Over the past decades, the increasing developments of X-ray computed microtomography have enabled high-resolution imaging of internal anatomical structures such as the trabecular bone (or cancellous bone). This technique provides several advantages over the traditional histological sections inasmuch as it is nondestructive and noninvasive (e.g., Fajardo et al., 2002), and as it allows for three-dimensional analyses and quantification of the trabecular structure and its mechanical properties. Indeed, the three-dimensional arrangement, fabric, or textural anisotropy of trabeculae in limb or girdle bones can be used as an indicator of differences in the orientation and magnitudes of external loads engendered during various locomotor behaviors [e.g., Odgaard, 1997; Fajardo and Müller, 2001; Viola, 2001; Fajardo et al., 2002; MacLatchy and Müller, 2002; Ryan and Ketcham, 2002a,b, 2005, 2010; Hill and Richtsmeir, 2008; Volpato et al., 2008; but see Ryan and Walker (2010)]. Giving access to the structural adaptation of bones, the trabecular bone provides a potentially useful analytic tool for reconstructing the locomotor behavior of extinct taxa.
The trabecular bone consists of interconnected rods and osseous plates that form a complex three-dimensional network, which can vary in density, orientation, and complexity, in relation to the functional loading pattern to which the bone is subjected (e.g., Biewener et al., 1996). Basically, when specific loads associated with different locomotor types are strong, bones are required to be strengthened in three dimensions via the development of a complex internal trabecular network. Body size (body mass) has also some effects on the trabecular bone structure. Rafferty (1996) showed that across primates ranging from 3 to 90 kg, the trabecular volume seems to scale isometrically with body mass. In studying the proximal femora of moderately large primates, such as the cercopithecoids, Viola (2001) showed a correlation between the body mass and the degree of anisotropy of the trabecular network. However, in small-bodied primates (strepsirhines), body mass has appeared unimportant as a factor driving the bone volume fraction and the fabric anisotropy (e.g., MacLatchy and Müller, 2002; Ryan and Ketcham, 2002a). Other factors can affect the architecture of the trabecular network, for instance, intrinsic parameters, such as age and sex, may cause variations among individuals of the same species. It has been shown in human that bone restructuring reduces during the aging process and that phenomena of osteopeny/osteoporosis weaken the trabecular structure in aged women (e.g., Lanyon, 1996). In addition, sexual hormones, especially in females (estrogen), contribute to the formation of the trabecular network by stimulating the osteogenic response related to compressive loads (e.g., Lanyon, 1996). Finally, variations in the trabecular architecture between males and females may be observed, notably in the case of strong sexual dimorphism involving significant differences in the locomotor repertoires between both sexes.
Living primates exhibit a wide range of activities and positional behaviors, which are associated with various types of locomotion [quadrupeds (arboreals, terrestrials), leapers, climbers, etc.; see Table 1; see also Napier and Napier (1967), Rollinson and Martin (1981), Martin (1990), Fleagle and Mittermeier (1980), Fleagle and Meldrum (1988), and Hunt et al. (1996)]. Primates have therefore attracted much attention for the study of the relationships between the trabecular bone structure and functional constraints (loading environment) generated during locomotion. Most of analyses on this internal structure have primarily focused on proximal femora (i.e., head and neck) and proximal humeri, but also on pelvis, vertebrae, calcanei or forefoot, and mandibular condyles (e.g., Macchiarelli et al., 1999; Viola, 2001; MacLatchy and Müller, 2002; Ryan and Ketcham, 2002a,b, 2005; Ryan and Rietbergen, 2005; Maga et al., 2006; Volpato et al., 2008; Ryan et al., 2010; Griffin et al., 2010; Ryan and Walker, 2010), as these investigations on primates have direct implications regarding human clinical applications. However, few analyses have focused on tarsal bones, notably on the ankle bone (i.e., talus or astragalus), which is one of the most critical tarsal bone involved during locomotion. Indeed, as it connects the leg to the foot, the talus plays a pivotal role in terms of posture and locomotion, because it is primarily responsible for dorsiflexion and plantarflexion of the foot at the talocrural joint, and it contributes to the eversion and inversion of the foot at the subtalar joint (grasping foot). The external morphology of the talus is mostly determined by mobility demands at these joints. The morphofunctional external anatomy and joint mechanics of the talus are particularly well understood, notably among living but also extinct primates (e.g., Dagosto, 1983, 1986, 1993; Gebo, 1986, 1988; Beard et al., 1988; Fleagle and Meldrum, 1988; Dagosto and Gebo, 1994; Fleagle and Simons, 1995; Gebo et al., 2000; Seiffert and Simons, 2001; Marivaux et al., 2003, 2010, 2011; Dagosto et al., 2010; Boyer et al., 2010; Youlatos and Koufos, 2010; Youlatos and Meldrum, 2011). In contrast, the internal structure of this tarsal bone has not been intensively investigated among primates, except in humans (e.g., Takechi et al., 1982; Pal and Routal, 1998).
Table 1. Classification of locomotor categories among modern small to medium-sized arboreal primates (adapted from Napier and Walker, 1967; Rollinson and Martin, 1981; Martin, 1990)
The asterisks “*” indicate the taxa for which the trabecular bone structure of the talus was investigated here.
In this work, we analyze the trabecular network variations of the ankle bone in small to medium-bodied primates in using high-resolution X-ray computed tomography (HRXCT). Several small to medium-sized living species for which the locomotor behavior is known were selected among strepsirhines (lemurs and lorises) and haplorhines (Tarsius and anthropoids). With such a taxonomic sampling, we assess the relationships between the variations in the trabecular bone pattern of the tali and the functional strain pattern engendered during locomotion (postural activities and positional behaviors). Given the critical role of the talus in the foot motions, it may be expected that interspecific differences in the three-dimensional arrangement of the talar trabecular bone (Fig. 1) may indicate biomechanically fine-scale differences in function resulting from dissimilar locomotor behavior between taxa. These observations will then be applied to ankle bones belonging to some Eocene European primates [Adapis, Leptadapis (Adapiformes), and Necrolemur (Omomyiformes)], for which the trabecular network is reasonably well-preserved. This approach will allow to compare the functional signal deriving from the external anatomy of these fossil tali with that deriving from their internal anatomy and to discuss in fine the potential locomotor behavior of these Eocene primates.
MATERIALS AND METHODS
Selected Taxa and Locomotion
The taxa used in this study are listed in Table 2. The sampling of living species derives from the osteological collections of the Université Montpellier 2 (Laboratoire de Paléontologie, ISE-M), University of Zurich (Anthropological Institute and Museum), and Museum National d'Histoire Naturelle in Paris (Laboratoire Mammifères et Oiseaux). Twenty-eight living species were analyzed; the body mass of which ranged from 60 to 5000 g (Table 2). We selected a maximum number of species differing in locomotor behavior in order to cover at best the diversity of locomotion of these small to medium-sized primates. We followed the work of Rollinson and Martin [1981; revised in Martin (1990)], which addresses a classification of locomotor categories (in relation to body size) most frequently practiced among living primates (Table 1). This classification only groups data into meaningful categories that facilitates analysis and interpretation. However, the range of activities of a species is generally wider and not limited to a single activity. For instance, Pithecia pithecia, Saimiri sciureus, and Saguinus midas are classified among active arboreal quadrupeds after Rollinson and Martin (1981). These primates are somewhat generalists as they can climb, leap, walk, and run on the branches, although not particularly specialized for either of these activities. However, the locomotor repertoire of these species clearly differs in the frequencies of the leaping activity. Indeed, based on published reports of positional behavior and/or ecology, Pithecia pithecia, Saimiri sciureus, and Saguinus midas show a leap frequency of 75%, 42%, and 24%, respectively (Fleagle and Mittermeier, 1980). Therefore, it would be interesting to analyze if these differences in leaping activity, which probably imply differences in the loading environment at the ankle joints, involve different fine-scale three-dimensional rearrangements of the trabecular bone in the talus of these taxa.
A total number of 128 talus specimens were analyzed across 28 small to medium-sized species (16 strepsirhines and 12 haplorhines). For most species, we analyzed a minimum number of two or three tali belonging to different individuals in order to have a representation of the within-species variation of talar trabecular structure (although not statistically measurable). Only a few species (eight) were represented by only one specimen. In contrast, four species, Callithrix jacchus, Perodicticus potto, Saimiri sciureus, and Cebus albifrons, were analyzed with large sample sizes (38, 32, 31, and 8 specimens, respectively), which allowed performing statistical analyses regarding some aspects of the interspecific, intraspecific, and intraindividual variability in the trabecular structure of the ankle bone (see below).
Among the sampled species, we preferentially selected individuals that were wild-shot adults but also included breeding individuals in captivity (menageries of zoos). We were cautious not to select individuals that showed apparent pathologies. Information regarding the sex was available for most of the individuals. The effect of sex on the trabecular bone structure was therefore analyzed notably in monomorphic (Callithrix jacchus and Perodicticus potto) and dimorphic samples (Cebusalbifrons and Saimiri sciureus). We primarily analyzed left tali in this study. However, we assessed the intraindividual variation of the trabecular bone structure between left and right tali on eight individuals of Cebus albifrons. Information regarding age was not available for most of the individuals. However, we defined three broad categories of age: when deciduous premolars were still present, individuals were classified as “juveniles.” When the molars showed abrasion, they were classified as “old.” When the molars showed no sign of abrasion or only little abrasion, they were classified as “adults” (Table 2).
High-Resolution X-ray Computed Tomography and Image Treatments
The 128 tali were imaged using high-resolution X-ray computed tomography (HRXCT). The data regarding the talar trabecular structure were obtained from projection images using a SkyScan 1076 Micro-CT scanner at the Université Montpellier 2 (ISE-M, Montpellier RIO-Imaging) as well as a SCANCO UCT 80 Micro-CT at the University of Zurich (Anthropological Institute and Museum) at a resolution of 18 and 20 μm, respectively. The virtual slices were reconstructed and saved in TIFF 16-bit format. The obtained image stacks were imported in the software VGStudioMax v. 1.1 (Volume Graphics, Heidelberg, Germany). We used this software primarily for reorienting the virtual slices of each analyzed talus in anatomical position. The reoriented images were subsequently converted to BMP-8 bit format using ImageJ (Rasband).
Choice, Position, and Volume of Interest Extraction
The eight-bit-reoriented virtual slices were imported in the software CT-Analyser v. 220.127.116.11 (Skycan, Kontich, Belgium) for defining region of interests (ROI) within the talar body and then to extract a corresponding volume of interest (VOI). We selected the talar body as an ROI, because this region of the talus is particularly subjected to functional loads associated with leg movements and foot positions during locomotion (dominant loading regime) and because it supports and transmits a part of the body mass. Dorsally, the talar body bears the trochlea, which articulates medially with the tibia and laterally with the fibula (talocrural joint). Plantarly, the talar body bears the ectal facet and the proximal part of the subtantacular facet, which both articulate with the calcaneum (subtalar joint). Basically, the gross morphology of the talus differs from one species to another, and as such, the ROI in the talar body corresponds to an approximate functionally homologous region across species. To analyze and compare the trabecular architecture of the talus between different taxa, cubic VOIs of the three-dimensional reconstruction of the HRXCT scan data were extracted by segmentation within the talar body (under the trochlea) on each sample. We developed a VOI selection method for standardizing the location and the extraction across the studied specimens. For this, we located a central point in the talar body, which is a point situated at mid width and mid length of the trochlea, and projected at mid height in the talar body (Fig. 2). We consider here that the position of this central point is analogous from one talus to another. The talar body being often asymmetric mediolaterally, two symmetric and adjacent cubic volumes were extracted as VOIs on both sides (medial and lateral) of the talar body following the median anteroposterior long axis of the trochlea (Fig. 2). This has allowed investigating regional variation within the talar body. The central point of the trochlea represents the center of the lateral side of the medial VOI and the center of the medial side of the lateral VOI (Fig. 2). The size of both medial and lateral VOIs was scaled based on the width of the trochlea (mid-trochlear width). This method allows for extracting optimal cubes (maximal volumes) and for studying a proportional amount of trabecular bone within the talar body across each specimen. The cortical bone of the talar body (generally not heavily developed or even nonexistent in tali of taxa of this selected body size range) was not included in the VOIs. Medial and lateral cubes from the same individual were analyzed separately in order to get information regarding regional and intraindividual structural variation. Reconstructed volumes extracted from the medial and lateral sides of the tali of all taxa analyzed in this work are shown in Fig. 3.
Trabecular Bone Structural Analysis
For quantifying in three dimensions, that is, the trabecular structure on each of the extracted VOI, we used the morphometric analysis program QUANT3D (Ryan and Ketcham, 2002a; Ketcham and Ryan, 2004; Ketcham, 2005). The threshold value between bone and air within each individual's cubic VOI was calculated using the iterative method of Ridler and Calvard (1978) and Trussell (1979) (implemented in QUANT 3D). The trabecular bone morphometric parameters quantified in the VOIs included the fabric anisotropy degree of anisotropy (DA), the trabecular elongation (E), the bone volume fraction (BV/TV), the number of trabeculae (Tb.N), and the thickness of the trabeculae (Tb.Th). The star volume distribution (SVD) method was used here to calculate a second-rank fabric tensor that describes the distribution of the trabeculae in three dimensions. Details of the SVD method can be found in Ryan and Ketcham (2002a, 2005), Ketcham and Ryan (2004), and Ketcham (2005). In this study, linear intercepts were measured using 2,049 uniformly distributed orientations and 8,000 random points in three-dimensional space. The three-dimensional orientation of the trabecular fabric (primary, secondary, and tertiary material axes) was captured by the tensor eigenvectors (υ1, υ2, υ3), which represent the principal component directions. The corresponding eigenvalues (τ1, τ2, τ3) represent the relative magnitudes of each of the three material orientations. The SVD-DA is calculated as the ratio of the primary to the tertiary eigenvalues (τ1/τ3). The description of the trabecular elongation is given by the SVD-E as 1 − τ2/τ1. The bone volume fraction is defined as the number of bone voxels/total voxels (BV/TV) in the VOI. The Tb.N is an estimated parameter based on the number of intersections between a grid of lines and the bone, normalized by total grid line length (Odgaard, 1997). The Tb.Th was calculated using the intercept data collected for the SVD analysis by defining Tb.Th as the shortest measured intercept lying in bone at each of the 8,000 points [see Ryan and Ketcham (2002a, 2005) and Ketcham and Ryan (2004)].
For the genera Callithrix, Perodicticus, Saimiri, and Cebus, differences in the talar trabecular structure (Tb.Th, Tb.N, BV/TV, SVD-DA, and SVD-E) were assessed using paired-sample t tests. Differences between left and right feet on talar trabecular structure were assessed in Cebus albifrons using paired-sample t tests for the eight specimens for which both left and right tali were digitized. The effects of sex and age on the talar trabecular structure within the genera Callithrix, Perodicticus, Saimiri, and Cebus were tested using ANOVAs. The effect of captivity on trabecular structure was tested in Perodicticus and Saimiri, for which both wild-shot and captive specimens have been digitized (9 wild/13 captive for Saimiri and 11 wild/9 captive specimens for Perodicticus), using ANOVAs. The effect of the body mass on the trabecular bone structure was also assessed using linear regression. Data on adult body mass were provided after Smith and Jungers (1997). For each species, mean values of trabecular bone structure parameters were computed, and linear regression of these parameters with specific-wise body mass estimates was computed. Because body mass estimates vary between males and females in most taxa (see Table 2), for the species for which both males and females or sex undetermined individuals had been digitized, body mass was defined as the mean of male and female body masses provided by Smith and Jungers (1997). Variation of trabecular bone morphometric parameters within our sample was described using a principal component analysis (PCA) on the correlation matrix. A second PCA was also performed on species averages as a mean to describe interspecific variation of trabecular bone parameters. To assess the potential to infer the locomotion repertoire from trabecular structure, a linear discriminant analysis (LDA) at the specific level was computed, the categorical variable being locomotion as defined in Table 1. Scores of the specimens on the discriminant axes were used to reallocate a posteriori all the specimens to a locomotion category. All statistical tests and multivariate analyses were performed using R 18.104.22.168 (R Development Core Team, 2011) and JMP 9.0. (SAS Institute).
Species-wise mean values and variance for trabecular parameters can be found in Table 3.
Table 3. Trabecular bone morphometric parameters measured for left tali of living taxa digitized in this study
Mediolateral Intraindividual Variation and Interspecific Variations
Results of the paired t tests between the medial and lateral sides within the genera with large sample sizes, Callithrix, Cebus, Saimiri, and Perodicticus, are listed in Table 4. In these four genera, the BV/TV is significantly higher in the medial side than in the lateral side. The highest mean BV/TV values for the medial side are observed in Microcebus and Cebuella with about 31% of bone in the VOI. The lowest means BV/TV in the lateral side are found in Arctocebus and Saimiri with 12.7% and 13.7% of bone in the VOI, respectively (Table 3). In Perodicticus and Cebus tali, the trabecular thickness (Tb.Th) in the medial side is significantly greater than in the lateral side (see Table 4). However, similarity in medial–lateral trabecular thickness could not be rejected in Callithrix and Saimiri (see Table 4), and no general trend in the difference between lateral and medial trabecular thickness can be assessed statistically in other taxa (see Table 3). Values for Tb.Th range from 0.056 to 0.138 mm in Microcebus (lateral side) and Propithecus verreauxi (medial side), respectively (Table 3).
Table 4. Differences in talar trabecular structure between medial and lateral VOIs
Tb.Th, trabecular thickness; Tb.N, trabecular number; BV/TV, bone volume fraction; SVD-DA, directional anisotropy; SVD-E, elongation; d.f., degrees of freedom.
t = 1.1; P = 0.13
t = 9.06; P < 0.01
t = 9.34; P < 0.01
t = 0.8; P = 0.79
t = 8.23; P < 0.01
t = 5.72; P < 0.01
t = 5.86; P < 0.01
t = 6.51; P < 0.01
t = 2.35; P = 0.01
t = 0.94; P = 0.18
t = 1.1; P = 0.14
t = 5; P < 0.01
t = 3.91; P < 0.01
t = 1.03; P = 0.16
t = 2.5; P < 0.01
t = 3.24; P < 0.01
t = 1.17; P = 0.13
t = 2.11; P = 0.03
t = 0.47; P = 0.32
t = 2.32; P = 0.02
The estimated Tb.N is significantly higher in the medial side than in the lateral side of tali in Callithrix, Perodicticus, and Saimiri (see Table 4). These values suggest that higher BV/TV values result from an increase in the Tb.N in the medial VOI. Interestingly, in Perodicticus, the high value of BV/TV in the medial VOI is associated with a large Tb.N, which also appears to be thicker than that in the lateral side. The highest numbers of trabeculae in the medial side are observed in Microcebus and Cebuella with 2.83 and 2.46/mm, respectively. The lowest numbers of trabeculae in the lateral side are in Arctocebus and Saimiri with 0.82 and 0.96/mm, respectively (Table 3).
Equality of anisotropy (SVD-DA) in the medial and lateral sides could not be rejected in Cebus, Saimiri, and Callithrix (Table 4). The degree of anisotropy is, however, significantly greater in the lateral side than in the medial side for Perodicticus, but the values remain well-inferior to those measured for Saimiri and Callithrix (Table 3). Values for anisotropy in the medial side range from 2.14 to 35.4 in Daubentonia and Tarsius syrichta, respectively (Table 3). The SVD-E values are significantly higher in the lateral side than in the medial side in Cebus, Saimiri, and Callithrix, but not in Perodicticus (Table 4), in which the elongation is quite similar in both sides with lower values (Table 3). The highest mean elongation values for the lateral side are in Leontopithecus and Callimico with 0.788 and 0.774, respectively. The lowest mean elongation values in the medial side are in Daubentonia and Cebusalbifrons with 0.285 and 0.323, respectively (Table 3).
Left Versus Right Tali Intraindividual Variation
In Cebus albifrons, significant differences in right and left tali were found: the Tb.N was found to be higher in left tali than in right tali (Table 5). The lateral medial bone volume ratio was found to be significantly greater in the left feet. Also, anisotropy was found to be significantly lower in left tali than in right tali. But this result would require further specimens (>8) to assess properly this variation and its significance.
Table 5. Differences on trabecular structure between left and right feet of Cebus albifrons
Computed on the eight Cebus albifrons specimens for which both left tali and right tali were digitized.
Medial trabecular thickness (mm)
Lateral trabecular thickness (mm)
Medial trabecular number (mm−1)
Lateral trabecular number (mm−1)
Medial bone volume ratio
Lateral bone volume ratio
Age, Sex, Wild/Captivity Effects
No significant effect of the age and sex on the trabecular structure was found in Cebus, Saimiri, Callithrix, or Perodicticus. The only exception is for SVD-E in the lateral side in Callithrix, which is significantly lower in older specimens (F = 5.628, P = 0.02); but again, this result would require further specimens to assess properly this variation. No significant effect of captivity was found in Perodicticus or Saimiri.
Body Mass Effect
Trabecular thickness (lateral and medial) is positively and significantly correlated with the logarithm of body mass (F = 9.72, P < 0.01; F = 9.8, P < 0.01). The trabecular number (lateral and medial) is negatively and significantly correlated with the logarithm of body mass (F = 6.5, P = 0.02; F = 13.2, P < 0.01). Bone volume fraction, degree of anisotropy, and elongation are not significantly correlated with the logarithm of body mass.
The trabecular bone structure within the talar body of practically all the sampled primates is mostly oriented in a dorsoplantar direction. In both lateral and medial VOIs (Fig. 4A,B), the primary eigenvector orientation within each taxon is roughly similar and demonstrates that there is no strong difference in the orientation of the trabecular fabric across the locomotor categories. Practically, all taxa (except slow climbers) are relatively tightly clustered around the dorsoplantar direction. However, slow climbers exhibit more variation, notably in the medial VOIs, in which the primary axis is oriented slightly dorsally to the anterior direction. In the medial VOIs, practically, all taxa are clustered around the anteroposterior direction (more posteriorly) for the secondary eigenvector orientation and in the mediolateral direction for the tertiary eigenvector orientation. In contrast, in the lateral VOIs, all taxa are more dispersed across the stereonet, ranging from the anteroposterior to mediolateral axis for the secondary and tertiary eigenvector orientations, respectively. The nonleaper climber and generalist quadruped taxa display variation in secondary and tertiary eigenvector orientations in both lateral and medial VOIs.
Principal Component Analyses
The results of the principal component analysis (PCA) are shown in Fig. 5. The first axis (31.46%) describes mostly variation in the bone volume fraction and in the number and thickness of trabeculae. Specimens projecting to higher values of PC1 tend to have a larger Tb.N, a higher bone volume ratio, and Tb.Th in both medial and lateral sides of their tali (e.g., Lepilemur, Eulemur, and Lemur). The second axis (22.18%) describes mostly negative correlation in the variation in the Tb.N and Tb.Th: specimens projecting to higher values of PC2 tend to have more and thinner trabeculae (e.g., Tarsius, Microcebus, and Cebuella), while specimens projecting to lower values of PC2 have less and Tb.Th (e.g., Arctocebus, Cebus, and Aotus). The third axis (13.57%) describes mostly variation in medial and lateral degree of anisotropy and lateral elongation. Specimens projecting to higher values of PC3 tend to have a higher degree of anisotropy and high values of elongation (e.g., most Callithrix specimens and Leontopithecus). Projection of the specimens in PC1–PC3 space shows that there is a large intraspecific variation regarding the trabecular structure. This is particularly clear when observing the projection scores of the specimens belonging to the genera Saimiri, Cebus, Callithrix, and Perodicticus for which a larger number of specimens had been sampled. However, there are trend differences between these four genera. For instance, specimens of Perodicticus potto exhibit a wide range of PC1 scores but tend to project values close to 0 on PC2 and tend to express low PC3 scores. Specimens belonging to the genus Cebus tend to project low value of PC2, indicating that they tend to have a low Tb.N, relatively to their thickness. Specimens of Saimiri sciureus project to low scores on PC1 and PC2. Specimens of Callithrix jacchus project to high values on PC2. Regarding locomotion, no difference between arboreal quadrupeds and vertical clingers and leapers can be seen in PC1–PC3 space. However, all small agile quadrupeds (B and C) and small vertical clingers and leapers (A1) tend to project positive values on PC2, whereas medium-sized branch runners and walkers (E1 and E2) project low values. Slow climbers (D) also tend to project low values on PC3, as they exhibit a small degree of anisotropy and a low-lateral elongation. Most medium-sized clingers and leapers (A2) project positive values on PC1.
The results of the PCA at the specific level are shown in Fig. 6. The first axis explains 29.26% of total variance and describes mostly negative correlation in the variation of trabecular thickness and trabecular number. This axis is significantly and negatively correlated with the logarithm of body mass (r2 = 0.37; P < 0.001). The largest species selected, such as Propithecus diadema and Propithecus verreauxi, Cebus albifrons and Cebus apella, and Aotus trivirgatus, tend to show low-projection values on PC1 and show a higher degree of trabecular thickness relatively to their Tb.N. The smallest species selected, such as Microcebus murinus and Cebuella pygmaea, tend to project higher values on PC1 and show thinner trabeculae relatively to trabecular number. The second axis (23.84% of total variance) describes mostly negative correlation in the variation of bone volume and anisotropy. Species, such as Galago demidovii, Tarsius bancanus, Pithecia pithecia, and Saimiri sciureus, which project to low values on PC2, tend to show a high degree of anisotropy relatively to the bone volume ratio. Species that project to higher values on PC2, such as Saguinus midas, Lepilemur ruficaudatus, and Lepilemur leucopus, tend to show a smaller degree of anisotropy relatively to their bone volume ratio. PC3 (12.12% of total variance) describes mostly negative correlation in the variation of medial and lateral elongation in the talus. Species that show low-PC3 values (such as Hapalemur griseus or Varecia variegata) exhibit relatively higher values of elongation in the medial side than in the lateral side, while those that project to higher PC3 values (such as Leontopithecus rosalia or Callimico goeldii) exhibit relatively higher values of elongation in the lateral side than in the medial side. No clear separation between locomotor categories can be seen within PC1–PC3 space.
Linear Discriminant Analysis
The linear discriminant analysis (LDA) successfully discriminates primate species according to the seven locomotion categories defined earlier (see Table 1, after Rollinson and Martin, 1981). Twenty-five of the 28 species were correctly reallocated to their original locomotion category, suggesting that talar trabecular structure has a good potential for predicting locomotion in extinct species. Only Hapalemurgriseus was wrongly classified as a branch runner quadrupedalist (E1) rather than as a vertical clinger and leaper (A2). Also, Aotus trivirgatus was wrongly classified as a slow climber (D) rather than a branch-running quadrupedalist (E2), while Arctocebus calabarensis was classified as a branch-running quadrupedalist (E2) rather than a slow climber (D). The projection scores of the specimens on the first two discriminant axes of the LDA are presented in Fig. 7. These axes are both negatively and significantly correlated with the logarithm of body mass (LD1: r2 = 0.18, P = 0.03; LD2: r2 = 0.36, P < 0.001). The first axis discriminates small vertical clingers and leapers (A1) and small agile clawed quadrupeds (B; high-projection scores) from slow climbers (D) and branch-running quadrupeds (E1 and E2). Medium-sized lemuriform vertical clingers and leapers (A2) tend to project to the center of LD1–LD2 space. The second axis tends to discriminate species according to body mass.
These results highlight the issues of interspecific and intraspecific variations of trabecular bone architecture of the talus and notably the regional variation (medial vs. lateral sides) within this tarsal bone. Only the large sample sizes used here (Callithrix, Perodicticus, Saimiri, and Cebus) allow for an evaluation of the significance of the observed variations. The small sample sizes for most taxa prevent any definite conclusions regarding the differences in the internal bone structure; but together with the better sample, it is possible to address some trends regarding relationships between trabecular bone architecture and ankle joint loading associated with specific foot postures involved during locomotion.
Body Mass and Trabecular Bone Structure in Tali of Living Primates
The effect of body mass on the three-dimensional architecture of trabecular bone has never been clearly established, notably for small-bodied primate species. Swartz et al. (1998) found no significant relationship between trabeculae size and body size. Rafferty (1996) has shown that across primates ranging from 3 to 90 kg, the trabecular volume seems to scale isometrically with body mass. Viola (2001) has also shown a correlation between body mass and the degree of anisotropy of the trabecular network in the proximal femur (neck) of large primates, such as cercopithecoids. However, for small-bodied primates (≤ 1 kg), no significant relationship between body size and trabecular structure, either in bone volume fraction or fabric anisotropy, has been demonstrated. These conclusions are derived from the studies of the trabecular bone of the femoral head of small-bodied strepsirhines (e.g., MacLatchy and Müller, 2002; Ryan and Ketcham, 2002a). Taking into consideration these previous observations, one could expect the existence of a size threshold. However, given that the postural activities and positional behaviors of primates are to some extent constrained or limited by body size (e.g., Biewener, 1989; Fleagle and Mittermeier, 1980; Polk et al., 2009), it seems difficult to identify precisely whether body mass per se or functional load constraints engendered during locomotion are more important in determining the variations of trabecular bone structure. In this study, we have analyzed the variations of the trabecular structure in the talus (talar body) across taxa (strepsirhines and haplorhines) ranging from 60 to 5,000 g, taxa for which we had some information regarding their postural activities and positional behaviors during locomotion. As we have mentioned previously, we have selected the talus (talar body) for these investigations of the structure and function of the trabecular bone, because the internal and external morphology of this tarsal bone are determined by both weight bearing and mobility demands. From the results of our analyses performed on small to medium-sized taxa, it appears that neither the bone volume fraction nor the fabric anisotropy is significantly correlated with body mass. These results corroborate the conclusions of Ryan and Ketcham (2002a) and MacLatchy and Müller (2002) who analyzed these possible relationships among small-bodied strepsirhine primates. However, we get new interesting information regarding the effect of body mass on some trabecular parameters, notably on the number and Tb.Th. In the talar body (both lateral and medial sides), trabecular thickness is positively and significantly correlated with body mass. Even more interesting, the Tb.N is negatively and significantly correlated with body mass. The smallest species sampled (e.g., Cebuella or Microcebus) record the highest Tb.N, which are very thin. In contrast, medium-sized species (e.g., Cebus, Aotus, and Propithecus diadema) show a higher degree of trabecular thickness relatively to their Tb.N. From the results performed on small to medium-sized primates, it seems that both the orientation and elongation of the trabecular fabric are essentially driven by the loading environment associated with locomotion. In contrast, if this loading environment has an effect on the number and thickness of trabeculae (see below), these two trabecular bone parameters seem to be also body mass-dependent. Regardless of the functional loading effects, it would be however particularly interesting to explore more in detail the biomechanical and/or physiological reasons for such an increase of the number of thin trabeculae in very tiny species.
Mediolateral Intraindividual and Interspecific Variations of the Trabecular Bone Structure in Tali of Living Primates and Their Relations to the Foot Motions
In Callithrix, Perodicticus, and Saimiri, for which we have large sample sizes, the BV/TV measured on the medial side of their tali is significantly higher than on the lateral side (Tables 3 and 4). In these taxa, the high-medial BV/TV values are associated with a significantly higher Tb.N. In taxa with small sample sizes (variations not able to be statistically assessed), this medial increase in the bone volume density is also observed, with taxa showing strong (Arctocebus, Microcebus, and Loris), moderate (Aotus, Cebuella, and Lemur), or slight (Lepilemur, Varecia, Callimico, Leontopithecus, and Cebus) mediolateral differences. A mediolateral difference in the bone volume fraction is particularly marked in tali of Perodicticus, which have furthermore significantly Tb.Th in the medial side like in Arctocebus and Loris [and also, but to a lesser extent in Microcebus, Lepilemur, and Cebuella (not statistically measurable)]. Surprisingly, some taxa, such as Pithecia and Nycticebus, have virtually no mediolateral difference in bone volume density, although the medial side of their tali records a noticeable increase in the Tb.N, but without thickening (Table 3). Some other taxa, such as Avahi, Daubentonia, Eulemur, Hapalemur, Tarsius, and also Galago (G.demidovii), show a slight to moderate mediolateral difference in BV/TV, but contrary to the other primates, the increase in bone volume density is recorded on the lateral side of their talar body (Table 3).
From the measures made on the taxa for which we do not have large sample sizes, it remains difficult to evaluate the significance of the variations observed in the talar bone density. However, some general trends can be addressed. Most primates are generalists and include climbing in their locomotor repertoire, a locomotor behavior that requires some mobility at the talocrural joint (i.e., trochlea slightly grooved [rather flat] and wedge-shaped) and some capabilities of foot inversion at the subtalar and transverse tarsal joints (i.e., grasping foot). The mediolateral variation in the bone volume fraction in the talus is therefore likely to be linked to these specific foot postures. During eversion–inversion foot motions, which are nonstereotypical but somewhat directional, the magnitude of forces at the talocrural and subtalar joints is seemingly stronger on the medial aspect of the talus. Such a loading pattern probably involves a three-dimensional strengthening of the trabecular structure, which is primarily characterized by an increase in the Tb.N (denser structure) without specific strong organization (i.e., less oriented structure). In highly specialized climbers, such as Perodicticus, Loris, or Arctocebus (slow and cautious climbers [D]), but also in some frequent climbers, the increase in the Tb.N in the medial aspect of the talus is also generally associated with an increase in the thickness of these trabeculae. This local increase in bone volume density in tali of small frequent climbers therefore reflects a specific reinforcement of the trabecular structure in response to the apparent strong magnitude of loads engendered in habitually sustained foot inversion. Given that body mass has a significant effect on the number and thickness of trabeculae, it may be expected that body mass is a component of the loading environment at the ankle joint during the eversion–inversion foot motions. Conversely, in medium-sized taxa characterized by less sustained or less frequent inverted foot posture (e.g., Cebus, Pithecia, Lemur, and Daubentonia), such a mediolateral variation in the trabecular volume density is not so marked or even is absent. In these more generalized quadruped taxa, which are capable of climbing and leaping, although not particularly specialized for either of these activities, the bone volume fraction is rather mediolaterally uniform. This could be explained by the fact that their feet are subjected more frequently to parasagittal motions (dorsi- and plantar flexions) than to transverse motions (eversion–inversion).
In small to medium-sized leaping primates, such as Tarsius,Galago, Avahi, and Propithecus, there is no strong mediolateral difference in the bone volume fraction either, but unlike climbers, they show a slight increase in the Tb.N (without thickening) in the lateral side of their talar body (Table 3). However, we must keep in mind that the increase in the Tb.N in tiny species may be due to an effect of body mass (see above). Tali of primates who are frequent leapers exhibit a different three-dimensional distribution of material, which consists of a more anisotropic structure (more oriented trabecular structure). However, it is worth noting that none of the specimens selected in the current study exhibits a fully nonoriented fabric structure (i.e., isotropic) in the talar body. In contrast, all taxa show an anisotropic trabecular bone, but the SVD-DA and SVD-E vary, notably across the locomotion categories (Table 3). From our results, it is clear that primates who leap frequently tend to have more anisotropic trabecular bone in the talar body than less frequent and nonleaping forms. Interestingly, data on the fabric orientation in both medial and lateral VOIs indicate that, practically, all taxa display a generally consistent pattern of orientation restricted primarily to a dorsoplantar direction (Fig. 4). Basically, locomotion in all primates implies primarily parasagittal movements of the feet. So, the fact that the primary orientation appears to be the same in all taxa is consistent with the observation that the talus (talar body) functions primarily in all taxa as a pulley at the talocrural joint (dorsal trochlea), which allows the dorsi- and plantar flexions of the feet. With such movements at the talocrural joint, the talus is therefore constantly loaded dorsoplantarly, which explains this primary directional orientation (dorsoplantar) recorded in all taxa. In all large sample sizes taxa measured (i.e., Callithrix, Cebus, Saimiri, and Perodicticus), the SVD-DA within the talar body differs between taxa, but that fabric anisotropy appears more or less conservative mediolaterally (difference not significant, except for Perodicticus; Table 3). However, the SVD-E is significantly higher in the lateral side than in the medial side (except for Perodicticus), which indicates a stronger primary orientation within the lateral talar body of Callithrix, Cebus, and Saimiri (Tables 3 and 4). For taxa with small sample sizes, similar trends in the orientation of the trabecular fabric, notably regarding the mediolateral difference in elongation, are observed in practically all small to medium-sized primates who are capable of leaping (Table 3). However, the assumption that the degree of anisotropy does not vary significantly between the lateral and medial sides of the talar body, as revealed from tali of taxa with large sample sizes, cannot be generalized. Indeed, in specialized vertical clingers and leapers (such as Tarsius and Galago), as well as in some taxa engaged in active arboreal quadrupedalism with some good abilities of leaping (such as Callimico, Leontopithecus, Saguinus, Hapalemur, Avahi, and Pithecia), there are well-marked mediolateral differences (although not statistically quantifiable) in the fabric anisotropy in their talar body (Table 3). However, while Callimico, Leontopithecus, Hapalemur, and Galago exhibit a noticeable anisotropic trabecular structure in the lateral side of their talar body, Saguinus, Pithecia, Avahi, and Tarsius have a more anisotropic trabecular structure in the medial side (Table 3). The tali of frequent or specialized leapers have generally limited talocrural mobility in the transversal plane (grooved trochlea) and function primarily in a parasagittal plane for providing dorsi- and plantar flexions of the feet. The different degrees of interspecific and within-species mediolateral variations in fabric anisotropy and elongation demonstrate how differently the species or individuals load their talus for leaping (locomotor specificity or individuality) and that the trabecular bone adapts to local load/stress conditions as a result.
Tali of Tarsius and Galago record very high values of SVD-DA (Table 3), which hence indicates a highly oriented trabecular structure. In these two small-bodied taxa, which are highly specialized vertical clingers and leapers, such a high degree of anisotropic trabecular structure in their tali seems particularly well-adapted for supporting the strong unidirectional and stereotypical loading conditions (stress) generated at the ankle joint by the push-off and landing phases of their powerful leap. In contrast, in highly specialized climbers, such as Perodicticus, Loris, and Nycticebus (as well as in Microcebus), which have generally no leaping activity in their locomotor repertoire (except Microcebus), the talar body displays low-SVD-DA values (among the lowest recorded; Table 3) and does not show any mediolateral difference in elongation, the values (SVD-E) recorded of which remain moderately low as well (Table 3). Compared to leaping primates, it is clear that this more generalized or less anisotropic trabecular structure in the talus of frequent climbers reflects a distinct loading regime, which is most probably less unidirectional (not exclusively in the parasagittal plane) due to the frequent use of transversal movements of the foot (inverted postures) for climbing and grasping (i.e., much subtalar and talocrural joint mobility).
As we have mentioned earlier, published reports of positional behavior for Pithecia pithecia, Saimiri sciureus, and Saguinus midas indicate a leaping frequency in these taxa of 75%, 42%, and 24%, respectively (Fleagle and Mittermeier, 1980). Pithecia and Saimiri, which leap more frequently than Saguinus, submit their ankle to more frequent stereotypical loading environments than Saguinus. Pithecia and Saimiri have indeed more oriented trabeculae in both the medial and lateral sides of their talar body than Saguinus has (low-SVD-DA value in its lateral side). In the talus of Saguinus, the values of elongation are also globally lower (especially in the lateral side and moderately in the medial side) than in Pithecia and Saimiri. In contrast, the values of the bone volume fraction recorded in both sides of their talar body as well as the number and thickness of trabeculae are well-inferior to those recorded in the talus of Saguinus, values which are otherwise particularly high in this taxon (but probably due to its small body size). The less-anisotropic trabecular structure in the talar body of Saguinus indicates less strong unidirectional loading conditions at the talocrural joint than it does in Pithecia and Saimiri, thereby indicating that leaping is not as important in the locomotor repertoire of Saguinus. This suggests more emphasis on quadrupedalism as a result. The locomotor behavior of Saguinus midas is indeed primarily quadrupedal branch running and walking and only secondarily bounding (Fleagle and Mittermeier, 1980).
Application to Eocene Primates
Our analyses of the trabecular structure in the talus of small to medium-sized living primates indicate that the number, thickness, three-dimensional distribution, and arrangement of the trabeculae in this tarsal bone are primarily linked to the types and magnitudes of loads associated with the postures and motions of the foot. Body mass seems to only affect the number and robustness (thin vs. thick) of the trabeculae, but seemingly not their orientation and elongation. In addition to the external functional anatomy, the study of the trabecular structure of this bone is therefore particularly interesting for extinct taxa, as it can contribute to better understanding their potential locomotor behavior. Here, we have quantified the three-dimensional architecture of the trabecular bone in the tali of three late Eocene primates from Southern of France (Quercy phosphorites): two Adapiformes (Leptadapis and Adapis) and one Omomyiformes (Necrolemur). For this study, we have analyzed two tali referred to Leptadapismagnus [Quercy Old Collections (locality unknown)], four tali referred to Adapisparisiensis [localities: Escamps (MP19) and Rosières 2 (MP19)], and ten tali referred to Necrolemurantiquus [locality: La Bouffie (MP17a)]. These fossils are housed in the Paleontological collections of the University Montpellier 2. The locomotor behavior of these taxa has been analyzed previously using various postcranial skeletal elements (e.g., Dagosto, 1983; Godinot and Dagosto, 1983; Godinot and Jouffroy, 1984; Godinot, 1991; Bacon and Godinot, 1998).
Adapis parisiensis was a medium-sized primate with an estimated body weight of about 1.3 kg (Fleagle, 1999). Most of the postcranial features referred to this Adapis species indicate increased mobility of joints, especially for movements out of the sagittal plane. Basically, A. parisiensis appears as a slow-moving arboreal-quadruped primate (climber), which did not include significant leaping activities in its locomotor repertoire (Dagosto, 1983). However, other postcranial analyses have shown the existence of two categories of activities in the locomotor repertoire of Adapis, thereby suggesting the existence of distinct but close species among the “Adapis parisiensis” group. From the functional morphology of the hand bones referred to Adapis from Rosières 2, Godinot and Jouffroy (1984) have shown that this arboreal quadruped primate was rather a branch-running and walking form, capable of climbing but without strong specialization. From the morphofunctional anatomy of the femora, Bacon and Godinot (1998) have shown that some Adapis femora belonged to individuals, which were branch walking and running forms, using climbing less frequently, and some others, which were less specialized for running and that used climbing more frequently.
Leptadapis magnus was a medium to large-sized primate with a body weight estimated between 4 and 5 kg (Fleagle, 1999). Although larger than A. parisiensis, L. magnus was probably also engaged in a kind of arboreal quadrupedalism, using climbing frequently. Indeed, its postcranial morphology, which was unusually robust, dictates less-efficient leaping abilities and consequently slower, more labored arboreal locomotion (Dagosto, 1983). Unlike Adapis and Leptadapis, Necrolemur antiquus was a small-sized primate, the body weight of which approximated 300 g. The postcranial morphology of N. antiquus (notably the talar features) indicates some importance of leaping in the locomotor behavior of this small primate (Godinot and Dagosto, 1983).
The set of fossil tali studied here are particularly well-preserved without postmortem distortion and alteration. For each fossil talus, we have followed the same VOI selection and extraction methods in the talar body than those applied for the tali of living species (Fig. 8). No dense inclusion was detected in the talar body of each specimen, which has therefore allowed us performing the segmentation of the imagery into bone and nonbone phases without difficulty. Both medial and lateral extracted VOIs were analyzed with QUANT3D. The trabecular bone morphometric parameters measured for these fossil tali are provided in Table 6.
Table 6. Trabecular bone morphometric parameters measured for tali of fossil taxa
BV/TV, bone volume fraction; Tb.Th, trabecular thickness; Tb.N, trabecular number; SVD-DA, directional anisotropy; SVD-E, elongation. Adapis parisiensis (ECA-936, ECA-1379, ROS2–108, and ROS2–106); Leptadapis magnus (ACQ-205, CG 1903–20); Necrolemurantiquus (BFI-812–821).
Compared to the tali of living primates, the BV/TV values recorded in the talar body of fossils appear noticeably high. This is particularly shown for the tali of Leptadapis, which record the highest BV/TV values (Table 6). However, a question arises as to whether such a strong density of bone in this taxon reflects natural condition, or if it is the result of an artificial increase due to diagenesis (i.e., increase related to the mineral replacement during the fossilization process). In Leptadapis, the trabeculae are indeed rather thick, and much thicker than in tali of certain living primates for which we have recorded strong Tb.Th values [e.g., Cebus and Propithecus (Table 3)], the body mass of which approximates that of Leptadapis [i.e., 3–5 kg (Table 2)]. The same seems to be true for the tali of Adapis, which exhibit lesser but noticeably thick trabeculae compared to living taxa of the same body mass category (1–2 kg). For the tali of Necrolemur, it seems difficult to claim an effect of diagenesis, because the values of all the trabecular bone parameters, notably those reflecting bone density, enter in the range of variation recorded in tali of living species of similar body mass. Regardless this possible diagenetic effect in tali of Leptadapis and Adapis (which would introduce a bias on the Tb.Th, and by extension on the measure of the bone volume fraction), the Tb.N in their tali remains, however, in the range of Tb.N values recorded in tali of certain living primates. The same is true for the values of SVD-DA and SVD-E, which are for the three fossils, in the range of the fabric anisotropy and elongation values recorded in tali of living primates. For the three taxa, there is no strong mediolateral difference in the bone volume fraction. Only the tali of Leptadapis record a slightly higher bone density in their lateral side. In this taxon, the Tb.N approximates that observed in Propithecus, Hapalemur, or Daubentonia (Tables 3 and 6). For the tali of Adapis and Necrolemur, the Tb.N is moderately high, in the range of that observed in Lepilemur, Loris, and Tarsius (Tables 3 and 6). However, the talar bone structure of Necrolemur compared to that of Adapis, as well as of Leptadapis, differs substantially in the pattern of the fabric anisotropy. The trabecular structure in tali of Necrolemur is clearly more oriented (especially in the medial side). The degree of anisotropy and elongation in Necrolemur approximate those observed in tali of Propithecus, Cebus, Cebuella, Saimiri, or Leontopithecus, which are primates engaged in a kind of arboreal quadrupedalism with some good abilities to leap. In contrast, in tali of Adapis and Leptadapis, the trabecular structure appears more randomly oriented as it is shown by the rather low values of SVD-DA-E, which are in the range of some primates who are arboreal quadruped, without leaping or frequent leaping activities in their locomotor repertoire (e.g., Loris, Perodicticus, and Nycticebus, but also Microcebus, Daubentonia,). In sum, it is clear that the trabecular bone structure of the tali of Necrolemur, with a moderate degree of anisotropy and elongation indicates that the ankle bone in this extinct taxon was subjected to moderately strong and directed loading conditions at the ankle joints, thereby demonstrating a leaping activity. This corroborates the functional conclusions deriving from the external morphology of the talus of this Eocene primate. However, although the external talar morphology in this fossil taxon is quite similar to that found in Tarsius and Galago, the degree of anisotropy indicates that Necrolemur did not exhibit extreme specialization for leaping, as it is the case for Tarsius and Galago. The values of SVD-DA-E recorded in tali of Necrolemur are elevated but clearly lower than those recorded in Tarsius and Galago. According to the whole trabecular bone parameters measured, compared to those in living taxa, it seems that Necrolemur was a small arboreal quadruped primate, which was capable of leaping but also climbing, although not particularly specialized for either of these activities. This locomotor behavior hypothesis is confirmed by the discriminant analysis (Fig. 9B), which primarily allocates Necrolemur to the vertical clingers and leapers (A1: P = 0.4; A2: P = 0.32), but also to the small arboreal agile quadrupeds (B: P = 0.28), albeit to a lesser extent. In contrast, the trabecular bone parameters measured in tali of Leptadapis and Adapis indicate that leaping was probably not an important component of their locomotor repertoire. The less anisotropic and dense (although possibly artificial) trabecular structure recorded within their talar body indicates that the bone was rather adapted for less strong directional loads, thereby suggesting more emphasis on quadrupedalism and climbing. However, in the tali of Adapis and Leptadapis, we have not recorded a strong mediolateral difference in the bone volume fraction, as it is the case in most primates who are frequent climbers. For living primates, we have hypothesized that the mediolateral variation in the bone volume fraction might be linked to the forces generated by sustained inverted foot postures, which seems to load the medial aspect of the talar body more strongly. The external morphology of the tali of these two fossil primates indicates that they were able to practice foot inversion (Dagosto, 1983). However, in Adapis and Leptadapis, the loading pattern at the ankle joints associated with the inverted foot position for climbing might have been more balanced mediolaterally, thereby involving a more uniform distribution of trabeculae (weakly oriented) in both sides of their talar body. Nycticebus, which is a nonleaping, slow-moving arboreal quadruped, exhibits a similar case of mediolateral uniform pattern of bone volume fraction, but it slightly differs in the pattern of the fabric anisotropy. The discriminant analysis (Fig. 9B) revealed strong affinities with the medium-sized lemuriform branch-running and walking type (E1: P = 1) for Adapis and Leptadapis. This would indicate that A. parisiensis and L. magnus exhibited some degree of proficiency for leaping (springing) as it does in the living species of this locomotor subcategory (e.g., Lemur, Eulemur, and Varecia, but also Daubentonia; Table 1). Both fossil taxa, especially Leptadapis, have been generally regarded as slow-moving arboreal-quadruped primate (climber with labored locomotion), which did not included very important leaping activities (less efficient abilities) in their locomotor repertoire (Dagosto, 1983). Regarding the locomotor repertoire of Adapis, the functional interpretation of the postcranial remains referred to this genus has proven to be more mixed in terms of postural activities and positional behaviors, which included branch walking and running forms, using climbing less frequently, and other forms less specialized for running that used climbing more frequently. For Adapis, we have therefore sampled here primarily the tali of the individuals, which were rather belonging to the branch-running and walking forms, as those evidenced by Bacon and Godinot (1998) from the femoral anatomy. For Leptadapis, the LDA results are somewhat surprising, given the rather plate-like but rather low anisotropy observed on tali of this taxon (very low value of SVD-E in the lateral side of the talar body). Leptadapis was certainly an arboreal quadruped who was able to run and walk on branches with some abilities for climbing (without extreme specialization), but most likely with less-efficient leaping abilities than the living primates classified as branch-running and walking type (E1). However, for the fossil tali of Leptadapis, we must keep in mind that a diagenesis effect might have biased the measure of the three-dimensional architecture of the trabecular bone.
The diversity of activities and positional behaviors characterizing primate locomotion includes various ways of running, walking, leaping, and climbing on different substrates, which produce distinct loading regimes. Our analyses of the trabecular bone architecture in the ankle bone of small to medium-sized living primates indicate that the variations in number, thickness, and three-dimensional distribution and arrangement of the trabeculae in this tarsal bone across taxa are seemingly linked to the loading environment associated with the various postures and motions of the foot. There are large intra- and interspecific variations in the trabecular architecture of the talar body, which reflect differences in the magnitude and orientation of loads at the ankle joints during locomotion both across individuals and across taxonomic groups. From our set of selected taxa (haplorhines and strepsirhines; 60–5000 g), it seems that body mass has no effect on the bone volume fraction as well as on fabric anisotropy. However, we have shown that both the number and robustness (thickness) of trabeculae seem to be body mass-dependent. The medium-sized species show a higher degree of trabecular thickness relatively to their Tb.N, while the smallest species show thinner trabeculae relative to trabecular number.
General trends can be addressed regarding the relationships between the variations in the bone structure and the diverse loading patterns of foot postures involved during locomotion. Small to medium-bodied primates engaged in a kind of arboreal quadrupedalism can show proficiency for climbing and leaping, although not particularly specialized for either of these activities. Given this range of activities involving diverse postures and motions of the foot, it is clear that the loading environment at the ankle joints may not appear strictly unidirectional, thereby generating a complex three-dimensional arrangement of the trabecular bone. We have observed a mediolateral difference in the bone volume fraction in most primates who are frequent, proficient, or specialized climbers. We have hypothesized that this local (medial) increase in bone volume density in tali of these climbers could reflect a specific reinforcement of the trabecular structure in response to the apparent strong magnitude of loads engendered in habitually sustained foot inversion. In contrast, tali of primates who are frequent, proficient, or specialized leapers rather exhibit a more highly oriented trabecular structure (more anisotropic), which rather reflect stronger unidirectional and stereotypical loading conditions generated at the ankle joints during a leap. The different degrees of interspecific and within-species mediolateral variations in the fabric anisotropy and bone volume density demonstrate that there are biomechanically differences in function resulting from dissimilar locomotor behavior between taxa or individuals and that the trabecular bone adapts to local load condition. Finally, as it provides some information regarding the loading environment pattern at the ankle joints, the talar trabecular structure has a good potential for predicting some aspects of the locomotion in extinct species. The talus is only a small part of the skeleton, but it occupies a pivotal position, as it is responsible for the foot movements during locomotion. These investigations on the trabecular structure must be therefore performed on several other joints as a means to better understand the global functional signal of the bone adaptations to the specific loading environment engendered by different kind of postural activities and positional behaviors characterizing locomotion. These relationships established for extant species might be particularly useful for paleontologists to infer the mechanical environment of the bones of extinct taxa. However, diagenesis remains a critical factor for fossil bones, because it can involve biases in the measures of the trabecular morphometric parameters, notably the bone volume fraction and the Tb.Th.
The authors thank Christoph PE Zollikofer and Marcia S Ponce de Léon (Anthropological Institute and Museum in Zurich), Christiane Denys and Jacques Cuisin (Muséum National d'Histoire Naturelle [MNHN] in Paris), and Suzanne Jiquel (ISE-M, Collections Université Montpellier 2 [UMC]) for access to their collections and permission to scan extant specimens. They acknowledge Bernard Marandat (ISE-M, UMC) and Marc Godinot (MNHN, Paris) for access to the fossil specimens of the Quercy (France). We are grateful to Virginie Volpato (Senckenberg Research Institute and Natural History Museum, Frankfurt) who offered valuable comments and advices on the trabecular architecture. This is ISE-M publication no. 2012-150.