Nonhuman anthropoid primate femoral neck trabecular architecture and its relationship to locomotor mode



Functional analyses of human and nonhuman anthropoid primate femoral neck structure have largely ignored the trabecular bone. We tested hypotheses regarding differences in the relative distribution and structural anisotropy of trabecular bone in the femoral neck of quadrupedal and climbing/suspensory anthropoids. We used high-resolution X-ray computed tomography to analyze quantitatively the femoral neck trabecular structure of Ateles geoffroyi, Symphalangus syndactylus, Alouatta seniculus, Colobus guereza, Macaca fascicularis, and Papio cynocephalus (n = 46). We analyzed a size-scaled superior and inferior volume of interest (VOI) in the femoral neck. The ratio of the superior to inferior VOI bone volume fraction indicated that the distribution of trabecular bone was inferiorly skewed in most (but not all) quadrupeds and evenly distributed the climbing/suspensory species, but interspecific comparisons indicated that all taxa overlapped in these measurements. Degree of anisotropy values were generally higher in the inferior VOI of all species and the results for the two climbing/suspensory taxa, A. geoffroyi (1.71 ± 0.30) and S. syndactylus (1.55 ± 0.04), were similar to the results for the quadrupedal anthropoids, C. guereza (male = 1.64 ± 0.13; female = 1.68 ± 0.07) and P. cynocephalus (1.47 ± 0.13). These results suggest strong trabecular architecture similarity across body sizes, anthropoid phylogenetic backgrounds, and locomotor mode. This structural similarity might be explained by greater similarity in anthropoid hip joint loading mechanics than previously considered. It is likely that our current models of anthropoid hip joint mechanics are overly simplistic. Anat Rec, 2007. © 2007 Wiley-Liss, Inc.

Debate exists over the functional significance of cortical bone structure (and distribution) in the human and nonhuman anthropoid femoral neck (Lovejoy,1988; Stern and Susman,1991; Ohman et al.,1997; Rafferty,1998; Kalmey and Lovejoy,2002; Lovejoy et al.,2002). This issue is significant because some researchers argue that the best indicator of a shift in the hominin lineage to facultative bipedalism lies in the femoral neck cortical bone structure of the well-known fossil Lucy (A.L.288-1) and other specimens of Australopithecus afarensis (e.g., MAK VP 1/1) (Lovejoy,1988,2005; Ohman et al.,1997; Rafferty,1998; Lovejoy et al.,2002).

The adult human superior cortical rim is approximately 1 mm thick at the midpoint of the femoral neck and the inferior cortical rim is approximately 2.5 mm thick or greater (Rafferty,1998; Mayhew et al.,2005). This structural asymmetry presumably results from the unique biomechanics of the human hip during the mid-stance phase of bipedal gait, when the femoral neck experiences a bending load with tension present along the superior cortical rim and compression along the inferior cortical rim (Meyer,1867; Koch,1917; Frankel,1960; Pauwels,1980; Wolff,1986). The tangential component of the bending load also imposes axial compression on the femoral neck. The lesser gluteal musculature, an active abductor of the hip joint during stance phase, provides a colinear force relative to the femoral neck's long axis and reduces the tensile stresses along the superior cortical bone rim. As a result, the superior cortex can be thin while the inferior cortical rim must be thick to sustain the high-magnitude compressive stresses. The cortical bone structural asymmetry exists through the length of the femoral neck but is reported to achieve a maximum at the neck-trochanteric junction (Ohman et al.,1997), where the stress differential is theoretically maximized (Frankel,1960; Lovejoy,1988,2005). The human condition contrasts with the more equally distributed cortical bone in the femoral neck of climbing/suspensory nonhuman anthropoids (Great Apes and atelines) (Lovejoy,1988; Ohman et al.,1997). This cortical bone structure in climbing/suspensory species presumably is a response to larger axially compressive loads on the femoral neck that result from the higher femoral neck angle (Rafferty,1996).

Quadrupedal anthropoids also exhibit strong femoral neck cortical bone structural asymmetry similar to that seen in humans at mid-neck. In fact, quadrupedal anthropoids and humans have similarly thin superior cortical bone thicknesses relative to body size at mid-neck (Rafferty,1998). During mid-stance, the quadrupedal anthropoid femoral neck is modeled as a cantilevered beam in bending similar to humans, but one key feature of humans is lacking in quadrupedal anthropoids. The lesser gluteals are primarily hip rotators in dedicated quadrupedal anthropoids (Vangor,1979; Stern and Susman,1981; Lovejoy et al.,2002) and presumably contribute a smaller muscle force colinear to the femoral neck axis and opposite in direction to the tension resulting from the bending moment imposed by the hip joint force. As a result, the human gait mechanisms reported to cause the cortical bone thickness asymmetry and a thin superior cortical shell may not also explain the thickness asymmetry observed in quadrupedal anthropoids. Even though similar analyses have not been completed at the neck-trochanteric interface (i.e., Lovejoy,1988,2005), the shared structural similarity between quadrupedal anthropoids and humans challenges prior functional explanations uniquely associating the human asymmetric condition with bipedalism.

Cancellous bone structure has largely been ignored in these functional morphology studies of the anthropoid femoral neck. Since biomechanical (Lanyon,1974; Goldstein et al.,1991; Kamibayashi et al.,1995; Biewener et al.,1996; Pontzer et al.,2006) and comparative studies strongly suggest that assessments of trabecular bone structure in the femoral neck will be able to discriminate anthropoid locomotor modes, this seems to be a large oversight (Ward and Sussman,1979; Oxnard and Yang,1981; Biewener et al.,1996; Macchiarelli et al.,1999; Rook et al.,1999; Fajardo and Müller,2001; MacLatchy and Müller,2002; Ryan and Ketcham,2002a,2002b,2005). A single study quantitatively examined the trabecular bone structure of the nonhuman anthropoid proximal femur (Fajardo and Müller,2001). Those quantitative data, similar to qualitative observations (Rafferty,1998), suggested that nonhuman anthropoid locomotor modes might be distinguished on the basis of the trabecular bone structure of the femoral neck. Interpretations of those quantitative microcomputed tomography data must be regarded with caution since they derived from a small pilot study that included four species and a single individual represented in each taxon. Subsequent studies of strepsirhine primate femoral head and neck suggest that analyses of trabecular bone structure among nonstrepsirhine primates might be sensitive to locomotor mode differences (MacLatchy and Müller,2002; Ryan and Ketcham,2002a,2002b,2005).

In this study, we assess the trabecular bone structure in several nonhuman anthropoid taxa and determine its relationship to locomotor mode. This analysis focuses on two structural variables, the trabecular bone degree of anisotropy and the bone volume fraction, in two locomotor groups: quadrupedal and climbing/suspensory anthropoids. The degree of anisotropy, a measure characterized in several other comparative primate publications (Fajardo and Müller,2001; MacLatchy and Müller,2002; Ryan and Ketcham,2002a,2005), reflects the quality of the mechanical loading to the extent that a repetitive or stereotypical loading pattern is associated with anisotropic trabeculae (Goldstein et al.,1991) and a heterogeneous or variable loading pattern is associated with an isotropic trabecular pattern (Goldstein et al.,1991). It would then be expected that quadrupedal primates would have anisotropic trabecular bone in the femoral neck since their hind limb excursion patterns during gait are presumably stereotyped, with movements primarily in the sagittal planes. Climbing/suspensory primates would have isotropic trabecular bone in the femoral neck since their hind limb excursion patterns are hypothesized to be variable (Fleagle,1976).

The trabecular bone volume fraction is the primary determinant of the yield strength and stiffness of cancellous bone. The bone volume fraction, which is a surrogate of the apparent density (Rice et al.,1988), is influenced by the magnitude of the loads experienced by the bone (Rubin and Lanyon,1985; Wolff,1986; Lee et al.,2002). Regulation of the volume fraction, however, is complex and multifactorial. Low-magnitude but high-frequency strains have been shown to have an anabolic effect on the volume fraction of cancellous bone (Rubin et al.,2001,2002; Judex et al.,2003). Furthermore, nonmechanical factors such as age (Weaver and Chalmers,1966; Parfitt et al.,1983; Grote,1995; Halloran,2002; Riggs et al.,2004), sex (Aaron et al.,1987; Kneissel et al.,1994), reproductive stage (Kalkwarf and Specker,1995; Lees and Jerome,1998), hormone levels (Jerome et al.,1994,2001; Hamrick et al.,2006), diet (Parsons et al.,1997), and disease (Kleerekoper et al.,1985; Crane et al.,1990; Fazzalari et al.,1992; Majumdar et al.,1997) can impact bone density. Because of the influence of several nonmechanical factors on bone density, it is likely that the degree of anisotropy will present the stronger functional signal.

Two basic hypotheses were developed to test whether trabecular bone structural differences exist among quadrupedal and climbing/suspensory nonhuman anthropoid primates. These hypotheses are based on the cantilevered beam biomechanical models of the proximal femur generally described above and characterizations of hind limb excursion patterns in these taxa.

H1: Cancellous Bone Volume Fraction in Anthropoid Primates

Among climbing and suspensory anthropoid primates, the cancellous bone volume fraction will be evenly distributed across the femoral neck (as measured from superior to inferior). In contrast, the cancellous bone volume fraction will be unevenly distributed across the femoral neck of quadrupedal anthropoid primates. This uneven distribution will present itself as a high bone volume fraction in the inferior half of the femoral neck relative to the superior half of the neck.

H2: Cancellous Bone Structural Anisotropy in Anthropoid Primates

Among climbing and suspensory anthropoid primates, the cancellous bone will be isotropic, or at minimum, less anisotropic than that of quadrupedal taxa.


Table 1 lists important details of the study sample including the species and their locomotor modes. One femur from each of 46 individuals was scanned with high-resolution X-ray computed tomography. Specimens were included if the proximal femoral epiphysis was fused, the sagittal suture was less than 50% obliterated, and, where observable, tooth wear was not excessive. None of the femora showed external signs of fracture or indications of bone disease (e.g., presence of osteophytes).

Table 1. Study sample
SpeciesNaBody Weight (g)Locomotor ModeSourcec
  • a

    sample size (N) and body weight listed by sex (male/female).

  • b

    data taken from Smith and Jungers (1997).

  • c

    AMNH, American Museum of Natural History, MCZ, Harvard Museum of Comparative Zoology, NMNH, National Museum of Natural History, USB, University at Stony Brook Primate Museum, UTPC, University of Texas at Austin Primate Collection.

Ateles geoffroyi27530bClimbing/SuspensoryAMNH, NMNH
Symphalangus syndactylus2/511900/10700Climbing/SuspensoryAMNH, NMNH
Alouatta seniculus5/56690b/5210bArboreal Quadrupedalism/SuspensoryAMNH, NMNH
Colobus guereza6/69421/7853Arboreal Quadrupedalism/LeapingNMNH
Macaca fascicularis6/64768/3590bArboreal Quadrupedalism/LeapingAMNH, MCZ
Papio cynocephalus−/3−/9676Terrestrial QuadrupedalismUTPC

High-Resolution X-Ray CT Scanning

Creating volumes of interest (VOIs) for three-dimensional analysis of the femoral neck trabecular architecture required several steps. First, we measured the biomechanical neck length (BNL) of each femur on digital frontal X-ray images (NIH Image version 1.61). A small aluminum bit of known dimensions was included in each image to calibrate linear measurements. Biomechanical neck length was defined as a line extending between the center of the femoral head and the intersection of the neck and diaphyseal long axes (Fig. 1a; also see Rafferty,1998). The BNL was used to determine scaled lengths for the VOIs (perpendicular to the neck axis) for each specimen.

Figure 1.

a: Measurements of BNL were taken from digital radiographs. BNL was defined as the distance between a point along the long axis the diaphyseal shaft (A) and the center of the femoral head (B). Dotted lines approximate the location of the volume of interest boundaries. Arrows also approximate the positions of trabecular bone bands similar in position to the human medial compressive band (black arrow) and lateral tensile band (white arrow). b: Image of originally acquired computed tomography data. Four garnet granules were placed on specimens (superiorly, posterioly, inferiorly, and distally to the neck to mark the end of the scan) to facilitate scanning and reorienting of image data. Posteriorly positioned granule visible in this image. c: Image representing placement of the volumes of interest. An anteroposteriorly extending line through the mid-section of the superoinferior diameter demarcated the boundary between the superior volume of interest and the inferior volume of interest.

Subsequently, high-resolution X-ray computed tomography scans were acquired on a system that has been shown to produce accurate images of cancellous bone (Fajardo et al.,2002). Specimens were scanned in the transverse plane of the diaphysis (Fig. 1b) and acquired with a 32 mm field of view (FOV), a 1,0242 pixel matrix, and 800 projections. These settings created 0.031 mm pixels and slices were 0.040 mm thick.

Prior to scanning, four small garnet grains were applied to the external surface of the specimens using nondamaging removable wax glue (Fig. 1b). Three markers were positioned at the head-neck transition, delimiting the anterior, inferior, and posterior surfaces of the transition. The three markers defined a plane perpendicular to the femoral neck and were used as landmarks for reconstructing the image data perpendicular to the long axis of the neck. A fourth marker was positioned to indicate the distalmost extent of the scan.

Image data were exported as 16 bit TIF files. An in-house-developed software package was used to reorient the image data. After data reorientation, a volume equivalent to 15.4% of the BNL, beginning from the plane defined by the three garnet markers, was extracted and saved. In the largest specimen, this volume equaled 5 mm in length. All image files included both cancellous and cortical bone. Image files were then exported to and analyzed on a Scanco micro-CT system (Scanco Medical, Bassersdorf, Switzerland) to perform three-dimensional trabecular bone morphometry.

VOI Determination

Two VOIs were created for each specimen in the proximal femoral neck: one included the trabecular bone of the superior femoral neck and the other the trabecular bone of the inferior femoral neck. The positions of the VOI (proximal neck) were chosen because preliminary work (Fajardo and Muller et al.,2001) indicated this region had dense trabecular bone and it appeared to contain bands of trabeculae analogous to the human medial compressive and tensile bands (Fig. 1a). To create these VOIs, the midpoint intersection of superoinferior and anteroposterior diameters was found using a linear measurement tool in the morphometry software. These lines identified the superior and inferior areas of interest in a single image slice of the femoral neck, and the collection of these areas through the image stack constituted the volume of interest (Fig. 1c). Prior to structural analysis, the pixel dimensions were dropped to create 0.040 mm3 voxels to conform to the cubic voxel requirements of the analysis software (Rüegsegger et al.,1996).

Image Thresholding

Threshold values distinguishing bone from the background were determined using an adaptive iterative algorithm described in the 1970s (Ridler and Calvard,1978; Trussell,1979) and since then incorporated into trabecular bone analyses (Ryan and Ketcham,2002a,2002b; Meinel et al.,2005; Rajagopalan et al.,2005; Maga et al.,2006). Each data set's threshold value was determined based on the characteristics of its image histogram.

Morphometric Variables

We measured the bone volume fraction (BV/TV) and mean intercept length degree of anisotropy (DA) in each VOI to test our hypotheses. We also measured five other three-dimensional nonmodel-based (Parfitt et al.,1983) structural variables to characterize femoral neck trabecular architecture in these species (Table 2): structural model index (SMI), trabecular thickness (Tb.Th), trabecular separation (Tb.Sp), trabecular number (Tb.N), and connectivity density (Conn.D). The methods used to measure these parameters in this study are well described in the biomedical (Hildebrand and Rüegsegger,1997; Müller et al.,1998; Hildebrand et al.,1999) and comparative morphology literature (Fajardo and Müller,2001; MacLatchy and Müller,2002).

Table 2. Measurements of trabecular architecture
bone volume fractionBV/TV%ratio of bone volume to total volume of interest
degree of anisotropyDA(−)extent to which trabeculae are similarly oriented
structure model indexSMI(−)measure of distribution of rod- to plate-like trabeculae
trabecular thicknessTb.Thmmmeasure of average strut thickness
trabecular separationTb.Spmmmeasure of average distance between struts
trabecular numberTb.Nmm−1measure of average number of trabeculae per millimeter
connectivity densityConn.Dmm−3relative quantity describing how well are the struts interconnected

VOI Precision

Volumes of interest precision tests were run on one specimen exhibiting variable endosteal geometric complexity throughout the length of the neck. In the test specimen's proximal femoral neck, the outline of the medullary cavity was irregular, with sudden shape changes along its length; numerous trabeculae were present throughout this region. The distal portion of the neck was typically circular or elliptical in shape with very few contour changes. Furthermore, trabeculae were sparse relative to the proximal portion of the neck. This process was repeated five times on a single image volume over 3 days and the data evaluated for each to determine the precision in VOI determination. Results indicated that there was less than 0.70% variability (coefficient of variation) in the total volume of the VOIs, suggesting our approach was highly repeatable. All other structural measurements (mentioned in previous section), such as the BV/TV and Tb.Th, showed less than 2% variability in repeat analyses.

Statistical Methods

Means, standard deviations, and coefficients of variation are reported for all measurements. Kolmogorov-Smirnov tests indicated that the data were normally distributed. We used paired t-tests on sex-pooled species groups to compare the structural results of the superior and inferior VOIs. The sexes of M. fascicularis were kept separate because independent t- and Mann-Whitney U-tests indicated significant structural differences between the sexes of this species. No other species showed sex differences.

We approached all other tests and comparisons more conservatively, using sex-specific groups for the size-dimorphic anthropoids (M. fascicularis, A. seniculus, and C. guereza) and a pooled sex group for the size-monomorphic taxon S. syndactylus. Superior/inferior VOI ratios (S/I) for BV/TV and DA were tested against the value of 1 using a one-sample t-test to determine whether these structural variables demonstrated an uneven distribution in the femoral neck. Also, these S/I ratios were compared interspecifically using one-way ANOVA with Tamahane posthoc pairwise comparisons. Individual interspecific comparisons of all structural parameters, in each VOI, were also carried out using one-way ANOVA with Tamahane posthoc pairwise comparisons. All statistical tests were performed with SPSS version 8.0 and used a significance threshold set at α = 0.05.


Regional Variation (Differences Between Superior and Inferior Subvolumes)

Tables 3 and 4 list the descriptive results of the trabecular bone morphometric analysis. Intraspecimen comparisons between the superior and inferior volumes of interest indicated that the two femoral neck subvolumes were often structurally distinct (Table 5). The BV/TV was significantly greater in the inferior region of the femoral neck in female M. fascicularis (t = −8.646; P < 0.001), C. guereza (t = −9.426; P < 0.001), and P. cynocephalus (t = −5.498; P < 0.01). In contrast, no regional BV/TV differences were observed in the quadrupedal male M. fascicularis or in either of the climbing/suspensory anthropoids, A. geoffroyi and S. syndactylus.

Table 3. Descriptive statistics (means with standard deviations, SD and coeffcicients of variation, CV) of superior volume of interest listed by species and sex
   A. geoffroyiS. syndactylusA. seniculusM. fascicularisC. guerezaP. cynocephalus
  • Abbreviations: see Table 2 for list of measurements, abbreviations, and their descriptions.

  • a

    sexes significantly different in independent t-test (p < 0.05).

  • b

    sexes significantly different in Mann-Whitney U test (p < 0.05).

Table 4. Descriptive statistics (means with standard deviations, SD and coeffcicients of variation, CV) of inferior volume of interest listed by species and by sex of species
   A. geoffroyiS. syndactylusA. seniculusM. fascicularisC. guerezaP. cynocephalus
  • Abbreviations: see Table 2 for list of measurements, abbreviations, and their descriptions.

  • a

    sexes significantly different in independent t-test (p < 0.05).

  • b

    sexes significantly different in Mann-Whitney U test (p < 0.05).

Table 5. Results of paired t-tests comparing superior and inferior volumes of interesta
  A. geoffroyiS. syndactylusA. seniculusM. fascicularisC. guerezaP. cynocephalus
  • Abbreviations: t, t-statistic, p, probability value, see Table 2 for complete description of variables.

  • a

    dash mark indicates that values were not significantly different.


Degree of anisotropy values were significantly higher in the inferior femoral neck region of female M. fascicularis (t = −5.594; P < 0.01), C. guereza (t = −11.820; P< 0.001), and S. syndactylus (t = −3.939; P < 0.01; Table 5). In contrast, DA values were similar in both femoral neck regions of A. seniculus, A. geoffroyi, male M. fascicularis, and P. cynocephalus.

All taxa except A. geoffroyi exhibited significant regional structural variation in other variables. S. syndactylus exhibited about as much regional variation as female M. fascicularis, C. guereza, and more than female P. cynocephalus.

S/I Ratios for Bone Volume Fraction and Anisotropy

The S/I ratios for BV/TV of four taxa were not significantly different from the test value of 1: S. syndactylus (0.99; t = −0.158), A. geoffroyi (0.88; t = −1.399), A. seniculus (males, 0.86, t = −1.206; females, 0.93, t = −1.657), and male M. fascicularis (0.94; t = −2.086; Fig. 2a). All other taxa in the sample had S/I ratios for BV/TV significantly lower than 1. The lowest S/I ratios calculated belonged to the two sexes of C. guereza (males, 0.79, t = −7.152; females, 0.77, t = −7.291). Interspecific comparisons of the S/I ratios for BV/TV indicated that the only significant pairwise differences were between S. syndactylus and both sexes of C. guereza (Fig. 2a).

Figure 2.

a: Superior/inferior VOI ratios for bone volume fraction. The trabecular bone is equally distributed across the two VOI in the two climbing/suspensory anthropoids, both sexes of A. seniculus, and male M. fascicularis. Interspecific comparisons reveal most taxa have similar superior/inferior ratios for BV/TV and that only two significant differences exist: S. syndactylus and each sex of C. guereza. The distribution of trabecular bone fails to show a correlation with locomotor mode in this sample of anthropoid primates. b: Superior/inferior VOI ratios for the degree of anisotropy. The superior/inferior ratio for DA is significantly different from one in S. syndactylus, both C. guereza sexes, and female M. fascicularis. Interspecific comparisons show that female C. guereza's superior/inferior ratio for DA is significantly lower than those of female A. seniculus, A. geoffroyi, and S. syndactylus. Also, male C. guereza's superior/inferior ratio for DA is significantly lower than the female A. seniculus ratio. Open bars indicate climbing/suspensory species and closed bars indicate quadrupedal taxa in our sample. Asterisk denotes a ratio that is significantly different from the test value of 1. Dagger denotes a significant interspecific comparison.

The S/I ratios for DA of the following species were not significantly different from the test value of 1: A. geoffroyi (0.76, t = −1.399), both sexes of A. seniculus (males, 1.03, t = 0.822; females, 1.03, t = 1.154), male M. fascicularis (0.91, t = −1.389), and P. cynocephalus (0.90, t = −3.354; Fig. 2b). In contrast, the S/I ratios for DA were less than 1 in both sexes of C. guereza (males, 0.81, t = −8.860; females, 0.78, t = −10.394), female M. fascicularis (0.87, t = −5.683), and S. syndactylus (0.92, t = −3.966). Tamahane posthoc comparisons of S/I ratios for DA indicated that S. syndactylus and both sexes of A. seniculus were significantly different from female C. guereza (Fig. 2b). In addition, female A. seniculus had a significantly higher S/I ratio than male C. guereza.

Interspecific Comparisons of Cancellous Bone Architectural Variables

Tables 6 and 7 summarize the results of the pairwise comparisons of each structural index (e.g., BV/TV, SMI) and metric variable (e.g., Tb.Th, Tb.N). Relative to the total number of pairwise comparisons, few significant differences exist and body size-related patterns are weak at best. As a result, a brief summary of these results follows. Only one BV/TV pairwise comparison was significantly different (superior VOI: S. syndactylus > female M. fascicularis, P < 0.05) and only three DA pairwise comparisons were significantly different (inferior VOI: S. syndactylus > female A. seniculus, P < 0.05; female C. guereza > female A. seniculus and M. fascicularis, P < 0.05). Female C. guereza, but not the larger male C. guereza and female P. cynocephalus, had greater Tb.Th than female M. fascicularis (P < 0.05).

Table 6. Results of Tamahane post hoc comparisons (ANOVA) of the superior VOI
 A. geoffroyiS. symphalangusA. seniculus malesA. seniculus femalesM. fascicularis malesM. fascicularis femalesC. guereza malesC. guereza femalesP. cynocephalus
  • Abbreviations: dash mark indicates no significant difference for pair-wise comparison, see Table 2 for an explanation of morphometry abbreviations.

  • 1

    p < 0.05

  • 2

    p < 0.01

  • c

    p < 0.001.

A. geoffroyi         
S. symphalangus        
A. seniculus males       
A. seniculus females      
M. fascicularis malesTb.Th1     
M. fascicularis femalesTb.N2 Conn.D1BV/TV1 SMI2 Tb.Th2 Tb.Sp2 Tb.N1 Conn.D1Tb.N2 Conn.D1Conn.D1    
C. guereza males   
C. guereza femalesTb.Th1Tb.Th1 Tb.Sp1 Tb.N2  
P. cynocephalus females 
Table 7. Results of Tamahane post hoc comparisons (ANOVA) of the inferior VOI
 A. geoffroyiS. symphalangusA. seniculus malesA. seniculus femalesM. fascicularis malesM. fascicularis femalesC. guereza malesC. guereza femalesP. cynocephalus
  • Abbreviations: dash mark indicates no significant difference for pair-wise comparison, see Table 2 for an explanation of morphometry abbreviations.

  • 1

    p < 0.05

  • 2

    p < 0.01

  • 3

    p < 0.001.

A. geoffroyi         
S. symphalangus        
A. seniculus males        
A. seniculus femalesDA1      
M. fascicularis malesTb.Th2Tb.Th1     
M. fascicularis femalesTb.Th3 Tb.N1Tb.Sp3 Tb.N2 Conn.D3Conn.D2Conn.D1    
C. guereza malesTb.N2 Conn.D3   
C. guereza femalesDA2Tb.Th1SMI1 DA2 Tb.Th1 Tb.N2 Conn.D3  
P. cynocephalus females 

Female M. fascicularis inferior VOI Tb.N was the highest measured in the sample and it was significantly higher than A. geoffroyi (P < 0.05), male C. guereza (P < 0.01), female C. guereza (P < 0.01), and S. syndactylus (P < 0.01), but the two cercopithecine taxa, M. fascicularis and P. cynocephalus, never exhibited significant differences in spite of the threefold range in body mass. Female M. fascicularis and P. cynocephalus had the highest measured Conn.D in the sample and were statistically similar. Female M. fascicularis Conn.D in the inferior femoral neck was significantly higher than male A. seniculus (P < 0.01), female A. seniculus (P < 0.01), male C. guereza (P < 0.001), female C. guereza (P < 0.001), and S. syndactylus (P < 0.001).


This study of the femoral neck trabecular architecture in nonhuman anthropoid primates aimed to determine whether correlations between locomotor mode and trabecular architecture exist and through these data evaluate the current concepts of nonhuman anthropoid hip joint loading that derive from the human cantilevered beam model. The results do not consistently satisfy the cancellous bone volume fraction distribution hypothesis. Even though C. guereza, P. cynocephalus, and female M. fascicularis have relatively more trabecular bone distributed inferiorly in the femoral neck and the climbing/suspensory species have an even distribution of trabecular bone, the fact that an even distribution of trabecular bone is also found in the quadrupedal male M. fascicularis, and interspecific comparisons of the S/I ratios for BV/TV (and BV/TV results, in general) overlap across almost all species, strongly contradict the trabecular bone distribution hypothesis. In addition, the overlap in BV/TV measurements across taxa may extend to humans as well. The bone volume fraction in the adult human proximal femoral neck's medial compressive band (site similar to our inferior VOI) is approximately 32% with at least 18% intraspecific variation (Whitehouse and Dyson,1974; Fazzalari et al.,1983,1989; Crane et al.,1990). Therefore, it appears that humans extend the range of anthropoid BV/TV values (38%–47% BV/TV in the inferior VOI) and that due to high intraspecific variation, some nonhuman anthropoid taxa would overlap with humans as well.

The structural anisotropy results also fail to support the degree of anisotropy hypothesis posited above. In contrast to our expectations, S. syndactylus, a climbing and suspensory primate, has anisotropy equal to or higher than the quadrupedal cercopithecines M. fascicularis and P. cynocephalus, and A. geoffroyi, another climbing/suspensory anthropoid, has the highest measured anisotropy value of all taxa in the sample. Our assumption, similar to that of other studies (Fajardo and Müller,2001; MacLatchy and Müller,2002; Ryan and Ketcham,2002a), was that differences in hind limb excursion patterns (stereotypy vs. variable) result in hip joint loading differences (stereotypy vs. variable). The results indicate that it is a mistake to assume linkage between kinematic stereotypy (or variability) and hip joint loading pattern stereotypy in nonhuman anthropoids. This interpretation is supported by recent kinetic evidence that suggests that stereotypy in locomotor gait does not equate with mechanical loading stereotypy in quadrupedal anthropoids (Demes et al.,2006). Furthermore, characterizations of locomotor repertoires are too simplistic and poorly correlate with diaphyseal structure within a species (Carlson et al.,2006). Interestingly, the A. geoffroyi inferior VOI DA result (1.71) is similar to the 1.68 DA value calculated for humans (intraspecific variation equaled 10.8%, which is comparable to values we report here), although this human value was not measured in exactly the same site as our inferior VOI (Hildebrand et al.,1999). A conclusion that can be drawn based on our own DA results combined with the overlapping results of the other trabecular bone parameters (e.g., Tb.Th, Tables 6 and 7) is that the taxa in this study share more similarities in femoral neck trabecular structure than differences and these similarities belie any correlation of structure with locomotor mode (Tables 6 and 7). The proximal femoral necks of all taxa in this study have trabeculae that are honey-combed in appearance with concave surfaces (Fig. 3) and that are characterized by a proximosuperiorly radiating band of trabeculae, similar in position and direction to the human medial compressive band (Kummer,1959; Pauwels,1980; Wolff,1986), as well as a proximoinferiorly radiating band, like the human lateral tensile band, of varying prominence (Fig. 4).

Figure 3.

Trabeculae of the proximal femoral neck in two anthropoids, the quadrupedal C. guereza and the climbing/suspensory S. syndactylus. The surfaces of the trabeculae are concave, which causes the SMI results to be negative. Generally, this structural design is present in all the species in this sample.

Figure 4.

Three-dimensional renderings of micro-CT data: a, Alouatta seniculus; b, Ateles geoffroyi; c, Papio cynocephalus; d, Colobus guereza; e, Macaca fascicularis (female); f, Symphalangus syndactylus. Top-row images are right femora and bottom-row images are left femora. Each image shows the full data set in the proximodistal direction and is cut in a mid-coronal plane of the femoral neck, showing the dorsal half of the femoral neck. The three garnet markers are visible in each image, marking the proximalmost (first) slice of each volume set. Each species has a band of trabeculae radiating from the inferior cortex in a proximosuperior direction (which is also called the medial compressive band). Its presence is strong in S. syndactylus, M. fascicularis, C. guereza, and P. cynocephalus. Furthermore, all species have a band of trabeculae radiating proximoinferiorly (also called the lateral tensile band) to some extent. The pattern of this proximoinferiorly extending band is weakest in A. seniculus. Interestingly, there appears to be a boundary between these two bands at the mid-height of the endosteal cavity of the femoral neck. In Colobus, the inferior trabecular band reaches a bit into the superior VOI, which might explain the lower DA values for this species in that upper VOI. The perspective shows that each species' cancellous bone is made up of somewhat circular porous spaces surrounded by plates with concave surfaces. Finally, this image shows that in the proximal region of the femoral neck, all data are characterized to some extent by a thicker inferior cortical shell than the superior cortical shell.

An interesting exception to this emerging pattern of trabecular bone structural similarity in the primate femoral neck is galagine vertical clingers and leapers. Galagine vertical clingers and leapers are structurally distinct from quadrupedal and climbing/suspensory strepsirhines (MacLatchy and Müller,2002; Ryan and Ketcham,2002a,2002b,2005). They have higher degrees of anisotropy and more anterosuperiorly oriented trabeculae (MacLatchy and Müller,2002; Ryan and Ketcham,2002a,2002b,2005), although the functional (mechanical) significance of differential anisotropy in these primates might be less clear than expected (Ryan and van Rietbergen,2005). It should be noted that, in our samples, as in the study by Ryan and Ketcham (2002a), quadrupedal and climbing/suspensory taxa were indistinguishable on the basis of the proximal femur trabecular architecture. The strepsirhine results suggest that only comparisons of extreme differences in locomotor mode, like vertical clinging and leaping and climbing/suspensory progression, are associated with structurally different architectures in the proximal femoral neck. Among anthropoids, vertical clinging locomotion is only known in Pithecia pithecia but no information is currently known about trabecular bone in this species.

Qualitative observations strongly suggest that trabecular bone structure may be similar across medium- to large-bodied mammals. Kummer's work (1959) shows a band of trabeculae, similar to the human medial compressive band, extends proximosuperiorly from the inferior femoral neck cortical shell in many mammals regardless of femoral gross morphology (external shape), locomotor mode, or phylogeny (Kummer,1959; Pauwels,1980; Wolff,1986). A proximoinferiorly directed band is also common but also more variable in prominence. These structural similarities may suggest hip joint loading similarities as well. Bending is most likely a major loading pattern for all anthropoid taxa, as well as other mammals, due to the morphology of the proximal femur (a neck offset obliquely from the diaphysis) and the ubiquity of a medial compressive band of trabeculae lends some support to this contention. In addition to this structural inference, in vivo hip joint contact force analyses suggest that hip joint loading regimes are similar in bipeds and quadrupeds, and possibly all mammals (Bergmann et al.,1993,1999,2001). Those data imply that the hip joint contact force tracks a similar small contact region on the proximal surface of the femur in humans and sheep and that the force components (e.g., in the ventral or coronal plane of the femur) are similar in direction (means and range of angles) during gait, in spite of the postural, hind limb flexion/excursion angle, and locomotor differences between these two species. [Force directions (components and resultant) in these studies are reported according to a femur-based coordinate system, where the x-, y-, and z-axes correspond to the mediolateral, ventrodorsal, and proximodistal axes of the femur. This coordinate system takes into account the different hindlimb flexion angles used by humans and sheep and reports the force angle with respect to the femur.] Bergmann et al. (1999) also indicate that different human activities (e.g., stair climbing, squatting) and hind limb postures matter little for the directionality of the human hip contact force (resultant and component forces), suggesting the same loading patterns might apply to anthropoid climbers and clamberers (e.g., Ateles and Symphalangus in this study) whose hind limb positions during gait are presumably more variable than in bipeds or quadrupeds. These data are preliminary but they further hint at an intriguing possibility that a common way to load the proximal femur exists among quadrupeds and bipeds, and possibly among mammals. Since trabecular bone architecture is sensitive to its loading environment (Lanyon,1974; Goldstein et al.,1991; Kamibayashi et al.,1995; Biewener et al.,1996; Pontzer et al.,2006), this provides a possible biomechanical mechanism that explains the similarity in nonhuman anthropoids reported here, as well as similarity across many mammals.

The results we report are at odds with the patterns of femoral neck cortical bone structure in nonhuman anthropoids. Femoral neck cortical bone is inferiorly thick and superiorly thin in quadrupedal anthropoids and homogeneously thick in climbing/suspensory anthropoids. At the present time, we cannot explain the presence of a locomotor-associated signal in the cortical bone consistent with nonhuman anthropoid hip joint loading models and the lack of a similar locomotor-associated signal in the trabecular architecture. These results echo recent brief comments by Lovejoy et al. (2002) that recognized possible contradictory functional signals in these tissues given their observations of ape and human radiographs. Through its mass and distance from the central axis of the femoral neck, cortical bone can significantly contribute to its bending strength and interspecific scaling data suggest that cortical bone structural design modifications are one strategy to reinforce the anthropoid femoral neck as body mass increases (Rafferty,1998). But this theoretical situation then begs the question that if hip joint loading during locomotion and trabecular architecture are similar across locomotor groups, respectively, then why there are not greater similarities in cortical bone structure in nonhuman anthropoids as well.

Clearly, we understand less about nonhuman anthropoid hip joint biomechanics than previously considered. In future studies that explore these issues, it will be important to collect more anthropoid comparative data and incorporate them into finite-element models like those presented by Ryan and van Rietbergen (2005), but that specifically focus on resolving the load sharing between cortical and trabecular bone in the femoral neck across taxa and locomotor groups. Even if the role of cancellous bone is primarily to transfer the hip joint load away from the joint and toward the cortical bone (Currey,2002) instead of load bearing per se, many studies indicate that it is best to consider the cortical and trabecular bone working synergistically to determine the mechanical properties of a whole bone or region of a bone. Loss of either tissue type (e.g., osteoporosis) will significantly alter the mechanical properties of the bone (Bohr and Schaadt,1985; Ito et al.,1993; Rogers and LaBarbara,1993; Lotz et al.,1995; Silva et al.,1997; Ciarelli et al.,2000; Grynpas et al.,2000; Jordan et al.,2003a,2003b; Riggs et al.,2004; Mayhew et al.,2005). In addition, if we look at primates and other mammals, it appears that we have yet to understand what mechanisms drive cortical bone structure in the femoral neck. An inferiorly skewed asymmetrical distribution of cortical bone in the femoral neck, with a very thin superior cortex, is ubiquitous in strepsirhines and common among other mammals (Kummer,1959; Demes et al.,2000). Although the loading models may be different, it is also worth pointing out that symmetrical distributions of cortical bone in the femoral neck are found in some rats (Bagi et al.,1997; Sun and Turner,2003) as well as hominoids (Lovejoy,1988; Ohman et al.,1997; Rafferty,1998; Lovejoy et al.,2002), and debate surrounds the cortical bone structure of Australopithecus africanus (Ruff et al.,1999) and Orrorin tugenensis (Pickford et al.,2002; Galik et al.,2004; Ohman et al.,2005). To complicate the issue further, adaptive computer simulations of human proximal femur bone density patterns suggest that a single structural solution can result from a variety of loading conditions (Fischer et al.,1996).

Volume of interest choice greatly influences the results of trabecular bone architecture studies (Ryan and Ketcham,2002a). We used volumes of interest that ranged in length from 5 mm (from first slice to last slice) to approximately 2 mm and were contoured to the endosteal border of the femoral neck. A 5 mm maximum volume length was chosen because prior histomorphometric and micro-CT studies typically used 4–5 mm2 areas or volume lengths in their analyses, respectively. Furthermore, in a preliminary study (Fajardo and Müller,2001), it appeared that a 5 mm long volume pushed the limits of manageable image files. As described above, these VOI were scaled to capture homologous regions and similar biomechanical zones in each species' femoral neck. The image data show that numerous trabeculae are present in the femoral neck and even within each subvolume of the neck. These VOI are positioned proximally in the neck and any further adjustment of the VOI proximally to capture more trabeculae would enter the femoral head. The VOI could have been extended distally but we cannot speculate how lengthening the VOI would have affected the results.

The determination of the volumes of interest for analysis would have only changed slightly had we considered an alternative femoral neck biomechanical model. Short beam theory has been proposed as an alternative model to explain strepsirhine femoral neck morphology (Rafferty,1998; Demes et al.,2000). This model might also apply to dedicated quadrupedal primates (e.g., M. fascicularis and P. cynocephalus) that have short femoral necks relative to width (Rafferty,1996). Although the nonlinear distribution of shear stresses in irregular-shaped objects makes it difficult to predict the distribution of stresses in a member, had we considered this model initially, we believe our VOI setup would have been the same. The use of superior and inferior subvolumes would not greatly affect a short beam model analysis because the amount of bone tissue would still be quantifiable and the expectation would be that the two VOIs would include similar amounts of bone. The use of subvolumes of interest, however, would be necessary to distinguish between more short beam appropriate and beam in bending appropriate biomechanical conditions since the distribution of the bone tissue is more important for the latter model.

We did not quantify the angular orientation of the trabeculae in this study (Ryan and Ketcham,2005). The similarities in BV/TV and DA in the nonhuman anthropoids suggest that the mechanical properties of the trabecular bone will be the same across taxa (Turner,1992), but future studies will endeavor to quantify trabecular orientation since it might be a better correlate of locomotor mode.

Anthropoid structural correlates of locomotor mode are lacking in the proximal femoral neck trabecular architecture. In fact, the structure of cancellous bone in the proximal femoral neck is strikingly similar across anthropoid phylogenetic and locomotor groups. This is an unexpected result considering the current models of hip joint loading among anthropoid primates. The structural similarity across the taxa studied here is suggestive of greater similarity in the mechanics of hip joint loading than differences. Our results highlight the need for more research in this area. Future studies will need to look at both bone tissue types to understand the respective roles of cortical and trabecular bone in this region. There has yet to be a comprehensive comparative functional (both bone tissue types) analysis of proximal femur bony morphology. This work suggests that future studies of the primate femoral neck will need to develop alternative models of hip joint loading. The current models of nonhuman primate hip joint biomechanics are likely to be simplistic at best.


The authors thank B. Demes, J. Stern, W. Jungers, and J. Kappelman for their insightful criticisms during this work. In addition, discussions with C. Heesy, P. O'Connor, A. Gordon, D. von Stechow, and T. Ryan helped advance the work at one stage or another. G. Hébert provided editorial comments. Supported by grants from the L.S.B. Leakey Foundation and the National Science Foundation (BCS-9904925; both to R.J.F.) and the Swiss National Science Foundation (SNF 620-58097.99; to R.M.)