York, 1998).

3. Burstein, A. H. & Wright, T. M. Fundamentals of Orthopaedic

Biomechanics (Williams & Wilkins, Baltimore, 1994).

4. Shadwick, E. D. J. Appl. Physiol. 68, 1033–1040 (1990).

5. Roberts, T. J. et al. Science 275, 1113–1115 (1997).

6. Susman, R. L. & Stern, J. T. Jr Am. J. Phys. Anthropol. 50,

565–574 (1979).

7. Thorogood, P. in Embryos, Genes, and Birth Defects (ed.

Thorogood, P.) 1–16 (Wiley, Chichester, 1997).

8. Frost, H. M. Anat. Rec. 255, 162–174 (1999).

9. Ogden, J. A. in Fundamental and Clinical Bone Physiology (ed.

Urist, M. R.) 108–171 (Lippincott, Philadelphia, 1980).

10.Hamrick, M. J. Theoret. Biol. 201, 201–208 (1999).

11.Karp, S. J. et al. Development 127, 543–548 (2000).

12.St-Jacques, B., Hammerschmidt, M. & McMahon, A. P. Genes

Dev. 13, 2072–2086 (1999).

13.Lovejoy, C. O., Cohn, M. J. & White, T. W. Proc. Natl Acad. Sci.

USA 96, 13247–13252 (1999).

Richmond and Strait reply — Dainton’sargument that knuckle-walking may haveevolved in parallel in Gorilla and thePan/Homo ancestor is possible, but unparsi-monious. Moreover, patterns of African-ape carpal growth1 do not provide an

adequate test of knuckle-walking homolo-gy, given evidence that development doesnot always faithfully reflect homology2. Sia-mangs and gibbons differ in skeletalgrowth3, but no one disputes the homologyof their suspensory specializations.

Regarding Dainton’s points about ourfunctional interpretation, an obtuse angula-tion between the scaphoid and lunate facetsresists compressive stresses as part of a com-plex that stabilizes the wrist in extensionduring knuckle-walking4. With respect tothe statistics, our cluster analysis (Fig. 2c),the Mahalanobis D2 distances, and posteriorprobabilities of group assignment show thatA. afarensis and A. anamensis radii are moresimilar to Gorilla and Pan than either are toPongo. Finally, although knuckle-walkingtraits may have arisen to serve extension-limiting functions in other forms of loco-motion, there is no evidence to supportthis. Our hypothesis meets Lauder’s defini-tion of a “best-case scenario”5, in which theextinct taxon whose behaviour is beingreconstructed lies within a clade thatincludes extant members for which a formis exclusively related to a function.

Lovejoy et al. misinterpret our discus-sion of radio-scaphoid joint function. Allthe anthropoids we examined havescaphoid notches. However, the morpholo-gy of the notch, in combination with fea-tures such as distal projection, distinguishesknuckle-walkers from other taxa (seewww.anthro.uiuc.edu/faculty/richmond/hominidradii.html). African apes differfrom other anthropoids in that the radiusand scaphoid conform posteriorly afterminimal extension, such that “no furthermovement is possible without disengagingthe articular surfaces … this part of thejoint achieves a ‘locked’ or ‘close-packed’position”4. Such terminology is convention-al6. Muscle and other tissues can play a rolein promoting joint stability, and we explicit-ly cited experimental evidence on the sub-ject. However, the digital flexors are notrecruited during knuckle-walking at slow-to-moderate speeds. Active recruitment iscrucial in maintaining tension to enable thetendons to store and return elastic energy7.Moreover, passive tension of the digital flex-ors is unlikely to be significant duringknuckle-walking because the fingers areflexed. Regardless, active or passive muscletension across the wrist does not necessarilydiminish the importance of joint morphol-ogy in providing stability.

Lovejoy et al. suggest developmentalhypotheses to explain variation in jointmorphology. We disagree with their impli-cation that species-specific aspects of jointshape have little hereditary basis. Duringearly development, joint morphology isheavily influenced by genetic factors, to theextent that normal articular topographywill appear even if the adjacent surface is

absent or if limb paralysis is induced8.Although the employment of abnormalpositional behaviours influences articulardevelopment9, it is unclear how, or whether,the absence of specific behaviours influ-ences joint growth. Current evidence sug-gests that bone length and aspects of jointshape are not heavily influenced by physio-logical variations in activity during growth,unlike trabecular structure, and corticalbone thickness, density and curvature10.Thus, functionally sensitive features such asfemoral bicondylar angle and phalangealcurvature suggest both bipedalism andarboreality in Australopithecus, whereasradiocarpal joint shape might be a productof evolutionary history.

Although Lovejoy et al. reject the possi-bility that functionally relevant traits mightbe retained through phylogenetic inertia,they have argued11, in reference to australo-pithecine joint morphology, that one can-not “expect an animal that has recentlyevolved bipedality to no longer have anyanatomical characters that are not alsofound in modern-day apes”. This thesis iscentral to their11,12 long-held argument thatearly australopithecines are obligate bipedsbut retain arboreal features without practis-ing arboreality. It might be concluded fromLovejoy et al.’s present developmental argu-ment that these very same fossil hominidsare knuckle-walkers. We doubt they believethis, and neither do we. Phylogenetic iner-tia remains the most plausible interpreta-tion of the morphology we identified. Howfunctionally to interpret primitive reten-tions is the central problem in reconstruct-ing behaviour in fossil hominids. We stilldo not fully understand the genetic andepigenetic influences on bone develop-ment, and need to identify bone morpholo-gy that provides a faithful record of anindividual’s biomechanical activity.Brian G. Richmond*†, David S. Strait†‡*Department of Anthropology, University of Illinois,Urbana, Illinois 61820, USAemail: brich@uiuc.edu†Department of Anthropology, George WashingtonUniversity, Washington DC 20052, USA‡Department of Anatomy, New York College of Osteopathic Medicine, Old Westbury, New York 11949, USA1. Dainton, M. & Macho, G. A. J. Hum. Evol. 36, 171–194 (1999).

2. Hall, B. K. (ed.) Homology: The Heirarchical Basis of

Comparative Biology (Academic, San Diego, CA, 1994).

3. Jungers, W. L. & Cole, M. S. J. Hum. Evol. 23, 93–105 (1992).

4. Jenkins, F. A. J. & Fleagle, J. G. in Primate Functional

Morphology and Evolution (ed. Tuttle, R. H.) 213–231

(Mouton, The Hague, 1975).

5. Lauder, G. V. in Functional Morphology in Vertebrate

Paleontology (ed. Thompson, J. J.) 1–18 (Cambridge Univ.

Press, Cambridge, 1995).

6. Tuttle, R. H. Am. J. Phys. Anthropol. 26, 171–206 (1967).

7. Roberts, T. J. et al. Science 275, 1113–1115 (1997).

8. Hamrick, M. W. J. Theor. Biol. 201, 201–208 (1999).

9. Nakatsukasa, M. et al. Folia Primatol. 64, 1–29 (1995).

10.Lanyon, L. E. J. Zool. 192, 457–466 (1980).

11.Oliwenstein, L. Science 269, 476–477 (1995).

12.Latimer, B. M. & Lovejoy, C. O. Am. J. Phys. Anthropol. 82,

125–133 (1990).

brief communications

326 NATURE | VOL 410 | 15 MARCH 2001 | www.nature.com© 2001 Macmillan Magazines Ltd