The Jaw Adductors of Strepsirrhines in Relation to Body Size, Diet, and Ingested Food Size

Authors


Abstract

Body size and food properties account for much of the variation in the hard tissue morphology of the masticatory system whereas their influence on the soft tissue anatomy remains relatively understudied. Data on jaw adductor fiber architecture and experimentally determined ingested food size in a broad sample of 24 species of extant strepsirrhines allows us to evaluate several hypotheses about the influence of body size and diet on the masticatory muscles. Jaw adductor mass scales isometrically with body mass (β = 0.99, r = 0.95), skull size (β = 1.04, r = 0.97), and jaw length cubed (β = 1.02, r = 0.95). Fiber length also scales isometrically with body mass (β = 0.28, r = 0.85), skull size (β = 0.33, r = 0.84), and jaw length cubed (β = 0.29, r = 0.88). Physiological cross-sectional area (PCSA) scales with isometry or slight positive allometry with body mass (β = 0.76, r = 0.92), skull size (β = 0.78, r = 0.94), and jaw length cubed (β = 0.78, r = 0.91). Whereas PCSA is isometric to body size estimates in frugivores, it is positively allometric in folivores. Independent of body size, fiber length is correlated with maximum ingested food size, suggesting that ingestive gape is related to fiber excursion. Comparisons of temporalis, masseter, and medial pterygoid PCSA in strepsirrhines of different diets suggest that there may be functional partitioning between these muscle groups. Anat Rec, 2011. © 2011 Wiley-Liss, Inc.

Recently, great strides have been made in our understanding of primate chewing muscle fiber architecture. Primate chewing muscle mass and fiber length (FL) scale isometrically to body size and muscle cross-sectional area scales isometrically or with slight positive allometry (Anapol et al.,2008; Perry and Wall,2008). Callitrichids demonstrate that the need for gape in gouging has driven FL and force production (Taylor and Vinyard,2004; Eng et al.,2009; Taylor et al.,2009); whereas, some primates have both great jaw adductor cross-sectional area and long muscle fibers, likely as an adaptation to processing large, resistant food items (Taylor and Vinyard,2009). Recent analysis of electromyographic (EMG) activity and fiber architecture of the jaw adductors in primates suggests that the constraints on the evolution of the temporalis might be different from those affecting the masseter (Vinyard and Taylor,2010).

In this contribution, we use fine-grained data on an expanded sample of strepsirrhine jaw adductors to test general hypotheses about how body size and diet might influence the dimensions and architecture of the jaw adductor muscles. In light of recent arguments about the strong influence of gape on jaw muscle architecture (Taylor and Vinyard,2004; Eng et al.,2009; Taylor et al.,2009), we also examine the relationship between jaw adductor FL and ingested food size in our sample.

We report on the mass, FL, physiological cross-sectional area (PCSA), and pinnation of the jaw adductor muscles of 24 species of strepsirrhine primates that span a large range of body sizes and diets (Table 1), and discuss how muscle dimensions relate to body size, diet, and experimentally determined ingested food size. We also discuss the influence of pinnation and the merits and drawbacks of adjusting PCSA based on pinnation.

Table 1. Specimens dissected
SpeciesSpecimenPreservation
  • a

    These specimens were initially frozen; then they were transferred to 10% formalin.

Avahi lanigerAMNH 17050170% ethanol
Cheirogaleus mediusDUPC 651mFrozen
Eulemur collarisDUPC 5800mFrozen
Eulemur coronatusDUPC 6251mFrozen
Eulemur macaco flavifronsDUPC 6394fFrozen
Eulemur mongozDUPC 6468fFrozen
Eulemur rubriventerDUPC 6631mFrozen
Galago moholiDUPC 2041mFrozen
Galago senegalensis braccatusASU Ef10% Formalin
Galago senegalensis braccatusASU Jf10% Formalin
Galagoides demidoffNCZP 510f10% Formalin
Galagoides demidoffNCZP 507f10% Formalin
Hapalemur griseusDUPC 1322fFrozen
Hapalemur griseusDUPC 1353fFrozen
Lemur cattaBAA C2f10% Formalina
Lepilemur leucopusAMNH 17079070% Ethanol
Microcebus murinusDUPC 889fFrozen
Microcebus murinusDUPC 874fFrozen
Mirza coquereliDUPC 363fFrozen
Nycticebus coucangBAA C6m10% Formalina
Nycticebus pygmaeusDUPC 1931fFrozen
Otolemur crassicaudatusORPC 1f10% Formalin
Otolemur crassicaudatusORPC 2f10% Formalin
Otolemur crassicaudatusDUPC 1731mFrozen
Otolemur garnettiiASU 1m10% Formalin
Otolemur garnettiiHaines Am10% Formalin
Perodicticus pottoAMNH 200640f70% Ethanol
Propithecus coquereliDUPC 6560mFrozen
Propithecus coquereliDUPC 6110fFrozen
Propithecus coquereliDUPC 5705fFrozen
Propithecus diademaDUPC 6563mFrozen
Propithecus tattersalliDUPC 6197mFrozen
Varecia rubraAMNH 20139570% Ethanol
Varecia rubraDUPC 6769mFrozen

Finally, we examine hypotheses on the functions of the different components of the jaw adductor musculature—or “division of labor” (Davis,1955; Smith and Savage,1959; Davis,1964; Cachel,1979; Anapol and Lee,1994). We examine these by comparing samples grouped by diet to detect differences in how the jaw adductor PCSA is partitioned across the individual jaw adductors.

SCALING PREDICTIONS

Previous research suggests that chewing muscle mass, FL, and PCSA in strepsirrhines scale isometrically with body size (Cachel,1979,1984; Anapol et al.,2008; Perry and Wall,2008). In an earlier study (Perry and Wall,2008), we assessed the scaling of chewing muscle dimensions in a sample of 12 strepsirrhine species and found isometric scaling of muscle mass (reduced major axis slope = 1.01) and FL (reduced major axis slope = 0.33) relative to body mass. The observed slope for PCSA (reduced major axis slope = 0.73) scaled greater than the expected slope of isometry (0.67), though the difference was not statistically significant. Since then, we have added new taxa to our sample, including many lemurids (Varecia and several species of Eulemur), and three species of Propithecus. We predict that the slopes for regressions of PCSA against body size will remain statistically indistinguishable from isometry.

HYPOTHESES ABOUT FIBER LENGTH AND INGESTED FOOD SIZE

Perry and Hartstone-Rose (2010) found that, in a captive population of strepsirrhines, experimentally determined maximum ingested food size (Vb: the largest piece of food that individuals ingest whole) scales isometrically with body mass, and that diet explains much of the variation in Vb that is not correlated with body mass. The size of food at ingestion likely has an adaptive influence on jaw adductor FL because fibers must stretch when the jaws are opened to accommodate the food and then must generate tension to break it. Clearly, this is not the only potential influence on FL. Other factors include head size, skull geometry, chew speed, and the location of the instantaneous center of rotation relative to the muscle parts and teeth (e.g., Herring and Herring,1974). We know from previous analyses that FL is highly correlated with body mass, as is Vb (Perry and Hartstone-Rose,2007a; Perry and Wall,2008; Perry and Hartstone-Rose,2010). Here, we predict that FL is also correlated with Vb independent of body mass; that is, strepsirrhines that ingest small food items (that is, foods that are small in at least on plane, such as leaves) likely have short jaw adductor fibers, whereas those that ingest large food items (e.g., large fruits) likely have long jaw adductor fibers.

HYPOTHESES ABOUT THE DIVISION OF LABOR IN THE JAW ADDUCTORS

Several hypotheses about the specific functions of the parts of the jaw adductor musculature have been proposed (Davis,1955; Smith and Savage,1959; Davis,1964; Turnbull,1970; Gingerich,1971; Cachel,1979; Hylander and Johnson,1985; Axmacher and Hofmann,1988; Solounias and Dawson-Saunders,1988; Cordell,1991; Anapol and Lee,1994; Solounias et al.,1995). Data generated by the present study can be used to examine whether there are morphological correlates that support these hypotheses as they pertain in strepsirrhines. In this study, we address only those hypotheses that consider muscle function in light of gross dietary differences because we are using comparative data on broad dietary categories, not performance data on food processing. Analyses of EMG, fiber architecture, and muscle leverage can be incorporated in a more fine-grained evaluation of hypotheses regarding division of labor in the masticatory system.

One goal of EMG studies of primate chewing muscles is to identify differences in the activity of the individual chewing muscles that track differences in food properties; this would indicate functional division of labor within the chewing musculature (Hylander and Johnson,1985; Hylander et al.,2005). There is recent evidence from baboons for functional partitioning in temporalis activity based on food properties (Wall et al.,2008). Also, chewing muscle architecture and electromyography are not equally linked in the temporalis and the masseter, suggesting that selection has operated differently on these two muscle groups (Vinyard and Taylor,2010).

Even if muscle activation patterns do not vary based on food properties, the sizes, architecture, and locations of the chewing muscles could do so. For example, if the masseter is better suited to breaking down leaves than the temporalis (e.g., because its fibers are oriented more horizontally), then a folivorous primate benefits from having a large masseter even if the pattern of activation of the chewing muscles is the same for all foods it consumes. Thus, the size of one muscle of mastication relative to the sizes of the others might signal diet. In this case, size should refer to PCSA because PCSA is a reflection of a muscle's ability to produce force. The following hypotheses pertain.

The first hypothesis is that the temporalis is especially large (in PCSA) in frugivorous primates. Herring and Herring (1974) suggested that the masseter muscle is especially vulnerable to stretch when the mouth is opened widely because, of all the masticatory muscle fibers, the anterior fibers of this muscle are at the greatest distance from the temporomandibular joint. Our observations suggest that the anterior fibers of the medial pterygoid are similarly vulnerable. If, as suggested by the bite size study of Perry and Hartstone-Rose (2010), frugivorous strepsirrhines ingest at larger sizes than folivores, then perhaps these strepsirrhines emphasize the less stretch-vulnerable temporalis over masseter and medial pterygoid. Furthermore, the ingestion of fruits likely requires wider gapes than does the ingestion of leaves (e.g., Cachel,1979). We predict that temporalis PCSA relative to total jaw adductor PCSA will be a significantly greater ratio in frugivores than in folivores.

A second hypothesis is that the masseter is large (in PCSA)—relative to the other jaw adductors—in folivorous primates. This hypothesis follows from the observation that estimated leverage for the masseter is especially great in platyrrhine folivores (Anapol and Lee,1994). Here, the hypothesis is evaluated in strepsirrhines. The temporalis is oriented more vertically than the masseter, therefore, enlargement of the temporalis likely produces a more vertically oriented muscle resultant while enlargement of the masseter would produce a more horizontally oriented resultant. The latter would be more effective at processing leaves given the occlusal patterns of primate molars. We predict that the ratio of masseter PCSA to temporalis PCSA will be significantly greater in folivorous strepsirrhines than in other ones. Because, like the masseter, the medial pterygoid is relatively horizontally oriented and likely plays an important role in processing leaves, we have chosen to compare masseter PCSA to temporalis PCSA rather than masseter PCSA to total jaw adductor PCSA.

A third hypothesis, suggested by Davis for carnivorans, relates to how the cross-sectional area of the masseter muscle group is divided up among its three component muscles (Davis,1955,1964). The hypothesis is that the zygomatico-mandibularis muscle is larger (in PCSA) in herbivores and the superficial masseter muscle is larger (in PCSA) in carnivores, whereas there is no specified expectation for the deep masseter (the intermediate layer of the group). Davis discovered this pattern in the masses of these muscles in carnivorans of varying diets and suggested it might pertain to mammals in general. The reason for this pattern, according to Davis, is that the fibers of the superficial masseter are more vertically oriented and thus ideally situated for producing fast vertical closure of the jaws, and that by comparison the fibers of the zygomatico-mandibularis are transversely oriented and ideal for providing transverse force during grinding. In our sample, we evaluate this hypothesis by comparing herbivorous strepsirrhines to insectivorous strepsirrhines. We predict that the ratio of superficial masseter PCSA to zygomatico-mandibularis PCSA will be greater in insectivorous strepsirrhines.

The fourth hypothesis follows from our consideration of the fiber architecture of medial pterygoid in strepsirrhines. We propose that the medial pterygoid is larger (in PCSA) in strepsirrhines that must generate a large component of transverse force during the power stroke of mastication, namely folivores and insectivores. Like leaves, the exoskeletons of insects are tough and require extensive processing if they are to be digested efficiently (Kay and Sheine,1979). Also, strepsirrhines have unfused or only partly fused symphyses (Beecher,1977,1979,1983; Ravosa and Hylander,1994; Ravosa,1996,1999; Ravosa and Hogue,2004). Therefore, they likely reduce the input from the balancing side masseters to reduce dorsoventral shear and wishboning. The working-side medial pterygoid is likely important in generating transverse force in strepsirrhines and the balancing-side medial pterygoid is in a position to reduce wishboning. We predict that the ratio of medial pterygoid PCSA to total jaw adductor PCSA will be greater in folivorous and insectivorous strepsirrhines than in frugivorous ones. Medial pterygoid likely acts to reduce wishboning for all strepsirrhines; however, wishboning is likely more severe in strepsirrhines that have a more horizontally oriented muscle resultant.

The fifth and final hypothesis is that the temporalis is emphasized over the masseter in insectivores and that the reverse is true in herbivores. This follows from Smith and Savage (1959) who suggested that the temporalis is emphasized in carnivores to resist struggling prey. This muscle has great leverage in many carnivores because the condyle is low (and the coronoid process is large); the masseter has great leverage in many herbivores because the condyle is high (and the angular process is large). Although we hesitate to compare prey capture in true carnivores such as carnivorans to prey capture in lemurs and lorises, we tentatively predict that the ratio of temporalis PCSA to masseter PCSA will be significantly greater in insectivorous strepsirrhines than in herbivorous strepsirrhines.

MATERIALS AND METHODS

Sample

We dissected the jaw adductor muscles of 24 strepsirrhine species, represented by a total of 36 specimens (Table 1). The jaw adductor musculature can be divided into the temporalis muscle group, the masseter muscle group, and the medial pterygoid muscle. There is disagreement about the nomenclature for the individual jaw adductors muscles (Toldt,19041905; Edgeworth,1935; Fiedler,1953; Saban,1968; Gaspard et al.,1973a,b,c; Cordell,1991). We have followed the nomenclature of Cordell (1991), who performed the most comprehensive analysis of the literature on the nomenclature for these muscles (see also Perry,2008); both sources provide detailed criteria for identifying zygomatico-mandibularis, for example.

The presence of fascial planes determines the manner in which we divide up the jaw adductor musculature. Here, a named muscle (e.g., superficial masseter) is a sheet of fibers with a similar orientation, bounded by fascial planes or by bony surfaces. The temporalis group consists of the zygomatic temporalis, superficial temporalis, and deep temporalis. The masseter group consists of the superficial masseter, deep masseter, and zygomatico-mandibularis. The medial pterygoid muscle of strepsirrhines is very complex: connective tissue sheets divide it into four major compartments of fibers that twist around one another (Gaspard et al.,1973a); the thickness and extent of these connective tissue sheets differ between species. The medial pterygoid muscle is impossible to divide into unipinnate layers without breaking its fibers; therefore, it was treated differently from the other muscles (see below).

To measure muscle dimensions, we followed a protocol described previously (Perry and Wall,2008). Some specimens were preserved in 10% formalin, some in 70% ethanol, and others had been frozen since the time of death (Table 1). Minimal fiber shortening is expected in the specimens we used as they were preserved whole with the muscles intact. All specimens had similar degrees of gape, close to centric occlusion. Because of uniform gape among subjects, we did not normalize measured fiber lengths to a standardized sarcomere length. Thus, FL and PCSA values reported here reflect FL and PCSA area at near-minimum gape.

Variables

We measured three primary variables (mass, FL, and thickness parallel to muscle orientation—nearly in a parasagittal plane) on each muscle and calculated three variables from them (pinnation, PCSA, and RPCSA—a corrected PCSA that is “reduced” in relation to pinnation). Mass was measured immediately after removing the muscle from its attachments and after patting it gently with a paper towel. Weight was recorded using a digital scale accurate to 0.001 g. Although “dry weight” has been advocated by some (e.g., Rayne and Crawford,1972; Cachel,1984), desiccation makes it impossible to perform the chemical dissection of the muscles necessary to accurately measure FL (described below).

Fiber length was measured using a protocol modified from Rayne and Crawford (1972). Each muscle (e.g., superficial masseter) was immersed in 10% sulfuric acid and cooked in an oven at 60°C. Cooking time varied between 1 and 6 hr depending on muscle size, the amount of connective tissue present, and preservational medium (with fresh specimens requiring less time to cook than preserved ones). Once sufficient connective tissue had been dissolved to allow fascicles to be manipulated without breakage, muscles were examined under a dissection microscope fitted with a calibrated ocular micrometer (reticle). Thirty to sixty representative fiber bundles were measured per muscle, ensuring that all regions of the muscle were represented.

Muscle thickness was measured perpendicular to the direction of pull of the muscle (bony origin to bony insertion) in the coronal plane. This measurement was taken for use in calculating pinnation and constitutes the perpendicular distance “a” between the surface of fiber origin and the surface of fiber insertion. Because of the way in which we dissected the muscles (e.g., superficial masseter removed separately from deep masseter), “a” is simply the thickness of the muscle (i.e., “a” for superficial masseter is the thickness of the superficial masseter—see Fig. 1).

Figure 1.

Pinnation and fiber length. A depicts a parallel-fibered muscle in long-section with five fibers. The line of action is vertical. B shows that a unipinnate muscle has more fibers, but the fibers pull obliquely to the vertical line of action. The thick black lines represent tendon sheets on either side of the muscle that are themselves anchored to the bony points of attachment. C shows the calculation of pinnation where the sine of the angle of pinnation (θ) is thickness divided by fiber length. “Thickness” is equivalent to a in Eq. (1).

Pinnation angle (θ) was calculated using Eq. (1) (Anapol and Barry,1996, see also Fig. 1) where a is the thickness of the muscle and l is mean fiber length.

equation image(1)

Because the fiber layers of medial pterygoid were inseparable, we measured medial pterygoid thickness in a different way. The four layers of medial pterygoid are approximately equal in thickness, so we used total medial pterygoid thickness divided by four for a in Eq. (1).

The PCSA of each muscle was calculated from Eq. (2) using a formula modified from Schumacher (1961):

equation image(2)

Here, q is PCSA, m is muscle mass, l is mean fiber length, and ρ is a constant: the specific density of muscle. The ρ value used was 1.0564 g/cm3 (Murphy and Beardsley,1974) and thus PCSA is in cm2, muscle mass is in g, and fiber length is in cm.

Reduced PCSA (RPCSA) is a calculated variable that provides information about the percentage of muscle cross-section that applies force parallel to the muscle's overall line of action. The equation for calculating RPCSA includes pinnation angle (θ) and as its name implies, RPCSA is always smaller than PCSA. We follow the method of Anapol and Berry (1996) in using the following equation:

equation image(3)

Data Analysis

Descriptive statistics for muscle mass, FL, PCSA, RPCSA, and body mass were calculated. For each individual specimen, total adductor mass is the sum of the masses of all seven jaw adductors (superficial temporalis, deep temporalis, zygomatic temporalis, superficial masseter, deep masseter, zygomatico-mandibularis, and medial pterygoid). Mean FL is the grand mean of the mean fiber lengths for all seven jaw adductors. Total PCSA (or RPCSA) is the sum of the PCSA (or RPCSA) values for all seven jaw adductors.

To evaluate scaling relationships, we regressed Total Muscle Mass, Mean FL, Total PCSA, and Total RPCSA against three different variables: (1) body mass, (2) skull size (a geometric mean of skull measurements—see below), and (3) mandible length. In most cases, the last known living weight was used for body mass. However, for the galagos, species mean weights (from Smith and Jungers,1997; Sussman,1999) were used because living weights were not known. For most of the American Museum of Natural History specimens, cadaver weights were used because living weights were not known. For the AMNH Avahi specimen, species mean weight (from Sussman,1999) was used because only the head was preserved. Table 2 provides the body mass estimate for each specimen.

Table 2. Body size estimates for specimens in this study
SpecimenBody mass (g)aWeight sourceJaw length (cm)Geometric mean (cm)
  • a

    Species mean weights for G. s. braccatus are from Smith and Jungers (1997), others are from Sussman (1999).

Avahi laniger1,178sp. mean3.452.10
Cheirogaleus medius268living2.881.37
Eulemur collaris2,300living6.453.14
Eulemur coronatus1,440living5.862.91
Eulemur macaco flavifrons2,120living5.952.90
Eulemur mongoz1,580living5.272.74
Eulemur rubriventer1,940living6.053.05
Galago moholi141living2.381.41
Galago senegalensis braccatus (Ef)250sp. mean2.52n/a
Galago senegalensis braccatus (jf)250sp. mean2.52n/a
Galagoides demidoff (510f)61sp. mean1.96n/a
Galagoides demidoff (507f)61sp. mean1.96n/a
Hapalemur griseus (1322f)1,030living4.482.36
Hapalemur griseus (1353f)940living4.722.49
Lemur catta2,207sp. mean5.942.67
Lepilemur leucopus742cadaver3.341.87
Microcebus murinus (889f)63living2.081.12
Microcebus murinus (874f)58living2.041.02
Mirza coquereli320living2.961.65
Nycticebus coucang679sp. mean3.842.09
Nycticebus pygmaeus318living3.191.81
Otolemur crassicaudatus (1f)1,258sp. mean4.59n/a
Otolemur crassicaudatus (2f)1,258sp. mean4.59n/a
Otolemur crassicaudatus (1731m)1,684living5.292.74
Otolemur garnettii (ASU)822sp. mean4.17n/a
Otolemur garnettii (Haines)822sp. mean4.17n/a
Perodicticus potto1,100sp. mean4.322.32
Propithecus coquereli (6560m)2,780living5.262.78
Propithecus coquereli (6110f)3,700living6.003.15
Propithecus coquereli (5705f)2,134living5.853.10
Propithecus diadema4,270living6.393.47
Propithecus tattersalli3,620living6.343.41
Varecia rubra (AMNH)3,865cadaver7.463.60
Varecia rubra (DUPC)3,650living7.513.56

For skull size, the geometric mean was calculated from five cranial measurements, four mandibular measurements, and one measurement of the articulated skull (Tables 2 and 3). Table 2 provides the calculated geometric mean for each specimen. Table 3 describes the individual measurements that make up the geometric mean.

Table 3. Measurements that make up the geometric mean
  • a

    This landmark is defined as the posterior edge of the mandibular condyle.

Mandible load arm length from condylarea to anterior edge of mandibular symphysis
Mandible length from posterior edge of angular process to anterior edge of mandibular symphysis
Length of lower postcanine tooth row (includes all teeth that are not part of the tooth comb)
Greatest width of lower postcanine tooth row
Rostrum length from nasion to alveolare
Bizygomatic breadth at postorbital bars (maximum distance at outside edge of bone)
Height of zygomatic arch at mid temporozygomatic suture
Breadth at deepest part of postorbital constriction
Length of skull in midsagittal plane
Height of articulated skull at angle of mandible

Mandible length is a mechanical variable often used in studies of masticatory scaling (Hylander,1979; Vinyard et al.,2003) as a “mechanical denominator” that approximates the bite force load arm and the bending moment acting on the mandibular corpus during incisal biting (Hylander,1979,1985). Mandible length was measured from the posterior edge of the mandibular angle to the anterior-most point on the mandible, excluding the teeth. The latter point, infradentale, lies between the lower central incisors (Bass,1995). Table 2 provides the measured mandible length for each specimen.

Skull size and mandible length were cubed prior to logging to enable easier comparison to body-mass-based regressions.

Regressions were performed in log space using the reduced major axis regression (RMA). Regressions were performed in (S)MATR (Version 1, Falster DS, Warton DI & Wright IJ, www.bio.mq.edu.au/ecology/SMATR).

The F test in (S)MATR was used to test the significance (at α = 0.05) of the difference between observed slope and predicted slope of isometry. Because (S)MATR does not produce plots, all plots shown here are from Microsoft Excel 2003 (Copyright 1985–2003, Microsoft Corporation).

To analyze the strength of the functional relationship of Vb to FL, independent of body mass, we first ran a least squares (LS) regression of FL on body mass. Then, we ran a LS regression of Vb on body mass. The residuals for the first regression were plotted against the residuals for the second regression and a RMA line was fitted to the plot. This procedure was performed in JMP 7 (JMP, Version 7. SAS Institute, Cary, NC, 1989–2007) using data from Perry and Hartstone-Rose (2010) for three experimental foods.

To test hypotheses on the division of labor among the jaw adductors, we used one-tailed Wilcoxon tests (JMP 7), comparing sample means where the species in the sample were grouped by diet.

RESULTS

Tables 4–8 provide the data for muscle mass, FL, PCSA, RPCSA, and pinnation for the seven mandibular adductor muscles. The regression results on the entire sample conform to predictions of isometry (see also Cachel,1984; Shahnoor et al.,2005; Anapol et al.,2008; Perry,2008; Perry and Wall,2008). Altering the independent variable (body mass, skull size, or mandible length) produces very few differences in the overall scaling patterns.

Table 4. Musclea mass (g) for species in this study
SpeciesAbbrDMaDTMPSMSTZMZTSUM
  • a

    Muscle abbreviations are as follows: DM, deep masseter; DT, deep temporalis; MP, medial pterygoid; SM, superficial masseter; ST, superficial temporalis; ZM, zygomatico-mandibularis; ZT, zygomatic temporalis.

Avahi lanigerA.l.0.420.410.370.310.420.280.172.38
Cheirogaleus mediusC.m.0.070.200.110.200.220.070.050.92
Eulemur collarisE.col.0.653.430.702.501.850.560.4810.18
Eulemur coronatusE.cor.0.742.750.742.371.390.740.619.34
E. macaco flavifronsE.m.f.0.342.560.811.450.990.720.597.47
Eulemur mongozE.m.0.381.430.741.410.840.440.385.61
Eulemur rubriventerE.r.0.463.381.001.861.380.820.509.41
Galago moholiG.m.0.040.260.160.130.150.020.050.82
G. senegalensis braccatusG.s.0.040.380.120.250.220.050.051.09
Galagoides demidoffG.d.0.010.060.020.050.030.010.010.18
Hapalemur griseusH.g.0.361.100.651.310.700.270.344.73
Lemur cattaL.c.0.522.180.831.611.010.280.707.13
Lepilemur leucopusL.l.0.340.440.210.200.580.140.192.10
Microcebus murinusM.m.0.030.180.050.140.080.020.030.52
Mirza coquereliM.c.0.030.140.090.190.160.080.060.75
Nycticebus coucangN.c.0.431.590.440.850.980.280.304.87
Nycticebus pygmaeusN.p.0.220.630.280.260.380.090.202.06
Otolemur crassicaudatusO.c.0.352.840.852.462.120.290.529.42
Otolemur garnettiiO.g.0.592.210.632.121.980.240.378.14
Perodicticus pottoP.p.0.261.250.441.000.620.320.464.35
Propithecus coquereliP.c.0.821.751.401.670.990.800.728.15
Propithecus diademaP.d.2.233.672.122.482.461.981.9016.84
Propithecus tattersalliP.t.1.854.081.843.153.371.310.9816.59
Varecia rubraV.r.0.703.961.102.102.031.170.8511.90
Table 5. Fiber length (cm) for species in this study
SpeciesDMaDTMPSMSTZMZTMEAN
  • a

    See Table 4 for abbreviations of muscle names and species names. We cannot explain the extremely long fibers of Eulemur collaris; more specimens must be dissected for this species.

A.l.0.520.800.390.560.740.530.550.58
C.m.0.330.510.320.420.550.310.620.44
E.col.1.171.200.871.371.770.791.181.19
E.cor.0.780.960.580.821.120.620.720.80
E.m.f.0.630.950.680.991.240.841.090.92
E.m.0.650.880.571.031.000.450.360.71
E.r.0.660.930.530.811.060.650.760.77
G.m.0.340.860.260.470.540.320.400.46
G.s.0.490.610.480.570.650.420.620.55
G.d.0.330.360.300.410.370.360.420.36
H.g.0.770.870.490.940.970.480.700.75
L.c.0.991.130.660.981.130.750.980.95
L.l.0.500.830.360.550.900.570.860.65
M.m.0.320.450.260.360.400.310.340.35
M.c.0.340.320.380.380.480.290.520.39
N.c.0.630.980.470.871.020.650.960.80
N.p.0.400.640.300.450.680.330.680.50
O.c.0.610.830.500.760.830.570.850.71
O.g.0.880.740.570.850.880.700.900.79
P.p.0.501.220.591.161.070.950.900.91
P.c.0.760.770.571.040.650.610.760.74
P.d.0.690.890.430.680.660.550.690.66
P.t.0.741.000.471.101.100.490.860.82
V.r.0.971.390.841.241.490.941.111.14
Table 6. Physiological cross sectional areas (cm2) for species in this study
SpeciesDMaDTMPSMSTZMZTSUM
  • a

    See Table 4 for abbreviations of muscle names and species names.

A.l.0.770.480.910.520.540.500.294.01
C.m.0.200.370.330.450.380.210.082.02
E.col.0.532.710.761.740.990.670.397.79
E.cor.0.902.701.212.741.171.130.8010.65
E.m.f.0.512.561.131.390.750.810.527.67
E.m.0.551.541.221.290.790.921.017.33
E.r.0.663.451.782.181.231.190.6311.12
G.m.0.110.280.590.260.270.070.131.71
G.s.0.080.590.240.420.320.110.071.82
G.d.0.020.150.070.110.070.020.020.47
H.g.0.441.191.321.290.720.530.475.95
L.c.0.501.831.181.560.840.350.676.94
L.l.0.650.500.550.350.610.230.213.09
M.m.0.080.370.170.380.200.050.091.35
M.c.0.080.420.220.470.310.260.111.88
N.c.0.641.540.880.930.910.410.305.61
N.p.0.510.930.880.550.530.260.273.94
O.c.0.553.021.722.882.360.500.5611.60
O.g.0.632.841.052.362.140.320.399.71
P.p.0.490.970.710.810.550.320.494.34
P.c.1.122.182.381.741.511.330.9111.17
P.d.3.053.904.703.433.543.422.6324.65
P.t.2.383.883.682.712.892.541.0819.16
V.r.0.732.711.251.671.281.190.719.53
Table 7. RPCSA (cm2) for species in this study
SpeciesDMaDTMPSMSTZMZTSUM
  • a

    See Table 4 for abbreviations of muscle names and species names. Most of the galago specimens had been dissected for a prior study (Cordell,1991), and since we did not have the required measurements to perform a pinnation correction on them we report RPCSA for only Galago moholi and a single specimen of Otolemur crassicaudatus (DPC 1731m) for which we performed the entire dissection.

A.l.0.720.480.900.500.530.470.273.86
C.m.0.200.370.330.430.380.210.081.98
E.col.0.512.500.711.650.990.620.387.35
E.cor.0.872.641.202.611.151.030.7310.23
E.m.f.0.502.461.051.350.750.790.507.41
E.m.0.531.531.221.250.770.750.436.47
E.r.0.643.151.542.041.211.110.6010.30
G.m.0.110.280.310.240.260.070.121.39
H.g.0.431.150.961.210.700.490.435.36
L.c.0.491.811.181.460.840.350.636.74
L.l.0.610.500.540.340.600.220.203.01
M.m.0.080.360.150.350.190.050.081.27
M.c.0.080.400.220.450.310.250.111.83
N.c.0.601.420.880.900.870.390.295.34
N.p.0.470.900.710.520.520.230.273.62
O.c.0.542.951.482.762.350.480.5511.11
P.p.0.480.950.710.800.540.320.474.27
P.c.0.932.062.131.581.371.040.879.99
P.d.2.933.712.853.243.442.802.4121.38
P.t.2.133.653.252.482.802.121.0617.49
V.r.0.722.611.211.621.271.140.699.25
Table 8. Pinnation angles (degrees) for species in this study
SpeciesDMaDTMPSMSTZMZT
  • a

    See Table 4 for abbreviations of muscle names and species names. Note that many of the greatest pinnation angles are found in folivores (H.g., P.c., P.d., and P.t.).

A.l.22971492123
C.m.1265209911
E.col.1323221862213
E.cor.1512418112425
E.m.f.1116221391313
E.m.186515133665
E.r.1524302092116
G.m.1775818151622
H.g.1416361892421
L.c.111032181221
L.l.20881371715
M.m.13143021121520
M.c.12147164176
N.c.2223315181816
N.p.2416362011269
O.c.1719402492517
P.p.9105108915
P.c.26182619203217
P.d.16185319143523
P.t.26202824153410
V.r.1015121181713

Scaling to Body Mass

Total Adductor Mass is isometric with respect to body mass (β = 0.99, CI = 0.85–1.15, Table 9, Fig. 2). The observed slope for Mean FL is slightly negatively allometric with respect to body mass, but the 95% confidence interval for the slope includes isometry (β = 0.28, CI = 0.21–0.37, Table 9, Fig. 2). There is considerable scatter in this regression (r = 0.85). Fiber length appears to have a weaker relationship to the independent variables than do the other dependent variables. Alternatively, its measurement is perhaps subject to greater observer error. Yet another possibility is that there is a strong but complex functional signal in fiber length that is causing lot of interspecific variation—variation too complex to be analyzed using our current dataset. Similarly, the confidence interval for Total PCSA includes isometry, but the observed slope is slightly positively allometric with respect to body mass (β = 0.76, CI = 0.64–0.90, Table 9, Fig. 2). Total RPCSA follows the same pattern (Table 9).

Figure 2.

RMA regressions (A) muscle mass, (B) fiber length, and (C) cross-sectional area against body mass. In C, the solid line and the upper symbols represent PCSA whereas the dashed line and the lower symbols represent RPCSA. Circles represent frugivores, asterisks represent insectivores, and triangles represent folivores. See Table 4 for abbreviations of species names.

Table 9. Descriptive statistics for regressions against body mass
y-variableSlope (β)y-int.rβ lower C.L.β upper C.L.Sample
  • a

    All variables were converted to base 10 logarithms for regressions.

Muscle massa0.99−2.330.950.851.15All Species
0.88−1.960.940.681.14Frugivores
1.35−3.170.930.732.47Insectivores
1.23−3.240.940.712.14Folivores
Fiber length0.28−0.990.850.210.37All Species
0.32−1.090.920.240.43Frugivores
0.34−1.110.770.150.76Insectivores
0.16−0.680.370.060.47Folivores
PCSA0.76−1.510.920.640.90All Species
0.61−1.060.870.440.86Frugivores
1.04−2.170.940.661.66Insectivores
1.14−2.770.960.791.62Folivores
RPCSA0.75−1.510.920.630.89All Species
0.61−1.070.870.430.86Frugivores
1.02−2.150.960.701.51Insectivores
1.08−2.650.970.771.53Folivores
Jaw length cubeda0.97−0.970.960.861.09All Species
0.98−0.920.980.861.11Frugivores
0.86−0.720.970.611.22Insectivores
1.18−1.820.920.711.98Folivores

Regressions were also run on data grouped by broad dietary preference. For most species, we used Kay's dietary assessments (Kay,1975). These assessments have been useful in identifying relative shearing crest length as a dietary signal in strepsirrhines (e.g., Kay et al.,2004). We have updated some of these dietary categorizations based on recent studies of primate feeding ecology (e.g., Nekaris and Rasmussen,2003). We have labeled Nycticebus coucang and Nycticebus pygmaeus as insectivores rather than frugivores or gummivores. Despite the fact that the slow lorises consume a lot of fruit and some gum, predation on insects and vertebrates is an important component of their dietary behavior (Tan and Drake,2001; Nekaris,2005). We have also labeled Lemur catta as a frugivore, though this is somewhat arbitrary as ring-tailed lemurs eat about as much foliage as they do fruit (Sussman,1999). Otolemurcrassicaudatus was treated as a frugivore even though tree exudates are a major component of the diet of this species, especially in the dry season (Doyle and Bearder,1977).

We performed diet-group RMA regressions on muscle dimensions and body mass using the categories “frugivore,” “folivore,” and “insectivore.” Scaling patterns differ slightly by dietary category (Table 9).

In regressions of jaw adductor mass against body mass, although the slopes are higher for insectivores and folivores than for frugivores, no observed slope is significantly different from the slope of isometry (Table 9).

In the regressions of FL against body mass, the slope for folivores is relatively low whereas the slopes for frugivores and for insectivores are relatively high. The confidence intervals are wide for all dietary groups and again the slopes are not significantly different from isometry (Table 9).

We assessed the relationship between average pinnation angle (for all jaw adductors) and log body mass by performing an angular-linear correlation test (Zar,1999). There is no correlation when the entire sample is considered together (r = 0.249). However, there is a good correlation (r = 0.908) for folivores. Pinnation increases with body size in folivorous strepsirrhines.

For the regression of PCSA against body mass, the line of fit for folivores and the line for insectivores have higher slopes than the line for frugivores. Furthermore, the line for folivores is shifted to the right of that for insectivores; this makes sense as the folivores are larger than the insectivores. The slope of the frugivore line is isometric; whereas, the slope of the folivore line is positively allometric (Table 9). Despite being very high, the slope for insectivores is also not significantly different from isometry, (Table 9).

The results for the regression of RPCSA against body mass are very similar to those for PCSA (Table 9). Correlations are higher for RPCSA. Also, for large folivores RPCSA is low compared to PCSA because pinnation is especially great in large strepsirrhine folivores (Table 8).

Scaling to Skull Size

Regressions on the geometric mean of ten skull measurements (cubed) yielded similar results to regressions on body mass (Tables 9 and 10). In general, correlations are higher when skull size is used as the independent variable. The slopes are different for insectivores because the skulls of two members of the insectivore sample were not available (Galago senegalenis, Galagoides demidoff). When these two species are left out of both sets of regressions, the results for regressions against skull size are similar to those against body mass. Another difference between regressions against body mass and regressions against skull size is that all folivores are shifted to the left in the latter, reflecting the fact that folivores have smaller skulls (especially in their length) relative to body mass.

Table 10. Descriptive statistics for regressions against the geometric mean cubeda
y-variableSlope (β)y-int.rβ lower C.L.β upper C.L.Sample
  • a

    The geometric mean was cubed to allow for easy comparison of slopes against the slopes that were generated when body mass was used as the x variable.

  • b

    All variables were converted to base 10 logarithms for regressions. Galago senegalensis braccatus, Galagoides demidoff, and Otolemur garnettii were excluded from these regressions.

  • c

    Confidence intervals were not calculated for any sample size less than six.

Muscle massb1.04−0.520.970.931.17All Species
0.93−0.380.990.841.04Frugivores
1.78−1.060.91n/an/aInsectivores
1.18−0.700.990.971.43Folivores
Fiber length0.33−0.530.840.250.42All Species
0.34−0.510.910.250.46Frugivores
0.62−0.730.78n/an/aInsectivores
0.16−0.360.570.060.42Folivores
PCSA0.78−0.100.940.660.92All Species
0.640.040.950.510.80Frugivores
1.17−0.360.94n/acn/aInsectivores
1.09−0.440.990.891.32Folivores
RPCSA0.78−0.120.950.660.91All Species
0.640.020.950.510.79Frugivores
1.25−0.460.970.473.35Insectivores
1.04−0.420.990.841.28Folivores

Scaling to Mandible Length

Substituting mandible length cubed (JLC) for body mass in scaling analyses of muscle mass reduces the slope for folivores, increases the slope for insectivores, and the slope for frugivores is barely different (Tables 9 and 11, Fig. 3). None of these differences is significant.

Figure 3.

RMA regressions (A) muscle mass, (B) fiber length, and (C) cross-sectional area against body mass. In C, the solid line and the upper symbols represent PCSA whereas the dashed line and the lower symbols represent RPCSA. Circles represent frugivores, asterisks represent insectivores, and triangles represent folivores. See Table 4 for abbreviations of species names.

Table 11. Descriptive statistics for regressions against jaw length cubeda
y-variableSlope (β)y-int.rβ lower C.I.β upper C.I.Sample
  • a

    Fiber length was cubed to allow for easy comparison of slopes against the slopes that were generated when body mass was used as the x variable.

  • b

    All variables were converted to base 10 logarithms for regressions.

Muscle massb1.02−1.350.950.891.17All Species
0.90−1.130.930.701.16Frugivores
1.56−2.040.930.952.55Insectivores
1.04−1.340.990.821.31Folivores
Fiber length0.29−0.710.880.230.35All Species
0.32−0.780.910.250.43Frugivores
0.39−0.830.740.170.90Insectivores
0.14−0.440.650.060.35Folivores
PCSA0.78−0.750.910.660.94All Species
0.63−0.480.870.440.88Frugivores
1.21−1.290.950.781.87Insectivores
0.96−1.030.970.691.33Folivores
RPCSA0.77−0.760.920.650.92All Species
0.63−0.500.860.440.88Frugivores
1.19−1.290.960.821.73Insectivores
0.92−0.980.970.651.29Folivores

Regressing FL against JLC yields the same effects, though they are even more subtle (Fig. 3).

Regressing pinnation against JLC instead of body mass or skull length has little effect on the regression statistics. There is a good correlation between pinnation and JLC for folivores, but no correlation for insectivores or frugivores. Nevertheless, the correlation for folivores is strongest when pinnation is regressed against JLC rather than the other independent variables (r = 0.969). The slope is also lower; the smaller folivores are shifted to the left.

When PCSA is scaled against JLC, the line of fit for frugivores is similar to the line of fit when body mass is the independent variable (Tables 9 and 11, Fig. 3). The line of fit for folivores has a lower slope when the independent variable is JLC (β = 0.96); the slope is not significantly different from isometry (CI = 0.69–1.33). This difference in slope relative to regressions on body mass accompanies a leftward shift in data points for the small folivores Lepilemur and Avahi that have particularly short jaws for their body mass.

To assess the relationship between mandible length and body mass in our sample, we regressed JLC against body mass (Fig. 4). The slope is isometric (β = 0.97, CI = 0.86–1.09) and the correlation is high (r = 0.96). However, diet appears to have a strong effect on the scatter. In general, folivores lie below the line of fit and frugivores lie above it.

Figure 4.

RMA regression of jaw length cubed and body mass. For descriptive statistics, see Table 9. Circles represent frugivores, asterisks represent insectivores, and triangles represent folivores. Note that most folivores are below the line and most frugivores are above it.

Fiber Length and Ingested Food Size

Residuals from a least squares regression of log Vb against log body mass were plotted against residuals from a least squares regression of log mean jaw adductor FL against log body mass. A reduced major axis line of fit was applied to each plot of residuals. This was done for data on three experimental foods: carrot, melon, and sweet potato (Perry and Hartstone-Rose,2010).

For all foods, the RMA regression of residuals was significant (carrot P = 0.004, melon P = 0.010, sweet potato P = 0.009). This suggests that the variation in Vb not correlated with body mass is correlated with the variation in FL not correlated with body mass, supporting the hypothesis that FL tracks the size of ingested foods. The coefficient of correlation is low in every case (carrot r = 0.722, melon r = 0.605, sweet potato r = 0.714).

Folivorous strepsirrhines tend to fall below the line of fit while frugivores are above it. Thus, folivores take small bites for their FL and frugivores take large bites. It remains to be seen whether or not this reflects a reconfigurSation of the spatial relationships of the chewing muscles and teeth—relative to the axis of rotation of the mandible, for example. If the axis of rotation of the mandible is low, then chewing muscle excursion is even further constrained in folivores due to their especially deep mandibles.

Division of Labor

Hypothesis 1: The ratio of temporalis PCSA to total jaw adductor PCSA is significantly greater for frugivorous strepsirrhines than for folivorous ones (Table 12: P = 0.0015). With the caveat that there is no evidence for differential recruitment of temporalis or masseter during incision in humans and macaques (Hylander and Johnson,1985), this supports an association between relative temporalis PCSA and frugivory. However, incisor-use need not be the key difference between frugivores and folivores to explain this pattern (see discussion).

Table 12. Results of comparisons to determine individual jaw adductor role
Muscle comparisonDietary groupsWilcoxon One-tailed P valueCitation for original hypothesis
  • a

    Significant P values are highlighted in bold.

Temporalis PCSA/Sum of Jaw Adductor PCSAsFrugivores versus folivores0.0015aHerring and Herring,1974; Perry and Hartstone-Rose,2010
Masseter PCSA/Temporalis PCSAFolivores versus frugivores and insectivores0.0033Anapol and Lee,1994
Zygomatico-mandibularis PCSA/Superficial Masseter PCSAInsectivores versus frugivores and folivores0.2571Davis1955,1964
Medial pterygoid PCSA/Sum of Jaw Adductor PCSAsFolivores and insectivores versus frugivores0.0262This study
Temporalis PCSA/Masseter PCSAInsectivores versus frugivores and folivores0.0113Smith and Savage,1959

Hypothesis 2: The ratio of masseter PCSA to total jaw adductor PCSA is significantly greater for folivorous strepsirrhines than for other strepsirrhines (Table 12: P = 0.0033). This result provides support for the hypothesis that the masseter is emphasized in folivorous primates (c.f., Anapol and Lee,1994).

Hypothesis 3: The ratio of zygomatico-mandibularis PCSA to superficial masseter PCSA is greater—but not significantly so—in frugivorous and folivorous strepsirrhines than in insectivorous ones (Table 12: P = 0.2571). These results lend only weak support to Davis' hypothesis about diet and division of labor within the masseter group (Davis,1955,1964).

Hypothesis 4: The ratio of medial pterygoid PCSA to total jaw adductor PCSA is significantly greater in insectivorous and folivorous strepsirrhines than in frugivorous ones (Table 12: P = 0.0262). These results provide support for our hypothesis that insectivorous and folivorous strepsirrhines emphasize the medial pterygoid in PCSA more than do frugivorous strepsirrhines.

Hypothesis 5: The temporalis to masseter ratio in PCSA is significantly greater in insectivorous strepsirrhines than it is in frugivorous and folivorous ones (Table 12: P = 0.0113). Here, frugivores and folivores were grouped together. This result provides support for the hypothesis that strepsirrhine insectivores emphasize the temporalis more than do strepsirrhine herbivores (c.f., Smith and Savage,1959). In most of the insectivores (but in few herbivores), temporalis PCSA exceeds masseter PCSA.

DISCUSSION

Scaling Models

Jaw adductor PCSA scales with isometry or slight positive allometry relative to body mass in strepsirrhines in our sample. Furthermore, when split by dietary group isometry is the pattern for frugivores and insectivores, whereas positive allometry is the pattern for folivores.

In our sample, regression of PCSA against body mass yielded wide 95% confidence intervals for the slope (>0.3). However, the r2 value exceeds 0.85 in every case. This suggests that much of the variation in PCSA is explained by variation in body mass. When species are grouped by diet, the r2 values increase, suggesting that jaw adductor PCSA is better correlated with body mass within dietary categories.

Positive allometry of jaw adductor PCSA is supported by recent studies of RPCSA in a sample that includes both strepsirrhines and haplorhines (Shahnoor et al.,2005; Anapol et al.,2008). However, this is only when RPCSA is scaled against craniobasal length, and only for the total primate sample. Perhaps this is an effect of shorter skulls in anthropoids as compared to prosimians. Within prosimians, platyrrhines, and catarrhines individually, the slopes are not significantly different from isometry, and RPCSA scaled isometrically when body mass was used as the independent variable.

Lucas (2004) predicted strong negative allometry (slope of 0.5) of jaw adductor cross-sectional area relative to body mass. To date, studies of PCSA and RPCSA in primates have not recovered a pattern of strong negative allometry (Shahnoor et al.,2005; Anapol et al.,2008; Perry,2008; Perry and Wall,2008).

There are several possible explanations for why PCSA data do not to support the predictions of fracture scaling. For one, PCSA is an estimator of muscle force, whereas fracture scaling actually predicts bite force, the force needed to break food. Bite force differs from the input force because it is modulated by several factors such as muscle activity, mechanical advantage, and occlusal morphology. Another explanation is that one (or more) of the assumptions of fracture scaling is violated in our sample. Lucas (2004) specifically acknowledged that several assumptions of the model are violated in most real primate foods: primate foods may not be homogeneous, linearly elastic, and of uniform toughness,. Some of these assumptions may be violated in a fashion that covaries with body size; for example, larger primates in general might feed on tougher foods (e.g., Kay,1975; Clutton-Brock and Harvey,1977).

PCSA Scaling in Relation to Diet

There is a difference in the scaling pattern in different dietary groups. Large-bodied folivores have relatively greater PCSA than small-bodied folivores. This suggests that there is a need for greater chewing force as folivores get larger; perhaps they need to chew more times or chew relatively more resistant foods. This allometry is not present in frugivorous strepsirrhines. Slopes are high for insectivores, but not significantly different from isometry.

Large frugivores can presumably take larger bites than small frugivores. This might provide the large frugivores with some “force discount.” That is, as the fracture scaling model suggests (Lucas,2004), larger frugivores do not have to use as much force (relative to their body mass) to fracture their foods as small frugivores, even if Vb scales to body size. Large folivores presumably can eat a larger bolus of leaves than those consumed by small folivores. However, a thick stack of foliage behaves differently from a thick chunk of fruit: fracture cracks will not propagate from one layer of foliage to the next (e.g., Sanson,2010). Therefore, larger folivores will not experience the same force discount based on ingesting a larger stack of leaves. This may be why folivores increase PCSA at a greater rate with increasing body size than frugivores: ingesting one large bolus of leaves does not reduce the labor of the masticatory system very much compared to ingesting two half-sized boluses of leaves.

It is hard to explain the pattern we see in the PCSA data for insectivores. The observed slope for PCSA versus body mass is nearly as high as for folivores, but the confidence interval is so wide that it includes isometry. More data are needed on the jaw adductors of insectivorous primates to improve the statistical power of this set of regressions. It is also possible that, given the narrow body size range of strepsirrhine insectivores, a tight fit is statistically less likely than for other dietary categories.

When PCSA is scaled against mandible length, it becomes obvious that, relative to body mass, jaws are shorter in folivores than in frugivores. This is probably an adaptation to increase bite force leverage by decreasing the length of the bite force load arm. Because increasing PCSA is one way to increase bite force, and decreasing jaw length is another way to increase bite force, PCSA and jaw length are not selectively independent variables. Another explanation for the short jaws of folivores is that they aid in reducing bending moments along the jaw (Bouvier,1986; Daegling and Hylander,1998). Small folivorous strepsirrhines have especially short jaws relative to body mass, even when compared to large folivorous strepsirrhines. Small folivores may be at a disadvantage compared to large folivores if the properties of the leaves they eat are similar. Having very short jaws might compensate by decreasing the bite force load arm and by reducing bending moments.

Fiber Length and Diet

Fiber length scales isometrically with body mass in strepsirrhines; however, much of the variation in FL fails to be explained by variation in body mass (r2 = 0.723). Nevertheless, all regressions here are significant and thus we favor the conclusion that variation in FL does have a relationship to variation in body mass (see Antón,1999,2000).

It is not surprising that body mass variation explains only some of the variation in FL. Fiber length should bear a strong relationship to the range of motion experienced by the lower jaw (Herring and Herring,1974; Herring et al.,1979). Thus, FL is a possible indicator of the demand for gape (Herring and Herring,1974; Smith,1984; Taylor and Vinyard,2004; Perry and Hartstone-Rose,2007b). There is a well-documented relationship between length and tension in fibers of vertebrate muscles (e.g., Herring and Herring,1974; Gans,1982). If a food item is large enough, it could potentially stretch the jaw adductors beyond the point where they can produce sufficient force to fracture the food. Gape may be under selection for other behaviors too, such as yawning, incision, grooming, and gouging (Smith,1978,1984). However, Vb presumably influences both gape and FL to an important degree.

Mean FL increases isometrically with body mass among frugivorous and insectivorous strepsirrhines but with negative allometry in folivores. This suggests that perhaps gape does not need to increase in step with body mass in folivores or that the demand for increased PCSA influences muscle architecture more than the need for large ingestive gapes. Large folivores probably do not increase their intake relative to small folivores by ingesting larger leaves; rather, they probably ingest more leaves of the same size, although this requires them to ingest and chew more times (e.g., Hylander,1979). Mammalian folivores adopt these strategies to differing degrees, depending on the mode of ingestion and the nature of the forage; nevertheless, Vb in folivores is likely less tightly related to mouth dimensions than it is in frugivores (Clutton-Brock and Harvey,1983). Ross et al. (2009) demonstrated positive allometry of daily feeding time across primates and suggested this might be due to a decrease in nutrient quality as primate body size increases. Those finding are consistent with the hypothesis that larger folivores do not ingest larger leaves, but rather ingest more of them; however, Ross et al. found no significant relationship between daily feeding time and body size in strepsirrhines (only in haplorhines).

Mean FL increases with Vb independent of each variable's correlation with body mass. This suggests that ingestive behavior has a selective effect on jaw adductor FL. This makes sense because FL should be tailored in such a way that fibers can produce sufficient tension in their stretched state to break up the ingested bolus of food. Furthermore, since tetanic tension in fibers is related to their degree of stretch, strepsirrhines should vary Vb based on the amount of force needed to break down the food. This appears to be the case based on our experimental results of Vb in strepsirrhines (Perry and Hartstone-Rose,2010).

Interspecific differences in the spatial relationships between the muscles and hard tissue structures (e.g., the teeth) of the masticatory system would signify differences in the degree to which strepsirrhine bite size is constrained by FL. For example, shifting the teeth posteriorly (or shifting the muscles anteriorly) increases the degree to which a muscle fiber must stretch to accommodate a given size of food placed between upper and lower first molars. To compensate for this, some primates might shift the location of the instantaneous center of rotation of the mandible to minimize muscle stretch during jaw opening (Weijs et al.,1989) with concomitant increased movements at the TMJ (Wall,1995). Given that the frugivores in this sample take large bites relative to FL and the folivores take small ones, it is very likely that there are differences in the arrangement of jaw structures and/or in jaw kinematics between these frugivores and folivores.

Fiber length is related to pinnation. When all else is equal, increasing pinnation has the effect of increasing PCSA and decreasing mean FL. The large folivores in our sample have short-fibered, highly pinnate muscles (see below); this likely has the effect of limiting gape while increasing PCSA and bite force.

Correcting for Pinnation

Gans (1982) summarized the theoretical expectations of the performance of muscle fibers in a pinnate muscle. In a pinnate muscle, much of the contractile work of the fibers is lost either in swiveling or in pulling against other fibers (obliquely to the muscle's line of action).

Anapol and Barry (1996) provided a correction to reduce physiological cross-sectional area (RPCSA) by the fraction of force that is generated perpendicular to the line of action of the muscle. It is a useful modification of the PCSA equation when one is interested in the component of muscle force parallel to the tendon of insertion and is reasonable for muscles that have long tendons and/or very clear lines of action (e.g., the limb muscles used in that study). However, correcting for pinnation is complicated by several factors.

To list a few of these factors: (1) the chewing muscles have different lines of action relative to one another (Fig. 5), (2) these lines of action are oblique to anatomical reference planes, (3) feeding behaviors include forces that have different and changing directions, (4) the orientation of a fiber changes as gape changes, and (5) the bite force acts on food at the occlusal table—a complicated surface with planes at different orientations.

Figure 5.

The complexity of correcting for pinnation in jaw adductors. A shows an idealized coronal section through the skull illustrating pinnation in the jaw adductors. Broken lines represent aponeuroses; the thickness of the lines roughly corresponds to the thickness of the aponeuroses. Unbroken black lines represent thin fascia. Gray lines (unbroken) represent fascicles within the muscle. “m” is the mandible; “z” is the zygomatic arch. The lateral pterygoid is included for completeness. B shows the deep masseter considered alone. Using Anapol and Barry's (1996) method corrects PCSA according to the line of action of the muscle. This is oblique to a vertical bite force. C shows that correcting PCSA to a vertical bite force requires knowledge of the in situ orientation of the muscle.

Anapol and Barry (1996) method of calculating reduced physiological cross-sectional area does not actually help with any of these complications. It only corrects for pinnation (the offset between the line of action of a fiber and the line of action of the muscle). Therefore, RPCSA is not a vast improvement over PCSA in approximating the force potential of a chewing muscle, especially given that pinnation angles are generally less than 30° in this system (Table 8). To fully correct for all of the offset vectors in this system, the ideal would be a digital means of quantifying fiber orientations (and their changes in orientation throughout the chewing cycle) in three dimensions and in live subjects (see Van Spronsen et al.,1989; Raadsheer et al.,1999).

Division of Labor in the Jaw Adductors—Support for Hypotheses

Our data support the hypothesis that the ratio of temporalis PCSA to total jaw adductor PCSA is greater for frugivores than for folivores. This result supports the suggestion that the temporalis is an important muscle during processing of foods that are ingested at wide gapes (e.g., many fruits) (Perry,2008). It is also possible that there is no direct functional signal related to frugivory. Perhaps because of the role the masseter and medial pterygoid play in transverse chewing, the temporalis is relatively large in the frugivores when frugivores are compared to folivores that have large masseters/medial pterygoids.

The ratio of masseter PCSA to temporalis PCSA is greater in folivorous strepsirrhines than in frugivorous strepsirrhines; therefore, strepsirrhines conform to expectations based on the patterns of leverage seen in platyrrhines (Anapol and Lee,1994) and other herbivores (e.g., Turnbull,1970). One mechanism that might explain this correlation is that, given the more transverse orientation of much of the masseter, this muscle is more important in primates that require a strong transverse component to the power stroke of mastication (e.g., folivores).

Davis' prediction about the partitioning of muscle PCSA within the masseter complex was only weakly supported by the results of this study (Davis,1955,1964). This may be due to differences between carnivorans and primates in the gross anatomy and fiber architecture of these muscles. It might also be due to insufficiently fine-grained selection on the masticatory muscles. Without more detailed studies of activity within the masseter muscle group, it is difficult to know whether or not the different muscles within the masseter group function independently. For example, we do not know whether motor units are segregated within the muscle compartments or whether they cross compartments.

The hypothesis that the medial pterygoid is emphasized in folivorous and insectivorous strepsirrhines compared to frugivorous strepsirrhines is supported by the results of this study. This suggests that the medial pterygoid has an important role to play in generating transverse force in primates that must process tough foods and that lack a fully ossified mandibular symphysis.

The prediction that the temporalis is emphasized over the masseter in insectivores (as compared to herbivores) proved true in our sample of strepsirrhines. This suggests that prey struggle might play a role in insect predation by strepsirrhines. However, as the loads experienced by the strepsirrhine skull in this situation are currently unknown, this possibility requires much more study. As with the first hypothesis, it might be that the pattern we see here is purely a byproduct of the emphasis on masseter in folivores.

Taken together, these results suggest that there is functional partitioning between the temporalis, masseter, and medial pterygoid sufficient to be detected using these coarse-grained morphological analyses. A similar conclusion was reached with respect to the temporalis and masseter by analyzing the correlation between muscle anatomy and EMG activity in eleven species of primates including three species of strepsirrhines (Vinyard and Taylor,2010). However, if there is functional partitioning between the muscles of the masseter group, we did not detect it in a statistically significant way.

CONCLUSIONS

The mandibular adductors of strepsirrhines fit the predictions of geometric scaling relative to body mass. Muscle mass and FL scale isometrically to body mass, though for FL, the confidence interval is very wide. Adductor PCSA scales isometrically or with slight positive allometry to body mass. RPCSA scales in the same fashion as PCSA. However, folivores (especially large-bodied ones) tend to have very pinnate muscles and therefore have relatively low values for RPCSA.

Scaling patterns are very similar when the independent variable is a geometric mean of skull measurements. When jaw length is used as the independent variable, most lines of fit are similar to the lines of fit for body mass and for the cranial geometric mean. However, folivores have relatively short jaws (especially the small ones), and this affects the lines of fit. Because short jaws likely provide better leverage for bite force, jaw length is not simply a proxy for body size.

Jaw adductor mass is largely a function of PCSA and FL. Therefore, the patterns we see in the scaling of jaw adductor mass follow from the patterns seen in the scaling of PCSA and FL. All slopes are statistically indistinguishable from isometry. Because the slopes for PCSA scaling and for FL scaling are both high for insectivores, the slope for muscle mass scaling is particularly high in insectivores.

Jaw adductor FL scales isometrically with body mass for all dietary groups. The slopes for insectivores and frugivores are much higher than the slope for folivores. One explanation for this is that Vb has an effect on FL (Perry and Hartstone-Rose,2007a,2010). This possibility is supported by the results of this study which suggest that, independent of body mass, FL is correlated with Vb.

Jaw adductor PCSA scales isometrically with body mass in frugivores and insectivores, but it scales with positive allometry in folivores. Thus, large folivores are likely capable of producing relatively high forces with their chewing muscles.

The current method of correcting for pinnation is straightforward in terms of measurement but unsatisfactory for fine-grained biomechanical inferences. Measurements of the orientations of the chewing muscles (and their longitudinal tendons) as well as the orientation of the occlusal plane, both relative to a reference anatomical plane, are difficult in terms of generating a standard, easily reproduced methodology but would be great improvements. Such improvements are only touched on here and are worth exploring in much greater detail in future work.

Some of the published hypotheses about the role of the individual jaw adductors were supported by the results of this study. Our data support the following hypotheses regarding individual jaw adductor roles: the ratio of masseter PCSA to total adductor PCSA is greater in folivorous strepsirrhines than in other strepsirrhines; the ratio of medial pterygoid PCSA to total adductor PCSA is greater in insectivorous and folivorous strepsirrhines than in frugivorous strepsirrhines; and the ratio of temporalis PCSA to masseter PCSA is greater in insectivorous strepsirrhines than in herbivorous strepsirrhines.

Receiving only weak support is the hypothesis that the ratio of zygomatico-mandibularis PCSA to superficial masseter PCSA is greater in strepsirrhine herbivores than in strepsirrhine carnivores.

We believe that our data on the division of labor in the masticatory system support the notion that the temporalis, masseter, and medial pterygoid potentially have different functions and may be subject to independent selection. Although there are no consistent patterns in EMG activity demonstrating divisions of labor, the muscle architecture does reveal some potential dietary signals. We could not detect these signals in the finer anatomical divisions within the chewing musculature.

Body size and food properties (material and structural) appear to have driven adaptation in the hard tissues of the masticatory system (e.g., Hylander,1979; Ravosa,1991; Daegling,1992). It appears that they have also driven adaptation in the musculature to generate specific, potentially predictable patterns of variation in fiber architecture.

Acknowledgements

The authors thank the Duke Lemur Center (especially S. Combes, J. Ives, and S. Zehr) and the members of the Department of Evolutionary Anthropology (Duke University) for their help, support, and for the use of cadaver material during both the aspects of data collection. They thank the American Museum of Natural History (especially E. Westwig), the Department of Anatomical Sciences at Stony Brook University, N.N. Cordell, and C. Terranova for access to additional cadaver material. They thank R.F. Kay and W.L. Hylander for the use of laboratory space and materials. N.N. Cordell, R.C. Fox, W.L. Hylander, K.R. Johnson, R.F. Kay, A.B. Taylor, C.J. Vinyard, and A.C. Zumwalt provided helpful advice. This publication is Duke Lemur Center publication number 1191. The experimental protocol was approved by the Duke University Institutional Animal Care and Use Committee (protocol A109-06-03) and by the Duke Lemur Center Research Committee (protocol O-2-06-3).

Ancillary