Growth and development of trabecular structure in the calcaneus of Japanese macaques (Macaca fuscata) reflects locomotor behavior, life history, and neuromuscular development

Abstract Bone structure dynamically adapts to its mechanical environment throughout ontogeny by altering the structure of trabecular bone, the three‐dimensional mesh‐like structure found underneath joint surfaces. Trabecular structure, then, can provide a record of variation in loading directions and magnitude; and in ontogenetic samples, it can potentially be used to track developmental shifts in limb posture. We aim to broaden the analysis of trabecular bone ontogeny by incorporating interactions between ontogenetic variation in locomotor repertoire, neuromuscular maturation, and life history. We examine the associations between these variables and age‐related variation in trabecular structure in the calcaneus of Japanese macaques (Macaca fuscata). We used high‐resolution micro‐computed tomography scanning to image the calcaneus in a cross‐sectional sample of 34 juvenile M. fuscata aged between 0 and 7 years old at the Primate Research Institute, Japan. We calculated whole bone averages of standard trabecular properties and generated whole‐bone morphometric maps of bone volume fraction and Young’s modulus. Trabecular structure becomes increasingly heterogeneous in older individuals. Bone volume fraction (BV/total volume [TV]) decreases during the first month of life and increases afterward, coinciding with the onset of independent locomotion in M. fuscata. At birth, primary Young’s modulus is oriented orthogonal to the ossification center, but after locomotor onset bone structure becomes stiffest in the direction of joint surfaces and muscle attachments. Age‐related variation in bone volume fraction is best predicted by an interaction between the estimated percentage of adult brain size, body mass, and locomotor onset. To explain our findings, we propose a model where interactions between age‐related increases in body weight and maturation of the neuromuscular system alter the loading environment of the calcaneus, to which the internal trabecular structure dynamically adapts. This model cannot be directly tested based on our cross‐sectional data. However, confirmation of the model by longitudinal experiments and in multiple species would show that trabecular structure can be used both to infer behavior from fossil morphology and serve as a valuable proxy for neuromuscular maturation and life history events like locomotor onset and the achievement of an adult‐like gait. This approach could significantly expand our knowledge of the biology and behavior of fossil species.

To understand how variation in trabecular structure arises within and between species, it is imperative to understand how it forms during growth and development (Gosman & Ketcham, 2009;Ryan et al., 2017;Ryan & Krovitz, 2006;Saers et al., 2020). Indeed, alterations to ontogenetic trajectories are the principal ways in which evolutionary changes in life history and morphology occur (Gould, 1977;Hallgŕimsson & Hall, 2005;Kardong, 2018;Woronowicz & Schneider, 2019). Recent methodological and technological advances in the analysis of trabecular bone structure have opened new possibilities for studying the development of trabecular bone structure (DeMars et al., 2020;Gross et al., 2014). Here we apply these new techniques to analyze the ontogeny of trabecular structure in the calcaneus of Japanese macaques (Macaca fuscata).
The musculoskeletal system undergoes striking changes throughout growth and development. Movements starting in utero and continuously changing throughout development, generate loads that shape bone morphology into the general adult form (Carter & Beaupré, 2001). These processes are essential for generating the adult morphology that is required for typical speciesspecific gait and posture (Tardieu, 1999). Mammalian locomotion usually develops in a stereotypical species-specific sequence of events that dramatically change how their skeletons are loaded (Doran, 1997;Lacquaniti et al., 2012;Sarringhaus et al., 2014).

| Trabecular bone ontogeny
The connections between trabecular morphology and habitual loading patterns have been demonstrated experimentally (Barak et al., 2011;Biewener et al., 1996), but trabecular bone ontogeny is still not yet thoroughly understood, particularly how the degree of plasticity versus genetic canalization varies throughout development and into adulthood (Cunningham & Black, 2009;Gorissen et al., 2016;Raichlen et al., 2015;Reissis & Abel, 2012;Ryan et al., 2017;Saers et al., 2020). Bone growth occurs via the transformation of growth plate cartilage into bone through a series of cell and matrix changes (Burr & Organ, 2017;Byers et al., 2000;Parfitt et al., 2000). The transformation from growth plate cartilage to trabecular bone is similar among mammals, indicating a highly conserved process (Byers et al., 2000;Frost & Jee, 1994). This process sets up a basic trabecular structure which can later be modified through metabolic and mechanical factors. Trabecular bone is laid out orthogonal to the growth plate in a dense and anisotropic structure which is later refined into bone-and species-specific heterogeneous adult states. Frost and Jee (1994) argue that the effects of mechanical usage during periods of rapid bone growth in early ontogeny explain many of the features observed during the ossification process. They propose that mechanical strain is the controlling mechanism for endochondral ossification, in which the underloaded elements of the dense bone structure during the first years of life are removed and bone is added in strained areas, resulting in a mechanically adapted state (Frost & Jee, 1994). This model correctly predicts observations of bone loss at early stages of ontogeny and explains it as the result of the removal of redundant material below a certain species would show that trabecular structure can be used both to infer behavior from fossil morphology and serve as a valuable proxy for neuromuscular maturation and life history events like locomotor onset and the achievement of an adult-like gait. This approach could significantly expand our knowledge of the biology and behavior of fossil species.

| The brain-bone connection
Brains and trabecular bone have more in common than one might initially think. Both are made up of complex, interconnected 3D structures and broadly share developmental patterns. At birth, both trabeculae and neurons are overproduced (Collin & Van Den Heuvel, 2013;Rabinowicz et al., 1996). Structures are refined to a more heterogeneous state during ontogeny through modeling in bone and synaptic pruning in neurons (Sakai, 2020). This happens under the influence of some input, presumably mechanical in terms of trabecular bone (Carter & Beaupré, 2001;Huiskes et al., 2000), and through neural activity in the brain (Sakai, 2020;Shatz, 1990). In both cases, there is a long history of debate as to how much of its respective morphology is genetically canalized versus plastic in response to its environment, that is, nature versus nurture, and in both cases, the consensus is "both." While starting with an excess of connections to remove many of them later may seem inefficient, the result is a state that is adapted to an individual's specific environment. Indeed, this process is so efficient that it is found in many other tissues as well including connective tissues like ligaments and tendons (Grinnell, 2000) to the nervous system (Sakai, 2020).
The patterns of the growth and development of trabecular structure reviewed above are consistent with a model where a generalized trabecular structure is formed by dynamic adaptation to local, bone-and region-specific loading patterns. These loading patterns are generated by neural circuits that develop in parallel to increases in physical size and weight of a growing organism (Forssberg, 1985). Locomotor patterns are transformed from an immature state to increasingly adult-like patterns during development. During the early ontogeny of gait, infants are mainly focused on minimizing the risk of falling. When individuals increase in strength, stability improves, and postural constraints are reduced (Vaughan & Langerak, 2003). It is thought that development subsequently proceeds to select the most optimal neural networks (Forssberg, 1999), resulting in a reduction in the variability of muscle activation and co-contraction, and the adult gait pattern emerges (Okamoto et al., 2003). If trabecular structure is a reliable reflection of gait mechanics, then changes in trabecular structure during growth should reflect gait mechanics, which in turn reflects the degree of neurological maturation of locomotion, as well as an animal's degree of precociality. If this link can be demonstrated, then trabecular structure could be a valuable proxy for neuromuscular maturation in fossil species.
Across mammals (Garwicz et al., 2009) and birds (Iwaniuk & Nelson, 2003), adult brain size strongly predicts time to locomotor onset after conception. In addition, the onset of walking is strongly correlated with the timing of several important aspects of brain development. In humans, locomotion is not just a developmental precursor to numerous psychological changes but plays a causal role in their formation Campos et al., 2000;Dahl et al., 2013;Uchiyama et al., 2008). The onset of human independent locomotion is followed by a revolution in perception-action coupling, spatial cognition, memory, and social and emotional development . Research indicates that neural function and structure reciprocally influence one another throughout development Campos et al., 2000), placing the activity of locomotor development in the center of development, rather than being just a consequence of neural maturation. In other words, the onset of independent locomotion is an important life history event related to adult brain size and the timing of neuromuscular development. If we can detect bony markers of locomotor development, this would be able to provide a unique insight into fossil locomotion as well as aspects of life history (Zihlman, 1992).

Japanese macaques
The basic locomotor characteristics of Japanese macaques appear in the first 2 months after birth (Dunbar & Badam, 1998;Kimura, 2000;Nakano, 1996;Torigoe, 1984). Newborns cannot walk, stand, or sit on their own and reflexively hang on to their mother for transport. Initial, somewhat poorly coordinated quadrupedal movements emerge in the second half of the first month (Nakano, 1996;Torigoe, 1984). Macaques begin to locomote primarily by walking in both diagonal and lateral sequences, followed after 4 weeks by occasionally running and trotting (Nakano, 1996). Independent locomotion away from the mother becomes regular after 2 months of age.
Coordinated walking appears after 3 months. Prior to this, their style of walking is limited by the immature development of their musculoskeletal system (Nakano, 1996). Locomotion becomes increasingly refined and independent throughout the first year of life (Dunbar & Badam, 1998). Locomotor activities include unskilled locomotion between 1 and 6 months when monkeys still frequently lose their footing. After 6 months, the macaques are skilled at both terrestrial and arboreal locomotion (Kimura, 2000;Nakano, 1996;Torigoe,, 1984).
Between the age of 1-and 3-years macaques enter the juvenile phase which contains the most diverse range of posture and locomotion. Juveniles have a well-developed musculoskeletal system which, combined with a small body size, enables juvenile macaques to exploit terrestrial and arboreal environments to their fullest potential (Dunbar & Badam 1998). After the juvenile phase macaques are considered adults but they continue to grow in size, albeit at a decreasing rate, until around 10 years of age (Hamada et al., 2004). The postural and locomotor repertoires of adults are reduced compared to juveniles, potentially due to increases in body size. The largest reduction is among play behaviors in the small-branch setting of trees, and below-branch postures and locomotion disappear (Dunbar & Badam, 1998). Passive joint mobility of macaques declines rapidly between 6 and 15 months and more gradually afterward (DeRousseau et al., 1983;Turnquist & Wells, 1993;Wells & Turnquist, 2001). These studies on the ontogeny of joint mobility, postural, and locomotor behaviors indicate that the most substantial changes in locomotor anatomy and behavioral control occur within the first 18 months of life (Turnquist & Wells, 1993;Wells & Turnquist, 2001).

Aims
If trabecular bone markers of behavior, neuromaturation, and life history variables such as the onset of independent locomotion can be established, reconstructions of the biology and behavior of fossil animals could be substantially improved. Our aim is to first document how trabecular structure of the calcaneus of Japanese macaques varies with age and body mass. We then test whether landmark events in the development of locomotion (independent locomotion, achievement of adult-like locomotor repertoires) in macaques coincide with clear signals in the trabecular structure. We do this by analyzing whole-bone averages of standard trabecular properties ( Table 1) as well as regional variation in the distribution of these properties throughout the calcaneus. Additionally, we aim to broaden the analysis of bone structure beyond pure locomotor mechanics by proposing a new way to incorporate the interactions between behavior and neuromuscular development, body size, and life history.

| Predictions
If the development of trabecular structure is largely or partially mediated through mechanics (Carter & Beaupré, 2001;Huiskes et al., 2000) rather than genetic programming (Lovejoy et al., 2003), one would predict that the minimal locomotor-related loading during the first month might lead to either bone resorption or no change in bone volume (BV) relative to total volume (TV), while increases in loading after the onset of locomotion should result in bone formation (Frost, 2003;Pivonka et al., 2018). New mechanical stimulation after the onset of locomotion combined with increases in body size are predicted to initiate reorganization of the trabecular architecture throughout the calcaneus. Redundant trabeculae are expected to be removed, while trabeculae oriented in directions involved in the distribution of loads associated with locomotion are preserved or enlarged, resulting in a reorganization of the primary direction of bone stiffness, and increases in bone volume fraction and average trabecular thickness (Tb.Th). After the onset of locomotion, trabeculae are expected to increase in regional variation in the amount of bone, bone stiffness, and average orientation of trabeculae. The highest bone volume fraction (BV/TV) is expected to be found where loads are applied to the calcaneus, including under joint surfaces (posterior talar facet and calcaneocuboid joint) and the attachment sites of the Achilles tendon and the plantar ligaments (Giddings et al., 2000;Saers et al., 2020).
We predict the following events to invoke the following associated morphological signals: 1. Onset of locomotion: Whole-bone average BV/TV is expected to decrease before the onset of locomotion and increase afterward. As such, the slope of the relationship between BV/ TV and age will shift from negative to positive, trabeculae will become thicker, and primary stiffness will align within the direction of joint loading (Barak et al., 2011;Ryan et al., 2017;Saers et al., 2020).

Appearance of adult-like locomotor repertoire: After locomotion
has matured the only remaining effects on trabecular structure should be allometry as body size continues to increase. The slope of the relationship between BV/TV and age should become more shallow and follow allometric scaling with still increasing body size (Doube et al., 2011;Mulder et al., 2020;Ryan & Shaw, 2013;Saers et al., 2019). Based on our review of the ontogeny of joint mobility, postural, and locomotor behaviors above, we predict this change to a shallower slope after 18 months (Turnquist & Wells, 1993;Wells & Turnquist, 2001).
3. Neuromuscular maturation of gait: age-related variation in loading patterns is generated by neural circuits that develop in parallel to increases in physical size and weight of a growing organism. As such, trabecular properties should be predicted by an interaction between body mass and neuromaturation. Here we use the estimates of the percentage of adult brain size for age as a proxy for neuromuscular maturation.

| Sample
We used high-resolution micro computed tomography (μCT) scanning to image the calcaneus from the skeletal remains of 34 juvenile male Japanese macaques (Macaca fuscata fuscata) from a

Abv. Description
Bone volume fraction BV/TV Ratio of bone volume to total volume of interest  (Torigoe, 1984). The enclosure was designed to mimic the natural environment of Japanese macaques and contains plentiful trees and climbing installations.

| Calculation of whole bone average trabecular properties
The 3D structure of trabecular bone was quantified using standard trabecular properties (  Kohler et al. (2007), following recommendations by (Bouxsein et al., 2010). We calculated the average bone volume fraction (BV/TV), Tb.Th, and trabecular separation (Tb.Sp).

| Three-dimensional mapping of trabecular structure
Segmented scans were categorized into three regions (cortex, trabeculae, and internal region of the bone) using Medtool 4.0 (www. dr-pahr.at, Figure 1). Morphometric maps of BV/TV and primary Young's modulus were generated following Gross et al. (2014). A 3D tetrahedral mesh was created of the internal region of the bone using CGAL (http://www. cgal.org). A mesh size of 0.6 mm was used.
A 3.5 mm background grid was applied in three dimensions to the trabecular, and BV/TV and Young's modulus were quantified at each node of the background grid using a 7 mm sampling sphere. The values from each sampling sphere were interpolated and applied to elements of the 3D tetrahedral mesh to generate morphometric maps.
Changes in the orientation of average primary Young's Modulus were investigated using a spherical volume of interest with a diameter of 50% of the maximum posterior talar facet length ( Figure 8).

| Statistical analysis
Linear regressions and interactions between trabecular properties, age, and body weight were performed in R version 4.0.2 (R Core Team, 2021). Alpha level was set to 0.05 for all statistical tests.
When comparing various regression models, the model with the lowest Akaike Information Criterion (AIC; Akaike, 1974) and highest R 2 was chosen as indicating the highest model quality. Adding additional variables to a regression always increases the fit (R 2 ) due to spurious correlations, this process is called overfitting and causes the model to learn too much from the data, resulting in poor predictive power for non-measured samples. AIC measures the degree to which a model is overfit with lower values indicating a greater model quality (Akaike, 1974;McElreath, 2015). In addition to regular linear regression, we run piecewise regressions using the 'segmented' R package (Muggeo, 2008). Piecewise regression is a useful technique for finding significant changes in slope in the relation between F I G U R E 1 Medtool workflow example showing a sagittal slice through a calcaneus a dependent and independent variable. The technique uses dummy variables and an interaction term to split a linear regression into multiple segments. The least-squares method is applied separately to each segment, by which the two regression lines are made to fit the dataset as closely as possible while minimizing the sum of squares of the differences between observed and predicted values of the dependent variable. We compared models with 0, 1, or 2 segments and chose the model with the highest R 2 and lowest AIC as the highest quality model.

| Bone properties with age
Mean trabecular properties calculated in the whole calcaneus are plotted against age in Figure 2

| Bone properties with body weight
The relationship between age and body weight at death is plotted in  Regression parameters for models where trabecular properties are predicted by the interaction between body weight and locomotor onset are given in Table 4. Individuals younger than 1 month are partially dependent on their mothers for locomotion and therefore coded as pre-locomotor onset, and those older than 1 month were coded as post-locomotor onset (Torigoe, 1984). The linear models with an interaction between body mass and locomotor onset given in  Figure 7 shows the heterogeneous distribution of BV/TV and primary Young's modulus throughout sagittal cross sections of a subset of specimens. When BV/TV is scaled between the sample minimum of 0.14 and maximum of 0.70, trabecular structure appears relatively homogeneous at birth and becomes more varied regionally with age.

| Morphometric maps
When the colormaps are scaled to the minimum and maximum values for each individual, regional variation is evident at all stages. BV/TV is   Figure 9 shows the close overlap between attainment of adult brain size and adult-like BV/TV. While macaques reach their maximum adult body size between 10 and 12 years of age (Hamada, 1994), 95% of maximum brain size is reached at 2 years of age and 100% at 5 years. BV/TV continues to increase slightly after the age when adult brain size has been obtained and body weight continues to increase. We can explain the patterns in Figure 9 with the following model: while the brain is still growing, increases in neuromuscular control and locomotor experience make loading environment of the calcaneus, and by proxy trabecular structure, increasingly like that of adults. When the brain has reached its full size, neuromuscular control of locomotion, and locomotor loading conditions also approach the adult-like pattern. This model suggests that the steep early age-related variation in BV/TV found in the macaque calcaneus may be related to increasing neuromuscular control of gait with a slight positive allometry after gait has matured but when body mass continues to increase. We test this model using percent adult brain size as a proxy for neuromuscular control.

| Bones and brains
In Table 4 we compare various types of models to assess which model predicts trabecular properties best (lowest AIC, highest R 2 ).
Variation in BV/TV is best explained by a three-way interaction between body mass, percentage of adult brain size, and locomotor onset. Trabecular thickness and separation are also explained by a two-way interaction between body mass and locomotor onset but differences between interaction models are limited as AIC and R 2 vary little between them.

| DISCUSS ION
We made the following predictions: 1. Onset of locomotion: we predicted that the slope of BV/TV with age in our sample should shift from negative to positive, that trabeculae would become thicker on average, and that Young's modulus should change direction from orthogonal to the ossification center to joint surfaces. All these predictions are supported.
2. Appearance of adult-like locomotor repertoire: the greatest changes in locomotor anatomy and repertoire occur in the first 12-18 months of life. We predicted that the slope of the relationship between age and BV/TV would become shallower after 18 months of age and largely follow allometric scaling with still increasing body size. This prediction is supported.
3. Neuromuscular maturation of gait: We predicted that trabecular properties at different ages should be predictable by an interaction between body mass and percent adult brain size. This prediction is supported in BV/TV after adding in an extra interaction term for locomotor onset. The prediction is not supported for average trabecular thickness or separation.
The developmental trajectories of trabecular properties in the calcaneus of Japanese macaques resemble those of other mammals (Colombo et al., 2019;Ragni, 2020;Tsegai et al., 2018;Wolschrijn & Weijs, 2004) including humans (Gosman, 2007 Saers et al., 2020), indicating a generally shared mechanism of growth (Carter & Beaupré, 2001). The distribution of BV/TV is substantially more homogenous in younger individuals and becomes increasingly heterogeneous with age, like results reported by Tsegai et al. (2018) for the chimpanzee postcranium. The development of trabecular structure of the calcaneus of Japanese macaques follows the same patterns as the human calcaneus, but with differences in the timing of stages , as well as other postcranial elements (Acquaah et al., 2015;Gosman & Ketcham, 2009;Raichlen et al., 2015;Ryan & Krovitz, 2006;Saers et al., 2020). In both species, bone is overproduced during early development with high BV/TV and struts largely oriented perpendicular to the ossification center in the calcaneus, or the growth plate in long bones. In both species, BV/TV reduces after birth and begins to increase again at the same time when individuals typically begin independent locomotion. At the same time the primary Young's modulus shifts in direction from orthogonal to the ossification center to orthogonal to the joint surfaces. This reorientation in the direction in which the bone is loaded helps to more efficiently distribute the loads placed upon the calcaneus during locomotion (Maquer et al., 2015;Roux, 1881;Wolff, 1867;Zysset, 2003). Contrary to our findings, Tsegai et al. (2018) did not find the initial overproduction of trabecular bone, followed by a drop in BV/TV. However, their sample included two individuals between 0 and 5 months of age and it is unclear if they were newborns or already locomoting five-month-olds.
Previous work in the human calcaneus found that Tb.Th was largely predictable by increasing body mass throughout ontogeny ,  Wells, 1993;Wells & Turnquist, 2001), and the eruption of the first permanent molars (18 months, Smith et al., 1994).
The relationship between Tb.Sp and age in the macaque calcaneus resembles that reported for the human calcaneus , humerus, and femur (Ryan et al., 2017). The average distance between trabeculae increases rapidly after birth (and the start of ossification). After this initial increase Tb.Sp continues to increase slowly with increasing age and body size. However, there is substantial individual variation in Tb.Sp of which roughly half can be explained by allometry (R 2 = 0.54). The large amount of variation in Tb.Sp relative to Tb.Th and BV/TV areh consistent with results reported throughout the postcranium of humans (Gosman & Ketcham, 2009;Raichlen et al., 2015;Ryan et al., 2017;Saers et al., 2020), apes (Ragni, 2020), and ungulates (Gorissen et al., 2016).

| The link between bones and brains
Locomotor patterns are transformed from an immature state to increasingly adult-like patterns during development. Changes in loading patterns with advancing age are generated by neural circuits that develop in parallel to increases in physical size and weight of a growing organism. Our results lend strong, but indirect support, for a model where trabecular structure is the product of age-related variation in loading conditions (Figure 9). These changes in loading conditions are a product of the development of gait which, in turn, is the product of neural maturation with age and experience. If this link is demonstrated experimentally, and in other species, trabecular structure could be used as a proxy not only for just development of locomotion but also neural maturation in fossil species. This brainbone connection would then serve as a powerful life history marker in fossil species.

| Developmental trajectories of trabecular structure as a life history marker
We used piecewise regression analysis to locate potential inflection points in the relationship between age and trabecular properties. All identified inflection points occur around two age points: one around  & Wells, 1993;Wells & Turnquist, 2001) and the eruption of the first permanent molars (18 months, Smith et al., 1994).
While some aspects of human bone morphology are genetically determined, others are environmentally induced. For example, the human lateral patellar lip is present already at birth (Lovejoy, 2007;Scheuer & Black, 2004), whereas the human bicondylar angle develops postnatally in response to mechanical loading associated with bipedal locomotion (Tardieu, 1999(Tardieu, , 2010Tardieu & Trinkaus, 1994).
In terms of trabecular bone, Barak et al. (2011) showed experimentally that differences in peak loading angle as well as magnitude alter trabecular bone orientation and BV/TV in sheep. Our ontogenetic data also suggest that age-related variation in trabecular structure corresponds to variation in loading conditions during landmark events in the maturation of gait (onset of locomotion, achievement of adult-like gait). As such, our results suggest that trabecular structure can potentially be used to infer the timing of locomotor onset and the achievement of adult-like locomotor repertoires. The advent of independent locomotion coincides with important aspects of mammalian brain size and neuromuscular development Campos et al., 2000;Garwicz et al., 2009). Patterns of locomotor development may therefore provide unique insights into the evolution of locomotor mode, how locomotion develops, neuromaturation, and the onset of independent locomotion in fossil species ( Figure 9).

| Limit ations
The data presented here correspond to a model where brain maturation increases neuromuscular control of gait, which, in turn, affects the mechanics of gait, which then shape loading patterns of the foot to which trabeculae dynamically adapt. However, our study is crosssectional with a skeletal sample of individuals of whom we do not know the individual behavior during life. As such we cannot quantify the changes in mechanical loading over time directly. This study design cannot be used as evidence of a causal mechanism. Confirming our proposed model will require controlled experiments resulting in longitudinal data, ideally in several species.
The percentage of adult brain size is a very rough proxy for neural development and numerous changes in brain composition and wiring occur after adult brain size has been reached (Lebel & Beaulieu, 2011). However adult brain size in humans (calculated from Cofran & Desilva, 2015) is perfectly correlated (R 2 = 0.99) with experimentally derived measures of neuromaturation (Vaughan & Langerak, 2003). This simple measure of percentage adult brain size is all that is available in paleontological contexts, and it is, therefore, encouraging that we can report such tight correlations.

| CON CLUS IONS
The developmental trajectories of trabecular properties in the calcaneus of Japanese macaques are similar to other species, indicating a broadly shared mechanism of growth. Trabeculae are overproduced at birth, followed by refinement leading to adaptation to local conditions and resulting in a species and joint-specific

ACK N OWLED G M ENTS
Many thanks to Dr. Takeshi Nishimura for providing access to the skeletal collection, databases, and CT scanner at the Primate Research Institute, Inuyama, Japan.

CO N FLI C T O F I NTE R E S T
The authors have no conflicts of interest to declare.

AUTH O R CO NTR I B UTI O N S
Jaap P. P. Saers: concept/design, funding acquisition, data acquisition and analysis, drafting of the manuscript. Adam D. Gordon: critical revision of the manuscript and statistical advice. Timothy M.
Ryan, Jay T. Stock: critical revision of the manuscript.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available on request from the corresponding author.