Regional variability in secondary remodeling within long bone cortices of catarrhine primates: the influence of bone growth history

Authors



Shannon C. McFarlin, Center for the Advanced Study of Hominid Paleobiology, Department of Anthropology, The George Washington University, 2110 G Street, N.W., Washington, D.C. 20052, USA. T: +1 202 9945923; F: +1 202 9946097; E: mcfarlin@gwu.edu

Abstract

Secondary intracortical remodeling of bone varies considerably among and within vertebrate skeletons. Although prior research has shed important light on its biomechanical significance, factors accounting for this variability remain poorly understood. We examined regional patterning of secondary osteonal bone in an ontogenetic series of wild-collected primates, at the midshaft femur and humerus of Chlorocebus (Cercopithecus) aethiops (n = 32) and Hylobates lar (n = 28), and the midshaft femur of Pan troglodytes (n = 12). Our major objectives were: 1) to determine whether secondary osteonal bone exhibits significant regional patterning across inner, mid-cortical and outer circumferential cortical rings within cross-sections; and if so, 2) to consider the manner in which this regional patterning may reflect the influence of relative tissue age and other circumstances of bone growth. Using same field-of-view images of 100-µm-thick cross-sections acquired in brightfield and circularly polarized light microscopy, we quantified the percent area of secondary osteonal bone (%HAV) for whole cross-sections and across the three circumferential rings within cross-sections. We expected bone areas with inner and middle rings to exhibit higher %HAV than the outer cortical ring within cross-sections, the latter comprising tissues of more recent depositional history. Observations of primary bone microstructural development provided an additional context in which to evaluate regional patterning of intracortical remodeling. Results demonstrated significant regional variability in %HAV within all skeletal sites. As predicted,%HAV was usually lowest in the outer cortical ring within cross-sections. However, regional patterning across inner vs. mid-cortical rings showed a more variable pattern across taxa, age classes, and skeletal sites examined. Observations of primary bone microstructure revealed that the distribution of endosteally deposited bone had an important influence on the patterning of secondary osteonal bone across rings. Further, when present, endosteal compacted coarse cancellous bone always exhibited some evidence of intracortical remodeling, even in those skeletal sites exhibiting comparatively low %HAV overall. These results suggest that future studies should consider the local developmental origin of bone regions undergoing secondary remodeling later in life, for an improved understanding of the manner in which developmental and mechanical factors may interact to produce the taxonomic and intraskeletal patterning of secondary bone remodelling in adults.

Introduction

The microscopic structure of bone represents an important source of information concerning the various phylogenetic, ontogenetic, environmental, and local/functional factors that influence its formation and subsequent remodeling during life (Enlow, 1963; Ricqlès, 1993; Martin et al. 1998). Bone microstructural analyses therefore have the potential to complement studies of whole bone morphology, and provide unique insight into behavior and life history. Secondary remodeling of bone exhibits considerable variation among vertebrate taxa, within single skeletons, and among individuals of different ages (e.g. Amprino & Godina, 1947; Enlow & Brown, 1958; Enlow, 1962a; Kerley, 1966; Schaffler & Burr, 1984; Paine, 1994; Currey, 2002; Lieberman et al. 2003). However, although recent efforts have focused on its biomechanical significance, the factors accounting for this variability remain poorly understood (Enlow, 1962b; Martin et al. 1998; Currey, 2002). The current study provides important baseline information on regional variability in secondary remodeling of bone at the midshaft femur and humerus of three species of Old World anthropoid primates, and considers the manner in which this variability may reflect the influence of relative tissue age and other circumstances of bone growth.

Secondary bone remodeling is carried out by coordinated populations of osteoblasts and osteoclasts, together constituting a basic multicellular unit (BMU) whose activities at single bone sites involve the resorption of an existing bone volume and its subsequent replacement by bone of more recent origin (Frost, 1986, 1990; Parfitt, 1994a; Enlow & Hans, 1996). When secondary remodeling occurs on intracortical surfaces, the BMU tunnels through bone to form a roughly cylindrical resorption space; concentric deposition of lamellar bone follows, thus refilling this space up to and leaving only a central vascular canal. The result of this process is the formation of a cylindrical structure, termed a secondary osteon, delimited from adjacent bone tissue by a reversal cement line (Fig. 1). The budding vascular supply of the secondary osteon provides progenitor cells for the advancing BMU, a passageway for mineral exchange between the skeleton and the bloodstream, and nourishment for embedded osteocytes and bone lining cells within the completed secondary osteon (Parfitt, 1994a; Enlow & Hans, 1996).

Figure 1.

Secondary osteonal bone. Left: Circularly polarized light (CPL) image showing the collagen fiber organization of bone. Arrows point to bone lamellae (appearing as alternating bright and dark bands), concentrically arranged around the Haversian canal of a secondary osteon. Right: Brightfield (LM) image of same field of view, showing a Haversian canal (HC) and cement line (CL), the latter delimiting a secondary osteon from adjacent tissue. Scale bar = 0.5 mm. Upper right: LM image with overlay of tissue type map, indicating bone regions occupied by secondary osteons.

Secondary intracortical remodeling (hereafter referred to as ‘intracortical remodeling’) exhibits a highly regional distribution within vertebrate skeletons and single skeletal sites, and it has been proposed to fulfill a variety of functions. Enlow (1962b; also see Enlow, 1963, 1976; Enlow & Hans, 1996) observed that intracortical remodeling occurs in predictable locations in the young skeleton where it is associated with certain developmental circumstances, most notably in cortical regions undergoing inward (i.e. endosteal) growth and at sites of soft tissue attachment. In this latter case, he proposed that the sequential resorptive and depositional activities of remodeling serve as a mechanism for the relocation of muscles and other soft tissue attachments along growing bone surfaces. The role of secondary remodeling in mineral homeostasis is thought to be fulfilled primarily by remodeling at cancellous bone surfaces, particularly in regions of the axial skeleton and pelvis adjacent to hematopoietic marrow, although remodeling in the peripheral skeleton may also contribute to temporary adjustments in blood calcium levels (Parfitt, 1994a, 2001, 2002a). Further, investigators have noted that intracortical remodeling is often concentrated in inner and/or older bone regions (e.g. Amprino & Godina, 1947; Enlow & Brown, 1958; Smith, 1960; Pfeiffer et al. 1995; Skedros et al. 2001; Currey, 2002), an observation which has been attributed to its function in older bone areas in the replacement of more highly mineralized or necrotic bone tissue for metabolic purposes (Enlow, 1962b; Parfitt, 2001, 2003), or to its role in microdamage repair in regions subject to biomechanical fatigue (Bouvier & Hylander, 1981; Burr, 1993).

A major focus of research on secondary intracortical remodeling in recent decades has been on its biomechanical significance (see reviews by Martin et al. 1998; Currey, 2002; Pearson & Lieberman, 2004). Although intracortical remodeling may serve to maintain bone mechanical competence in several ways, considerable attention has focused on its role in the prevention and/or repair (i.e. by replacement) of accumulated microdamage due to fatigue (e.g. Frost, 1960, 1990; Burr, 1993; Martin et al. 1998; Martin, 2003, 2007). Several aspects of the mechanical strain stimulus may influence the fatigue behavior and thus, intracortical remodeling, of bone [see Martin et al. (1998) for a review]. Degradation of bone material properties and microdamage accumulation is more pronounced under cyclic loading at higher strain magnitudes, strain rates, and over an increasing number of load cycles (e.g. Carter et al. 1981; Burr et al. 1985; Schaffler et al. 1989, 1990). Further, many studies have demonstrated increased microdamage accumulation and/or intracortical remodeling in skeletal regions experiencing elevated strains under both normal and experimental loading conditions (Lanyon et al. 1979, 1982; Bouvier & Hylander, 1981, 1996; Burr et al. 1985; Burr & Martin, 1993; Mori & Burr, 1993; Bentolila et al. 1998; Verborgt et al. 2000; Lee et al. 2002; Lieberman et al. 2003).

Although there is evidence that mechanical strains and/or microdamage serve as important stimuli for site-specific intracortical remodeling of bone, the manner in which its spatial patterning may also reflect the influence of other factors remains poorly understood (Enlow, 1962b, 1976; Parfitt, 2001, 2002b; Burr, 2002; Skedros et al. 2004). For instance, Bouvier & Hylander (1996) reported that while the regional distribution of intracortical remodeling in the macaque face coincided with peak mechanical strains in immature animals, it was observed in both high and low strain regions in adults. Further, it was recently shown that intracortical remodeling at the midshaft radius of immature and adult goats showed no significant regional patterning within cross-sections, and lacked correspondence with measured strain distributions (Main, 2007). These studies suggest that mechanical stimuli, metabolic factors, bone growth dynamics, and other unidentified factors interact in complex ways to influence the distribution of intracortical remodeling within and among vertebrate skeletons.

Primates vary considerably in the degree to which they undergo intracortical remodeling of their limb skeletons. This variability reflects, in part, species differences in age-related remodeling dynamics (Kerley, 1966; Burr et al. 1989; Burr, 1992; Mulhern & Ubelaker, 2003, 2006; Havill, 2004), body size and skeletal function (Schaffler & Burr, 1984; Paine, 1994; Paine & Godfrey, 1997; Warshaw, 2007). However, regional variability in intracortical remodeling within primate skeletons is poorly understood, and studies to date have focused largely on adult animals (for an exception, see Newell-Morris & Sirianni, 1982). In nonhuman primates, Paine & Godfrey (1997) found that indices of relative amounts of remodeling in the femur vs. the humerus varied in relation to locomotor behavior in cercopithecid and galagonid taxa, as did regional patterning of intracortically remodeled bone across four anatomically defined quadrants within midshaft skeletal sites. However, variability across proximal, midshaft and distal section levels of the femur and humerus was independent of behavioral differences (Paine, 1994). More recently, Warshaw (2007) found no consistent relationship between locomotor behavior and regional patterning of secondary osteonal bone across quadrants within the midshafts of limb elements in strepsirhine and platyrrhine primates. Further, Newell-Morris & Sirianni (1982) demonstrated that the distribution of intracortical remodeling at the midshaft humerus of neonatal pig-tailed macaques was strongly influenced by local bone growth dynamics associated with cortical drift. Collectively, these results suggest that studies examining the influence of developmental factors on intracortical remodeling may lead to an improved understanding of its regional distribution in adult skeletons.

Objectives of the current study

In this study, we begin to examine regional variability in the intracortical remodeling of bone in an ontogenetic series of wild-collected Old World anthropoid primates characterized by differences in positional behavior and body size (summarized in Table 1): the midshaft femur and humerus of Chlorocebus (Cercopithecus) aethiops and Hylobates lar and the midshaft femur of Pan troglodytes. We address the following question: Does secondary osteonal bone exhibit a preferential distribution in inner cortical regions within bone cross-sections? Tissue age is expected to have an important influence on the distribution of intracortical remodeling in bone, for both metabolic and mechanical reasons (Enlow, 1962b; Bouvier & Hylander, 1981; Burr, 1993; Parfitt, 2001). As tissues deposited at the periosteal surface are sequentially relocated to inner cortical regions during growth, older bone areas comprising inner and middle cortical regions are expected to exhibit higher proportions of secondary osteonal bone than outer cortical regions. Mechanical strain history may also influence this pattern, as tissues deposited at the periosteal surface are subjected to elevated strain environments early in their developmental history (i.e. prior to being relocated inward during growth), due to their distance from the neutral axis of bending (Biewener, 1992).

Table 1.  Locomotor behavior of study taxa and sample representation across biological age classes (femur/humerus)
Adult locomotor behaviors, listed in order of their frequency*SexAdult mass (kg)Total NDECIDM1M2M3EPIPH
  • *

    References: Doran, 1996; Doran & Hunt, 1994; Fleagle, 1980; Rose, 1979. The most frequent behaviors are listed first; least frequent behaviors are listed last.

  • References: Bolter & Zihlman, 2003; Carpenter, 1940; Smith & Jungers, 1997, based on P. troglodytes.

  • Includes one individual for which associated craniodental material was missing; assigned to M3 age class based on diaphyseal length and epiphyseal union.

  • §

    Includes two individuals for which associated craniodental material was missing; assigned to M3 age class based on diaphyseal length and epiphyseal union.

  • Includes one individual having a questionable association with a skull having only permanent M1s emergent and unemergent M2s; reassigned to M2 age class based on diaphyseal length, as described in McFarlin (2006).

Chlorocebus (Cercopithecus) aethiops
quadrupedal walking/running, climbing, leapingF3.514/131/14/52/14/33/3
M5.816/174/41/14/54/43/3
Hylobates lar
brachiation, climbing, leaping, bipedalismF5.313/122/22/22/14/4§3/3
M5.915/142/21/14/45/43/3
Pan troglodytes
knuckle-walking quadrupedalism, quadrumanous climbing/scrambling, suspension, bipedalismF45.8510130
M59.7200101
unknown 511201

This question is addressed by partitioning each cross-section into three circumferential rings (inner, middle and outer cortex), as a means of quantifying intracortical remodeling across bone regions deposited during different growth stages. As this method does not permit the discrimination of bone areas of periosteal vs. endosteal origin or circumstances of lateral cortical drift, observations of primary bone microstructural development provide an additional context in which to consider regional variability in intracortical remodeling across rings (McFarlin, 2006, in prep.).

Materials and methods

Sample materials

The midshaft femur and humerus were chosen for study. As prior researchers have shown that substantial bending strains are generated at the midshaft of long bones during locomotion (e.g. Biewener, 1991), midshaft sites are commonly included in examinations of bone structural and histocompositional variables (e.g. Schaffler et al. 1985; Paine & Godfrey, 1997; Goldman et al. 2003). To utilize both adults and immature specimens lacking fused epiphyses, midshaft was determined based on a measure of diaphyseal length, as described more fully by McFarlin (2006; also see Warshaw, 2007).

As chronological age was unknown, we grouped individuals into five biological age classes based on dental alveolar emergence and long bone epiphyseal union. Individuals having one or more emergent deciduous teeth and lacking permanent molars in the process of eruption were assigned to the infant (‘DECID’) age class. Individuals were assigned to three juvenile age classes (‘M1’, ‘M2’, ‘M3’) based on alveolar emergence of one or more first, second or third permanent molars, respectively. As individuals continue skeletal development after full eruption of the dentition (observations of this study; Watts, 1990; Bolter & Zihlman, 2003), individuals were considered fully mature only after complete fusion of all long bone epiphyses (‘EPIPH’). Biological age classification and methods for assessing dental emergence and epiphyseal union were modified from Bolter & Zihlman (2003). All individuals lacked obvious evidence of skeletal fracture and pathology. Older adults were excluded from the sample on the basis of advanced occlusal wear of first permanent molars, as described in McFarlin (2006) (Table 1).

Sample preparation

Blocks of bone 1 cm in thickness were extracted from the midshaft cross-section of each element, cleaned of internal non-mineralized organic components and adhering soft tissue in a 1% Terg-A-Zyme solution, dehydrated, and then refluxed in a Soxhlet apparatus for 1–2 weeks (depending on size) in a 50 : 50 solution of isopropanol and heptane for further dehydrating and defatting prior to embedding in polymethylmethacrylate. Embedded block surfaces were exposed on the cut surface nearest to midshaft by grinding through a graded series of emory papers, and then prepared and mounted on glass slides using Dentsply dental adhesives. Histological thin sections were produced according to the method of Goldman et al. (1999) with some modifications (see McFarlin, 2006). Thin sections were ground to a uniform section thickness of 100 ± 5 µm at a final surface topography of 1200 grit, and temporarily cover-slipped with ethylene glycol for optimum optical performance. This method results in the production of histological thin sections that are compatible with imaging in both transmitted light and scanning electron microscopy (SEM) modalities, and thus will permit a future correlative study of mineralization density on these same specimens utilizing backscattered electron imaging in SEM.

Image acquisition, processing and quantification

Same field-of-view images were acquired in brightfield (LM) and circularly polarized light (CPL) microscopy using a Leica DMRX/E Universal microscope configured with an automated Marzhauser stage (0.1-µm accuracy in ‘X’, ‘Y’). Whole cross-section images were obtained using a JVC KY55BU (5× objective; 1 pixel = 1.4970 µm) or KY55BE (5× objective; 1 pixel = 1.2563 µm) color CCD camera, integrated with a Syncroscopy Montage Explorer system (Synoptics Ltd., Cambridge, UK).

LM and CPL images were imported into Adobe Photoshop 6.0 software, and superimposed for visualization and quantification of bone areas represented by secondary intracortical remodeling. First, most non-bone areas (e.g. medullary canal, resorption spaces) were assigned to a background color black so they could be deleted in later quantitative analyses. Secondly, bone areas comprising intact and fragmentary secondary osteons were manually traced and color-coded, according to the following criteria: 1) presence of a complete or partial Haversian canal (HC); 2) concentric lamellae; and 3) presence of a complete or partial reversal cement line (CL), delimiting a secondary osteon from adjacent tissue (Fig. 1). Finally, as we wished to relate intracortical remodeling to other aspects of bone microstructural development, primary bone tissue types were also manually traced as described more fully in McFarlin (2006, in prep.), following published classifications (Enlow, 1963; Francillon-Vieillot et al. 1990). This color-coded tissue type map was then converted to gray-scale for analysis. All intracortically remodeled bone areas were assigned a single gray-level value along a 256 scale (arbitrarily set at 114), and imported into Optimas 6.5 Image Analysis software (Media Cybernetics Inc.) for quantification. Using a customized macro, each cross-section was divided into three circumferential rings (inner cortical, middle cortical and outer cortical) and eight radial sectors, comprising 24 bone segments in total (described by Feik et al. 2000; Goldman et al. 2003) (Fig. 2). Within each bone segment, the number of pixels at each gray level (or its corresponding tissue type) was calculated, and results were transferred via a dynamic data exchange link to a Microsoft Excel spreadsheet. In this study, we report results from analysis of regional variability among rings.

Figure 2.

Segmentation analysis. Partitioning of bone cross-sections into three circumferential rings (labeled ‘Inner’, ‘Middle, and ‘Outer’) and radial sectors for quantitative analysis of spatial variation in intracortical remodeling. Variability among rings is considered in the current study.

Data analysis

We examined variability in the percentage of cortical area comprising secondary osteonal bone (%HAV; including both intact and fragmentary secondary osteons), both among whole cross-sections and among circumferential rings within cross-sections. %HAV was calculated from raw pixel data generated in Optimas according to the following equation:

%HAV = (pixels represented by HAV/pixels represented by bone area)*100 ((Eqn 1))

Although %HAV does not provide information about dynamic parameters of bone remodeling (e.g. the rate at which new remodeling events are initiated), it does provide a cumulative measure of the overall extent to which cortical regions at the midshaft femur and humerus have undergone intracortical remodeling (e.g. Paine & Godfrey, 1997).

The focus of the current study was on regional variability in %HAV within skeletal sites, and all statistical analyses were therefore restricted to intra-specific comparisons. Due to small sample sizes (n < 5) within age classes, data were analysed using nonparametric techniques in Statistica 6.0 (Statsoft, Inc.) software. A probability value of P ≤ 0.05 was considered statistically significant. Mann–Whitney, Kruskal–Wallis tests and Spearman rank order correlations were performed to determine the effect of sex and age class on whole cross-section %HAV within taxa. To analyse regional variability in %HAV, the effect of bone element and circumferential ring on %HAV was determined using nonparametric tests for dependent samples. Where Friedman anova revealed significant variability among rings within cross-sections, post-hoc Wilcoxon matched pairs comparisons were performed to identify significant pair-wise contrasts. Probability values arising from post-hoc tests were corrected using the Bonferroni method, in which P was divided by the number of total comparisons (Sokal & Rohlf, 1995).

Results

General patterns: whole cross-section %HAV

With all age classes combined and regional data pooled, secondary osteonal bone in C. aethiops constituted a mean 7.7% (range = 0.0–22.4, SD = 5.4527) of cortical area at the midshaft femur and 21.2% (range = 2.6–36.5, SD = 8.4245) at the midshaft humerus; in H. lar, a mean 38.2% (range = 25.5–57.0, SD = 8.3367) of cortical area at the midshaft femur and 49.2% (range = 35.4–66.8, SD = 8.8862) at the midshaft humerus; and in P. troglodytes, a mean 54.0% (range = 37.7–78.1, SD = 11.8610) of cortical area at the midshaft femur. %HAV showed a significant positive correlation with age class in C. aethiops (femur, Spearman r = 0.6175, P = 0.0003; humerus, Spearman r = 0.4776, P = 0.0076). In both C. aethiops and H. lar, the humerus exhibited a significantly greater %HAV compared to the femur (Table 2). Inspection of raw data revealed considerable overlap among sexes, and statistical analyses indicated that sex was not a significant factor contributing to variation in %HAV in C. aethiops and H. lar. Variability in intracortical remodeling within and among sites can be appreciated in Fig. 3.

Table 2.  Results of Wilcoxon matched pairs tests for variability in %HAV among bone elements. Significant contrasts in bold (P < 0.05)
TaxonAge classNTZP
Chlorocebus (Cercopithecus) aethiopsTotal sample2814.59980.000
DECID502.02260.043
M1511.75290.080
M2502.02260.043
M3702.36640.018
EPIPH602.20140.028
Hylobates larTotal sample26124.15260.000
DECID440.36510.715
M1311.06900.285
M2502.02260.043
M3802.52050.012
EPIPH602.20140.028
Figure 3.

Examples of tissue type maps. Color-coded tissue type maps derived from the midshaft femur and humerus of individuals representing the M1, M2, M3 and EPIPH age classes are shown for each taxon. Purple = secondary osteonal bone; green shades = primary periosteal tissues; red shades = endosteal circumferential tissues; light orange = endosteal compacted coarse cancellous bone; light yellow = Sharpey's fiber bone. Scale bars = 1.0 mm.

Variability among circumferential rings within cross-sections

Considerable variability in the proportion of intracortically remodeled bone was observed within rings among conspecifics of similar dental emergence and epiphyseal union status. Standard deviations ranged from 0% to 42.53% within age classes, when taxa and bone elements were considered separately. In P. troglodytes, the two individuals constituting the EPIPH age class differed markedly from one another, and standard deviations for %HAV ranged from 0.36% to 23.46% in this taxon when the EPIPH age class was excluded. Despite differences among age-matched individuals in the amount of intracortically remodeled bone within regions, however, it was also the case that consistent and statistically significant variability was observed among circumferential rings at the midshaft femur and humerus of all taxa (Fig. 4).

Figure 4.

Regional variability in the proportion of secondary osteonal bone across circumferential rings, plotted across age classes.

Although age class was a significant factor only in C. aethiops when whole cross-section %HAV was considered, age class was a significant factor contributing to variability in %HAV within rings in both C. aethiops and H. lar (Table 3). (Sample sizes within age classes did not permit statistical analysis in P. troglodytes.) Examination of regional variability in %HAV among rings within cross-sections was therefore undertaken with age classes considered separately within taxa, as shown in Table 4. However, power to discriminate significant post-hoc contrasts among rings was limited due to small sample sizes. Visual inspection of tissue type maps and bivariate plots revealed that regional patterning broadly typical of adults was evident in the M2 and all older age classes within taxa. Therefore, results reported below focus on intraspecific analyses undertaken in two subsets of each taxon sample: the DECID–M1 age classes (DE–M1, representing ‘early’ ontogeny) and the M2–EPIPH age classes (M2–EP, representing ‘late’ ontogeny), respectively. Results are also reported for age classes considered separately, to demonstrate their general correspondence with results based on sample subsets.

Table 3.  Results of Kruskal–Wallis tests for variability in %HAV among age classes by ring. Significant results in bold (P < 0.05). Post-hoc contrasts reported here are significant at P < 0.05 after Bonferroni correction
RingdfnHPPost-hocs (P′= 0.005)
Chlorocebus (Cercopithecus) aethiops: midshaft femur
 Outer43017.518710.0015 
 Middle43013.222510.0102DECID*M3
 Inner430 8.6756990.0697 
Chlorocebus (Cercopithecus) aethiops: midshaft humerus
 Outer43013.00480.0113 
 Middle430 8.74380.0678 
 Inner430 4.95840.2916 
Hylobates lar: midshaft femur
 Outer428 4.62600.3279 
 Middle428 8.90440.0635 
 Inner42811.28860.0235 
Hylobates lar: midshaft humerus
 Outer426 8.74490.0678 
 Middle426 0.60770.9622 
 Inner42613.97810.0074 
Table 4.  Results of Friedman's anova for variability in %HAV among rings. Significant results in bold (P < 0.05). Post-hoc contrasts reported are significant at P < 0.05 after Bonferroni correction
 nOUTERMIDDLEINNERPPost-hocs (P’= 0.0167)
MeanSDMeanSDMeanSD
  • Ring abbreviations: Outer (O); Middle (M); Inner (I).

Chlorocebus (Cercopithecus) aethiops: midshaft femur
 DE-M1100.00.152.54.3011.19.980.000O*M, O*I, M*I
 M2-EP202.11.739.47.0820.111.030.000O*M, O*I, M*I
 DECID50.00.000.60.495.65.690.039 
 M150.10.224.45.8916.610.780.007 
 M260.81.335.63.2516.710.200.003 
 M382.51.6110.06.4520.213.180.001O*M, O*I
 EPIPH62.91.6912.49.6623.29.490.002 
Chlorocebus (Cercopithecus) aethiops: midshaft humerus
 DE-M11113.410.4823.515.4118.49.690.029 
 M2-EP1916.14.6930.09.2824.214.700.000O*M, M*I
 DECID54.16.7515.511.5016.68.450.022 
 M1621.24.9530.115.9019.911.170.069 
 M2616.66.2825.06.1616.812.070.135 
 M3715.95.2229.011.9725.820.030.066 
 EPIPH615.62.5936.24.6629.66.830.006 
Hylobates lar: midshaft femur  
 DE-M1710.95.9970.116.1052.08.670.001 
 M2-EP218.46.6146.514.6367.49.500.000O*M, O*I, M*I
 DECID413.57.0571.120.1649.510.960.018 
 M137.51.2968.712.7455.24.240.050 
 M264.74.6942.211.6365.513.290.002 
 M399.16.7347.418.8766.16.810.000O*M, O*I
 EPIPH611.07.4349.610.8071.49.130.002 
Hylobates lar: midshaft humerus
 DE-M1737.015.7164.514.0637.316.140.005 
 M2-EP1919.311.5767.713.3273.312.870.000O*M, O*I
 DECID444.717.4761.314.6128.315.790.039 
 M1326.83.1068.914.9549.35.510.050 
 M2520.816.7064.317.2575.110.130.015 
 M3821.79.9969.512.8772.616.340.010O*M
 EPIPH614.99.1468.212.3372.711.650.009 
Pan troglodytes: midshaft femur
 DE-M1326.69.6970.911.4663.46.080.050 
 M2-EP930.520.6364.012.6983.07.500.000 
 DECID232.20.3677.42.3966.90.64  
 M1115.4 57.8 56.4   
 M2422.84.7765.211.1079.69.23  
 M3337.023.4666.614.9088.55.05  
 EPIPH236.242.5357.519.0381.51.55  

Intraspecific analyses demonstrated significant variability in %HAV among rings within midshaft femur and humerus cross-sections. %HAV was usually higher in the inner and middle cortical rings than in the outer cortical ring, though post-hoc contrasts were not significant in many comparisons. Exceptions to this trend were the C. aethiops midshaft humerus (M1 and M2 age classes) and the H. lar midshaft humerus (DECID age class). However, of particular interest, regional patterning among inner vs. middle cortical rings was more variable across skeletal sites, age classes and taxa examined. Results are summarized by taxon below.

Chlorocebus (Cercopithecus) aethiops

At the midshaft femur, post-hoc contrasts were significant in all comparisons for each subsample; %HAV was significantly higher in the inner ring, intermediate in the middle ring, and lower in the outer ring in all age classes. At the midshaft humerus, considerable overlap in %HAV among rings was observed across age classes, although significant contrasts in the late ontogeny subsample reflected a tendency for increased %HAV in the middle ring.

Qualitative observations of the midshaft femur revealed that secondary osteons were largely absent from more recently deposited cortical regions of periosteal origin (i.e. located near the periosteal surface), except in the posterior cortex, where secondary osteons occurred in association with Sharpey's fibers, extrinsic collagen fiber bundles indicative of local muscle attachment sites (Fig. 3). In DECID and M1 age classes, scattered secondary osteons in anterior and adjacent periosteal cortices were associated with older primary vascular bone. In older age classes, secondary osteons outside of the posterior cortex were largely restricted to cortical regions of endosteal origin showing a compacted coarse cancellous organization, and regions adjacent to the reversal boundary between endosteal and periosteal cortex. At the midshaft humerus, secondary osteons in cortex of periosteal origin were more commonly observed anterolaterally and laterally, where they were distributed throughout the thickness of the cortex in some individuals. Evidence of intracortical remodeling in regions of endosteal origin was prevalent, and associated with interstitial tissues having both circumferential and compacted coarse cancellous organizations (Fig. 5).

Figure 5.

Secondary osteons in regions of endosteal compacted coarse cancellous bone: Chlorocebus (Cercopithecus) aethiops midshaft humerus. Left: CPL image. Right: LM image of same field of view. Upper right: CPL image with overlay, showing the distribution of secondary osteons. PLB, periosteal lamellar bone to the far left of this image. Secondary osteons are concentrated along the reversal boundary (**) between periosteal and endosteal cortex. To the right of this reversal, secondary osteons in endosteal cortex are associated with interstitial tissues having a compacted coarse cancellous organization. (Note the convoluted appearance of interstitial bone.) Secondary osteons are recognized in such regions on the basis of their reversal cement line border and the presence of a Haversian canal. Scale bar = 0.5 mm.

Hylobates lar

In both elements, significant variability in %HAV was observed across rings in all age classes, although post-hoc contrasts were significant only in the late ontogeny subsample. At the midshaft femur, %HAV was consistently lowest in the outer ring, but inner and middle rings showed a variable pattern with age: DECID and M1 age classes showed higher mean %HAV in the middle ring, whereas mean %HAV was highest in the inner ring in all older age classes. At the midshaft humerus, %HAV was lower in the outer ring in all but the DECID age class. However, variable patterning across inner and middle rings was observed with age, as also reported for the midshaft femur. Mean %HAV was highest in the middle ring in DECID and M1 age classes, whereas inner and middle rings showed considerable overlap in older age classes.

Qualitative observations revealed that secondary osteons in both elements were most commonly distributed in the inner one-third to two-thirds of the periosteal cortex, in cortical regions of endosteal origin and along the endosteal-periosteal reversal boundary (Fig. 3). In DECID and M1 age classes, endosteal tissues constituted a considerable proportion of cortical area at the midshaft femur and humerus. Secondary osteons were concentrated in the mid-cortical region in several individuals, appearing less densely distributed in more recently deposited periosteal and endosteal tissues. However, due to the density of intracortical remodeling, it was often difficult to assess the microstructural organization of interstitial bone in these mid-cortical regions; in some individuals, interstitial areas of endosteal compacted coarse cancellous bone were observed, particularly at the midshaft humerus. Large resorption spaces were also common in DECID and M1 individuals, and secondary osteons were often eccentric in shape. Older age classes showed a pattern at the midshaft femur in which secondary osteons were observed in more recently deposited periosteal regions of the posterior cortex, often in association with Sharpey's fibers. Evidence of intracortical remodeling in cortical regions of periosteal origin was more extensive at the midshaft humerus.

Pan troglodytes

At the midshaft femur, significant variability in %HAV was observed across rings in both subsamples, though post-hoc contrasts were not significant, likely due to small samples. Whereas mean %HAV was consistently lowest in the outer ring, its distribution among inner and middle rings showed a variable pattern with age. Increased %HAV was observed in the middle ring of DECID individuals, whereas %HAV was similar in inner and middle rings of the single M1 individual. In older age classes, the inner ring showed the highest mean %HAV.

Qualitative observations revealed that secondary osteons were abundantly distributed in cortical regions of endosteal origin, and along the endosteal-periosteal reversal boundary (Fig. 3). Endosteally deposited cortex was most prominent in individuals of the M2 age class, where secondary osteons were associated with interstitial bone having both circumferential and compacted coarse cancellous organizations. Second ary osteons were also commonly distributed in inner cortical regions of periosteal origin, and in certain local regions of the outer periosteal cortex. The latter was most notable posteriorly and posterolaterally, where secondary osteons were distributed throughout the thickness of the periosteal cortex in some individuals. Sharpey's fibers were also observed in these regions. Individuals of the M3 and EPIPH age classes were especially variable in the overall extent to which the periosteal cortex was intracortically remodeled; in those individuals showing the highest proportions of secondary osteonal bone, this remodeling appeared to be preferentially distributed in the posterior, lateral and anterior cortices.

Discussion

We undertook a quantitative examination of regional variability in secondary osteonal bone in an ontogenetic series of wild-collected primates, focusing on the midshaft femur and humerus of Chlorocebus (Cercopithecus) aethiops (vervet monkeys) and Hylobates lar (white-handed gibbons), and the midshaft femur of Pan troglodytes (chimpanzees). Our results confirm prior research indicating that primates vary considerably in the overall degree to which they undergo intracortical remodeling of their limb skeletons, and provide the first quantitative data on intracortical remodeling in the femur and humerus of gibbons. When all age classes were combined and regional data pooled, secondary osteonal bone in C. aethiops constituted a mean 7.7% of cortical area at the midshaft femur and 21.2% at the midshaft humerus; in H. lar, a mean 38.2% of cortical area at the midshaft femur and 49.2% at the midshaft humerus; and in P. troglodytes, a mean 54.0% of cortical area at the midshaft femur.

Whole cross-section proportions of secondary osteonal bone in H. lar were in agreement with the findings of Schaffler & Burr (1984) based on other primate taxa, in which primates adapted for suspensory behaviors (Ateles species and P. troglodytes in the latter study) showed higher proportions of secondary osteonal bone at the midshaft femur than did quadrupedal primates (including C. aethiops), despite similarities in body size. Further, increased intracortical remodeling in the humerus compared to the femur, as shown here, has been previously described for a number of primates, including Old World monkey quadrupeds (Paine & Godfrey, 1997; Warshaw, 2007). These taxon and inter-element patterns have been attributed by prior researchers to differences in fore- and hindlimb kinetics among arboreal vs. more terrestrial primates, as revealed by force-platform analyses (for examples of the latter, see Kimura et al. 1979; Kimura, 1985, 1992; Reynolds, 1985; Demes et al. 1994; Schmitt & Hanna, 2004). Although they did not examine intracortical remodeling in the forelimb, Schaffler & Burr (1984) attributed increased femoral remodeling in suspensory primates to their more pronounced functional limb differentiation compared to terrestrial primates, and prior evidence indicating that suspensory primates experience significantly greater vertical loads (i.e. associated with body mass support) and propulsive force components in the hindlimbs compared to the forelimbs during quadrupedal locomotion. Paine & Godfrey (1997) suggested that increased humeral relative to femoral remodeling in Old World monkey quadrupeds reflected a requirement to resist increased forelimb loadings associated with braking movements and shock-absorption on less flexible terrestrial substrates. Warshaw (2007) also observed increased intracortical remodeling in forelimb elements compared to hindlimb elements in strepsirhine and platyrrhine primates characterized by a diverse range of positional behavior repertoires. However, she suggested that the intraskeletal distribution of intracortical remodeling in primates may be influenced in part by patterns of primary bone microstructural development.

Only a few studies of nonhuman primates thus far have quantified regional patterning in the distribution of secondary osteonal bone within single limb elements (Paine, 1994; Paine & Godfrey, 1997; Warshaw, 2007), and these have focused largely on adults (but see Newell-Morris & Sirianni, 1982). The current study differs from prior research on nonhuman primate bone remodeling by focusing on a more detailed study of ontogenetic and regional variability within skeletal sites; only by examining intracortical remodeling within an ontogenetic framework can we begin to understand the manner in which developmental factors may also influence its taxonomic and intraskeletal distribution in adults (Enlow, 1962a,b; Newell-Morris & Sirianni, 1982; Bromage, 1992; Bouvier & Hylander, 1996; Bromage & Boyde, 1998).

Variability across circumferential rings within bone cross-sections

Prior investigators have observed that secondary osteonal bone in many animals is concentrated in inner cortical regions within cross-sections, adjacent to the endosteal surface (Amprino & Godina, 1947; Enlow & Brown, 1958; Smith, 1960; Enlow, 1962b; Burton et al. 1989; Pfeiffer et al. 1995; Skedros et al. 2001; Currey, 2002). This pattern has been attributed to preferential intracortical remodeling of older bone areas, for the purposes of maintaining fatigue life, replacement of more highly mineralized bone for metabolic reasons, and/or to serve in the replacement of necrotic bone regions (Enlow, 1962b; Bouvier & Hylander, 1981; Burr, 1993; Parfitt, 2001, 2003; Currey, 2003). Although regional variability in intracortical remodeling was also appreciated around the circumference of bone cross-sections, in the current study we quantified the proportion of secondary osteonal bone across inner, middle and outer circumferential cortical rings within cross-sections, as a means of examining its distribution across bone regions deposited during different growth stages. We predicted that secondary osteonal bone would occur in higher proportions in inner and middle rings than in outer cortical rings (the latter comprising more recently deposited cortex), consistent with preferential remodeling of older bone areas.

Results of this study provided partial support for this prediction. Proportions of secondary osteonal bone were usually higher in inner and middle rings compared to the outer ring within cross-sections. However, of particular interest, regional patterning of intracortical remodeling across the inner vs. middle ring was not consistent across all skeletal sites, age classes or taxa examined. As predicted, the midshaft femur of C. aethiops, the least intracortically remodeled among all skeletal sites examined here, showed higher proportions of secondary osteonal bone in the inner ring than in the middle and outer rings within cross-sections. However, the midshaft humerus showed considerable overlap among all three rings, with a tendency for increased intracortical remodeling in the middle ring. Elements of H. lar and P. troglodytes, which exhibited more intracortical remodeling overall, showed more overlap and age-related variability in the proportion of secondary osteonal bone in inner vs. middle rings; proportions of secondary osteonal bone tended to be higher in the inner ring than in the middle ring during later ontogeny, whereas the reverse was true during early ontogeny.

In interpreting these results, an important limitation of the current study warrants consideration. The automated segmentation macro in Optimas provided the most reliable method available for the quantitative analysis of spatial variability in secondary osteonal bone within cross-sections. However, this method did not discriminate between cortical regions of periosteal vs. endosteal origin. Further, the size of individual bone segments was influenced by overall cross-section size and shape (as discussed previously by Goldman et al. 2003). As a result, bone segments defined using this procedure did not represent developmentally homologous bone regions in all comparisons; tissues of endosteal and periosteal origin were variably represented in inner and middle rings, depending upon specimen age, element and taxon. Due to this limitation, observations of aspects of primary bone microstructural development in this sample provide an additional and relevant qualitative context in which to consider variability in secondary osteonal bone across circumferential rings.

Prevalence of secondary osteonal bone in cortical regions of endosteal origin

Consistent with predictions of this study, qualitative assessment revealed that secondary osteons were concentrated in earlier-deposited regions of the periosteal cortex (i.e. located deeper in the cortex), although they were also observed locally in outer cortical regions adjacent to the periosteal surface in some taxa and skeletal sites (e.g. the lateral femoral cortex in some older P. troglodytes individuals, and the posterior femoral cortex of all taxa examined here). However, another interesting pattern emerged; the presence and regional distribution of endosteally deposited tissues also had an important influence on the distribution of secondary osteons within bone cross-sections. Secondary osteonal bone was often concentrated in cortical regions of endosteal origin, particularly in younger age classes where evidence of bone deposition at the endosteal surface was often prevalent (McFarlin, 2006, in prep.). Further, where the density of secondary osteons permitted assessment of microstructural organization in interstitial bone areas, they were often observed in regions of endosteal compacted coarse cancellous tissue. When present, this latter tissue type always exhibited at least some intracortical remodeling, even in those skeletal sites and age groups which showed comparatively low proportions of secondary osteonal bone overall. As endosteal tissues may be older or more recent in their depositional history than adjacent periosteal cortex (Enlow, 1962b, 1963), the significance of tissue age for understanding intracortical remodeling of endosteal regions as a whole is not entirely clear.

Results of this study are in agreement with prior work by Enlow (Enlow, 1962b, 1963, 1976), which demonstrated that intracortical remodeling in developing long bones was predictably located in regions undergoing endosteal growth, most prominently in regions of compacted coarse cancellous bone. This latter tissue type is formed in bone regions originally comprising coarse cancellous trabeculae; available spaces between trabeculae are in-filled or ‘compacted’ by subsequent endosteal deposition, giving this tissue type its characteristically convoluted appearance. The presence of endosteal compacted coarse cancellous bone is indicative of specific developmental circumstances, including cortical drift (i.e. involving coordinated patterns of resorption and deposition on complementary bone surfaces to effect bone movement through tissue space) and/or growth remodeling of former metaphyseal regions as they are incorporated into the diaphyseal cortex during linear bone growth (i.e. accomplished by periosteal resorption and endosteal deposition). Subsequent endosteal deposits are organized circumferentially as trabeculae projecting from the endosteal surface become less numerous and, as a result, compacted coarse cancellous bone areas are displaced towards mid-cortical regions.

In the current study, we observed that older-deposited endosteal tissues (i.e. those located deeper in the endosteal cortex, adjacent to the periosteal reversal boundary) commonly exhibited a compacted coarse cancellous organization and more secondary osteonal bone than endosteal circumferential bone of more recent depositional origin (i.e. located adjacent to the endosteal surface) (see examples in Fig. 3). This pattern contributed to the increased proportion of secondary osteonal bone observed in the middle ring compared to inner and outer rings during early ontogeny, particularly in H. lar. With expansion of the medullary canal in older age classes as a result of bone resorption from the endosteal surface, intracortically remodeled areas formerly located in the middle cortex were sequentially relocated to inner cortical regions; thus, proportions of secondary osteonal bone increased in the inner cortical ring, and endosteal compacted coarse cancellous bone was generally not observed in interstitial areas.

Qualitative observations of C. aethiops were consistent with a prior investigation by Paine (Paine, 1994; Paine & Godfrey, 1997), which reported significantly increased proportions of secondary osteonal bone in anterior and lateral quadrants of the midshaft humerus, and in the posterior quadrant of the midshaft femur, of vervets and other Old World monkeys. Results of current study reveal the manner in which bone growth history may influence these patterns. The prevalence of secondary osteonal bone in endosteally deposited cortex was especially obvious in C. aethiops, where secondary osteons appeared less densely distributed overall and compacted coarse cancellous bone was often prominent. Regional patterning of intracortical remodeling observed at the midshaft humerus was associated with a highly asymmetric bone growth pattern, characterized by cortical drift in a predominantly posteromedial direction during early ontogeny (as was also reported for the midshaft humerus of pig-tailed macaques by Newell-Morris & Sirianni, 1982). Tissues deposited at the endosteal surface formed a large proportion of cortical area in anterior, lateral and anteromedial cortices (depending upon age); as a result of continued endosteal growth in these cortices, intracortical remodeling along the endosteal-periosteal boundary was relocated towards mid-cortical and outer cortical regions. In this manner, pronounced cortical drift at the midshaft humerus contributed to the overlap reported among rings in the proportion of secondary osteonal bone, and it appeared to have a lasting influence on the distribution of secondary osteonal bone in adults. Secondary osteonal bone in anterior, lateral and anteromedial cortices was found in regions of endosteal origin and regions of periosteal cortex interpreted to be of older depositional history (McFarlin, 2006).

At the midshaft femur of C. aethiops, older age classes revealed a pattern in which secondary osteons in the inner ring were concentrated in endosteal compacted coarse cancellous regions, whereas secondary osteons in the outer ring were restricted in large part to periosteally deposited regions of the posterior cortex adjacent to the linea aspera. This pattern of increased intracortical remodeling at the linea aspera of the midshaft femur, a prominent muscle attachment site, also characterized H. lar and P. troglodytes, and has been reported previously in humans and other primates (Goldman et al. 2003; Warshaw, 2007). It is interesting to note that in both elements of C. aethiops examined here, secondary osteons were observed in regions of endosteal compacted coarse cancellous bone irrespective of the position of this latter tissue type around the circumference of the endosteal margin. (Note individual variability in the distribution of this tissue type, evident in Fig. 3). However, intracortical remodeling of periosteally deposited cortex appeared to be more consistent in its distribution around the cortex (e.g. posteriorly at the midshaft femur, anterolaterally at the midshaft humerus of C. aethiops).

Although Enlow (1962b, 1963, 1976) discussed the role of intracortical remodeling of compacted coarse cancellous bone in regions adjacent to soft tissue attachments (e.g. muscle attachment sites), the underlying factors accounting for the prevalence of secondary osteonal bone in endosteal regions more generally remained poorly understood. Few investigators have since commented specifically on the presence of secondary osteons in bone regions of endosteal origin (Bouvier & Hylander, 1981; Bromage, 1992; Currey, 2002; Warshaw, 2007). Warshaw (2007) also reported a tendency for endosteal compacted coarse cancellous bone to exhibit evidence of intracortical remodeling in strepsirhine and New World anthropoid primates, even in those taxa and/or bone elements showing few secondary osteons elsewhere in the cortex. Further, a similar prevalence of intracortical remodeling in cortical regions of endosteal vs. periosteal origin was noted for the midshaft femur of human juveniles (Goldman et al. in press). Whereas tissue age may contribute to the prevalence of intracortical remodeling in bone regions of endosteal origin, other factors may also influence this pattern, as considered below.

Possible factors contributing to intracortical remodeling of endosteal cortical regions

A number of explanations, in addition to tissue age, have been put forth to account for the prevalence of intracortical remodeling in inner cortical regions more generally, or endosteal regions specifically. It has been proposed that intracortical remodeling of inner bone areas may represent an elevated secondary remodeling response to mechanical strains existing below a theoretical strain threshold (e.g. Frost, 1990; Skedros et al. 2001). Such a response in low-strain regions would increase porosity and reduce bone mass, thereby serving to increase mechanical strains to levels appropriate for normal bone functioning. However, in the current study, secondary osteons were not always located immediately adjacent to the endocortical surface, where strain magnitudes are predicted to be lowest in bones experiencing pure bending (Biewener, 1992); in some age classes and skeletal sites examined, they occupied a mid-cortical location. Alternatively, intracortical remodeling of inner bone areas may serve to realign collagen in response to prevailing strain conditions (e.g. Ascenzi & Bonucci, 1967, 1968; Portigliatti Barbos et al. 1984; Boyde & Riggs, 1990; Bromage, 1992; Riggs et al. 1993). This may be particularly important in regions of endosteal compacted coarse cancellous bone, which is typically more disordered in its microstructural arrangement and characterized by reduced impact strength and work to fracture compared to other tissue types (Currey, 2002). However, if such remodeling were mechanically mediated, this might be more important in regions of high strains adjacent to the periosteal surface. Mechanical interpretations of increased secondary osteonal bone in inner and/or endosteal bone regions are further complicated by observations of prior studies, which have reported a lack of concordance between measured trans-cortical mechanical strain gradients and regional patterning of intracortical remodeling and other microstructural features across endosteal vs. outer cortical regions (Skedros et al. 1996, 2001). Bromage & Boyde (1998, unpublished data) reported that compacted coarse cancellous bone in the mandibular corpus of macaques exhibited no evidence of intracortical remodeling, despite observations that preferred collagen orientation in these regions contrasted sharply with the mechanical strain environment. These authors suggested that species and bone-specific developmentally constrained construction rules governing bone microstructure during growth may influence the propensity for different tissue types to undergo mechanically mediated intracortical remodeling later in life. This is consistent with observations that other mechanically relevant bone microstructural variables (namely collagen fiber orientation) appear to show important differences among primary and secondary bone tissue types (Riggs et al. 1993; McMahon et al. 1995; Goldman et al. 2003).

Metabolic factors and/or proximity to the marrow may also be relevant for understanding the preferential distribution of intracortical remodeling in inner bone regions (Frost, 1986, 1998; Parfitt, 2001; Skedros et al. 2001), where it may have only minimal impact on measures of bone strength (e.g. Burr et al. 2001). The role of secondary remodeling in mineral homeostasis is primarily fulfilled at cancellous and endocortical bone surfaces adjacent to hematopoietic marrow, although intracortical remodeling may contribute to mineral homeostasis during periods of increased demand, such as may occur during growth (Parfitt, 1994b, 2001, 2002a). The factors underlying the association between secondary remodeling and hematopoiesis are not fully understood, although they have been the focus of recent attention (e.g. see van Dyke, 1967; Compston, 2002; Kacena et al. 2006). However, the unique developmental circumstances under which endosteal compacted coarse cancellous bone is formed suggests that metabolic factors may be particularly relevant for understanding the tendency of this tissue type to undergo intracortical remodeling after its initial compaction. Hematopoietic bone marrow is distributed throughout the skeleton at birth, although its distribution is confined to the axial skeleton, pelvis and the proximal ends of long bones in adult mammals (van Dyke, 1967; Compston, 2002). Intracortical remodeling for the purposes of mobilizing calcium stores may also be expected to vary with age and skeletal site, accordingly.

One problem that has received very little attention concerns the possible role of intracortical remodeling in providing vascular continuity. The dual blood supply to the cortex of long bone diaphyses, from both medullary vessels (i.e. derived from the nutrient artery) and vessels entering from the periosteal surface, has received considerable attention. However, the relative importance of these two blood supplies has been debated, and it has been suggested that the vascular supply of cortical bone exhibits flexibility depending on a number of factors, including age, pathology, and local circumstances (e.g. proximity to muscle attachments and their blood supply) (Rhinelander, 1968; Skawina et al. 1994; Bridgeman & Brookes, 1996). A vascular network in bone is formed by anastomoses between medullary and periosteal circulations, and with vessels contained within secondary osteons (e.g. also see Trias & Fery, 1979; Marotti & Zallone, 1980; Skawina & Gorczyca, 1984; Arsenault, 1990). However, the nature of the vascular interface at the reversal boundary between periosteally deposited and endosteally deposited cortex remains poorly understood, as is the fate of blood vessels contained within primary vascular canals that are subsequently removed from resorptive bone surfaces, as may occur during metaphyseal remodeling and cortical drift. In the current study, we observed that secondary osteons were often concentrated along the endosteal-periosteal reversal boundary. Whether secondary osteons may serve an important and basic growth function in maintaining vascular continuity on resorptive bone surfaces, and across the reversal boundary between cortical regions of endosteal and periosteal origin, warrants further study.

Summary and implications of this research

This study demonstrated significant regional variability in the proportion of secondary osteonal bone among three circumferential rings (representing the inner, middle and outer cortex) at the midshaft femur and humerus in ontogenetic series of Chlorocebus (Cercopithecus) aethiops and Hylobates lar, and at the midshaft femur of Pan troglodytes. Within cross-sections, proportions of secondary osteonal bone were usually higher in inner and middle cortical rings than in the outer cortical ring, providing support for the hypothesis that tissue age is an important factor influencing the distribution of intracortical remodeling in bone. However, the distribution of secondary osteonal bone across inner vs. middle cortical rings showed a more complicated pattern, varying with age, taxon and skeletal site. The distribution of endosteally deposited cortex had an important influence on the regional patterning of secondary osteonal bone.

One of the most interesting patterns to emerge in this study was the prevalence of secondary osteonal bone in cortical regions of endosteal origin during ontogeny, most notably in regions of endosteal compacted coarse cancellous bone. When present, the latter tissue type always exhibited some degree of intracortical remodeling irrespective of its cortical location, even in those taxa and age groups which exhibited comparatively low proportions of secondary osteonal bone overall. The distribution of endosteal compacted coarse cancellous bone reflects local element-specific circumstances of cortical drift and metaphyseal remodeling (Enlow, 1962b, 1963, 1976); its presence varies among taxa, with age, within and among bone elements of differing shape and overall size, and it may have a lasting influence on bone microstructure and the distribution of secondary remodeling in the adult skeleton (Enlow, 1963; Bromage & Boyde, 1998; McFarlin, 2006; Warshaw, 2007).

Prior researchers have shown that intracortical remodeling serves an important function in maintaining the biomechanical integrity of bone (Frost, 1990; Martin et al. 1998; Currey, 2002; Pearson & Lieberman, 2004). Results of the current study lend support to previous arguments that local circumstances of bone growth may also influence the distribution of intracortical remodeling in ways that are not related in an obvious manner to local mechanical strain history (Enlow, 1962b). The metabolic and/or developmental function of intracortical remodeling has yet to be studied across many different skeletal locations, and in a broad range of vertebrate taxa. Future studies would benefit by incorporating more proximal and distal section levels along long bone diaphyses (i.e. regions expected to show more endosteal compacted coarse cancellous bone), vital fluorescent labeling protocols to examine dynamic parameters of bone growth and remodeling, and consideration of the developmental origin of primary bone regions that undergo intracortical remodeling later in life. The extent to which such developmental factors may interact with local mechanical influences to produce regional patterning of secondary osteonal bone within and among bone elements warrants future research.

Acknowledgements

This research benefited greatly from discussions with and assistance from a number of people, including Johanna Warshaw, Haviva M. Goldman, Debra Bolter, Eric Delson, Fred Szalay, and Chet Sherwood, and from students and other researchers of the Hard Tissue Research Unit. We also thank Haviva Goldman, Chet Sherwood and two anonymous reviewers for comments on earlier drafts of this manuscript. Skeletal materials from the Sherwood L. Washburn collection at the University of California Santa Cruz, the Museum of Comparative Zoology of Harvard University, and the Museum für Naturkunde of the Humboldt University were kindly made available for the current study; numerous individuals at the latter institutions also provided valuable assistance, including Maria Rutzmoser, Judy Chupasko, Manfred Ade, Peter Giere, Andrea Mess, Detlef Wilborn, and Irene Thomas. This research was supported by the National Science Foundation (SBE-DDIG; funding to NYCEP), Leakey Foundation, and the City University of New York (Robert E. Gilleece Fellowship; Dissertation Fellowship). Support to S.M. was also provided by funding from The George Washington University's Selective Academic Excellence Initiative to the Center for the Advanced Study of Hominid Paleobiology.

Ancillary