Cortical and trabecular bone structure of the hominoid capitate

Abstract Morphological variation in the hominoid capitate has been linked to differences in habitual locomotor activity due to its importance in movement and load transfer at the midcarpal joint proximally and carpometacarpal joints distally. Although the shape of bones and their articulations are linked to joint mobility, the internal structure of bones has been shown experimentally to reflect, at least in part, the loading direction and magnitude experienced by the bone. To date, it is uncertain whether locomotor differences among hominoids are reflected in the bone microarchitecture of the capitate. Here, we apply a whole‐bone methodology to quantify the cortical and trabecular architecture (separately and combined) of the capitate across bipedal (modern Homo sapiens), knuckle‐walking (Pan paniscus, Pan troglodytes, Gorilla sp.), and suspensory (Pongo sp.) hominoids (n = 69). It is hypothesized that variation in bone microarchitecture will differentiate these locomotor groups, reflecting differences in habitual postures and presumed loading force and direction. Additionally, it is hypothesized that trabecular and cortical architecture in the proximal and distal regions, as a result of being part of mechanically divergent joints proximally and distally, will differ across these portions of the capitate. Results indicate that the capitate of knuckle‐walking and suspensory hominoids is differentiated from bipedal Homo primarily by significantly thicker distal cortical bone. Knuckle‐walking taxa are further differentiated from suspensory and bipedal taxa by more isotropic trabeculae in the proximal capitate. An allometric analysis indicates that size is not a significant determinate of bone variation across hominoids, although sexual dimorphism may influence some parameters within Gorilla. Results suggest that internal trabecular and cortical bone is subjected to different forces and functional adaptation responses across the capitate (and possibly other short bones). Additionally, while separating trabecular and cortical bone is normal protocol of current whole‐bone methodologies, this study shows that when applied to carpals, removing or studying the cortical bone separately potentially obfuscates functionally relevant signals in bone structure.

the internal structure of bones has been shown experimentally to reflect, at least in part, the loading direction and magnitude experienced by the bone. To date, it is uncertain whether locomotor differences among hominoids are reflected in the bone microarchitecture of the capitate. Here, we apply a whole-bone methodology to quantify the cortical and trabecular architecture (separately and combined) of the capitate across bipedal (modern Homo sapiens), knuckle-walking (Pan paniscus, Pan troglodytes, Gorilla sp.), and suspensory (Pongo sp.) hominoids (n = 69). It is hypothesized that variation in bone microarchitecture will differentiate these locomotor groups, reflecting differences in habitual postures and presumed loading force and direction. Additionally, it is hypothesized that trabecular and cortical architecture in the proximal and distal regions, as a result of being part of mechanically divergent joints proximally and distally, will differ across these portions of the capitate. Results indicate that the capitate of knuckle-walking and suspensory hominoids is differentiated from bipedal Homo primarily by significantly thicker distal cortical bone. Knuckle-walking taxa are further differentiated from suspensory and bipedal taxa by more isotropic trabeculae in the proximal capitate. An allometric analysis indicates that size is not a significant determinate of bone variation across hominoids, although sexual dimorphism may influence some parameters within Gorilla. Results suggest that internal trabecular and cortical bone is subjected to different forces and functional adaptation responses across the capitate (and possibly other short bones).
Additionally, while separating trabecular and cortical bone is normal protocol of current whole-bone methodologies, this study shows that when applied to carpals, removing or studying the cortical bone separately potentially obfuscates functionally relevant signals in bone structure.

K E Y W O R D S
cancellous bone, functional morphology, locomotion, primates, wrist

| INTRODUC TI ON
Primates use their hands in a diverse set of postures to manipulate and navigate their environment (Fragaszy & Crast, 2016). The many articulations within the wrist are central to the capacity of the hand to move through multiple planes of space and, in combination with soft tissue morphology, joint congruence determines the degree of stability, flexibility and dexterity within the wrist and hand (Orr, 2010). The capitate articulates proximally with the scaphoid and lunate and distally with the trapezoid, hamate, and metacarpals 2, 3, and, sometimes, 4 (Kivell, 2016a). As such, the external morphology of the capitate plays a key role in the range of motion at the wrist as it is a central component of the midcarpal joint proximally and the carpometacarpal joints distally (Crisco et al., 2005;Jenkins & Fleagle, 1975;Jouffroy & Medina, 2002;Kijima & Viegas, 2009;Lewis, 1989;Orr, 2017;Orr et al., 2010).
The external morphology of the hominoid capitate has featured in hypotheses about the locomotor behavior in the last common ancestor of Pan and Hominini (Begun, 2004;Dainton & Macho, 1999;Kivell & Schmitt, 2009;Richmond et al., 2001;Tocheri et al., 2007) and the evolution of hominin dexterity and tool-related behaviors (Marzke, 1983;Niewoehner et al., 1997;Rein & Harvati, 2013;Wolfe et al., 2006). However, drawing locomotor or postural inferences based on external morphology is potentially confounded by the retention of primitive features that are no longer functionally significant (Kivell, 2016b;Lieberman, 1997;Pontzer et al., 2006;Ruff et al., 2006;Ward, 2002;Zeininger et al., 2011). Furthermore, making biomechanical inferences from external morphology requires in-depth knowledge of the formfunction relationship of the bone as well as its surrounding soft tissue and articular environment. This is a particular challenge for carpal research as understanding the kinematics and kinetics of the wrist is inherently difficult due to complications in imaging and analyzing the small, closely compacted bones without disrupting the normal kinematic chain (Brainerd et al., 2010;Gatesy et al., 2010;Orr, 2016;Wolfe et al., 2000Wolfe et al., , 2006. Although advances in 3D imaging and computational techniques have begun to improve our knowledge of human wrist biomechanics (see Orr, 2016 for review), our understanding of nonhuman ape biomechanics remains more limited (but see Orr, 2017Orr, , 2018. Moreover, the functional relationship between cortical and trabecular tissue within short bones is not well understood, and it is not clear how they may interact to provide whole bone functionality under the high mechanical loads of locomotion and manipulation. To date, it has yet to be determined whether the internal bone structure of the capitate might reflect differences in hand and wrist use across extant hominoids. Here we apply a whole-bone methodology to investigate how the internal cortical and trabecular bone structure potentially varies within the capitate in a broad sample of Homo (recent humans), Pan (chimpanzees and bonobos), Gorilla, and Pongo (orangutans).

| Trabecular bone: The relationship between behavior and structure
In addition to some important clade specific synapomorphies (Lewis, 1989;Sarmiento, 1988;Tocheri et al., 2008), the wrists of extant hominoids are adapted to their specialized behaviors and are habitually loaded in different ways. The Homo hand is conspicuous among the ape clade as the only species not to habitually utilize the hand for locomotion. Instead, the wrist experiences forces generated predominantly during tool use and other forms of manipulation. High compressive loads are imposed across the wrist by muscle contractions arising from the strong and forceful human thumb as well as flexion of the fingers (Bardo et al., 2018;Marzke, 1997;Napier, 1956;Tocheri, 2007). Bones must also withstand and transmit sheer and tensional strains as force is transferred radio-ulnarly across the wrist (Marzke, 2013;Tocheri, 2007). There is an abundance of clinical evidence to support the hypothesis that the Dart Throwers Motion (DTM) is the functional axis of human wrist movement (Brigstocke et al., 2014;Crisco et al., 2005Crisco et al., , 2011Schuind et al., 1994). The DTM runs from radial deviation in extension to ulnar deviation in flexion and is used across numerous activities from throwing an object to pouring water from a jug (Brigstocke et al., 2014). During this movement, the capitate is very mobile against a stabilized proximal row, with the rotation axis perpendicular to the wrist movement (Crisco et al., 2005).
In contrast, nonhuman apes utilize their forelimbs during locomotion. Pongo utilize a range of torso orthograde suspensory and climbing postures in an almost exclusively arboreal environment (Manduell et al., 2011;Thorpe & Crompton, 2006. In these positions, the wrist experiences substantial tensile loading from gravitational forces and stabilising ligaments, as well as compressive stress from muscle contractions (Isler & Thorpe, 2004;Swartz et al., 1989). Gorilla and Pan are primarily terrestrial knuckle-walkers but also engage in various types and frequencies of arboreal locomotion depending on the species, population or sex (Doran, 1993;Hunt, 1992;van Lawick-Goodall, 1968;Neufuss et al., 2017;Remis, 1995Remis, , 1998Thompson et al., 2018). During knuckle-walking, the wrist must resist compressive loading from both muscle contractions stabilizing the joints and gravitation forces acting on the body mass (Carlson & Patel, 2006). However, the knuckle-walking posture differs somewhat between the two genera. When compared to Gorilla, Pan typically use more variable hand and forelimb postures, do not bear weight as evenly across the digits, and more frequently engage a palm-in forelimb posture (Finestone et al., 2018;Inouye, 1994;Matarazzo, 2013;Wunderlich & Jungers, 2009). Gorilla typically knuckle-walk on digits 2-5 and more regularly utilize a palm-backwards forelimb posture (Inouye, 1994;Matarazzo, 2013;Tuttle, 1969), although hand postures in the wild are more variable (Thompson et al., 2018). Although Gorilla are hypothesised to use a more neutral, columnar wrist posture than Pan (Kivell & Schmitt, 2009), recent kinematic studies of captive African apes found that Gorilla and Pan were generally similar in their degree of wrist of extension during knuckle-walking (Finestone et al., 2018;Thompson, 2020).
Bone functional adaptation describes the biological process of bone altering its structure to optimize resistance against peak mechanical loads habitually experienced throughout the lifetime of the individual (Barak et al., 2011;Doube et al., 2011;Martin et al., 1998;Ruff et al., 2006). Numerous experimental studies suggest that variation in structure reflects, at least in part, load experienced during life (see Kivell, 2016b for review) and thus provides an opportunity to draw behavioral inferences better linked to actual, rather than potential, behavior (Frost, 1987;Ruff & Runestad, 1992). Bone functional adaptation research cannot only facilitate a greater understanding of the joint loading and kinematics of extant species but may also provide an informative avenue for behavioral reconstruction in fossil taxa (DeSilva & Devlin, 2012;Dunmore et al., 2020;Georgiou et al., 2020;Griffin et al., 2010;Kivell et al., 2018;Skinner et al., 2015;Su & Carlson, 2017). Previous studies of primate trabecular bone structure within the capitate have used a volume of interest (VOI) sampling sphere but have found limited functional correlation with locomotor behavior (Ragni, 2020;Schilling et al., 2014).
However, inferring a form-function relationship between bone microarchitecture and behavior is not always straightforward due to several potentially confounding variables (for a comprehensive review and discussion see Kivell, 2016b). Firstly, bone modelling (sensu Barak, 2019) is influenced by the genetic blueprint of the individual, as well as life history factors such as lactation or pregnancy (Kalkwarf & Specker, 1995;Lieberman, 1996;Lovejoy et al., 2003;Parsons et al., 1997;Paternoster et al., 2013;Pettersson et al., 2010;Tsegai et al., 2017;Yeni et al., 2011).
Systemic features such as these potentially undermine our ability to differentiate between functional and nonfunctional patterns expressed in bone structure across different individuals or species. Secondly, there is a higher capacity for functional adaptation to occur during the juvenile and young adult periods and the extent to which bone microarchitectural patterns can be linked to adult behavior has been debated (Bertram & Schwartz, 1991;Pearson & Lieberman, 2004;Ruff et al., 2006). This is particularly salient when analyzing African apes because locomotor behavior is known to differ across age categories (Doran, 1992(Doran, , 1997. Finally, there is also uncertainty regarding the loading frequency and magnitude necessary to induce a functional adaptation response (Barak et al., 2011;Frost, 1987;Ruff et al., 2006;Wallace et al., 2015). Consequently, microarchitecture will never represent the mechanical ideal of the bone as competing demands on bone tissue will result in a compromise morphology (Ruff et al., 2006).

| Cortical bone: Contribution to bone structure and functional adaptation
Carpal bones function within an intricate biomechanical environment.
The bones and ligaments are interdependent and work together making minor adjustments and movements in concert to create overall hand motion (Kijima & Viegas, 2009;Lewis, 1989;Orr, 2017). Among the carpus, the central role of the capitate within the midcarpal joint and its articulation with the metacarpus makes it an ideal bone to investigate functional differences in wrist loading. The distal capitate is not only compressed via its carpometacarpal articulations but it also receives tensional strain via the attachment of several extrinsic (between carpals and other hand bones) and intrinsic (between carpal bones) ligaments (Kijima & Viegas, 2009;Regal et al., 2020;Schuind et al., 1995). In contrast, the proximal capitate does not receive any ligaments but forms the "ball" component of the ball and socket midcarpal joint within the highly mobile proximal row and is thus loaded predominantly in compression Kivell, 2016a;Lewis, 1989;Orr, 2017).
Unlike long bones, short bones like carpals generally have a thin cortical shell and the entire internal space is filled with trabeculae (Currey, 2002;Schilling et al., 2014). During movement, short bones are likely to bear a significant portion of the load imposed upon the region as they resist against compressive forces and transfer load through the bone from one joint articulation to another, while also being strained via tensional loads from attached ligaments (Currey, 2002;Yao et al., 2020). Cortical and trabecular bone have divergent material properties due to differences in porosity, mineralization and cellular constitution (Currey, 2002). Cortical bone is stiffer and stronger than trabecular bone (Guo, 2001;Martin et al., 1998), but due to its lower porosity, it is slower than trabecular bone to model and is less compliant (Hart et al., 2017;Martin et al., 1998). While the two tissues work together to create the functionality of the whole bone, their relative contributions to stiffness, strength and homeostasis differs across regions of the same bone (Barak et al., 2010;Doube et al., 2009). It is not currently understood how cortical and trabecular bone work together to meet the mechanical demands of the carpus, particularly under the high mechanical demands of locomotion.
By quantifying the internal bone architecture of the hominoid capitate using a whole-bone methodology, this study aims to investigate whether differences in trabecular and cortical architecture among hominoids may relate to the divergent hand use across the clade. We also examine the proximal and distal segments of the capitate separately, due to the differences in the soft tissue and articular relationships with the surrounding bones.

| Allometry: Body size and bone structure
As functional adaptation research aims to identify markers of behavior rather than body size, analyzing bone parameters for allometric effects has been integral to interspecific analyses (Ruff, 1984). Decades of research across various species has yet to find consistent patterns; however, some research suggests there may be a general pattern across mammals and birds whereby bone volume to total volume (BV/TV) and degree of anisotropy (DA) are independent of body mass (Barak et al., 2013;Christen et al., 2015;Cotter et al., 2009;Doube et al., 2011;Kivell et al., 2018;Komza & Skinner, 2019;Schilling et al., 2014;Tsegai et al., 2017) while trabecular thickness (Tb.Th), trabecular number (Tb.N) and trabecular separation (Tb.Sp) scale with negative allometry (Barak et al., 2013;Kivell et al., 2018;Ragni, 2020;Ryan & Shaw, 2013). Cortical thickness (Ct.Th) is often reported to be isometric or slightly positively allometric (Demes et al., 2000;Fajardo et al., 2013;Runestad, 1997). However, not all studies find BV/TV and DA to be independent of body mass (for example Fajardo et al., 2013;Mielke et al., 2018;Ragni, 2020;Ryan & Shaw, 2013)  These conflicting results may be due in part to methodological differences for sampling trabeculae or calculating size. Nevertheless, the effects of allometry on the hominoid capitate remain unclear.

| Distal capitate
In all hominoids, the distal capitate is bound to the surrounding bones via strong ligaments which are often described as a unit that moves in unison during extension and flexion (Crisco et al., 2005;Moojen et al., 2003;Orr, 2010;Richmond, 2006;Richmond et al., 2001;Tang et al., 2011). The capitate articulates distoradially with the trapezoid (although this articulation can be absent in Gorilla) and second metacarpal (Mc2), and distally with the third and sometimes fourth metacarpals (Kivell, 2016a;Lewis, 1989). The topology of the metacarpal joint surfaces in the distal row is more complex and irregular in Pan and Gorilla compared to the smoother surfaces in Pongo; however, the capacity for extension is linked to the range of movement at the midcarpal joint rather than at the carpometacarpal junction (Begun, 2004;Orr, 2017;Richmond et al., 2001). The distal capitate in modern Homo sapiens is considered to have several derived features linked to committed manipulation and increased efficiency of radio-ulnar force transfer (Tocheri, 2007;Tocheri et al., 2008). A distally oriented capitate-Mc2 articulation allows pronation of the second finger towards the thumb facilitating precision grip, while a palmarly positioned and expanded capitate-trapezoid articulation is thought to better resist high radio-ulnarly oriented forces incurred by the thumb during toolrelated activities (Marzke, 1997;Tocheri, 2007;Tocheri et al., 2008). Furthermore, the disto-dorso-radial corner is truncated to accommodate the third metacarpal (Mc3) styloid process, providing a stable joint for transmitting high forces and resisting subluxation of the third ray during tool use (Marzke & Marzke, 1987;Niewoehner et al., 1997;Riley & Trinkaus, 1989;Tocheri et al., 2008;Ward et al., 2014). In nonhuman apes, load transfer also occurs radio-ulnarly as bones of the distal carpal row are compressed against one another. However, in contrast to humans, the orientations of the articular surfaces of the capitate (and distal carpal row more generally) indicate the wrist is better adapted to withstand and transfer proximo-distally oriented forces, which arise during use of the forelimb in locomotion (Tocheri, 2007;Tocheri et al., 2008). Only a small proportion of the dorsal surface of the distal capitate is without articular surfaces. In this distal segment, compression is induced at the distal, radial and ulna articular surfaces, while tension is induced by the supporting intrinsic ligaments surrounding these articulations. Tension further arises from the several intrinsic and extrinsic ligaments attached to the palmar and dorsal surfaces (Kijima & Viegas, 2009;Regal et al., 2020).

| Proximal capitate
In great apes, the rounded proximal surface of the capitate articulates with the bones of the proximal row to form the crux of the midcarpal joint (Kivell, 2016b). No ligaments attach directly onto the proximal capitate thus compared to the distal row, the bones of the midcarpal joint are able to move more independently of one another (Crisco et al., 2005;Kijima & Viegas, 2009;Moojen et al., 2003;Regal et al., 2020). In Pongo, the proximal capitate is radio-ulnarly narrow in comparison to the other great apes (Figure 1; Richmond et al., 2001). Notably, the os centrale is not fused to the scaphoid as it is in the other hominids and thus excludes the scaphoid from articulating with the capitate resulting in relatively greater freedom of movement at the midcarpal joint (Begun, 2004;Orr, 2018). In Pan and Gorilla, the proximal capitate is enlarged on the radial aspect, which produces a "waisted" mid-region forming an embrasure with the trapezoid (Kivell, 2016a;Orr, 2018;Richmond et al., 2001;Wolfe et al., 2006). There is also a notable radio-ulnar ridge along the distal extent of the dorsal articular surface that extends onto the hamate (Richmond et al., 2001). These features contribute to the so called "screw-clamp mechanism" that describes the functional complex limiting extension at the midcarpal joint. During extension, the scaphoid is wedged in between the capitate and trapezoid, providing stability between the proximal and distal row (Jenkins & Fleagle, 1975;Orr, 2005Orr, , 2017Richmond, 2006;Richmond et al., 2001;Tuttle, 1969).
Homo also exhibits the fused scaphoid-os centrale and radially expanded proximal capitate; however, an enlargement of the bone in the radial-palmar region results in a less dramatic "waisting" of the bone, resulting in a range of extension intermediate between the other hominoids (Lewis, 1977;Lewis, 1989;Orr, 2017). Notably, the proximal capitate is the crux of the functional axis of the DTM (Crisco et al., 2005). During motion, the rotation axis of the capitate is perpendicular to the movement of the DTM as it moves across a virtually motionless scaphoid and lunate (Crisco et al., 2005). Thus, although a small bone, the proximal and distal portion of the capitate functions within notably different ligamentous and articular environments.

| Hypotheses
This research centers on three interrelated hypotheses for both trabecular and cortical bone that are summarized in Table 1 and elaborated below.

| Trabecular bone architecture
We predict that the capitate of knuckle-walking Gorilla and Pan will have high relative BV/TV and high DA (Table 1, Hypothesis 1) due to the presumed high compressive forces and reduced mobility from their more extension-limiting midcarpal joint. In contrast, we predict that the Pongo capitate will have intermediate BV/TV and low DA due to their predominantly suspensory behavior, resulting in reduced compression but greater mobility. We expect Homo to exhibit low BV/TV and intermediate DA because their capitate is not loaded during locomotion and presumably has the least compressive loading but more predictable movement along the DTM axis.
Given the differences in mobility and presumed loading between the proximal and distal portions of the capitate, we hypothesize that there will be differences in the trabecular bone structure between these segments (measured as ratios). It is predicted that the distal aspect will have higher BV/TV and DA compared to the proximal aspect across all species (Table 1, Hypothesis 2). As there are no previous studies that have addressed this question for the capitate, we test the null hypothesis that these ratios will be similar among the study taxa. Although we report Tb.Th, Tb.N and Tb.Sp, we do not make explicit predictions about these parameters because all contribute, potentially in a variety of different combinations, to BV/TV.

| Cortical bone thickness
The contribution of cortical bone to the functional adaptation of the capitate in hominoids has never been investigated. Given the assumed loading differences described above, we predict that the cortical bone will be thickest in Gorilla and Pan, followed by Pongo, with Homo exhibiting the thinnest cortex (Table 1, Hypothesis 1). Also following the predictions for trabecular bone, it is predicted that the cortex of the distal capitate should be significantly thicker than the proximal capitate for all genera.
In long bones, the joint surface tends to have a thin layer of cortical bone covering a dense trabecular network that transfers load towards the thicker and stronger diaphyseal cortex (Currey, 2002). In short bones, the cortex is similarly described as thin; however, the relationship between cortical and trabecular bone has never been quantified among hominoids. Additionally, it is unclear whether the behavioral differences among ape genera will result in different ratios of cortical to trabecular bone. Therefore, this study will investigate the relative contribution of cortical bone to total bone volume, testing the null hypothesis that these ratios will be similar among the study taxa (Table 1, Hypothesis 2).

| Interspecific and intraspecific allometry in internal bone structure
As this study incorporates hominoids of diverse body size, interspecific and intraspecific allometry is also investigated. Predictions are outlined in Table 1
Capitates from nonhuman apes were wild-shot adults with no obvious TA B L E 1 Summary of the hypotheses, predictions, and statistical tests used in this study  (Table S1).
This resolution is below the suggested range for minimal error detection (Christen et al., 2016;Isaksson et al., 2011). Post-scanning, each capitate was positioned into approximately the same orientation using Avizo 6.0

| Data collection
This study uses the medtool 4.3 software package (http://www. dr-pahr.at/medto ol/) to quantify bone parameters throughout the entire capitate utilizing the method outlined in Gross et al. (2014). In brief, medtool utilizes a series of morphological filters to identify the cortical, trabecular, internal (marrow), and background elements of the segmented CT scans. After MIA segmentation, medtool projects a series of rays onto outside of the bone (Figure 2b) that continue to move internally through the bone until a nonbone voxel is reached (Pahr & Zysset, 2009a). By using a value of average trabeculae thickness, morphological filters fill and close small holes present in the porous cortex allowing a smooth boundary contour between cortical and trabecular bone to be identified (Gross et al., 2014;Pahr & Zysset, 2009a, 2009b. Two Gorilla, one Pan and two Pongo specimens were excluded from the sample as the internal cortical-trabecular boundary could not be confidently defined due to extreme cortical porosity (an example is provided in Figure S1). Medtool then super- 2.5-mm spaced nodes is then superimposed on an image stack and a 5-mm sampling sphere moves from node to node to measure parameters across the entire bone ( Figure 2f) (Pahr & Zysset, 2009a).
BV/TV is calculated as the ratio of bone to non-bone voxels. DA is calculated via the mean intercept length (MIL) method (Whitehouse, 1974) and is calculated as 1-(min. eigenvalue/max. eigenvalue) which produces a number limited between 1 and 0, with 1 being complete anisotropy and 0 being complete isotropy. Tb.Th, Ct.Th, and Tb.Sp are computed in a similar way to the more well-known BoneJ ® plugin (Doube et al., 2010) for ImageJ. Spheres are grown within the trabecular or cortical bone and medtool calculates the diameter of the largest sphere that fits within the bone (Hildebrand & Rüegsegger, 1997   A Kruskal-Wallis one-way ANOVA and pairwise Wilcoxon rank-sum test examined interspecific differences in the ratios.

| Cortical bone hypotheses
To test for differences in cortical bone, mean differences in total BV/ TV and Ct.Th were compared interspecifically in the proximal and distal segments using a Kruskal-Wallis one-way ANOVA and pairwise Wilcoxon rank-sum tests using the Holm p adjust method (R Core Team, stats package v3.6.1).
Within each genus, a distal to proximal ratio was calculated for each parameter and a Wilcoxon signed-rank test was applied to test whether mean values of the ratio were statistically significant. Additionally, we examined taxonomic differences in these ratios using a Kruskal-Wallis one-way ANOVA and pairwise Wilcoxon rank-sum tests.
Two additional ratios were calculated in order to test for taxonomic differences in the relative proportion of cortical and trabecular bone. These ratios were compared between species, using a Kruskal-Wallis one-way ANOVA and pairwise Wilcoxon rank-sum tests using the Holm p adjust method (R Core Team, stats package v3.6.1).

| Interspecific and intraspecific allometry
To test for allometric trends in the capitate, each whole-bone cortical and trabecular parameter was interspecifically and intraspecifically analyzed in a reduced major axis regression (RMA). As a proxy for body mass, the volume (

| Trabecular bone
Cross-sections of each genera in Figure 3 provide an example of the internal structure of the capitate within three planes of view. The red dotted line in Figure 3d indicates where the capitate was partitioned into the proximal and distal VOIs.

| Bone volume to total volume
Proximal and distal trabecular BV/TV differ significantly across the study sample (p ≤ 0.001 for both tests, Table S3). Gorilla has the highest proximal and distal BV/TV followed by Pan, then Pongo, with Homo having the lowest BV/TV values (Table S2)
Notably, DA is the only trabecular parameter which has a different significance result for the proximal and distal VOI. Homo and Pongo have the highest proximal DA with 0.30 followed by Pan and Gorilla, both with 0.24 (Figure 4b, Table S2). Distal DA differs by only 0.02 between the genera, with the highest value from Gorilla at 0.28 and lowest from Pan at 0.26 (Table S2). Pairwise comparisons reveal that proximally, Homo and Pongo are differentiated from Pan and Gorilla (p ≤ 0.001 for all four significant tests). Distally, there are no significant pairwise results (Figure 4b, Table S3).

Both Gorilla and Pan have a higher DA in the distal VOI whereas
Homo and Pongo both have higher DA in the proximal and the difference between the proximal and distal VOIs is significant for all genera (Figure 5b, Table S4). The DA ratio differs significantly across the genera (p ≤ 0.001) and pairwise comparisons reveal that Homo and Pongo are differentiated from Pan and Gorilla (p ≤ 0.001 for all four significant tests, Table S4).

| Trabecular thickness
Tb.Th differs significantly across both the proximal and distal capitate of the study sample (p = <0.001 for both tests, Table S3). Gorilla has the highest mean thickness followed by Pongo, with Homo having the thinnest (Table S2). Distally, all pairwise comparisons are significant except between Homo and Pan. Proximally, Gorilla is differentiated from all other taxa (Figure 4c, Table S3).
Homo, Pan, and Gorilla have thicker trabeculae in the proximal aspect and Pongo in the distal aspect (Tables S4 and S5). The difference between the two segments is statistically significant for Pan, Gorilla, and Pongo but not for Homo (Figure 5c, Table S4). The Tb.Th ratio differs significantly among the study sample (p ≤ 0.001) and all pairwise comparisons are significant except between Homo and Gorilla (Table S4).

| Trabecular number
Proximal and distal Tb.N differs significantly among the study sample (p ≤ 0.001 for both tests, Table S3). Gorilla has the lowest trabecular number while Pan has the highest number (Table S2)

| Trabecular separation
Tb.Sp differs significantly in the distal (p ≤ 0.001) and proximal (p = 0.038, Table S3) capitate of the study sample. Gorilla has the most widely spaced trabeculae, while Pan has the most tightly packed (Table S2). Pairwise comparisons indicate that distally, Pan is differentiated from all other taxa (Table S3). Proximally, the only significant pairwise result is between Pan and Gorilla (Figure 4d).
The Tb.Sp ratio shows that Homo, Gorilla, and Pongo have greater trabecular separation in the distal capitate whereas Pan has greater separation in the proximal (Figure 5d, Table S5). The difference between the separation in the distal and proximal capitate is significant for all genera (Table S4). The Tb.Sp ratio differs significantly among the study sample (p ≤ 0.001) and the results of the pairwise comparisons mirror those of the distal segment as the only significant tests are between Pan and the other taxa (p ≤ 0.001 for the three significant tests, Table S4).

| Total relative bone volume
Total BV/TV, which incorporates both trabecular and cortical bone, differs significantly across the study sample for both the proximal and distal capitate (p ≤ 0.001 for both tests,  (Figure 6c, Table S4). The Wilcoxon signed-rank tests indicate that the differences in the total BV/TV between the two segments is statistically significant for all genera. As outlined in the above section, trabecular BV/TV is consistently higher in the proximal segment compared to the distal segment in all genera (Figure 4a, Table   S2). However, when total BV/TV is measured, Pan, Gorilla, and Pongo show significantly higher values in the distal capitate (Figure 6a, Tables S2 and S4). In contrast, Homo maintains the trabecular BV/TV pattern, with higher total BV/TV in the proximal segment.
In the proximal capitate, the ratio of cortical bone to trabecular bone is similar among all genera, and pairwise comparisons reveal no significant results (Tables S4 and S5). In this segment, the inclusion F I G U R E 6 (a, b) Split violin plots showing the distribution of total BV/TV (a) and Ct.Th (b) results in the proximal and distal VOI of each genus. Images are generated using ggplot2 in R (v.  (Table S4). The relative portions of distal cortical and trabecular bone are similar among the nonhuman apes with cortical bone contributing 59% of total BV/TV in Pan and Pongo and 58% for Gorilla. In Homo, cortical bone represents 38% of distal total BV/TV.

| Cortical bone thickness
Ct.Th differs significantly among the study genera in both proximal and distal capitate (p ≤ 0.001 for both tests, Table S3). In both segments Gorilla has the thickest mean cortical bone, followed by Pan, Pongo, and finally Homo (Figure 6b, All genera have thicker cortical bone in the distal VOI and the difference between the proximal and distal segments is statistically significant in all genera (p ≤ 0.001 for all tests) (Figure 6d, Tables   S4 and S5). Pongo has the greatest relative cortical thickening in the distal VOI with the distal cortex being 79% thicker than the proximal, followed by Gorilla (62% thicker), Pan (52% thicker) and finally Homo (12% thicker). Pairwise comparisons of the ratio indicate that Homo is differentiated from all nonhuman apes (p ≤ 0.001 for all tests, Table S4). There are no significant pairwise comparisons between the nonhuman apes. The relative thickness of the cortex and trabeculae is visualized in Figure 7. In nonhuman apes, the thickest bone is consistently seen within the distal cortex. In Homo, the cortex and trabeculae have a similar thickness across the entire bone.

| Allometry
The results of the allometry tests are reported in Table 3

| DISCUSS ION
This study quantified the internal bone structure of the hominoid capitate using a whole-bone methodology to examine (1) whether relative and absolute differences in trabecular and cortical parameters across hominoid taxa could be correlated to inferred habitual behavior and (2) how the parameters differed interspecifically and intraspecifically across the proximal and distal portion of the capitate.

| Allometry in the capitate
Interspecifically, the predictions for Tb.N and DA were supported while all others were rejected. The two parameters most strongly correlated with size were Tb.Th and Tb.N. This was particularly true for Gorilla, which had relatively strong positive scaling for Tb.Th, Tb.N and Tb.Sp, with r-squared values between 0.60 and 0.69. This suggests these parameters may be linked to sexual dimorphism, which is extreme in Gorilla (Smith & Jungers, 1997). Indeed, the largest Tb.Th and Tb.Sp, and smallest Tb.N values among the Gorillas were from males. Pan was the only genus that did not report at least one significant intraspecific allometric test. This indicates that capitate size differences (as a proxy for body mass differences) between Pan troglodytes and Pan paniscus have not influenced results.
The positive relationship found in BV/TV does not corroborate results of either previous study on allometry in the primate capitate (Ragni, 2020;Schilling et al., 2014) or the talus (Tsegai et al., 2017). Differences in results between this study and others may be driven by the variation in the methodologies for calculating size or body mass. While this study used the cube root of the capitate, other studies have used the geometric mean (Schilling et al., 2014;Tsegai et al., 2017), body mass (Barak et al., 2013;Cotter et al., 2009), or linear dimensions of the bone (Ryan & Shaw, 2013). Furthermore, this study used a whole-bone mean of trabecular parameters whereas other studies have used a VOI sampling sphere (Cotter et al., 2009;Ragni, 2020;Ryan & Shaw, 2013;Schilling et al., 2014). Results are likely also affected by the species constituting the study sample or the bone used for analysis (Doube et al., 2009;Ruff, 1987;Ryan & Shaw, 2013;Tsegai et al., 2017). Nevertheless, as BV/TV is widely reported as being independent of body mass/size, results here may indicate carpals are more likely than other skeletal elements to increase BV/TV in response to size, across hominoids. However, given the similarity in capitate size between Homo, Pan and Pongo, the positive relationship found here is likely driven by the larger size of Gorilla, rather than reflecting a hominoid trend.
Ct.Th also scaled positively with size across hominoids and within Homo, Gorilla, and Pongo. Notably, the r 2 value for Pongo  (Fajardo et al., 2013), positive allometry with confidence intervals incorporating isometry in the femoral neck (Demes et al., 2000) or negative allometry in the radius and humerus (Doube et al., 2009).
BV/TV and Ct.Th are a primary component of bone strength and are thus critical to inferring function and functional adaptation from form (Maquer et al., 2015).  (Schilling et al., 2014) and other skeletal elements, including the talus (Desilva & Devlin, 2012;Tsegai et al., 2013), humerus  and femur (Georgiou et al., 2019 (Cotter et al., 2009;Hammond et al., 2018;Hart et al., 2017.) In Homo, DA in the proximal capitate may be explained by load predictability as the DTM constitutes the path of motion in a large proportion of daily activities (Brigstocke et al., 2014;Crisco et al., 2005;Kaufman-Cohen et al., 2019;Moritomo et al., 2014;Schuind et al., 1994). However, the relatively high DA in the Pongo proximal capitate was unexpected as it was assumed that the highly mobile joint and presumed variability in wrist postures adopted during ar-  (Georgiou et al., 2018;Kivell et al., 2018;Matarazzo, 2015;Tsegai et al., 2013) others have also found higher than expected values (Dunmore et al., 2019;Georgiou et al., 2019). Although arboreal locomotion is associated with mobile joints capable of receiving load from multiple directions, our knowledge of Pongo hand and wrist kinematics and kinetics is limited (but see Orr, 2010Orr, , 2017Orr, , 2018. The few studies of captive apes have provided invaluable data on the kinematics of vertical climbing (Isler, 2005;Isler & Thorpe, 2004) and quadrupedal walking (Finestone et al., 2018;Watson et al., 2011), but these behaviors constitute a small proportion of the Pongo locomotor repertoire (Cant, 1987;Thorpe & Crompton, 2006). Additionally, we currently lack manual pressure research on Pongo similar to that by Wunderlich and Jungers (2009)  apes to be more isotropic than predicted. While African apes are categorized as terrestrial knuckle-walkers, they also utilize arboreal substrates variably across their lifetimes to nest and exploit high quality food resources (Neufuss et al., 2017;Remis, 1995;Thorpe & Crompton, 2006 This study also investigated potential differences in ratios of bone parameters across the proximal and distal capitate, testing the null hypothesis that these ratios would be similar across hominoids.
This hypothesis was generally not supported as only two ratios were statistically similar across all genera: distal trabecular BV/TV relative to proximal trabecular BV/TV and proximal total BV/TV relative to proximal trabecular BV/TV. Thus, although proximal Ct.Th in Homo and Pongo was significantly thinner than that of Pan and Gorilla, the relative proportion of cortex to trabeculae is similar across all taxa.
Similarly, although eight of the 12 pairwise comparisons indicated statistically different trabecular BV/TV across the taxa (Figure 4a), the way trabecular volume differs between the two segments is consistent across hominoids. Although it was not predicted that ratio calculations would differentiate locomotor groups, three ratios distinguished Homo from the suspensory and knuckle-walking taxa: (1) distal total BV/TV relative to proximal total BV/TV, (2) distal total BV/TV relative to distal trabecular BV/TV, and (3)  This distinctive cortical morphology in nonhuman apes may reflect arboreal behaviors. All nonhuman apes engage in suspensory locomotion and climb vertical supports (Neufuss et al., 2017;Remis, 1995;Thorpe & Crompton, 2006), and in both behaviors the forelimbs are loaded in tension (Hanna et al., 2017;Hunt et al. 1996;Swartz et al., 1989). The distal capitate has numerous ligament attachments that induce tensional strain onto the capitate (Kijima & Viegas, 2009;Regal et al., 2020). Bones loaded in tension have a lower failure point than those loaded in compression (Caler & Carter, 1989;Pattin et al., 1996) and therefore greater BV/TV or Ct.Th would be required to prevent failure at ligament attachment sites (Doube et al., 2009

| The relationship between trabecular and cortical bone in the capitate
This study reveals the importance of considering both cortical and trabecular bone in functional adaptation research, rather than investigating each tissue separately. As exhibited in Figures 6 and 7, and discussed above, the cortical bone of the nonhuman ape capitate varied substantially from that of humans. Thus, the null hypothesis that the ratios of cortical to trabecular bone would be similar across the hominoids was not supported. However, there was one notable exception, namely, that all the study taxa had similar cortical to trabeculae ratios in the proximal capitate.
The differences between the proximal and distal Ct.Th across the locomotor groups provide support for the hypothesis that thick distal cortex in the nonhuman apes is a result of functional adaptation.
However, research indicates modern Homo sapiens have systemically low BV/TV and Ct.Th, which has been hypothesized to correlate with increased sedentism after the transition to an agricultural lifestyle (Chirchir et al., 2015;Ruff et al., 2006;Ryan & Shaw, 2015;Saers et al., 2016;Tsegai et al., 2018). Thus, it would be valuable to assess the distal Ct.Th of pre-Holocene Homo sapiens to further interrogate whether thick distal Ct.Th can be correlated simply with higher loading more generally, or, as hypothesized here, is related to forelimb involvement in arboreal behavior among the nonhuman apes. Further, there are important limitations to our interpretation of cortical bone functional adaptation in short bones. Although cortical bone does model its structure during adulthood in response to load, the genetic blueprint and the process of modelling during ontogeny greatly determines cortical bone geometry (Lovejoy et al., 2003;Martin et al., 1998). Investigation on the changes to cortical bone geometry as a result of functional adaptation have predominantly focused on changes at the mid-shaft of long bones (for examples and summary see Ruff et al., 2006 and references therein). In short bones there is unlikely to be the same capacity for the cortical bone to substantially change its geometry with modelling processes because, unlike the diaphysis of a long bone, there is not substantial room to expand (Martin et al., 1998). During adulthood, cortical bone commonly adapts its mechanical properties via changes to porosity, apparent mineral density or cellular anisotropy (Currey, 2002;Martin et al., 1998), changes that require different methodologies to assess (e.g., histology). Finally, when segmenting different bone tissues, it can be challenging to identify the boundary between cortex and trabeculae, particularly when the cortex is porous or trabeculae are especially thick. This was a particular challenge in some of the nonhuman ape capitate specimens (see Figure S1) and will likely be a limitation for many short bones, depending on the question being addressed.

| CON CLUS ION
The capitate of knuckle-walking African apes and suspensory Pongo was differentiated from bipedal Homo, primarily, by thick distal cortical bone. African apes were further differentiated from Pongo and Homo by relatively isotropic trabeculae in the proximal capitate, which was not expected given the (presumably) more stereotypical loading of the wrist during knuckle-walking. However this higher than expected DA of the capitate head in Homo may indicate preferential alignment of trabeculae along the DTM. Although the wrist is often conceptualized as broadly being under compression or tension, the differentiated bone architecture in the proximal and distal regions of the capitate suggests that the loading environment can differ significantly even within the small bones of the carpus and highly localized functional adaptation responses may be taking place. Further, differences in cortical bone were critical for differentiating Homo from nonhuman apes. While an unexpected positive relationship was found between bone volume and capitate size, the low coefficient of determination indicated size did not strongly influence group differences in bone microstructure. Given the complex biomechanical environment, and our limited understanding of intercarpal motion, (particularly in nonhuman apes) functional adaptation research of the carpals should take a holistic approach, including incorporated analysis of cortical bone.

DATA AVA I L A B I L I T Y S TAT E M E N T
The authors confirm that the data supporting the findings of this study are available within the article and its supplementary materials.