constraints on masticatory system evolution in anthropoid primates

Constraints on Masticatory System Evolutionin Anthropoid Primates

MARK A. SPENCER*Department of Biological Anthropology and Anatomy, Duke UniversityMedical Center, Durham, North Carolina 27710

KEY WORDS jaw biomechanics; bite force; temporomandibularjoint; gape; diet; scaling

ABSTRACT It is well established that some observed patterns of forceproduction in the primate masticatory system match those predicted by asimplified lever model. This model is also commonly invoked in adaptiveexplanations of craniodental diversity. However, systematic studies of thepredictive power of this model are missing, leaving open the possibility thatfactors not traditionally included in the model alter the function and evolutionof the masticatory system. One such factor was proposed for mammalsgenerally by Greaves ([1978] J. Zool. (Lond.) 184:271–285), who argued thatthe temporomandibular joint (TMJ) was poorly suited to being pulled apart.In this constrained lever model, the avoidance of joint distraction leads tolimitations on masticatory system form and function. The goal of the presentstudy was to quantify masticatory system diversity in anthropoid primates forcomparison with these predictions.

Results indicate that all sampled taxa exhibit a form that is consistent withselection against regular distraction of the TMJ. Also apparent from observedpatterns of scaling is a regular interaction among a limited set of cranial anddental dimensions, in accordance with the constrained model. However, thedata indicate that specific positional relationships among the muscles, joints,and teeth differ from those predicted by Greaves (1978). The pattern ofdeviation suggests that selection has favored a conservative masticatorysystem configuration that safeguards the TMJ from distraction during thedynamic processing of irregular foods. The resulting buffered model leads toalternative hypotheses regarding the response of the masticatory system todietary selection pressures. It may, therefore, improve our understanding ofthe adaptive significance of primate craniofacial form. Am J Phys Anthropol108:483–506, 1999. r 1999 Wiley-Liss, Inc.

Evolutionary changes in the structuralrelationships within the masticatory system(i.e., among the muscles, joints, and teeth)influence many aspects of facial form inprimates. Such changes are commonlyviewed as adaptive responses to selectionpressures associated with the production ofocclusal forces (Du Brul, 1974, 1977; Carl-son and van Gerven, 1977; Hylander, 1977,1979a; Smith, 1978; Ward and Molnar, 1980;Jablonski, 1993; Rak, 1983; Bouvier, 1986;

Ravosa, 1990; Demes and Creel, 1988; Spen-cer and Demes, 1993; Anapol and Lee, 1994),the resistance to masticatory stresses (Cart-mill, 1974; Hylander, 1977, 1979a; Rak,

Grant sponsor: National Science Foundation; Grant sponsor:L.S.B. Leakey Foundation.

*Correspondence to: Mark A. Spencer, Department of Biologi-cal Anthropology and Anatomy, Box 3170, Duke UniversityMedical Center, Durham, NC 27710.E-mail: [email protected]

Received 6 March 1998; accepted 3 September 1998.

AMERICAN JOURNAL OF PHYSICAL ANTHROPOLOGY 108:483–506 (1999)

r 1999 WILEY-LISS, INC.

1983, 1986; Bouvier, 1986; Demes, 1987;Daegling, 1989, 1992), the production ofadequate gape (Hylander, 1979a; Smith,1984; Ravosa, 1990), and the spatial de-mands of other neurofacial components(Moss and Young, 1960; Cheverud, 1982;Smith and Paquette, 1989; Ravosa, 1991;McCollum, 1994). Since selection may favoropposing adaptive responses, the configura-tion of the masticatory system is presumedto be a function of the balance among compet-ing and varying demands (e.g., Hylander,1979a; Smith, 1984; Ravosa, 1990). In thisview, limits on diversity simply reflect thisbalance. However, Greaves (1978, 1982,1983, 1985, 1988; see also Druzinsky andGreaves, 1979; Werdelin, 1986, 1987, 1988;Spencer and Demes, 1993; Spencer, 1995)proposed a mechanical constraint that couldhave a consistent limiting influence on mas-ticatory system function and evolution: thetemporomandibular joint (TMJ) should notbe loaded so that the mandibular condyle isregularly or forcefully pulled away from thearticular eminence. The avoidance of suchdistractive forces is achieved, in Greaves’model, through a combination of 1) changesin masticatory muscle activity by bite pointand 2) limitations on the configuration thatthe masticatory components might assumethrough evolution. Recently reported electro-myographic data from humans are consis-tent with the first of these expectations(Spencer, 1998) (see below). The goal of thepresent research is to address the secondhypothesis by quantifying the diversity ofmasticatory system configuration among liv-ing anthropoid primates to determine if itappears limited by a restriction on distrac-tive joint loading.

CONSTRAINED MODEL OFMASTICATORY FORCE PRODUCTION

The forces applied to the mandible by themasticatory adductor muscles (i.e., masse-ter, temporalis, and medial pterygoid) areresisted by reaction forces in three regions ofcontact with the cranium: the bite point, theworking (biting) side TMJ, and the balanc-ing side TMJ (Gysi, 1921; Greaves, 1978;Smith, 1978; Walker, 1978; Wolff, 1984;Hylander, 1985; Spencer, 1998). The relativemagnitudes of these forces are determined

largely by their spatial relationships andare commonly examined in the context of asimplified lever model (Weijs and van Spron-sen, 1992; Weishample, 1993; Greaves, 1995;Spencer, 1998). Basic predictions derivedfrom this model regarding force productionare supported by diverse in vivo studies(e.g., Mansour and Reynik, 1975; Hylanderand Bays, 1979; Hylander, 1979b; Pruim etal., 1980; van Eijden, et al., 1988). Addition-ally, it serves as the basis for most adaptiveinterpretations of masticatory system con-figuration in relation to dietary selectionpressures (e.g., Du Brul, 1977; Hylander,1977, 1979a; Rak, 1983; Bouvier, 1986;Demes and Creel, 1988; Spencer and Demes,1993; Anapol and Lee, 1994). The goal of thecurrent study is not to examine the viabilityof the lever model. Instead, it addressesGreaves’ (1978) argument that, within thecontext of this model, mechanical con-straints are at work that limit muscle activ-ity patterns, bite force production, and theevolution of the masticatory system.

In his theoretical treatment of the mastica-tory system of selenodont artiodactyls,Greaves (1978) presented a general analysisin which masticatory forces are examined inan occlusal view (Fig. 1). As is commonamong other studies of this system, only thevertical (i.e., perpendicular to the occlusalplane) components of each force are includedin the analysis, and they are seen end-on inthe occlusal view. Through vector addition,the forces applied by all adductors are com-bined into a single muscle resultant force.The location of this muscle resultant force isdetermined by the positions and relativeforce contributions of the various muscles.When the balancing and working sidemuscles are equally active (e.g., during maxi-mum force production by both sets ofmuscles), the muscle resultant force lies inthe midline. However, differential activity ofthe balancing and working side musclesproduces mediolateral movement of themuscle force resultant (Hylander, 1985;Throckmorton et al., 1990; Weijs and vanSpronsen, 1992; Spencer, 1998).

Because the bite force and joint reactionforces resist the upward pull of the muscles,they lie at the corners of what Greaves(1978) termed the triangle of support (see

484 M.A. SPENCER

Fig. 1). The magnitudes of these forces aredetermined by the position of the muscleforce resultant relative to this triangle. Dur-ing biting at more anterior positions, thetriangle of support is relatively large andencloses a midline muscle resultant force(Fig. 1a). Under this loading regime, theforces at the corners of the triangle of sup-port will all be compressive (i.e., they willbring the opposing surfaces together). How-ever, during biting on more posterior teeth,the triangle of support is smaller and shiftedtoward the working side. A midline muscleresultant force may therefore fall outsidethe triangle (Fig. 1b), rotating the mandiblearound the bite point and the balancing sidejoint so that the working side mandibularcondyle is pulled downward (creating a dis-tracting force).

The force patterns described above do notdiffer from those of other analyses in whichboth the balancing and working side forcesare considered (e.g., Smith, 1978; Walker,1978; Wolff, 1984; Hylander, 1985). How-ever, Greaves (1978) assumed that the TMJshould not experience the distractive forcesthat may result during biting on more poste-

rior teeth. He therefore hypothesized thatthe muscle resultant force moves so that italways falls within the triangle of support,maintaining compressive forces at bothjoints. In his model, this repositioning isbrought about by shifting the muscle forceresultant toward the working side through areduction in the relative activity of the bal-ancing side muscles (Fig 1b).

Changes in joint loading and muscle activ-ity lead to the recognition of three zones ofpotential bite points that have been termedRegions I, II, and III (Spencer and Demes,1993; Spencer, 1995, 1998). Regions I and IIare separated by an oblique line passingthrough the balancing side joint reactionforce and a midline muscle resultant force(Fig. 2). Bite points anterior to this line inRegion I will be associated with triangles ofsupport that enclose a midline muscle resul-tant force. Movement of the resultant forcetoward the working side is therefore unnec-essary in this region, and there are no limitson maximum muscle forces. Maximum biteforce magnitudes in Region I are expected toincrease as the bite point is moved posteri-orly along the tooth row due to shortening of

Fig. 1. Occlusal view of mandible showing a verticalmidline muscle resultant force (j), which acts to pullthe mandible upward out of the plane of the page.During biting at more anterior points (a), this midlinemuscle resultant force passes through the triangle ofsupport (shaded zone). The corners of this triangle arepositioned at the bite force (r), the balancing side joint

reaction force (●), and the working side joint reactionforce (C). Distraction of the working side joint will occurif the muscle resultant force passes outside of thetriangle of support (b); Greaves (1978) argued that themuscle resultant would be repositioned toward theworking side in such loading situations to avoid tensionin the joint.

485PRIMATE MASTICATORY SYSTEM EVOLUTION

the bite force moment arm. Regions II andIII are separated by a transverse line pass-ing through the muscle resultant force (seeFig. 2). Bite points in Region II yield tri-angles of support through which a midlinemuscle resultant force will not pass, and thisforce must shift toward the working side toavoid distraction of the working side TMJ.The decrease in balancing side muscle forcethat brings about this shift leads to lowermaximum bite forces than predicted by anunconstrained model. Furthermore, sincethe muscle force resultant must shift farthertoward the working side as the bite point ismoved posteriorly within Region II, the ef-fects on bite forces are more pronouncedposteriorly. Maximum magnitude bite forcesare therefore predicted to be of relativelyhigh but equal magnitude along that portionof the tooth row that falls within Region II.Bite points posterior to the muscle resultantforce will produce triangles of supportthrough which the muscle resultant cannotpass, even through mediolateral reposition-ing. These bite points lie within Region III,and biting at them will be unavoidablyassociated with distraction of the TMJ.

Greaves (1978) further speculated that itwould be adaptive for specific functionalregions of the dentition to maintain consis-tent positional relationships with the re-

gions defined above. First, Greaves arguedthat no teeth should lie within Region III(i.e., posterior to the muscle resultant force),since biting on them will lead to joint distrac-tion. Second, he argued that powerful grind-ing teeth should always be located withinwhat is here termed Region II (i.e., directlyanterior to the muscle resultant force), sinceit is in this region that the highest magni-tude bite forces can be produced. Takentogether, these predictions require that themuscle resultant force lie immediately poste-rior to the most distal molar. Furthermore,it is expected that the position and mesiodis-tal length of the powerful grinding dentitionwill be correlated with the distribution ofRegion II.

PATTERNS OF COVARIATION IN THECONSTRAINED MODEL

Three parameters interact to determinethe distribution of Region II and its relation-ship to the dentition in Greaves’model (Spen-cer and Demes, 1993; Spencer, 1995, 1998):1) the distance of the balancing side jointreaction force from the midline, 2) the dis-tance anterior to the TMJ of the point ofintersection of the muscle resultant forcewith the occlusal plane, and 3) the mediolat-eral position of the tooth row. First, evolution-ary changes in the mediolateral position ofthe balancing side TMJ leads to a reorienta-tion of the line demarcating Regions I and II(Fig. 3a), altering the anterior border ofRegion II. If Greaves’ (1978) constrainedmodel is correct, therefore, lateral move-ment of the TMJ should be accompanied byan evolutionary reduction in the mesiodistallength of the dentition (and vice versa).Second, both boundaries defining Regions I,II, and III pass through the midline muscleresultant force and are repositioned withanteroposterior movements of the resultant(Fig. 3b). Posterior muscle resultant forcemigration is associated with both a posteriormigration of Region II and a change in itsshape. The molar dentition should adapt bymigrating posteriorly and decreasing in me-siodistal length (either by changes in toothsize or by the elimination of teeth from themolar-like functional unit.) Finally, RegionII tapers medially, and its anteroposteriorlength is therefore shorter close to the mid-

Fig. 2. Occlusal view of mandible showing the pre-dicted distributions of Regions I, II, and III (see text fordefinitions).

486 M.A. SPENCER

line. A more medially positioned tooth rowshould therefore be shorter than a morelaterally positioned tooth row, all other pa-rameters being equal (Fig. 3c).

The primate masticatory system differsfrom the model of Greaves (1978) in that theTMJs are usually positioned above the occlu-sal plane and the muscle resultant vector isprobably rarely perpendicular to this plane.These parameters together can act to ex-pand Region II by reorienting its boundarywith Region I. They do so by altering thepositional relationship of the muscle resul-tant vector to the triangle of support. Be-cause the bite point forms one corner of thetriangle of support, raising the other cornersof the triangle (those at the balancing andworking side TMJs) above the occlusal planecauses the triangle of support to becomeinclined (Fig. 4). This reorientation sepa-rates the point where the muscle resultantforce vector intersects the occlusal plane(point A in Fig. 4) from its intersection withthe triangle of support (point B in Fig. 4). InGreaves’ model, these points are alwayscoincident. However, when separated theyare independent; a reorientation of the resul-tant force vector can alter where it passesrelative to the triangle of support withoutaffecting where it pierces the occlusal plane.Therefore, a muscle resultant that appears

to pass through the triangle of support inGreaves’ occlusal view model might actuallypass anterior to the triangle if its vectorwere inclined anterosuperiorly. Since such aforce vector no longer passes through thetriangle of support, it must be shifted to-ward the working side to avoid joint distrac-tion, in compliance with Greaves’ constraints.Bite points that were not in Region II accord-

Fig. 4. Lateral view of mandible illustrating theeffects of positioning the TMJ above the occlusal planeand inclining the muscle resultant force. In this configu-ration, which is common among primates, the muscleresultant intersects the triangle of support (seen edge onas a dashed line) more anteriorly (B) than its intersec-tion with the occlusal plane (A). As a result, a muscleresultant force that appears to pass through the triangleof support in the occlusal view might actually lieanterior to it. This will influence the distribution ofRegion II.

Fig. 3. Diagrams showing the effects of changes(represented by states A, B, and C) in the relativepositions of the muscle and joint forces on the distribu-tion of Region II. a: Mediolateral movement of thebalancing side joint force (●) leads to a reorientation ofthe boundary between Regions I and II (diagonal line) sothat more or less of the tooth row (shaded boxes) fallswithin Region II. b: Anterior movement of the midline

muscle resultant force (j) causes Region II to be posi-tioned more anteriorly; its anteroposterior length (at agiven mediolateral position) will also increase. c: Be-cause Region II tapers medially, its anteroposteriorlength is shorter near the midline, and it should accom-modate fewer (or shorter) molar-like teeth.


ing to Greaves’ predictions will therefore liewithin Region II in this modified model. Thisanterior extension of Region II along thetooth row is proportional to the combineddegree of muscle inclination and TMJ height;if the muscle resultant force is vertical or theTMJs lie within the occlusal plane, no ante-rior expansion will occur (for further discus-sion see Spencer, 1995).

APPLICABILITY OF CONSTRAINEDMODEL TO PRIMATES

The fundamental assumption of the con-strained model is that the TMJ is poorlystructured for resisting distractive forces. Inprimates, it is likely that the TMJ canwithstand such forces, at least at low tomoderate magnitudes. Based on patterns ofsubcondylar bone strain in a macaque,Hylander (1979b) inferred the existence ofintermittent distractive forces in the work-ing side TMJ during unilateral isometricbiting on the third molar. Distractive forceswould presumably be resisted by tension inthe temporomandibular ligament (Hylander,1979b). However, the ability of this ligamentto resist gross dislocation of the TMJ doesnot imply that it regularly serves to main-tain joint congruity during function. To thecontrary, the structure of this ligament sug-gests that it acts to limit extremes of motion(particularly in the posterior direction) butthat it cannot hold the mandibular condyleagainst the articular eminence in normalloading positions (Bell, 1990). Given theimportance of joint stability to the guidanceof tooth contacts (Bakke and Møller, 1992), itseems unrealistic to suppose that repetitiveseparation of the condyle from the articulareminence occurs during routine mastica-tion. In support of this, Hylander (1979b)inferred only compressive joint reactionforces during mastication in several ma-caques. Furthermore, clinical studies in hu-mans suggest that injury to this ligament orthe structurally contiguous joint capsule canpermanently impair normal joint function(Bell, 1983, 1990; McKay, 1992). Given thefundamental role of this joint in food process-ing, selection against morphologies that in-crease the chances of such injuries is likelyto be high.

Additional experimental data relevant totesting the predictions of the constrainedmodel are few. Because these predictions areformulated for maximum force production,any experimental test must involve record-ing forces during heavy unilateral loading atseveral known bite points. Few studies havemet these requirements. For example, whilethe distinctive pattern of maximum magni-tude bite forces predicted by the constrainedmodel should be quantifiable, few studiesreport data from multiple bite points alongthe tooth row, and none report data for allpostcanine teeth. Furthermore, the avail-able studies differ in their conclusions, withmaximum voluntary bite forces in humanshaving been reported to increase posteriorlyat the rate predicted by the unconstrainedmodels (van Eijden et al., 1988; van Eijden,1991), increase posteriorly but more slowlythan expected (Mansour and Reynik, 1975),and decrease posteriorly (Pruim et al., 1980).

A more direct method of testing the con-strained model would be to compare relativebalancing and working side muscle activityduring maximum bite force production atpoints along the tooth row. As discussedabove, this model requires that, during highmagnitude biting, the activity of the balanc-ing side muscles will decrease relative tothat of the working side muscles as the bitepoint is shifted posteriorly along the molardentition. Recent electromyographic work inhumans (Spencer, 1998) supports this cen-tral prediction. This pattern of activity wasrecorded from the superficial masseter andthe anterior temporalis muscles of nine hu-man subjects. These data are difficult tointerpret outside of the context of the con-straints proposed by Greaves (1978) andoffer indirect support for the hypothesis thatdistraction of the TMJ is to be avoided.

The goal of the present study is to deter-mine if the observed position and dimen-sions of the postcanine dentition covary withthe distribution of Region II (the zone ofhighest magnitude bite force production), aspredicted by the constrained model de-scribed above. If the model has been influen-tial during the evolution of anthropoid pri-mates, masticatory form in this group shouldbe limited in accordance with its predic-

488 M.A. SPENCER

tions.1 If conformity to this limitation wereobserved in all anthropoid taxa, it wouldsuggest that either the predictions of theconstrained model are correct or that evolu-tion has simply not produced a form that isincompatible with the model (an unlikelyoccurrence, given the broad phenotypic rangeof anthropoid taxa). If observed diversity isnot so limited, the model must be rejected ormodified.

MATERIALS AND METHODSSample

Representative species of most extant an-thropoid genera were included in this study(Table 1) to increase the probability of sam-pling a taxon that is incompatible with theconstrained model. A total of 876 individualswas measured representing 39 species and31 of the 37 anthropoid genera recognized byFleagle (1988). Only adult crania (maxillarycanine fully erupted) and their associatedmandibles were used. Specimens were ob-tained from the following sources: the Ameri-can Museum of Natural History (New York,NY), the National Museum of Natural His-tory (Washington, DC), the Field Museum ofNatural History (Chicago, IL), the Museum

of Comparative Zoology (Cambridge, MA),and the British Museum of Natural History(London). No material from captive animalswas used.

Measurements

Three sets of measurements were re-corded from the above sample: 1) distancesthat represent the observed position andmesiodistal length of the postcanine denti-tion, 2) the dimensions needed to calculate apredicted position and length of the postca-nine dentition using the described model,and 3) estimates of the positions and orienta-tions of the primary masticatory adductormuscles. Both three-dimensional (3-D) coor-dinate data and 2-D linear and angular datawere quantified using MacMorphr (Spencerand Spencer, 1993), a computer-driven video-imageanalysispackage for theAppleMacintoshcomputer. Details of the 3-D data collectionmethod can be found in Spencer and Spen-cer (1995).

The landmarks for which 3-D coordinatedata were collected are shown in Figure 5.These data formed the basis for the calcula-tion of most dimensions used in this study.Five dimensions together determine the dis-tribution of Region II: 1) the distance of thebalancing side joint reaction force from themidline, 2) the distance of the working sidepostcanine tooth row from the midline, 3)the height of the TMJ above the occlusal

1Prosimian primates were not included in this study because,unlike anthropoids, all extant taxa possess an unfused mandibu-lar symphysis, a feature that is thought to change the wayworking and balancing side forces are produced (Beecher, 1977,1979; Hylander, 1979a, 1984; Greaves, 1988; Ravosa, 1991).

TABLE 1. Taxa included in study

Taxon M F Taxon M F

Aotus azarae 5 5 Macaca fascicularis 15 15Callicebus torquatus 15 15 Macaca arctoides 12 10Callicebus moloch 15 15 Cercocebus albigena 15 15Alouatta seniculus 15 15 Cercocebus torquatus 15 11Alouatta palliata 15 15 Cercopithecus nictitans 15 15Lagothrix lagothricha 5 5 Cercopithecus cephus 15 15Ateles geoffroyi 5 5 Erythrocebus patas 14 5Brachyteles arachnoides 2 1 Papio anubis 15 15Saguinus oedipus 15 15 Mandrillus sphinx 8 3Leontopithecus rosalia 1 1 Theropithecus gelada 13 3Callithrix jacchus 12 14 Colobus guerza 15 15Cebuella pygmaea 12 2 Colobus polykomos 15 15Cebus apella 15 15 Presbytis cristatus 5 5Cebus albifrons 15 15 Nasalis larvatus 5 5Saimiri sciureus 5 5 Simias concolor 5 5Pithecia pithecia 15 13 Pygathrix nemaeus 5 5Chiropotes satanas 15 15 Hylobates hoolock 5 5Cacajao melanocephalus 13 8 Pan troglodytes 15 15Cacajao calvus 12 15 Gorilla gorilla 15 8

Homo sapiens 46 22


plane, 4) the distance anterior to the TMJthat a midline muscle resultant force inter-sects the occlusal plane, and 5) the orienta-tion of a midline muscle resultant force inthe sagittal plane. Of these, the first threecould be directly quantified from the 3-Dlandmark data (see Appendix). However,quantifying muscle resultant position andorientation is severely hampered by ourpoor knowledge of the comparative myologyand function of the masticatory adductormusculature among primates (Throckmor-ton, 1985, 1989). It was therefore assumed

that the maximum magnitude muscle resul-tant force crosses the occlusal plane at apoint in the midline directly at the posteriorend of the tooth row. This position corre-sponds to that hypothesized by Greaves(1978) and therefore provides a directlyquantifiable value that is consistent withthe assumptions of the constrained model.As such, it should be regarded as a liberalestimate since the model is viewed as beingcorrect with regard to resultant position.

The accuracy of the assumed muscle resul-tant force position was tested by comparingit to estimated masticatory muscle forcevectors. Although the combination of indi-vidual forces into a resultant vector is prob-lematic, the possible locations of this resul-tant can be bracketed by examining theseparate positions of the component forces.Each muscle force vector intersects the occlu-sal plane at some position anterior to theTMJ. Since the resultant of two (or more) ofthese vectors is intermediate in orientation,it must cross the occlusal plane betweentheir separate points of intersection. Thepoints where the estimated force vectors forthe anterior temporalis, superficial masse-ter, and medial pterygoid muscles intersectthe occlusal plane were therefore quantifiedand compared to the position of the muscleresultant force assumed in the constrainedmodel.2 Approximate centroids of the attach-ment areas for all muscles were first markedon each specimen. Lateral view images ofeach specimen were then displayed withinMacMorph, and the marked cranial andmandibular centroids were connected as anestimate of the force vector for each muscle.The points where each ‘‘vector’’ intersectedthe occlusal plane were then measured forcomparison to the assumed intersectionpoint.

Muscle resultant orientation could not bereliably estimated from the data collectedfor this study since this value is dependenton unknown muscle force magnitudes. Sam-ple means for the quantified orientation ofestimated muscle force vectors are as follows

2The constrained model of Greaves (1978) assumes staticequilibrium, a scenario that is most similar to isometric biting.The deep masseter and posterior temporalis muscles were there-fore excluded from the present analysis because they are thoughtto contribute only a weak force during isometric biting (Ahlgren,1966; Møller, 1966; Hylander and Johnson, 1994).

Fig. 5. Landmarks for which three-dimensional coor-dinate data were collected. 1,24, center of articularsurface of articular eminence; 2,23, inferior edge ofmalar at most anterior point of attachment of superficialmasseter muscle; 3,22, point of intersection of temporalline and frontozygomatic suture (frontomalare tempo-rale [White, 1991]); 4,21, sphenopalatine suture atintersection of medial and lateral pterygoid plates; 5–9and 20–16, the center of the trigon basins of eachmaxillary molar and premolar; 10,15, maxillary caninetip; 11–14, center of occlusal surface of each maxillaryincisor.

490 M.A. SPENCER

(90° 5 perpendicular to the occlusal plane(in the sagittal plane), ,90° 5 inclinedanteriorly): anterior temporalis, 91.90 6 6.9°;superficial masseter, 77.5 6 6.5°; medialpterygoid, 81.2 6 8.0°. Based on these val-ues, a fixed orientation of 80° was assumedfor all taxa in this study. This value allowsvariation in the height of the TMJ above theocclusal plane to have only a moderate influ-ence on the calculation of Region II distribu-tion (see above).

The five parameters described above wereused to calculate the length of the tooth rowthat should fall within Region II. This pre-dicted length will be referred to as theeffective length of Region II. The procedureby which this length was calculated is pro-vided in the Appendix.

Test criteria

Greaves (1978) proposed the constrainedmodel for application to selenodont artiodac-tyls, in which the grinding tooth row iseasily identified due to the morphologicalsimilarity of the teeth of which it is com-posed. However, the diversity of premolarform among anthropoid primate species sug-gests these teeth experience varied loadingpatterns (Rosenberger, 1992). Therefore, itmay be selectively advantageous for sometaxa to include the premolars within RegionII, but it is difficult to predict a priori whichtaxa should do so. However, the most basicpredictions of Greaves’ (1978:276–277) modelwith regard to expected tooth position arethat ‘‘regardless of the muscle resultant’sposition, powerful grinding teeth are notexpected either posterior to the most poste-rior position of the muscle resultant or ante-rior to the region of maximum force applica-tion [Region II in the present work] wheretooth force rapidly decreases.’’

These requirements give rise to two predic-tions tested here. First, the effective lengthof Region II should be equal to or greaterthan the observed distance between thetrigon basins of the anterior- and posterior-most molar teeth. If observed molar rowlength were found to exceed the effectivelength of Region II in any taxa, it wouldimply that a tooth need not lie in Region II tofunction effectively as a molar. Second, theestimated position of Region II should corre-

spond to that predicted by the model. This istested by comparing the muscle resultantforce position assumed in the model to theobserved points of intersection between indi-vidual muscle force vectors and the occlusalplane. If the assumed resultant force is notbracketed by the estimated muscle vectorsin all taxa, it implies that the muscle resul-tant cannot be positioned as predicted dur-ing isometric loading. Failure of this testwould suggest either that some teeth fallposterior to Region II, where it is expectedthat loading the teeth would produce tensileforces in the working side joint or that someportion of Region II is not occupied by teetheven though it is well suited for the applica-tion of loads to molars.

Statistical procedures

The predictions described above wereevaluated through examination of plots ofpredicted and observed dimensions. For com-parison of the observed and predicted val-ues, it was useful to standardize the data sothat differences between taxa due to sizewere reduced. In most instances the relativesize of a limited set of variables was thesubject of interest, and the data were stan-dardized against some parameter that makesthe plot easier to interpret. This approach isnot meant to imply that the chosen standard-izing variables are indicative of overall size;they are used merely as indications of localsize.

Broad patterns of scaling among the di-mensions that influence the distribution ofRegion II were examined as a way to exploretheir potential interaction across anthro-poids. Based on model predictions regardingcovariation among these dimensions, it isexpected that allometric change in one pa-rameter (i.e., change in shape with size) willbe accompanied by compensatory allometricchanges in other parameters so that theeffective length of Region II scales withpostcanine dimensions. For the purpose ofcomparing dimensions to overall facial size,a geometric mean (Darroch and Mosimann,1985) of five dimensions was calculated(these dimensions were calculated using the3-D coordinate data set and include biarticu-lar breadth (landmark 1 to landmark 24),palate breadth (landmark 7 to landmark


18), temporal foramen length (landmark 1 tolandmark 2), tooth row length (landmark 20to landmark 16) and infratemporal height[landmark 4 to landmark 3]). Scaling rela-tionships were established by logging thedata and regressing dimensions indicativeof shape on the geometric mean. Examina-tion of the slope of a regression line in logspace allows inferences to be made regard-ing the scaling of variables, and, when anestimate of size is used, this approach pro-vides an indication of shape changes associ-ated with size (Mosimann and James, 1979).To test if an observed slope differed signifi-cantly from isometry (no change in shapewith size), 95% confidence limits for theslope of the reduced major axis (RMA) regres-sion line were calculated in NEWRMA (Cole,1997) and compared to a slope of 1.0. Slopesfor separate regression lines were comparedusing the Clarke (1980) test, and, when theywere found to be parallel, elevations werecompared using a modification of Tsutakawaand Hewitt’s (1977) quick test. These testswere performed in NEWRMA.

RESULTSRegion II distribution

Figure 6 compares the predicted effectivelength of Region II to actual postcaninetooth row dimensions. In this figure, threedistances are shown: 1) the observed dis-tance (measured between the trigon basins)separating the most mesial premolar andthe most distal molar, 2) the observed dis-tance between the first and last molars, and3) the calculated effective length of RegionII. All values are scaled to the distancebetween the first and last molars. The shadedzone between 0.0 and 1.0 therefore repre-sents the molar dentition, with M1 posi-tioned at 1.0 for all specimens and the mostdistal molar positioned at 0.0. The positions,in relation to the molar dentition, of themost anterior premolar and the anterior endof Region II are indicated by box plots;Region II is therefore predicted to extendfrom 0.0 to this anterior position.

The data in Figure 6 show systematicdifferences in the relative lengths of themolar and premolar dentition betweengroups that differ in postcanine dental for-

mula. However, within groups with the samedental formula, the relationship betweenpremolar row length and molar row length isgenerally consistent. Additionally, this rela-tionship is maintained despite differences inbody size; taxa with very different bodyweights (e.g., Aotus azarae (800 g) and Al-ouatta palliata (5,700–8,000 g) [Ford andDavis, 1992]) show a similar length of theirpremolar tooth rows relative to their molarrows.

The anterior border of Region II was con-sistently predicted to intersect the tooth rowin the premolar region (see Fig. 6). There-fore, when the muscle resultant force posi-tion assumed by Greaves (1978) is used,Region II envelopes the entire molar denti-tion and some portion of the premolar denti-tion in all taxa. Additionally, Region II doesnot extend anterior to the most mesial pre-molar in any taxa, though it does closelyapproach this position in the Callitrichidsand in Mandrillus. With the exception ofthese few taxa, therefore, the effective lengthof Region II appears to maintain a reason-ably uniform relationship with observedtooth row dimensions within broad taxo-nomic groups. As anticipated, there is varia-tion within more restricted groupings thatdoes not always match that seen in eithermolar row length or total postcanine toothrow length. For example, the effective lengthof Region II differs between Alouatta senicu-lus and Alouatta palliata even though thesetaxa differ little in relative premolar rowlength.

The above results suggest that the param-eters that influence the distribution of Re-gion II covary among anthropoid primatesso that, minimally, all molar teeth are main-tained within this region. Not demonstratedin these plots is the range of variation inthese parameters within this group. Werethey found to differ widely among taxa, theobservation that Region II envelops a consis-tent portion of the tooth row would assumegreater significance. A casual considerationof the cranial form of some of the includedtaxa makes it clear that the determinants ofthe distribution of Region II do differ sub-stantially across anthropoids (e.g., Cebus,Alouatta, Papio, Gorilla, and Homo). A quan-

492 M.A. SPENCER

Fig. 6. Plot comparing quantified postcanine dimen-sions to effective Region II length (as calculated using amuscle resultant force positioned at the posterior end ofthe tooth row). All values are standardized againstM1–M3/2 intertrigon distance. M1 therefore lies at avalue of 1.0 (bold line), and the molar dentition fills the

distance between 0.0 and 1.0 (shaded zone) in all taxa.Variation in the position of the most anterior premolarand the predicted anterior edge of Region II is repre-sented by a box plot. The box encloses the central 50% ofthe data, the horizontal bar encloses the central 80% ofthe data, and the vertical bar indicates the median.


titative way of exploring this issue is toexamine how each configurational param-eter changes with size.

Table 2 summarizes the RMA regressionparameters for bivariate log-log plots of sex-specific mean values for the relevant vari-ables regressed on the geometric mean. In-cluded are data showing how postcaninemesiodistal dimensions (measured betweenthe trigon basins of maxillary teeth) scalewith masticatory system size. Because it isevident in Figure 6 that relative molar/premolar dimensions vary among groupswith different dental formulae, separateregression lines are calculated for catar-rhines, three-molared platyrrhines, and two-molared callitrichids. Despite differences inmolar proportions among groups (as evi-denced by significantly different interceptvalues among the groups), overall postca-nine tooth row length scales isometricallyrelative to facial size and appears roughlysimilar in all anthropoids, with one excep-tion: the length of the postcanine dentitionscales with slight positive allometry in non-callitrichid platyrrhines. This pattern corre-

sponds to previously reported dental dimen-sions and probably relates to the strongtrend toward greater folivory with increas-ing body size in platyrrhines (Kay andHylander, 1978). When the scaling of theeffective length of Region II is compared tothat found for postcanine tooth row lengthand molar row length, no significant differ-ences in slope are found in any of the threeexamined groups; the effective length ofRegion II scales at a rate comparable topostcanine tooth row length.

Roughly isometric scaling of the effectivelength of Region II could result from shapemaintenance across all sizes in the param-eters that influence this length. However,the data in Table 2 indicate that several ofthe dimensions used to predict the effectivelength of Region II scale allometrically withmasticatory system size. Both palate breadthand biarticular breadth scale with negativeallometry, while glenoid height scales withstrong positive allometry.As discussed above,the theoretical effect of reducing palatebreadth is to shorten the calculated effectivelength of Region II. Were palate breadth the

TABLE 2. Reduced major axis regression parameters for sex-specific mean values of each dimensionplotted against the geometric mean1

Dimension Taxon R2 Y-interceptRMAslope

Lowerconfidence

limit

Upperconfidence

limit

Molar row length 1) Catarrhines 0.904 21.097 1.051 0.913 1.2112) Noncallitrichid platyr-

rhines0.941 21.958 1.250 1.097 1.424

3) Callitrichids 0.890 21.761 0.974 0.627 1.513Slope comparison2 1 5 2 5 3 5 1Intercept comparison3 1 : 2 : 3 9 1

Postcanine length 1) Catarrhines 0.918 20.460 1.029 0.903 1.1732) Noncallitrichid platyr-

rhines0.974 21.036 1.209 1.108 1.319

3) Callitrichids 0.976 20.749 1.084 0.874 1.344Slope comparison2 1 , 2 5 3 5 1Intercept comparison3 14 2 5 3 , 1

Effective region II length 1) Catarrhines2) Noncallitrichid platyr-

rhines

0.9220.951

21.41221.496

1.1921.235

1.0491.096

1.3551.392

3) Callitrichids 0.991 21.129 1.220 1.067 1.395Slope comparison2 1 5 2 5 3 5 1Intercept comparison3 1 5 2 9 3 : 1

Biarticular breadth All 0.993 0.573 0.951 0.926 0.977Palate breadth All 0.991 0.193 0.896 0.869 0.923Muscle resultant AP posi-

tionAll 0.972 20.332 1.035 0.981 1.092

Glenoid height All 0.919 24.200 1.932 1.766 2.1131 5, no significant difference; ., significant at P , 0.05; :, significant at P , 0.01; *no test for elevation performed due to significantlydifferent slopes.2 Results of tests for differences in slope using NEWRMA (Cole, 1997). Numbers correspond to listed taxa.3 Results of tests for differences in elevation using NEWRMA (Cole, 1997). Numbers correspond to listed taxa.4 No test for elevation performed due to significantly different slopes.

494 M.A. SPENCER

only parameter to scale allometrically withsize, the effective length of Region II shouldalso scale with negative allometry. However,the size-related shape changes in biarticularbreadth and glenoid height are predicted tolead to an increase in the effective length ofRegion II. Thus, these additional trendsmoderate the influence of changing palatebreadth with the result that the effectivelength of Region II scales in concert withobserved tooth row dimensions.

Test of assumed muscle resultantforce position

The data presented in the preceding sec-tion regarding the effective length of RegionII were based on the assumption that theresultant force of all jaw adductor musclescrosses the occlusal plane directly posteriorto the tooth row (as proposed by Greaves[1978]). This assumption was tested by com-paring this muscle resultant position to esti-mates of individual muscle force vector posi-tions. The point where each muscle forcevector intersects the occlusal plane was mea-sured as a distance anterior to the articulareminence within the occlusal plane. Theresulting values are shown in Figure 7 forthe superficial masseter, anterior tempora-lis, and medial pterygoid muscles. In thisfigure, the point where each muscle vectorintersects the occlusal plane is shown inrelation to the anteroposterior (A-P) posi-tions of the TMJ and most distal molar. Allvalues are standardized so that the articulareminence lies at 0.0 on the horizontal axisand the distal molar lies at 1.0. This methodof standardization allows for easy compari-son of muscle force vector/occlusal planeintersection points to the assumed muscleresultant position, which lies in the sameA-P position as the distal molar (i.e., at avalue of 1.0).

Examination of Figure 7 shows that noneof the three muscle force vectors intersectsthe occlusal plane at a point anterior to themost distal molar. While these intersectionpoints do vary, the position of the intersec-tion point for each muscle vector is visuallyquite consistent across all taxa. On average,the anterior temporalis vector intersects theocclusal plane most anteriorly (mean stan-dardized A-P distance 5 0.66 6 0.15 stan-

dard deviation). The medial pterygoid forcevector intersects the occlusal plane mostposteriorly (mean standardized A-P dis-tance 5 0.40 6 0.08). The superficial masse-ter intersection point lies between those ofthe other two muscles (mean standardizedA-P distance 5 0.57 6 0.09). The relativelylow values for all taxa suggest that in anthro-poids the resultant of the three force vectorsthat are active during isometric biting doesnot intersect the occlusal plane near thedistal molar, as assumed by Greaves (1978)for selenodont artiodactyls.

DISCUSSION

The configuration of the masticatory sys-tem in anthropoid primates has respondedto a variety of selection pressures during theevolutionary diversification of this group. Afundamental issue explored in this study iswhether this configuration is influenced pri-marily by the equilibrium reached amongselection pressures that vary among speciesor if it also is constrained by a universalneed to avoid distraction of the temporoman-dibular joint, as proposed by Greaves (1978).The distinction made between these possi-bilities does not imply that a constraint onjoint loading stems from some process inde-pendent of selection; it is selection, acting tofavor individuals that do not distract theirTMJ, that is expected to produce gaps in theset of observed phenotypes (for a discussionof the overlap between the concepts of con-straint and selection see Smith, 1993;Schwenk, 1995). Instead, the distinction liesbetween those selection pressures whosestrength is governed by varying aspects ofbehavior (such as gape or diet) and one thatis thought to be invariate among species.

The results of the broad-scale morphomet-ric analysis reported here suggest that phe-notypic diversity in cranial form within an-thropoids is limited by the need to avoiddistraction of the TMJ. Furthermore, thisdiversity conforms to several basic predic-tions of a modified version of Greaves’ (1978)constrained model (which includes numer-ous secondary elements beyond the basiclimit on joint loading); patterns predicted bythe model are evident throughout this group.However, it is also apparent that there is animportant incongruity between observed


morphology and this modified model. Thisdiscrepancy lies in the observation that themasticatory adductor muscles are not posi-tioned to produce a resultant force directlyat the posterior end of the tooth row duringforceful isometric biting. Instead, in all ofthe examined taxa they are positioned moreposteriorly (on average, approximately 60%

of the distance from the TMJ to the posteriormolar). Greaves (1978) argued that all molar-like teeth should lie directly anterior to themuscle resultant force within the specificregion where the highest forces can be pro-duced, here termed Region II. However, therelatively posterior positioning of the pri-mary jaw adductors found in anthropoids

Fig. 7. Three graphs showing the relative positions of the points where estimated force vectorsintersect the occlusal plane. Intersection points are shown in relation to the TMJ and most distal molar,with all values scaled to the distance of the distal molar anterior to the TMJ. The posterior end of the toothrow therefore lies at a value of 1.0 in all taxa. This figure indicates that the force vectors for the primaryjaw adductors all lie well posterior to the end of the tooth row. See Fig. 6 for box plot definitions.

496 M.A. SPENCER

suggests that natural selection has failed totake advantage of this seemingly more opti-mal region of bite force production.

Two lines of evidence, however, suggestthat the molar dentition of primates does infact lie within Region II and responds tochanges in the distribution of this region.The data presented above indicate that thetest criterion requiring all molars to liewithin Region II is uniformly complied withwhen it is assumed that the muscle resul-tant force crosses the occlusal plane directlyposterior to the tooth row. Indeed, the por-tion of the tooth row located in Region II isquite consistent, especially among taxa withsimilar dental formulae. These observationshold despite the wide range of cranial con-figurations found among anthropoid pri-mate taxa. Much of this variation is reflec-ted in the interspecific scaling relationshipsamong the parameters that influence effec-tive Region II length. The data reported inTable 2 indicate that most of these param-eters exhibit an allometric relationship withcranial size; the configuration of the mastica-tory system changes with size. Despite suchoverall trends, the portion of the tooth rowpredicted to fall within Region II is similarat all body sizes. This is true as a conse-quence of the specific relationships pre-dicted by the constrained model outlinedabove; most deviations from the observedpattern would lead to differential scalingbetween observed and predicted tooth rowdimensions.

Data on human muscle activity patternsduring forceful biting (Spencer, 1998) (seeabove) also provide support for the con-strained model that conflicts with the pat-terns of muscle position reported here. Aspredicted by the model, balancing sidemuscle activity declines relative to workingside activity as the bite point is moved fromM1 to M3. However, such changes in muscleactivity are unnecessary (and suboptimal)during maximum bite force productionwithin Region I. Therefore, if the molardentition does not usually lie within RegionII, there should be no need for the observedreduction of balancing side muscle activityas the bite point moves posteriorly along themolars. Since humans resemble other an-thropoid taxa in the relative positioning of

the adductor muscles and the relative effec-tive length and position of Region II, there isno evidence to suggest that this pattern ofmuscle activity is unique to them.

A central question presents itself whentrying to resolve the incongruity amongthese various observations: why does theestimated muscle resultant force lie moreposteriorly than it is theoretically requiredto by the constraints on distractive jointreaction forces? In the following section,several compatible answers to this questionare explored. These are related to alterna-tive loading conditions, gape, and the intro-duction of a safety factor against distractionof the TMJ.

Alternative loading conditions

The method used in this study to estimatethe position of the point where the muscleresultant force intersects the occlusal planeassumes the simple loading condition ofstatic isometric biting near centric occlu-sion. However, this intersection point mustchange during different types of loading. Ifthe avoidance of distractive joint reactionforces is potentially of selective importanceduring most types of loading, it is reasonableto expect the masticatory system to be config-ured so that such forces are not experiencedduring behaviors other than isometric biteforce production. Two such behaviors may beparticularly relevant.

First, Greaves (1978) modeled the muscleresultant force at a fixed anteroposteriorposition, the location of which is determinedby the positions of the maximum magnitudeforce vectors for each homologous musclegroup (i.e., the masseter, temporalis, andmedial pterygoid muscles). During submaxi-mum bite force production, however, a widen-ing range of anteroposterior positions mustbe available since 1) the muscles intersectthe occlusal plane at different A-P positions,2) single muscles may, through heterog-enous activity, produce forces of variableposition (Herring et al., 1979; Tonndorf etal., 1988; Blanksma et al., 1992), and 3) ithas been shown both theoretically and ex-perimentally that many combinations ofmuscle activity are adequate for the genera-tion of a bite force with a given magnitude(Koolstra et al., 1988; van Eijden et al.,


1988, 1990; van Eijden, 1991). Were theanterior temporalis muscle relatively moreactive than the masseter-medial pterygoidcomplex, for example, the muscle resultantforce would lie more anteriorly relative tothe position during maximum force produc-tion.

Rather than simply moving mediolater-ally along a transverse line, therefore, asubmaximum muscle resultant force canmore realistically be modeled at any A-Plocation within a specific zone (Fig. 8). Thesize of this zone will increase as lowermagnitude bite forces are required, and itsultimate size will be determined by thepositions of the individual muscle groups; itcannot extend anterior to the force vector forthe posterior temporalis or posterior to themedial pterygoid force. Any bite points lyingwithin this zone could fall posterior to themuscle resultant force during submaximumbite force production, resulting in joint dis-traction. Of course, the muscle resultantforce magnitude will be low near the bound-

aries of this region, and distractive jointforces are likely to be correspondingly weak.However, selection may favor a morphologyin which the dentition lies anterior to someor all potential submaximum muscle resul-tant positions rather than just the trans-verse line where the muscle resultant forceis maximum.

Mastication is a dominant type of loadingfor which these considerations may be rel-evant. Maximum magnitude bite forces arerarely produced during mastication(Hylander, 1979a), and the muscle resultantforce can therefore move both mediolater-ally and anteroposteriorly within the con-straints of the model. Furthermore, the pos-terior temporalis muscle is active duringmastication, both in the stabilization of theworking side joint and during retraction ofthe balancing side mandibular condyle nearthe end of the power stroke (Ahlgren, 1966;Møller, 1966). The force of this muscle maybe relatively high in magnitude, and it inter-sects the occlusal plane more anteriorly

Fig. 8. Occlusal view of mandible showing the zoneof potential muscle resultant force positions. Muscleresultant force magnitudes will be maximum at thecenter of this zone and will diminish with the distancefrom this position. With the muscle resultant force

positioned as assumed by Greaves (1978) (bold trans-verse line), more posterior molars will lie within thiszone; biting on them may lead to tensile joint reactionforces during submaximum force production.

498 M.A. SPENCER

than any other muscle vector (due to itsstrong posterior inclination). It has the po-tential, therefore, to pull the muscle resul-tant force anteriorly. While it probably doesnot do so during maximum isometric biting(Pruim et al., 1978; van Eijden et al., 1988,1990; van Eijden, 1990, 1991; Blanksma andvan Eijden, 1990), this possibility may berealized during mastication. If the maxi-mum magnitude muscle resultant force werepositioned as assumed by Greaves (1978),this anterior migration of the resultant dur-ing mastication would increase the possibil-ity that it would pass anterior to the biteforce, resulting in tensile joint reactionforces. The more posterior position of themuscle resultant force estimated in thisstudy may therefore be the result of selec-tion against distraction of the TMJ duringmastication or submaximum force produc-tion in general.

A second loading behavior that may influ-ence the position of the muscle resultantvector is the application of loads on moredistal teeth when the mandible is in anabducted position. The muscle vector/occlu-sal plane intersection points presented inFigure 7 were calculated under the assump-tion that the mandible is in centric occlu-sion. When applying force to a resistant foodobject, however, the mandible will be ab-ducted to a greater or lesser degree depend-ing on object size. Figure 9 illustrates thetheoretical effects of changes in mandibularposition on the relative positions of themuscle force vectors. Because the mandibu-lar insertion point of the temporalis muscleprobably lies well above the instantaneouscenter of rotation, it will move anteriorly(and inferiorly) as the mandible is abducted,while those for the masseter and medialpterygoid muscles move slightly posteriorly.The temporalis muscle vector will thereforeintersect the mandible at a more anteriorposition at wider gapes. Given the largemagnitude and more rapid rate of reposition-ing of the temporalis force vector, it is likelythat the resultant muscle force vector mi-grates anteriorly as gape increases. Itschances of falling outside of the triangle ofsupport will therefore be increased.

While very large objects are probablyrarely processed on distal molars, high mag-

Fig. 9. Representation of the changes in muscle forcevector orientation and position relative to the mandibleat differing degrees of gape. As gape increases, theanterior temporalis force vector intersects the mandiblemuch more anteriorly (r), while the masseter (andmedial pterygoid not shown) vector will intersect slightlymore posteriorly (e). These changes are based on anassumed center of rotation positioned at the circulararrow.


nitude forces could potentially be applied tothese teeth at less abducted positions (lessthan 20°), either intentionally or by inadver-tently biting on a resistant object. On theseoccasions, the degree of anterior migrationof the muscle resultant suggested in Figure9 could lead to joint distraction and possibleinjury. As with submaximum force produc-tion, this possibility could be expected toprovide selection pressures favoring a conser-vative masticatory system configuration inwhich the muscle resultant is positionedmore posteriorly than predicted by Greaves(1978). Both of these explanations rely onthe premise that the TMJ should not bedistracted.

Gape

An alternative hypothesis regarding theobserved posterior location of the muscles ofmastication in anthropoids derives from theneed to produce adequate gape. The impor-tance of abducting the mandible to someminimum gape during behaviors related tofood processing or display has frequentlybeen implicated as an important influenceon masticatory muscle morphology in mam-mals (Herring, 1972, 1975; Herring andHerring, 1974; Emerson and Radinsky, 1980;Smith, 1984; Ravosa, 1991). One of theprimary limits on gape is the length to whichthe adductor musculature can stretch. Pro-posed specializations of the masticatory sys-tem for producing wide gapes therefore in-clude factors related to both internal musclearchitecture (e.g., increased muscle fasciclelength [Herring and Herring, 1974; Herring,1975]) and aspects of muscle position andorientation. These latter may take the formof posterior migrations of the attachmentareas of the masticatory adductor muscles;when attached closer to the axis of mandibu-lar rotation, the muscles must stretch lessfor a given angle of gape.

Muscular and skeletal morphology consis-tent with predictions for specialized gapeangles have been well documented in hippo-potamids and tayassuids (Herring and Her-ring, 1974; Herring, 1975). Few such rela-tionships have been found within primates(Smith, 1984; Ravosa, 1991). However, it isclear that the need for some minimum gape

potential must be an important selectivefactor, even in animals not specialized forvery wide gapes; all primates need to opentheir mouths to get food in. Furthermore,many open them wide for display purposes.The degree to which this alternative require-ment has acted to constrain masticatorymuscle position and morphology is difficultto estimate, but it must be included as animportant potential influence on muscle po-sition.

Safety factors

The hypotheses presented above may par-tially explain why the masticatory musclesare positioned so far posteriorly, and whenconsidered together they make it unlikelythat the muscle resultant could be posi-tioned where Greaves (1978) assumed it tobe. However, on their own they cannot recon-cile the contradictory conclusions derivedfrom the experimental and comparative mor-phological data summarized here. None ofthese hypotheses explains the observed re-duction of balancing side muscle activityreported for humans (Spencer, 1998). Thesescenarios also fail to explain the observedcorrespondence between postcanine toothrow dimensions and the predictions of themodified constrained model.

There is an additional explanation, how-ever, for the posterior location of the mastica-tory muscles that is consistent with thesedata. Greaves’ (1978) constrained model pre-dicts that during maximum bite force produc-tion within Region II the muscle resultantforce will lie directly on the side of thetriangle of support that connects the bitepoint and the balancing side joint reactionforce. This position maximizes muscle resul-tant force magnitude by allowing the great-est balancing side muscle activity that doesnot lead to joint distraction. It therefore alsoresults in zero magnitude working side jointreaction forces. Slight deviations from thisposition will either bring the muscle resul-tant into the triangle of support or (probablywith equal likelihood) drop the muscle resul-tant outside of the triangle of support. Suchdeviations must be unavoidable during thedynamic process of loading even homoge-neously textured foods and may be quite

500 M.A. SPENCER

large during the processing of less predict-able objects. The optimal muscle resultantposition assumed by Greaves (1978) maytherefore be associated with relatively highfrequencies of distractive joint loading.

The frequency with which the muscleresultant falls outside of the triangle ofsupport can theoretically be reduced by posi-tioning the resultant further within thetriangle, in effect creating a buffer zonearound the borders of the triangle of supportand a ‘‘sweet spot’’ of more stable resultantpositions within the triangle (Fig. 10a). Theexistence of such a buffer zone could providean important safety factor and was brieflytheorized by Greaves (1983:360): ‘‘For in-

creased stability of the jaw . . . it would beadvantageous to have the muscle resultantlie well within the triangle of support ratherthan near the edge.’’ However, Greaves didnot recognize that the avoidance of a bufferzone changes our expectations regarding theinteraction between muscle resultant posi-tion and Region II distribution.

Region II is defined as the area of poten-tial bite points that will require a shift of themuscle resultant force toward the workingside (to avoid tensile working side jointreaction forces). As shown in Figure 10a,however, avoidance of the buffer zone willrequire the muscle resultant to move towardthe working side during biting at more

Fig. 10. Occlusal view of man-dible showing the effects on thedistribution of Region II of requir-ing a buffer zone around the perim-eter of the triangle of support. a: Ifno buffer zone exists, a muscle re-sultant force (j) positioned at theposterior end of the tooth row neednot move toward the working sideunless the bite point is positionedat or posterior to the closed dia-mond. However, if the muscle resul-tant force is required to lie wellwithin the triangle of support (inthe sweet spot), it will be forcedtoward the working side duringbiting at more anterior positions(at or posterior to the open dia-mond). b: Triangles of support forthe most anterior (r) and mostposterior (e) bite points in RegionII, with their associated bufferzones. A muscle resultant force po-sitioned as assumed by Greaves(1978) will lie within the bufferzone, while a more posteriorly posi-tioned muscle resultant force willnot. Due to the incorporation of abuffer zone, the predicted distribu-tion of Region II is the same evenwhen the muscle resultant force ispositioned far posterior to the lastmolar.


anterior bite points than predicted fromGreaves’ (1978) model. Furthermore, bitepoints positioned just anterior to the muscleresultant force (i.e., on the most distal molarin Greaves’ model) will be associated withtriangles of support whose buffer zone en-closes the muscle resultant. The posteriorend of Region II should therefore lie suffi-ciently anterior to the muscle resultant sothat the resultant will not typically be posi-tioned in the buffer zone. In this bufferedmodel, therefore, the portion of the toothrow that will fall in Region II will be shiftedanteriorly compared to the model proposedby Greaves (1978). Additionally, a gap willbe present between Region II and the A-Pposition of the muscle resultant force.

Figure 10b shows how the predictions ofGreaves’ (1978) model regarding the posi-tion and length of Region II can be main-tained with a modified, more posteriormuscle resultant position if the proposedbuffer zone and sweet spot exist. In thisfigure, Region II is distributed along thepostcanine dentition as predicted based on amuscle resultant positioned at the posteriorend of the tooth row. However, a more poste-riorly positioned muscle resultant is used,and it is maintained within the sweet spot ofthe triangle of support at all bite pointpositions. Thus, the predictions of the buff-ered model regarding the length and distri-bution of Region II (and therefore the postca-nine dentition) are similar to those ofGreaves’ (1978) constrained model, despitethe conclusion that overall positional rela-tionships within the masticatory system dif-fer in the two models.

Implications for dietary adaptation

The data presented in this paper suggestthat a constraint on distraction of the TMJlimits the evolution of the masticatory sys-tem in anthropoid primates. This constraintmay also alter the adaptive response of themasticatory system to specific dietary selec-tion pressures. Primates eat a wide varietyof foods and experience of range of selectionpressures related to generating masticatoryforces. In the unconstrained lever model,predictions regarding the relative adaptivevalue of specific phenotypes derive fromsimple lever mechanics. For example, selec-

tion for efficient or high magnitude forceproduction is expected to favor forms with arelatively high ratio of muscle force momentarm length to bite force moment arm length.This expectation holds for all bite pointlocations.

In some regards, the predictions of theconstrained lever model (and the bufferedversion of this model proposed here) regard-ing dietary adaptation are similar to those ofan unconstrained model. For biting on theanterior dentition (specifically in Region I),the predictions of the two models are thesame since the constraint on joint distrac-tion does not influence patterns of forceproduction in this region. However, duringpostcanine biting this constraint limits thechanges that may occur. For example, selec-tion for high magnitude force production onthe molars is commonly expected to lead tothe anterior migration of the masticatorymuscles and/or the posterior migration ofthe molars during evolution (Du Brul, 1977;Hylander, 1979a; Rak, 1983; Bouvier, 1986;Demes and Creel, 1988). However, either ofthese changes would alter the relative posi-tions of the postcanine dentition and Re-gions I, II, and III. They are therefore ex-pected to have additional mechanicalconsequences within the constrained model.In this example, the posterior portion of thetooth row would be forced into Region III,increasing the chances of distracting theTMJ. Furthermore, because the predictedmaximum magnitude bite forces do not in-crease at more posterior bite points withinRegion II, changes in tooth position withinthis region may provide little or no advan-tage for high magnitude biting. Species expe-riencing selection for high magnitude postca-nine force production are therefore notexpected to exhibit changes in the relativeanteroposterior positions of the muscles andteeth. Instead, the constrained model pre-dicts that selection will favor more mediallypositioned tooth rows, a configurationalchange that allows more balancing sidemuscle force to be produced during forcefulbiting at points in Region II.

A diverse range of morphological patternsevident within anthropoid primates maystem from the selective trade-off between

502 M.A. SPENCER

increasing bite force magnitudes and avoid-ing joint distraction. For example, it hasbeen proposed previously that a relativelyhigh rate of third molar agenesis amongInuit is related to a configurational special-ization for intensive incisor loading in thisgroup (Spencer and Demes, 1993). In thiscase, a more posterior position for the denti-tion (relative to behaviorally unspecializedpopulations) and a relatively anterior posi-tion of the superficial masseter and anteriortemporalis muscles have been attributed toselection for greater efficiency of force pro-duction at Region I bite points (Hylander,1977; Spencer and Demes, 1993). Thesechanges may have led to an increased fre-quency of joint distraction during distalmolar biting. Third molar agenesis mightthen be an adaptive response related tominimizing this occurrence. Similar configu-rational and dental changes characterizeother anthropoid groups that are behavior-ally specialized for intensive force produc-tion on the incisors or canines, such ascallitrichids, some pitheciines (Cacajao andChiropotes), and Cebus. These groups eachexhibit either relatively small third molarteeth or the evolutionary loss of their thirdmolars. Furthermore, Cacajao and Chiro-potes are characterized by molarized distalpremolars (Rosenberger, 1979; Kay, 1990),an observation that is curious given theirapparent reduction of molar occlusal area.These aspects of postcanine tooth form (andpresumably function) are consistent withthe altered position of the dentition relativeto Region II brought about by selection forintensive incisor or canine loading.

CONCLUSIONS

The anthropoid primate masticatory sys-tem functions as a lever during the genera-tion of masticatory forces. However, thissimple model may be inadequate in attempt-ing to explain many aspects of masticatorysystem diversity or function that are relatedto force production. For example, the datapresented here indicate that the positionalrelationships among the muscles, joints, andpostcanine dentition vary in finite and pre-dictable ways across the full range of anthro-poids. The consistency of this pattern sug-

gests that there has been some constantlimiting influence on the configuration of themasticatory system during the evolution ofthis group. Such an influence is difficult toderive from the standard unconstrained le-ver model or from any presumed balanceamong variable and competing selectionpressures.

The observed patterning within mastica-tory system diversity may be explained byGreaves’s (1978) assumption that the TMJshould not be loaded by distractive forces.Greaves argued that the avoidance of suchforces would be evident in changes in masti-catory muscle activity along the tooth rowand in a consistent positional interactionamong the masticatory muscles, the joints,and the postcanine dentition. Data on hu-man muscle activity support the first ofthese predictions (Spencer, 1998). The re-sults of the current study support the secondhypothesis. However, it is also apparent thatthe exact positional relationships predictedby Greaves are not realistic and that theconstrained lever model must also be incom-plete.

The observed discrepancy between mor-phological and experimental data suggeststhat selection favors the existence of safetyfactors that reduce the chances of joint dis-traction during diverse and dynamic loadingsituations. This buffered model of mastica-tory force production provides new expecta-tions regarding the adaptive response of theprimate masticatory system to dietary selec-tion pressures. It may therefore help toexplain a wide range of morphological pat-terns exhibited by living and extinct pri-mates.

ACKNOWLEDGMENTS

For their generous help during this re-search, I thank Dr. Brigitte Demes, Dr. JackStern, Jr., Dr. William Hylander, and Dr.William Jungers. Special thanks go to Dr.Lillian Spencer, Kirk Johnson, and twoanonymous reviewers for their helpful com-ments on this paper. Additionally, I amgrateful to the following people for allowingme access to the specimens in their care:Guy Musser at the American Museum ofNatural History, Bruce Patterson and Bill


Stanley at the Field Museum, Maria Rutz-moser at the Museum of Comparative Zool-ogy, Paula Jenkins at the Natural HistoryMuseum in London, and Richard Thoring-ton at the National Museum of NaturalHistory.

APPENDIX

The five dimensions used in the calcula-tion of effective Region II length are illus-trated in Figure 11. The algorithm used forthis calculation is as follows (refer to Fig. 11for a diagram of the parameters that are

calculated at each step in this algorithm):

P1° 5 1D

A 2 1(A 2 B)

2 22P2 5 tan (P1) p 1B2 2P3 5 1 P2

sin (E)2P4 5 ÎP3

2 2 P22

P5° 5 arctan 1D 2 P2

P4 1 C2P6 5 1 P2

sin (P5°)2P7 5 ÎP6

2 2 P22

Effective Region II Length 5 P7 1 P4.

LITERATURE CITED

Ahlgren J. 1966. Mechanism of mastication: a quantita-tive cinematographic and electromyographic study ofmasticatory movements in children, with special refer-ence to occlusion of the teeth. Acta Odont Scand24:1–109.

Anapol F, Lee S. 1994. Morphological adaptation to dietin platyrrhine primates. Am J Phys Anthropol 94:239–262.

Bakke M, Møller E. 1992. Craniomandibular disordersand masticatory muscle function. Scand J Dent Res100:32–38.

Beecher RM. 1977. Function and fusion at the mandibu-lar symphysis. Am J Phys Anthropol 47:325–336.

Beecher RM. 1979. Functional significance of the man-dibular symphysis. J Morphol 139:117–130.

Bell WE. 1983. Understanding temporomandibular bio-mechanics. J Craniomandibular Pract 1:27–33.

Bell WE. 1990. Temporomandibular disorders: classifica-tion, diagnosis, management. Chicago: Year BookMedical Publishers.

Blanksma NG, van Eijden TMGJ. 1990. Electromyo-graphic heterogeneity in the human temporalismuscle. J Dent Res 69:1686–1690.

Blanksma NG, van Eijden TMGJ, Weijs WA. 1992.Electromyographic heterogeneity in the human mas-seter muscle. J Dent Res 71:47–52.

Bouvier M. 1986. A biomechanical analysis of mandibu-lar scaling in Old World monkeys. Am J Phys Anthro-pol 69:473–482.

Carlson DS, van Gerven DP. 1977. Masticatory functionand post-pleistocene evolution in Nubia. Am J PhysAnthropol 46:495–506.

Cartmill M. 1974. Daubentonia, Dactylopsila, woodpeck-ers and klinorhynchy. In: Martin RD, Doyle GA,Walker AC, editors. Prosimian biology. London: Duck-worth. p 655–670.

Cheverud JM. 1982. Phenotypic, genetic, and environ-mental morphological integration in the cranium.Evolution 36:499–516.

Fig. 11. Frontal and lateral views of relevant masti-catory system components illustrating the five variablesand derivative parameters used in the calculation ofeffective Region II length. The shaded zone visible in thefrontal view is the triangle of support (viewed obliquely)for the most anterior bite point in Region II. Thebalancing side edge of the triangle intersects the midlinemuscle resultant force at the small closed circle. Thistriangle is seen edge-on in the lateral view as a diagonalline passing from the TMJ to the occlusal plane (horizon-tal line). A, biarticular breadth; B, palate breadth at M1;C, anteroposterior distance from articular eminence(● or C) to point of intersection of muscle resultant forceand occlusal plane (j); D, height of articular eminenceabove occlusal plane; E, angle of muscle resultant forcevector relative to occlusal plane (90° 5 perpendicular).

504 M.A. SPENCER

Clarke M. 1980. The reduced major axis of a bivariatesample. Biometrika 67:441–446.

Cole TM. 1997. RMA: software for reduced major axisregression. Baltimore: The Johns Hopkins UniversityDepartment of Cell Biology and Anatomy.

Daegling DJ. 1989. Biomechanics of cross-sectional sizeand shape in the hominoid mandibular corpus. Am JPhys Anthropol 80:91–106.

Daegling DJ. 1992. Mandibular morphology and diet inthe genus Cebus. Int J Primatol 13:545–570.

Darroch JN, Mosimann JE. 1985. Canonical and princi-pal components of shape. Biometrika 72:241–252.

Demes B. 1987. Another look at an old face: biomechan-ics of the Neandertal facial skeleton reconsidered. JHum Evol 16:297–303.

Demes B, Creel N. 1988. Bite force, diet, and cranialmorphology of fossil hominoids. J Hum Evol 17:657–670.

Druzinsky RE, Greaves WS. 1979. A model to explainthe posterior limit of the bite point in reptiles. JMorphol 160:165–168.

Du Brul EL. 1974. Origin and evolution of the oralapparatus. In: Kawamura Y, editor. Frontiers of oralphysiology. Basel: Karger. p 1–30.

Du Brul EL. 1977. Early hominid feeding mechanisms.Am J Phys Anthropol 47:305–320.

Emerson SB, Radinsky LR. 1980. Functional analysis ofsaber-tooth cranial morphology. Paleobiology 6:295–312.

Fleagle JG. 1988. Primate adaptation and evolution.New York: Academic Press.

Ford SM, Davis LC. 1992. Systematics and body size:implications for feeding adaptations in New Worldmonkeys. Am J Phys Anthropol 88:415–469.

Greaves WS. 1978. The jaw lever system in ungulates: anew model. J Zool (Lond) 184:271–285.

Greaves WS. 1982. A mechanical limitation on theposition of the jaw muscles of mammals: the one-thirdrule. J Mammal 63:261–266.

Greaves WS. 1983. A functional analysis of carnassialbiting. Bio J Linn Soc 20:353–363.

Greaves WS. 1985. The generalized carnivore jaw. Zool JLinn Soc 85:267–274.

Greaves WS. 1988. The maximum average bite force fora given jaw length. J Zool (Lond) 214:295–306.

Greaves WS. 1995. Functional predictions from theoreti-cal models of the skull and jaws in reptiles andmammals. In: Thomason JJ, editor. Functional mor-phology in vertebrate paleontology. Cambridge: Cam-bridge University Press. p 99–115.

Gysi A. 1921. Studies on the leverage problem of themandible. Dental Digest 27:74–84, 144–150, 203–208.

Herring SW. 1972. The role of canine morphology in theevolutionary divergence of pigs and peccaries. J Mam-mal 53:500–512.

Herring SW. 1975.Adaptations for gape in the hippopota-mus and its relatives. Forma e Functio 8:85–100.

Herring SW, Herring SE. 1974. The superficial masseterand gape in mammals. Am Naturalist 108:561–576.

Herring SW, Grimm AF, Grimm BR. 1979. Functionalheterogeneity in a multipinnate muscle. Am J Anat154:563–576.

Hylander WL. 1977. The adaptive significance of Es-kimo craniofacial morphology. In: DahlbergAA, GraberTM, editors. Orofacial growth and development. TheHague: Mouton Publishers. p 129–170.

Hylander WL. 1979a. The functional significance ofprimate mandibular form. J Morphol 160:223–240.

Hylander WL. 1979b. An experimental analysis of tem-poromandibular joint reaction force in macaques. AmJ Phys Anthropol 51:433–456.

Hylander WL. 1984. Stress and strain in the mandibu-lar symphysis of primates: a test of competing hypoth-eses. Am J Phys Anthropol 64:1–46.

Hylander WL. 1985. Mandibular function and temporo-mandibular joint loading. In: Carlson DS, McNamaraJA, Ribbens KA, editors. Developmental aspects oftemporomandibular joint disorders. Monograph 16,Craniofacial Growth Series, Center for Human Growthand Development. Ann Arbor, MI: University of Michi-gan. p 19–35.

Hylander WL, Bays R. 1979. An in vivo strain-gaugeanalysis of squamosal-dentary joint reaction forceduring mastication and incision in Macaca mulattaand Macaca fascicularis. Arch Oral Biol 24:689–697.

Hylander WL, Johnson KR. 1994. Jaw muscle functionand wishboning of the mandible during mastication inmacaques and baboons. Am J Phys Anthropol 94:523–547.

Jablonski NG. 1993. Theropithecus: the rise and fall of aprimate genus. Cambridge: Cambridge UniversityPress.

Kay RF. 1990. The phyletic relationships of extant andfossil Pitheciinae (Platyrrhini, Anthropoidea). J HumEvol 9:175–208.

Kay RF, Hylander WL. 1978. The dental structure ofmammalian folivores with special references to Pri-mates and Phalangeroidea (Mammalia). In: Montgom-ery GG, editor. The biology of arboreal folivores.Washington, DC: Smithsonian Institute Press. p 173–191.

Koolstra JH, van Eijden TMGJ, Weijs WA, Naeije M.1988. A three-dimensional mathematical model of thehuman masticatory system predicting maximum pos-sible bite-forces. J Biomech 21:563–576.

Mansour RM, Reynik RJ. 1975. In vivo occlusal forcesand moments: I. Forces measured in terminal hingeposition and associated moments. J Dent Res 54:114–120.

McCollum M. 1994. Mechanical and spatial determi-nants of Paranthropus facial form. Am J Phys Anthro-pol 93:259–273.

McKay GS. 1992. The structure and function of thetemporomandibular joint. Br Dent J 173:127–133.

Møller E. 1966. The chewing apparatus: an electromyo-graphic study of the action of the muscles of mastica-tion and its correlation to facial morphology. ActaPhysiol Scand 69(Suppl 280):1–129.

Mosimann JE, James FC. 1979. New statistical methodsfor allometry with application for Florida red-wingedblackbirds. Evolution 33:444–459.

Moss ML, Young RW. 1960. A functional approach tocraniology. Am J Phys Anthropol 18:281–292.

Pruim GJ, Ten Bosch JJ, De Jongh HJ. 1978. Jawmuscle EMG-activity and static loading of the man-dible. J Biomech 11:389–395.

Pruim GJ, De Jongh HJ, Ten Bosch JJ. 1980. Forcesacting on the mandible during bilateral static bite atdifferent bite force levels. J Biomech 13:755–763.

Rak Y. 1983. The australopithecine face. New York:Academic Press.

Rak Y. 1986. The Neanderthal: a new look at an old face.J Hum Evol 15:151–164.

Ravosa MJ. 1990. Functional assessment of subfamilyvariation in maxillomandibular morphology amongOld World monkeys. Am J Phys Anthropol 82:199–212.

Ravosa MJ. 1991. Structural allometry of the prosimianmandibular corpus and symphysis. J Hum Evol 20:3–20.

Rosenberger A. 1979. Phylogeny, evolution and classifi-cation of New World monkeys (Platyrrhini, Primates).PhD dissertation, City University of New York.


Rosenberger AL. 1992. Evolution of feeding niches inNew World monkeys. Am J Phys Anthropol 88:525–562.

Schwenk K. 1995. A utilitarian approach to evolutionaryconstraint. Zoology 98:251–262.

Smith FH, Paquette SP. 1989. The adaptive basis ofNeandertal facial form, with some thoughts on thenature of modern human origins. In: Trinkaus E,editor. The emergence of modern humans. Cambridge:Cambridge University Press. p 181–210.

Smith K. 1993. The form of the feeding apparatus interrestrial vertebrates: Studies of adaptation andconstraint. In: Hanken J, Hall B, editors. The skull,vol 3: functional and evolutionary mechanisms. Chi-cago: The University of Chicago Press. p 150–196.

Smith RJ. 1978. Mandibular biomechanics and temporo-mandibular joint function in primates. Am J PhysAnthropol 49:341–350.

Smith RJ. 1984. Comparative functional morphology ofmaximum mandibular opening (gape) in primates. In:Chivers DJ, Wood BA, Bilsborough A, editors. Foodacquisition and processing in primates. New York:Plenum. p 231–255.

Spencer MA. 1995. Masticatory system configurationand diet in anthropoid primates. PhD dissertation,State University of New York, Stony Brook.

Spencer MA. 1998. Force production in the primatemasticatory system: electromyographic tests of biome-chanical hypotheses. J Hum Evol 34:25–54.

Spencer MA, Demes AB. 1993. Biomechanical analysisof masticatory system configuration in Neandertalsand Inuit. Am J Phys Anthropol 91:1–20.

Spencer MA, Spencer GS. 1993. MacMorph data acquisi-tion package. Stony Brook, NY: State University ofNew York Department of Anthropology.

Spencer MA, Spencer GS. 1995. Technical note: video-based three-dimensional morphometrics. Am J PhysAnthropol 96:443–453.

Throckmorton GS. 1985. Quantitative calculations oftemporomandibular joint reaction forces—2. The im-portance of the direction of the jaw muscle forces. JBiomech 18:453–461.

Throckmorton GS. 1989. Sensitivity of temporomandibu-lar joint force calculations to errors in muscle forcemeasurements. J Biomech 22:455–468.

Throckmorton GS, Groshan GJ, Boyd SB. 1990. Muscleactivity patterns and control of temporomandibularjoint loads. J Prosthet Dent 63:685–695.

Tonndorf ML, Sasaki K, Hannam AG. 1988. Single-wirerecording of regional activity in the human massetermuscle. Brain Res Bull 23:155–159.

Tsutakawa R, Hewett J. 1977. Quick test for comparingtwo populations with bivariate data. Biometrics 33:215–219.

van Eijden TMGJ. 1990. Jaw muscle activity in relationto the direction and point of application of bite force. JDent Res 69:901–905.

van Eijden TMGJ. 1991. Three-dimensional analyses ofhuman bite-force magnitude and moment. Arch OralBiol 36:535–539.

van Eijden TMGJ, Koolstra JH, Brugman P, Weijs WA.1988. A feedback method to determine the threedimensional bite force capabilities of the human mas-ticatory system. J Dent Res 67:450–454.

van Eijden TMGJ, Brugman P, Weijs WA, Oosting J.1990. Coactivation of jaw muscles: recruitment orderand level as a function of bite force direction andmagnitude. J Biomech 23:475–485.

Walker AC. 1978. Functional anatomy of oral tissues:Mastication and deglutition. In: Shaw JH, SweeneyEM, Cappuccino CC, Meller SM, editors. Textbook oforal biology. Philadelphia: W.B. Saunders Co. p 277–296.

Ward SC, Molnar S. 1980. Experimental stress analysisof topographic diversity in early hominid gnathicmorphology. Am J Phys Anthropol 53:383–395.

Weijs WA, van Spronsen P. 1992. Variation in adulthuman jaw muscle size: computer models predictingthe biomechanical consequences of the variation. In:Davidovitch Z, editor. The biological mechanisms oftooth movement and craniofacial adaptation. Colum-bus, OH: The Ohio State University College of Den-tistry. p 549–557.

Weishampel DB. 1993. Beams and machines: modelingapproaches to the analysis of skull form and function.In: Hanken J, Hall BK, editors. The skull: vol 3.Chicago: University of Chicago Press. p 303–344.

Werdelin L. 1986. Comparison of skull shape in marsu-pial and placental carnivores. Aust J Zool 34:109–117.

Werdelin L. 1987. Jaw geometry and molar morphologyin marsupial carnivores: analysis of a constraint andits macroevolutionary consequences. Paleobiology 13:342–350.

Werdelin L. 1988. Circumventing a constraint: the caseof Thylacoleo (Marsupialia: Thylacoleonidae). Aust JZool 36:565–571.

White TD. 1991. Human osteology. San Diego: AcademicPress, Inc.

Wolff JEA. 1984. A theoretical approach to solve the chinproblem. In: Chivers DJ, Wood BA, Bilsborough A,editors. Food acquisition and processing in primates.New York and London: Plenum Press. p 391–405.

506 M.A. SPENCER

constraints on masticatory system evolution in anthropoid primates

Documents