The structure of pedicle and hard antler bone in the European roe deer (Capreolus capreolus): a light microscope and backscattered electron imaging study

Authors


Correspondence

Uwe Kierdorf, Department of Biology, University of Hildesheim, Marienburger Platz 22, 31141 Hildesheim, Germany. E: uwe.kierdorf@uni-hildesheim.de

Abstract

Deer antlers are deciduous bony structures that develop from permanent frontal outgrowths, the pedicles. While growth and bone architecture of antlers have been studied in greater detail, information on pedicle formation and structure is scarce. The present study provides information on the structure of pedicle and hard antler bone in the European roe deer. A pronounced seasonal variation in pedicle architecture was observed, with high porosity around antler casting and a very compact bone structure during the hard antler stage. These observations suggest a corresponding marked variation also in the biomechanical properties of the pedicles. The seasonally alternating extensive resorption and formation processes make the pedicles of older deer heavily remodeled structures. Pedicles increase in thickness by apposition of primary bone that subsequently becomes replaced by secondary osteons. The antler cortex of roe deer is largely composed of a tubular framework of woven bone trabeculae with some remnants of mineralized cartilage, and primary osteons that have filled in the intertrabecular spaces. Secondary osteons are scarce, denoting little remodeling in antlers, which can be related to their short lifespan. The occurrence of cement lines around primary osteons indicates resorption on the trabecular scaffold prior to infilling of the intertrabecular spaces. The outer cortex showed a higher autofluorescence and a more immature structure than the main cortex, suggesting that it was secondarily formed by periosteal activity. Pedicles and antlers constitute a functional entity, and future histological and/or biomechanical studies should therefore consider both components of the cranial appendages.

Introduction

Antlers are regularly replaced bony cranial appendages that are present in all extant deer species with the single exception of the Chinese water deer (Hydropotes inermis), whose lack of antlers is considered a derived condition (Goss, 1983; Groves, 2007). Antlers develop on top of permanent outgrowths of the frontal bones, referred to as pedicles (Goss, 1983), and are generally grown only by male deer. However, in the reindeer (Rangifer tarandus) also females bear antlers, which are smaller than those of bulls (Lincoln, 1992). Their rapid growth and periodic regeneration make antlers a good model for studying bone formation and regeneration in mammals (Goss, 1984; Bubenik, 1990; Price et al. 2005; Kierdorf et al. 2009; Kierdorf & Kierdorf, 2011).

Male deer use their antlers primarily for display, territorial marking and intraspecific fighting with other males during the rut (Clutton-Brock, 1982). When used for marking and fighting, antlers are bare bony structures, referred to as hard antlers, which have dried out (Currey et al. 2009) following the shedding of the skin (velvet) that envelops them during growth (Goss, 1983). Hard antlers are well suited mechanically for their function in intraspecific fighting, during which they are subjected to high impact loading and large bending moments (Chen et al. 2009; Currey et al. 2009; Launey et al. 2010).

Antler growth occurs by a specific form of endochondral ossification (Gruber, 1937; Banks, 1974; Banks & Newbrey, 1983; Kierdorf et al. 1995a; Szuwart et al. 1995). The tips of growing antlers are populated by rapidly proliferating mesenchymal cells (Wislocki, 1942; Banks, 1974; Kierdorf et al. 1995a, 2007; Price et al. 1996, 2005; Colitti et al. 2005; Cegielski et al. 2009). A mesenchymal growth zone is present at the tip of the main beam and of each of the tines that during antler growth successively branch off from the main beam.

Further proximally, the mesenchymal progenitor cells differentiate into chondroblasts, and a framework of hyaline cartilage is formed. This consists of longitudinally oriented, ramifying trabeculae that are separated by blood vessels and associated perivascular connective tissue (Gruber, 1937; Banks, 1974; Muir et al. 1988; Kierdorf et al. 1995a; Szuwart et al. 1995; Li et al. 2002, 2005; Price et al. 2005; Clark et al. 2006; Cegielski et al. 2009; Krauss et al. 2011). The latter contains mesenchymal cells that can differentiate into chondro- and osteoblasts (Banks & Newbrey, 1983; Kierdorf et al. 1995a). The cartilaginous trabeculae increase in thickness in more proximal locations by apposition of new cartilage onto their surface (Banks & Newbrey, 1983; Kierdorf et al. 1995a). The cartilage matrix is then mineralized and the cartilage resorbed, and woven bone is laid down on the scaffold provided by the remnants of the mineralized cartilage (Gruber, 1937; Banks, 1974; Muir et al. 1988; Kierdorf et al. 1995a; Szuwart et al. 1995, 1998; Price et al. 1996, 2005; Krauss et al. 2011). The combined chondroclastic and osteoblastic activities cause the replacement of the initially cartilaginous framework by a bony one exhibiting the same orientation. A contiguous growth plate and secondary ossification centers, which are typical features of endochondral ossification of mammalian long bones, are not present during antler growth (Gruber, 1937; Banks, 1974; Banks & Newbrey, 1983). In red deer (Cervus elaphus), remnants of mineralized cartilage have been observed within the trabecular framework of the cortex of hard antlers, demonstrating an incomplete replacement of cartilage by bone (Landete-Castillejos et al. 2012a). In addition to endochondral ossification, a bone sleeve is deposited along the periphery of the forming antler by perichondral/periosteal appositional (intramembranous) ossification (Banks, 1974; Kierdorf et al. 1995a, 2007; Price et al. 2005). In red deer, it has been demonstrated that late during antler growth an outer cortical layer is laid down sub-periosteally in the more proximal antler portions, while it is lacking in the distal antler (Gomez et al. 2013). Contrary to long bones, there is thus only very limited secondary growth in diameter of antlers.

Filling of the intertrabecular voids with primary osteons leads to the formation of a compact antler cortex (Rajaram & Ramanathan, 1982; Chen et al. 2009; Launey et al. 2010; Krauss et al. 2011; Landete-Castillejos et al. 2012a; Gomez et al. 2013). A recent labeling study in red deer demonstrated that the infilling process of primary osteons in the lower and middle thirds of the antler lasts only about 30–40 days due to a high mineral apposition rate (Gomez et al. 2013). Secondary osteons have been reported to be scarce in antlers (Chen et al. 2009; Launey et al. 2010; Krauss et al. 2011; Gomez et al. 2013), which reflects the short lifespan of the antler bone. Antlers lack a central, extended medullary cavity. Instead, the central portion of the antler shaft consists of cancellous bone (Muir et al. 1988; Chen et al. 2009). However, the tips of the main beam and tines are typically formed entirely of compact bone.

The developmental control of antler morphogenesis is still poorly understood. Regarding antler shape, it is probably the size (cell number) of the regeneration bud (‘blastema’) on the casting plane of the pedicle that ultimately determines the number of tines that branch off from the main beam during further growth. Circumstantial evidence in support of this view is the clear link between antler size and the number of points (tines) of an antler. As was already emphasized by Goss (1983), normally shaped antlers, i.e. antlers with typical numbers of points, are not known to occur in miniature. If antler size is reduced, so is the branching pattern, and therefore smaller antlers have fewer tines than larger ones. This suggests that splitting of an antler growth zone can only occur if the number of mesenchymal cells in this zone exceeds a certain threshold. A corresponding relationship between initial cell number in a (limb) bud and differentiation pattern has, for instance, been discussed also for the formation of extra digits in dogs (Alberch, 1985).

So far, studies on the bone architecture of hard antlers were predominantly conducted in larger species, especially red deer and fallow deer (Dama dama; Kierdorf et al. 1993, 2000, 2004a; Rolf & Enderle, 1999; Chen et al. 2009; Launey et al. 2010; Krauss et al. 2011; Landete-Castillejos et al. 2012a; Gomez et al. 2013), while information on the structure of hard antlers in smaller deer species is scarce. There is also only limited information on the structure of pedicle bone in deer. This may be because studying pedicles requires the death of the animal, while antlers can be obtained from living individuals. Most of the studies on pedicle bone in deer focused on the osteoclastic activity in the pedicle that precedes antler casting and the restorative events occurring after casting (Kölliker, 1873; Gruber, 1937; Wislocki, 1942; Waldo & Wislocki, 1951; Goss et al. 1992; Kierdorf et al. 2003), while the process of formation of pedicle bone and its mineralization received less attention (Gruber, 1937; Kierdorf et al. 2000). A comparison of the bone architecture of pedicles and antlers is of particular interest because pedicles are permanent structures consisting of wet bone, whereas functional antlers are composed of dry bone (Currey et al. 2009). Moreover, while there is little time for remodeling in antlers, there is no such time constraint for pedicles.

This paper reports our findings on the structure of hard antlers and pedicle bone in the European roe deer (Capreolus capreolus), using different light microscopic techniques as well as qualitative backscattered electron (BSE) imaging in the scanning electron microscope (SEM).

Materials and methods

Specimens

Samples of hard antler and pedicle bone were obtained from dry skull caps of five roe bucks (individuals nos 1–5) that had either been shot (n = 2) or killed in traffic accidents (n = 3) in Western Germany. Dates of death were known for four individuals (nos 1–4), while in individual no. 5 death was estimated to have occurred in summer/early autumn (Table 1). Based on the overall morphological appearance of pedicles and antlers and on pedicle diameters, one individual was classified as a yearling, two bucks were classified as 2 year olds, and two others as aged 3 years or older (Table 1). In addition, we studied the pedicles of an adult roe buck (3 years or older, individual no. 6) that had been killed shortly after antler casting. The exact date of death was also not available for this buck. All analyzed specimens were part of the osteological collection of the Department of Biology, University of Hildesheim, and had been stored at room temperature for several years prior to the study.

Table 1. Age class, date of death, pedicle diameter (anterior-posterior axis) and antler length of the roe bucks
Deer numberAge classDate of deathPedicle diameter (mm), left/rightAntler length (mm), left/right
1Yearling19 May12.8/12.7125/128
22 years15 June16.1/16.9158/162
32 years25 July13.9/13.8126/117
4≥ 3 years13 August25.0/25.0222/223
5≥ 3 yearsUnknown (estimated summer or early autumn)20.0/19.0220/185 (tip broken)
6≥ 3 yearsUnknown (estimated October to December)18.3/18.0No antlers

The European roe deer is the smallest deer species in Europe, with a live body mass of adult males typically ranging between 20 and 30 kg (Andersen et al. 1998). Roe bucks are unusual in that they grow their antlers during late autumn and winter, while in other deer species from temperate climates, antler growth occurs during spring and summer. The first set of antlers (tiny knobs), sometimes referred to as infant antlers, is produced in the autumn of the year of birth, and therefore the antlers carried by yearling roe bucks are already the second set (Stubbe, 1997). Adult roe bucks cast their antlers in the period October–December, and velvet shedding occurs in March–April (Stubbe, 1997). In captive individuals, antlers reached their final length after about 60 days of growth, the average growth rate being 3.3 mm day−1. This growth phase was reported to be followed by a 45-day period of ‘intense mineralization’ (Sempéré, 1990). This author provides no information on the histological events associated with this mineralization process, but it may be assumed that what he describes is related to cortical compaction caused by infilling of the intertrabecular voids with primary osteons. For free-ranging roe bucks ≥ 2 years old, Stubbe (1997) reported an average daily antler growth rate of 2.8 mm (range 1.3–5.6 mm). Roe bucks may show quite large variation in antler shape between ages, but there are usually not more than three tines per antler (Stubbe, 1997).

Microscopic analysis

For microscopic analysis, about 10-mm-thick, full-diameter cross-sectional discs of pedicle and antler bone (from the right side) were obtained with a fine-toothed saw from the skull caps of roe bucks nos 1–5. Pedicle sections were collected at about the midpoint of pedicle height (measured at the medial side), while antler samples were obtained from two sites (Fig. 1). These were located: (i) at approximately half the distance between the antler–pedicle junction and the origin of the anterior tine (‘proximal antler samples’); and (ii) close to the tip of the main beam (‘distal antler samples’). In roe buck no. 6, both pedicles were removed from the cranium at the level of the skull roof. In contrast to the other individuals, the pedicles of buck no. 6 showed signs of marked resorption. To study this in greater detail, three 10-mm cross-sectional slices were obtained from the left pedicle at different heights (distal, middle and proximal location), while the right pedicle was transversely cut at half height. The undecalcified samples were then either embedded in Biodur E12 epoxy resin (Biodur Products, Heidelberg, Germany) or in poly-methyl-methacrylate (PMMA). In roe bucks nos 1–5, each embedded bone disc was further bisected horizontally to yield two slices.

Figure 1.

Approximate location of the sampling sites of pedicle and antler bone in roe bucks nos 1–5. Note the bony protuberances present on the antler surface in the more proximal areas of the antler.

For BSE imaging, one of the cut surfaces was smoothed and polished using a series of silicon carbide papers (up to grade 4000), followed by a final polishing step using a leather cloth and a polishing compound (Menzerna, Ötigheim, Germany). The uncoated polished surfaces were examined in low vacuum mode with a solid-state BSE detector in an FEI Quanta 600 FEG SEM (FEI Company, Hillsboro, OR, USA) operated at an accelerating voltage of 20 kV. BSE imaging in the SEM is based on the detection of electrons from the primary electron beam accelerated into the specimen and backscattered in a surface layer after elastic collision with the atomic nuclei of the sample material. The probability of backscattering depends largely upon the energy of the primary electrons and the atomic number of the atoms hit by the electron beam and, at fixed primary electron energy, backscattering increases as the square of the atomic number (Skedros et al. 1993; Roschger et al. 1998; Karney et al. 2011). Several studies have demonstrated that gray-level variation in BSE images of mineralized tissues is largely dependent upon changes in mineral content, with brighter gray levels representing a relatively higher mineral content (Boyde & Jones, 1983; Skedros et al. 1993; Roschger et al. 1998). BSE imaging of polished bone surfaces in the SEM therefore provides simultaneous information on tissue architecture and mineralization at a high spatial resolution (Skedros et al. 1993, 2005). Individual gray-level images were converted to pseudocolor images using the 16-color lookup table of the software package ImageJ (NIH, USA).

For light microscopy, the Biodur-embedded antler and pedicle bone samples were polished and mounted with their polished sides down on glass slides, using the epoxy resin as glue. The mounted specimens were then sectioned to a thickness of about 300 μm, using a rotary saw with a water-cooled diamond-edged blade (Woko 50; Conrad Apparatebau, Clausthal-Zellerfeld, Germany), and subsequently ground and polished to a thickness of either 50 or 70 μm. Ground sections (50 μm) of the PMMA-embedded samples were prepared as described previously (Landete-Castillejos et al. 2012a).

Some of the antler and pedicle sections were stained with toluidine blue or toluidine blue-pyronin G after a brief etching pretreatment with 5% formic acid or decalcification in 1% periodic acid, some others were directly stained with a 5% aqueous silver nitrate solution (von Kossa method). Stained and unstained sections were examined in a Zeiss Axioskop 2 Plus microscope (Carl Zeiss, Jena, Germany) in normal transmitted light or in linearly polarized light (LPL), partly with the use of a full-wave (1λ) retardation plate (= first-order red plate), producing an optical path length difference of an entire wavelength (LPL + 1λ-plate). Unstained sections were also examined in a Keyence Biozero 8000 epifluorescence microscope (Keyence, Osaka, Japan) equipped with an OP-66836 GFP-BP filter set (excitation wavelength 470/40 nm, recorded emission wavelength 535/50 nm).

Results

Pedicle bone

Bone structure differed strongly between the pedicles of the individuals that had died during the hard antler period (roe bucks nos 1–5) and that of roe buck no. 6, which had died shortly after antler casting. The latter's pedicles exhibited evidence of marked bone resorption already on external inspection. Signs of resorption were not limited to the separation plane, but also present along parts of the pedicle shaft. Bone resorption had been particularly extended at the posterior surface of the pedicle, where traces of erosion were present along the whole length of the pedicle shaft down to the level of the skull roof (Fig. 2a). On the other sides, resorption had predominantly affected the more distal portions of the pedicle shaft.

Figure 2.

(a–d) Macroscopic and microscopic appearance of pedicles of roe buck no. 6 (≥ 3 years), killed shortly after antler casting. (a) Posterior-superior view of the pedicles; note the saucer-like depression of the casting plane and extension of the resorption zone along the entire pedicle shaft posteriorly. (b) Low-power micrograph (brightfield image) of decalcified transverse ground section of the left pedicle stained with toluidine blue-pyronin G. Due to the presence of numerous resorption spaces, the pedicle has a distinctly porous appearance. (c) BSE image of the outer portion of the transversely sectioned right pedicle demonstrating extended resorption spaces, some of which open onto the pedicle surface; thin arrow: scalloped bone surface from osteoclastic resorption; arrowhead: smooth surface indicating bone deposition onto the resorption surface; broad arrow: cement line; asterisks: peripheral rim of woven bone. (d) BSE image of the more central portion of the transversely sectioned right pedicle exhibiting extended resorption spaces; a mixture of older, more highly mineralized (brighter) and younger, less mineralized (darker) bone areas with cement lines (example marked by broad arrow) appearing as particularly bright (relatively hypermineralized) thin lines is present; thin arrows: scalloped bone surfaces; arrowhead: smooth surface indicative of bone deposition onto the former resorption surface.

Microscopic analysis of pedicle sections from individual no. 6 showed that extensive bone resorption had occurred not only along the outer (periosteal) surface but also in the interior of the pedicle (Fig. 2b–d). The numerous resorption spaces within the pedicle bone gave it a distinctly porous appearance (Fig. 2b). The percentage of the cross-sectional area covered by pores was 23% in the left pedicle of roe buck no. 6 (mean of the measurements on the three sections), compared with < 5% in the pedicles of roe bucks nos 1–5. In the pedicle of roe buck no. 6, the peripherally located resorption spaces frequently opened via broad connections onto the external surface, resulting in an irregular embayment of the resorption surface (Fig. 2c).

Parts of the surface of the resorption spaces exhibited an irregularly scalloped outline due to the presence of Howship's lacunae (Fig. 2c,d). Other stretches of these surfaces were smooth, indicating that following resorption some apposition of new bone had already occurred in these locations (Fig. 2c,d). In BSE images, the layers of newly deposited bone underneath these smooth surfaces appeared darker than the surrounding bone, indicating their relatively lower degree of mineralization. The findings indicate that in some places bone formation was already under way at the time of death of the buck, while in other locations this was not the case and resorption was either still ongoing or had only recently stopped.

The pedicle bone tissue of roe buck no. 6 consisted of a mixture of, mostly partly eroded, secondary osteons interspersed with areas of older, more highly mineralized (brighter on BSE images) interstitial bone of varying shape and size (Fig. 2c,d). Numerous cement (reversal) lines were present that represented the material initially deposited onto the resorption surfaces, which then could be referred to as reversal surfaces (cf. Skedros et al. 2005). In the BSE images, the cement lines appeared as particularly bright (relatively hypermineralized) lines with a typical scalloped course (Fig. 2c,d). In the periphery of the pedicle, woven bone was present (Fig. 2c). The overall impression was that the pedicle bone of roe buck no. 6 had undergone repeated rounds of resorption and subsequent infilling of resorption spaces with secondary osteons, and had just started a new period of infilling when the animal died.

Contrary to roe buck no. 6, the pedicles of roe bucks nos 1–5, which had all been killed during the hard antler phase, were of a dense structure. In all specimens, a relative thin outer zone (Figs 3 and 4) that represented the bone laid down onto the pedicle surface following the latest period of resorption could be distinguished from an inner zone (Fig. 5) that composed most of the pedicle.

Figure 3.

(a–f) Microscopic appearance of transversely sectioned pedicle of roe buck no. 1 (yearling). (a) BSE image of the outer pedicle zone with elongate vascular compartments (asterisks) partly filled with lamellar bone and centripetally adjacent zone composed of a bony network filled with primary osteons. Arrowheads: thin sheets of more highly mineralized bone. (b) Higher magnification BSE image of the outer pedicle zone shown in the upper left corner of (a). Asterisk: sheet of more highly mineralized bone; arrowhead: lamellar bone filling in a vascular compartment; arrow: signs of bone resorption (scalloped surface) in the deeper portion of the outer pedicle zone. (c) Outer pedicle zone showing sheets of woven bone (asterisks) and primary osteons (PO) that have filled in vascular compartments. SO, secondary osteon that has replaced fibrolamellar bone. Brightfield image of decalcified ground section stained with toluidine blue-pyronin G. (d) Outer pedicle zone showing woven bone (asterisk) as well as primary (PO) and secondary (SO) osteons. The latter have replaced the fibrolamellar bone. Undecalcified and unstained ground section, LPL + 1λ-plate. (e) Outer pedicle zone showing woven bone trabeculae (asterisks), primary osteons (PO), and more centrally located secondary osteons (SO) that have replaced the fibrolamellar bone. The secondary osteons are lined by cement lines (arrowheads). LPL image of decalcified ground section stained with toluidine blue-pyronin G. (f) Outer pedicle zone showing a primary osteon (PO) and secondary osteons (SO). The secondary osteon in the center of the image exhibits an elongate shape with three vascular canals and is delimited by a cement line (arrowheads) from the surrounding woven bone (asterisks). The arrow marks a young (wide vascular canal) second-generation secondary osteon that has replaced woven bone but also a part of the elongate first-generation secondary osteon. Undecalcified and unstained ground section, LPL + 1λ-plate.

Figure 4.

(a–f) Outer zone of transversely sectioned pedicles of older roe bucks. (a) Pedicle of roe buck no. 2 (2 years); the outer pedicle zone is composed of fibrolamellar bone. Arrowheads: vascular compartments at an early stage of infilling by primary osteons; asterisk: radially oriented bundles of collagen fibers (Sharpey's fibers). Brightfield image of decalcified ground section stained with toluidine blue-pyronin G. (b) BSE image of the outer pedicle zone of roe buck no. 2 (2 years) showing a primary osteon (PO) that has formed within an elongate cavity of the newly apposed bone layer, whose extension is indicated by the double-arrowed line. Asterisk: elongate primary osteon with multiple vascular canals. (c) BSE image of the pedicle of roe buck no. 4 (≥ 3 years); the extension of an outer zone of fibrolamellar appearance is indicated by a double-arrowed line. Asterisks: secondary osteons. (d) Higher magnification BSE image of the outer pedicle zone of roe buck no. 4 (≥ 3 years). The peripheral bone layer (extension indicated by double-arrowed line) apposed onto the former resorption surface, whose position is indicated by a thin bright (cement) line, exhibits a woven architecture. Arrows: secondary osteons (the leftmost with two vascular canals); arrowhead: primary vascular canal. (e) Pedicle of roe buck no. 5 (≥ 3 years). The outermost layer exhibits a fine lamellar structure (arrowhead). Asterisks: radially oriented bundles of collagen fibers (Sharpey's fibers) within the pedicle bone; arrows: secondary osteons. Undecalcified and unstained ground section, LPL. (f) Higher magnification of the outer pedicle zone of roe buck no. 5 (≥ 3 years). Note the outer zone of bone showing a fine lamellar structure (arrowhead) and bundles of radially oriented collagen fibers (Sharpey's fibers, asterisk). Arrow: secondary osteon. Undecalcified and unstained ground section, LPL + 1λ-plate.

Figure 5.

(a–d) Microscopic images of inner zones of transversely sectioned roe deer pedicles. (a) Pedicle of yearling roe buck (no. 1). The inner zone consists largely of primary osteons (PO), with only a few small secondary osteons being present (arrows). Undecalcified and unstained ground section, LPL + 1λ-plate. (b) Pedicle of roe buck no. 5 (≥ 3 years). The histological picture is dominated by secondary osteons belonging to different generations, indicative of several rounds of bone resorption and formation. The arrow indicates the most recent of a group of successively formed secondary osteons, with the partially resorbed older ones positioned to the left of the marked osteon. Undecalcified and unstained ground section, LPL. (c) Pedicle of roe buck no. 3 (≥ 3 years). Note the predominance of large, complex secondary osteons (asterisks), but also the occurrence of small secondary osteons (arrow) and interstitial bone (arrowheads). Brightfield image of undecalcified ground section stained with silver nitrate (von Kossa method). (d) BSE image of the pedicle of roe buck no. 3 (≥ 3 years) showing evidence of marked remodeling. Asterisk: complex secondary osteon; arrow: remnant of older osteon. Note numerous bright (relatively hypermineralized) cement lines (arrowheads).

In the yearling roe buck (no. 1, date of death 19 May), the outer pedicle zone consisted of fibrolamellar primary bone. This was composed of thin, more or less circumferentially oriented sheets of woven (woven-fibered) bone that had originally separated elongate vascular (soft tissue) compartments, and of lamellar bone that had been deposited onto these sheets. The circumferentially oriented bone layers were connected by short, radially oriented struts of bone. The centripetally progressing lamellar infilling of the initially relatively wide vascular compartments had caused a compaction of the outer pedicle zone (Fig. 3a,b). On BSE images, the sheets of woven bone appeared brighter (more highly mineralized) than the lamellar bone (Fig. 3a).

The lamellar bone that had filled in the vascular compartments consisted of peripheral lamellae oriented in parallel, which in the early stages of the infilling process had been deposited along the complete endosteal surface of the vascular compartments, and of more centrally located, more or less circular lamellar systems that had formed around the (axially oriented) blood vessels present in the compartments (Fig. 3b–d). Especially in the outermost pedicle zone, lamellar infilling had not been completed when the animal died. Each (primary) lamellar system formed within a vascular compartment is designated a primary osteon (for use of terminology, see 'Discussion'). The shapes of these primary osteons varied according to those of the vascular compartments that they had filled in and the number of blood vessels present in these compartments.

In the deeper layers of the outer pedicle zone, resorption spaces as well as secondary osteons were regularly observed, indicating remodeling activity (Fig. 3b–f). The secondary osteons that were of variable shape and frequently contained more than one blood vessel (Fig. 3c–f) had partly replaced the primary, fibrolamellar bone. In contrast to primary osteons, the secondary osteons were delimited by cement lines (Fig. 3e,f). Occasionally, first-generation secondary osteons were observed that had been partly replaced by second-generation ones, indicating the rapidity of the remodeling process in the pedicle (Fig. 3f).

The structure of the outer pedicle zones of roe bucks no. 2 (date of death 15 June) and no. 3 (date of death 25 July) consisted principally of fibrolamellar bone, with a few areas composed predominantly of woven bone (Fig. 4a,b). In the outermost layer, the infilling of the vascular compartments by primary osteons had often not been completed at the time of death (Fig. 4a). The outer pedicle zone of roe buck no. 4 (date of death 13 August) consisted partly of fibrolamellar, partly of woven bone (Fig 4c,d). The primary osteons possessed small vascular canals, indicating that the infilling process had been completed at the time of the buck's death. The position of the resorption surface onto which the new bone layer had been deposited was indicated by a cement line (Fig. 4c,d). Due to the complex topography of this surface, the cement lines often followed an involuted, meandering course (Fig. 4d). Contrary to the other individuals, in roe buck no. 5 (date of death unknown), the fibrolamellar bone present in deeper layers of the outer pedicle zone was overlain by an outermost layer exhibiting a lamellar structure (Fig. 4e,f). Numerous radially oriented bundles of collagen fibers (Sharpey's fibers) that had attached the periosteum to the bone were present in the outer pedicle zones of the bucks (Fig. 4a,e,f).

Secondary osteons were frequent in the deeper portion of the outer pedicle zone (Fig. 4c–f). They were delimited by cement lines and typically appeared darker (less mineralized) than primary osteons on BSE images, indicating the younger age of the secondary osteons. Most of the secondary osteons were rather small, had a round cross-sectional shape and possessed a single vascular canal (Fig. 4d–f). However, some exhibited an asymmetric shape and more than one vascular canal (Fig. 4c,d). Some secondary osteons were also observed within the newly deposited bone layer (Fig. 4f), indicating remodeling activity also in the outer pedicle zone.

The inner pedicle zone of roe buck no. 1 consisted of a trabecular framework of woven bone and of roughly circular lamellar systems (Fig. 5a). Those lamellar systems that were not delimited from the trabecular framework by cement lines were classified as primary osteons. They were of a variable shape and often contained more than one vascular canal. Secondary osteons were also present in this zone (Fig. 5a), indicating that remodeling activity had occurred also in the inner pedicle zone of the yearling buck. The secondary osteons were delimited by cement lines and typically of a smaller size than the primary osteons.

The inner pedicle zone of the older roe bucks (nos 2–5) was dominated by the presence of (intact and fragmentary) secondary osteons, while clearly identifiable remnants of the primary bone system were scarce (Fig. 5b–d). The histological picture was thus indicative of repeated alternation between periods of extensive resorption and infilling. In places, the occurrence of multiple cycles of bone resorption and formation was evidenced by the presence of different generations of secondary osteons, of which only the most recently formed was intact (Fig. 5b). The secondary osteons were partly of a ‘typical’ form, i.e. relatively small with a single vascular canal (Fig. 5b,c), while others clearly deviated from this type. For example, large secondary lamellar systems that had formed around two or more vascular canals were present in the inner pedicle zone (Fig. 5c,d). They were bordered by cement lines and had apparently filled the large and variably shaped resorption spaces described previously (Fig. 2c,d). To indicate their difference from the ‘normal’ secondary osteons, we will henceforth refer to these large secondary lamellar systems as complex secondary osteons.

The cross-sectional profiles of primary and secondary osteons indicated that they were basically oriented in the long axis of the pedicle (Figs 3-5). When viewed between crossed polars, the osteons typically appeared dark or showed up in magenta when using the 1λ-plate in addition (Figs 3c,d and 5a,b), thereby denoting that the collagen fibers within their lamellae were oriented largely in the direction of light propagation, that is, along the osteonal long axis. Areas with transverse or oblique orientation of collagen fibers appeared bright in polarized light, respectively, displayed yellow/orange and blue interference colors with the additional use of the 1λ-plate. In contrast to the antlers (see below), remnants of mineralized cartilage were not observed in the roe deer pedicles.

Antler bone

The proximal antler samples consisted of a dense cortex and an inner cancellous portion with a narrow transition zone between them (Fig. 6a,b). Using fluorescence microscopy, the cortex could be divided into a thin outer cortex with a relatively high degree of autofluorescence and a much thicker main cortex showing weaker autofluorescence (Fig. 6c). Where the antler surface exhibited the conspicuous protuberances (pearls) that are characteristic of the more proximal portions of roe deer antlers, the outer cortical zone of high autofluorescence was broadened (Fig. 6c).

Figure 6.

(a–d) Microscopic appearance of transversely sectioned antler bone. (a) Proximal antler of roe buck no. 5 (≥ 3 years). CZ, cancellous zone; MC, main cortex; TZ, transition zone. Arrow: osteoid seam. Brightfield image of undecalcified and unstained ground section. (b) BSE image of the proximal antler of roe buck no. 3 (2 years). Note the change in bone porosity from the MC to the TZ and CZ. The trabecular framework of woven bone (arrowheads) is evident by its higher brightness (relative hypermineralization) compared with the primary osteons that have filled in the tubular spaces lined by this framework. (c) Fluorescence image of the proximal antler of roe buck no. 2 (2 years). An outer cortical zone (OC) of higher autofluorescence can be distinguished from the MC with a lower degree of autofluorescence. In the region of a protuberance of the antler surface (asterisk), the zone of high autofluorescence is broadened. Undecalcified and unstained ground section. (d) BSE image of the distal antler of roe buck no. 1 (yearling). The antler tip region is composed entirely of compact bone.

The compact bone of the antler cortex was composed of two principal components. These were a tubular framework oriented in the antler's long axis, consisting of woven bone trabeculae with an often honeycomb-like (especially in the inner regions of the compacta) but sometimes (especially in more peripheral locations) more irregular cross-sectional shape, and a system of primary lamellar bone that filled in the intertrabecular spaces (vascular compartments) lined by this framework (Figs 6a,b, 7a–c and 8a,b). Small intertrabecular cavities had each become filled with a circular lamellar system with a single vascular canal (Fig. 7b). In contrast, the lamellar bone deposited within larger intertrabecular cavities was of a more irregular shape, with multiple, roughly circular systems (each centered on a vascular canal) that were not clearly delimited from each other (Figs 7a and 8a). As in the case of the pedicle, the complete lamellar system formed within a vascular compartment is classified as one primary osteon. On BSE images, the primary osteons appeared darker than the bone of the trabecular scaffold, indicating a lower mineral content of the osteons (Figs 6b and 7a,b).

Figure 7.

(a–f) Microscopic appearance of transversely sectioned antler bone. (a) BSE image of compact bone from the distal antler of roe buck no. 1 (yearling). Primary osteons have filled in the former vascular compartments lined by a trabecular framework (asterisks). The bone forming the trabecular scaffold is brighter (more highly mineralized) than the primary osteonal bone. Some particularly bright areas of the scaffold (arrow) represent remnants of mineralized cartilage. (b) BSE image of the outer cortex and part of the adjacent main cortex of the proximal antler of roe buck no. 1 (yearling) in the region of a protuberance. Note the relatively thick trabeculae of coarse woven bone (WB) and incomplete filling of the intertrabecular spaces with primary osteons (PO) in the outer cortex. There are no cement lines along the border between the primary osteons and the woven bone, indicating that no resorption on the trabecular scaffold occurred prior to the formation of the primary osteons (compare with Fig. 8a). The structure exhibiting a scalloped surface that is marked by the thick arrow is probably a remnant of mineralized cartilage in the peripheral area of the main cortex. Thin arrow: small primary osteon; arrowhead: area of wear at the antler surface. (c) Main cortex of the proximal antler of roe buck no. 3 (2 years). The image is dominated by primary osteons that are larger than the osteons classified as secondary ones (arrows). Undecalcified and unstained ground section, LPL. (d) Main cortex of the proximal antler of roe buck no. 3 (2 years). Small osteon classified as a secondary one. Higher magnification of the osteon marked by an arrow in the lower right corner of (c). Undecalcified and unstained ground section, LPL + 1λ-plate. (e) Distal antler of roe buck no. 5 (≥ 3 years). Remnant of mineralized cartilage with scalloped surface (arrows) indicative of previous resorption. Brightfield image of briefly etched ground section stained with toluidine blue. (f) Pseudocolor BSE image showing compact bone of distal antler of roe buck no. 1 (yearling). The insert gives the color coding for the 16 gray-level bands of equal width covering the range of the 256 gray levels from black (0) to peak white (255). Note the higher degree of mineralization of the bone forming the trabecular scaffold compared with the primary osteons. The central structure showing a particularly high mineral content and exhibiting a scalloped surface (arrow) is a remnant of mineralized cartilage (same as that arrowed in a).

Figure 8.

(a–f) Microscopic appearance of transversely sectioned antler bone. (a) BSE image of part of the main cortex (MC) and outer cortex (OC) from the proximal antler of roe buck no. 4 (≥ 3 years). Note the numerous bright (relatively hypermineralized) cement lines (arrowheads) lining the primary osteons present in the MC. The relatively thin OC is composed of woven bone. Asterisk: structure presumably representing a secondary osteon. It is completely encircled by a cement line and has replaced part of a large primary osteon centrally and woven bone peripherally. (b) Proximal antler of roe buck no. 5 (≥ 3 years). Primary osteons (PO) have filled in the spaces lined by the trabecular scaffold (asterisk). In places, signs of resorption are present on the scaffold (arrow). LPL image of briefly etched ground section stained with toluidine blue. (c) BSE image of the MC from the proximal antler of roe buck no. 4 (≥ 3 years). A relatively dark (low mineral content) osteon with a wide osteonal canal is completely encircled by a bright cement line (arrowhead). This osteon is diagnosed as a young secondary osteon. (d) The MC of the proximal antler of roe buck no. 5 (≥ 3 years). A primary (PO) and a secondary (SO) osteon occur next to each other. The secondary osteon is encircled by a cement line (arrows) and has partly replaced the primary osteon. Brightfield image of briefly etched ground section stained with toluidine blue. (e) BSE image of cancellous bone zone of proximal antler of roe buck no. 4 (≥ 3 years). Note the dark appearance (asterisks), indicative of low mineral content of the more recently formed bone lamellae. (f) Fluorescence image of the cancellous zone of proximal antler of roe buck no. 2 (2 years). Note the higher autofluorescence of the more recently formed lamellar bone and osteoid (asterisks). Undecalcified and unstained ground section.

Where it was thin, the outer cortex consisted (almost) entirely of woven bone (Fig. 8a). Where the thickness of the outer cortex was increased, that is, in the area of the protuberances, this zone was composed of a framework of coarsely structured woven bone onto which primary lamellar bone had been laid down (Fig. 7b). The infilling of the intertrabecular spaces with lamellar (primary osteonal) bone in the outer cortex was less complete than in the main cortex, so that frequently larger vascular compartments had remained in the bone. The structure of the outer cortex was, thus, more immature than that of the main cortex. Contrary to the proximal antler samples, the antler tip region (distal antler samples) consisted entirely of compact bone (Fig. 6d) and a structurally distinct outer cortex was missing.

Remnants of mineralized cartilage were present in the trabecular framework of the main cortex, transition zone and cancellous core (Fig. 7a,e,f), but were not observed in the outer cortex (Fig. 7b). As evidenced by BSE imaging, the cartilage matrix was more highly mineralized than the surrounding bone (Fig. 7a,f), and the surface of the mineralized cartilage matrix exhibited scalloped surfaces indicative of resorption (Fig. 7e,f). The presence of the mineralized cartilage remnants within the trabecular framework was most prominent in the distal antler samples and indicated an incomplete replacement of cartilage by bone during the process of endochondral ossification.

Interestingly, signs of resorption were observed not only in the remnants of mineralized cartilage but in places also on the bony trabecular scaffold of the main cortex (Fig. 8a,b). On BSE images, primary osteons of the main cortex (especially its more peripheral portions) were frequently observed to be partly or completely bordered by thin bright (relatively hypermineralized) lines with a scalloped appearance (Fig. 8a). These lines were diagnosed as cement lines, denoting osteoclastic resorption on the trabecular scaffold prior to the deposition of lamellar (primary osteonal) bone. The findings indicate that vascular compartments had become enlarged by resorption on the trabecular scaffold prior to infilling and that, in consequence, the primary osteons had partly replaced pre-existing bone (Fig. 8a,b). The reason for calling these osteons primary instead of secondary is that they had filled in the intertrabecular vascular compartments, i.e. they were predominantly (primary bone) structures laid down in former non-bone compartments and only to a lesser extent structures that had replaced previously existing bone tissue. In places, the trabecular scaffold had become considerably reduced by resorption prior to the lamellar infilling (Fig. 8a).

Secondary osteons were only rarely observed in the antler samples. As, according to our view, primary osteons could be (partially) delimited by cement lines, the distinction between primary and secondary osteons was sometimes not straightforward. Occasionally, a cement line was present between two adjacent osteons, indicating that an earlier formed osteon had partially been replaced by a later formed one. An example is the osteon marked with an asterisk in Fig. 8a. It is completely encircled by a cement line and has replaced the peripheral part of a centripetally adjacent primary osteon, while peripherally it borders on woven bone of the outer cortex that it has apparently partly replaced. The histological picture suggests that the osteon marked in Fig. 8a was formed late during antler growth and has replaced pre-existing (woven and primary osteonal) bone. It is therefore presumably a secondary osteon.

Small osteons with single vascular canals could be either primary or secondary ones. The osteon shown by an arrow in Fig. 7c and shown at higher magnification in Fig. 7d had apparently entirely replaced pre-existing bone and was thus classified as secondary. Figure 8c depicts another example of a small secondary osteon. It is completely encircled by a cement line, has a relatively wide osteonal canal and appears dark on the BSE image, the latter two features indicating its young age. A small primary osteon in the broadened area of the outer cortex is marked in Fig. 7b (thin arrow). It lacks a cement line as is the case for all the primary osteons in this zone, indicating that infilling of an intertrabecular void had occurred without prior resorption on the trabecular scaffold (Fig. 7b). In the main cortex, a differentiation between primary and secondary osteons might sometimes be difficult or even impossible, for example, in cases where a small intertrabecular void has been filled in by a primary osteon following some resorption on the trabecular scaffold, resulting in the presence of a cement line that completely encircles the osteon.

In the transition zone between antler cortex and inner cancellous bone, the size of the intertrabecular spaces increased towards the center (Fig. 6a,b). Like in the antler cortex, also in the cancellous antler core a system of woven bone trabeculae was lined by lamellar bone (Fig. 8e). A complete infilling of the intertrabecular spaces with primary osteons had, however, not occurred. The woven bone trabeculae onto which the lamellar bone had been deposited showed signs of previous resorption (Fig. 8e). The cancellous antler core and sometimes also the transition zone were characterized by a variable degree of hypomineralization of the more recently formed lamellar bone that was evident as darker areas in the BSE images (Fig. 8e) and relatively broad osteoid seams (Fig. 6a). The hypomineralized bone and the osteoid exhibited a marked autofluorescence (Fig. 8f).

In places the antler surface exhibited signs of natural wear (Fig. 7b). This wear may have occurred already in the process of velvet shedding, but more likely when the bucks used their antlers for fighting and for marking their territories, which occurs by thrashing of bushes, young trees or the ground.

Discussion

Pedicle bone

The present study, for the first time, provides a detailed description of the microscopic structure of both pedicle and hard antler bone in a cervid species, the European roe deer. Our results demonstrate pronounced seasonal changes in pedicle structure that can be related to the seasonal antler cycle. These findings corroborate and extend those of von Raesfeld (1919), who first described the seasonal variation in pedicle porosity based on macroscopic observations on longitudinally sectioned pedicles of roe deer.

Around antler casting, pedicle porosity is very high due to extensive osteoclastic resorption that has occurred in the pedicle bone. In addition to bone erosion within the pedicle, there is also extensive sub-periosteal resorption along the pedicle surface. Osteoclastic resorption is especially pronounced in the distal pedicle, where it causes the formation of extended, frequently confluent resorption spaces that were termed resorption sinuses by Kölliker (1873). During the pre-casting period, huge numbers of osteoclasts have been reported to occur first at the pedicle periphery and subsequently in progressively more central locations (Goss et al. 1992). In fallow bucks, in which premature antler casting was induced by castration, first osteoclasts were observed 3 days after castration along the sides of the pedicle, and 2 weeks after castration osteoclasts and associated resorption spaces were present along the full width of the pedicle bone (Goss et al. 1992). The weakening of the bony connections within the distal pedicle eventually leads to antler casting. Following antler casting, osteoclast numbers decline (Goss et al. 1992), but bone erosion continues for some time, causing a smoothing of the pedicle's separation plane (Gruber, 1937; Kierdorf et al. 2003).

Some apposition of new bone onto the resorption surfaces shortly after casting was observed in the present study. However, basically pedicle porosity remains high during the period of intense antler growth, and only during late antler growth the porosity becomes progressively reduced (von Raesfeld, 1919). As a consequence, pedicle bone is very compact during the hard antler phase. It may be hypothesized that maximum compactness is reached during the rutting period (mid-July to mid-August in Germany; Stubbe, 1997), when the antlers are used for fighting and the pedicles are therefore subjected to the strongest mechanical impacts within the seasonal cycle.

The seasonal variation in pedicle porosity is considered an adaptation to the varying functional demands on the pedicle during the antler cycle. During antler growth, pedicle vessels play a role in antler circulation in that the venous drainage of the antlers occurs internally through veins in the pedicle (Waldo et al. 1949). Also, arterial communications between pedicle and antler have been observed in this period, although at all stages of their growth the main blood supply of the antlers is through external arteries that ascend in the vascular layer of the velvet. As a result of the progressive reduction in antler porosity, venous drainage in the later stages of antler growth occurs largely via veins in the velvet (Waldo et al. 1949).

During the hard antler stage, especially the rutting period, the pedicles have to withstand the same forces as the antlers, and from a biomechanical point of view hard antlers and pedicles must be considered a functional unit. Previous studies demonstrated that the dry hard antler bone is very well suited for its function in sparring and fighting, as it exhibits a very high impact resistance and work to fracture (Currey et al. 2009). In contrast to the functional (dry) antlers, the pedicles consist of living, wet bone. It is presumed that, regardless of this difference, the mechanical properties of the pedicles during the hard antler period at least equal those of the antlers. It might even be hypothesized that the mechanical properties of the pedicles exceed those of the hard antlers during the rutting period. If this was the case, fractures would more likely occur in the antler than the pedicle. This would be selectively advantageous, as a fractured antler will be replaced by a normal one in the next growth phase, while a pedicle fracture will very probably cause a permanent malformation of the cranial appendage and may even result in the death of the animal. Thus far, studies on the mechanical properties of pedicles are lacking. Such studies have to take into account the strong seasonal variation in pedicle structure documented in the present and previous studies that will certainly be reflected in a corresponding variation of mechanical properties.

There is experimental evidence that casting of the dry hard antlers is prevented by higher levels of circulating androgens (Suttie et al. 1984, 1995; Goss et al. 1992; Lincoln, 1992; Kierdorf et al. 1993, 1995b). This suggests that higher androgen levels inhibit osteoclast formation in the pedicle, which would otherwise occur as a consequence of the death of the antlers due to insufficient blood supply and the attempt of the organism to sequester the dead bone (Goss, 1983). When androgen levels drop sharply after the rutting period, or experimentally following castration, this enables massive osteoclastogenesis and rapid bone resorption to take place in the pedicle (Goss et al. 1992). An inhibitory effect of androgens on osteoclastogenesis and osteoclast activity has been demonstrated experimentally in other mammals (Wiren, 2008).

The view that hard antlers are dead structures has been challenged by some authors, who claim that antlers maintain a functional blood supply through the pedicle after velvet shedding. In consequence, these authors posit that hard antlers survive until shortly before casting and that there is even some ongoing bone formation in them (Brocksted-Rasmussen et al. 1987; Rolf & Enderle, 1999; Rolf et al. 2001). Results from more recent studies, however, showed that hard antlers effectively dry out (being in equilibrium with the moisture content of the ambient air) and that there is no further mineralization of antler bone after velvet shedding (Currey et al. 2009; Gomez et al. 2013), thereby supporting the traditional view of hard antlers as dead bone. Moreover, the hypothesis that hard antlers are living structures cannot explain why premature casting of hard antlers, but not of antlers in velvet, can be induced by castration or treatment with anti-androgens (Kierdorf & Kierdorf, 2011).

It has been known for a long time that pedicles increase in diameter and shorten in length with age (Stubbe, 1997). The progressive shortening of the pedicles is due to the fact that the separation plane is located in the pedicle, and that with each casting event a portion of the distal pedicle is lost along with the cast antler (Berthold, 1831; Gruber, 1937; Goss et al. 1992; Kierdorf & Kierdorf, 2012; Li, 2013). There is some bone restoration in the distal pedicle, but this is insufficient to compensate for the loss at casting (Kierdorf et al. 2003; Li, 2013). A central biological function of pedicles is probably to ensure that the massive osteoclastic activity leading to antler casting occurs at a ‘safe distance’ from the skull roof. Moreover, pedicles provide a connection with a defined diameter between antler and skull that can effectively be eroded within the seasonal period of low androgen levels (Kierdorf et al. 2004b).

The increase in pedicle diameter due to periosteal bone apposition ensures that the pedicle can serve as a firm base for the regeneration of successively larger and heavier antlers, and as a firm connection between antler and skull when the antlers are used for fighting. In the yearling roe buck, the outer pedicle zone can be described as fibrolamellar bone, composed of mostly circumferentially oriented sheets of woven bone and of lamellar (primary osteonal) bone that fills in the elongate vascular compartments between these sheets. A similar structure of the outer pedicle zone (termed plexiform in that study) has previously been reported from yearling red deer stags, although in the red deer this zone was relatively broader than in the roe deer (Kierdorf et al. 2000). In the older roe bucks, the outer pedicle zone also consisted predominantly of fibrolamellar bone, with some areas showing a woven bone structure. Only in roe buck no. 5 was an outermost zone of lamellar bone present.

The formation of woven and fibrolamellar bone occurs at sites of rapid growth, while lamellar bone is laid down more slowly (Amprino, 1947; Currey, 2002; Huttenlocker et al. 2013; Lee et al. 2013). The fact that the structure of periosteal bone tissue reflects its rate of deposition is known as Amprino's rule (Lee et al. 2013). The finding that the outer pedicle zone of roe bucks (this study) and of young red deer stags (Kierdorf et al. 2000) consists mainly of fibrolamellar bone can thus be seen as indicative of a rapid seasonal growth process that is responsible for the increase in pedicle diameter. The observation of an outermost pedicle bone layer with a lamellar structure in roe buck no. 5 suggests that during the later stages of growth the rate of periosteal bone deposition onto the pedicle surface becomes considerably reduced. The results of the present study thus lead us to hypothesize that the seasonal pedicle growth can be divided into an early phase of rapid bone formation and a later phase of slower growth. To test this hypothesis, labeling studies on forming pedicle bone, similar to those recently performed to analyze bone formation in the antler cortex (Gomez et al. 2013), are encouraged.

Our characterization of the primary bone in the outer pedicle of the roe bucks as fibrolamellar and of the lamellar systems forming within the vascular compartments as primary osteons is in line with the use of these terms by other authors. Thus, Currey (2002) describes fibrolamellar bone as being composed of a thin scaffolding of woven (sometimes parallel-fibered) bone and lamellar bone that fills in the cavities (vascular compartments) within the scaffold, thereby forming primary osteons (Currey, 2002). Similarly, Huttenlocker et al. (2013, p. 21) characterize fibrolamellar bone as a ‘bone complex composed of a woven-fibered scaffolding and intervening primary osteons of varying orientations’. The initially large cavities (that may contain one or more blood vessels) between the woven bone layers are referred to as primary Haversian or osteonal cavities by these authors.

The primary nature of the circular lamellar systems forming within the vascular compartments of fibrolamellar bone was already emphasized by Enlow & Brown (1956, 1957, 1958) in their seminal study of fossil and recent bone tissues. However, in these papers, the authors use the term ‘primary osteons’ for primary vascular canals, while applying the term ‘protohaversian system’ to concentric primary lamellar systems deposited in primary vascular and marrow spaces. In a later paper, Enlow (1962b, p. 278) characterizes the primary osteon as ‘a variety of the Haversian system [..] formed by the deposition of concentric lamellae within tubular, anastomosing spaces located in surface deposits of fine-cancellous, non-lamellar bone’, and distinguishes between primary and secondary osteonal systems on the one hand and (‘non-Haversian’) primary vascular bone on the other. This is more in line with the way the term ‘primary osteon’ is used in the present paper and the works of other authors (Currey, 2002; Locke, 2004; Huttenlocker et al. 2013).

Stover et al. (1992), in their study of cortical bone growth in the metacarpal diaphysis of the horse, describe the formation of circumferential trabeculae of more highly mineralized woven bone by periosteal apposition and subsequent deposition of lamellar bone on the inner trabecular surfaces, resulting in the formation of what they describe as rows of primary osteons in the vascular compartments between woven-bone trabeculae. Likewise, when characterizing the growth in diameter of mammalian long bones, Locke (2004, p. 559) states that the predominantly axially oriented vessels present in the vascular compartments of the cortical diaphysis during later development become surrounded by cylindrical lamellae, which developmentally ‘are primary osteons in the sense that laminar bone is primary’. Note that this author applies the term ‘laminar bone’ to the primary bone formed during the growth in width of the long bone diaphysis. While Currey (2002) treats the terms fibrolamellar, plexiform and laminar bone as broadly synonymous, other authors distinguish between different sub-types of fibrolamellar bone based on vessel orientation in the vascular compartment or use terms like ‘laminar’ only for the orientation of the vascular system (Huttenlocker et al. 2013).

Secondary osteons in different stages of development were observed in the outer pedicle zone, where they replaced the primary (fibrolamellar or woven) bone. This indicates that erosion of the newly formed peripheral bone and its replacement by secondary osteonal bone laid down around one or more blood vessels occur rapidly. Occasionally different generations of secondary osteons were observed, likewise demonstrating the rapidity of the remodeling process. Replacement of laminar bone by secondary osteons also occurs in the shafts of mammalian long bones (Locke, 2004).

The inner pedicle zone of the yearling roe buck consisted of a trabecular framework of woven bone filled with primary osteons and secondary osteons. The latter were mostly small, with a single vascular canal. Formation of such typical secondary osteons can be ascribed to the action of basic multicellular units or bone remodeling units (Currey, 2002; Stout & Crowder, 2012). In this process, a relatively small volume of bone is first eroded by osteoclasts, and the resulting cavity (cylindrical in compact bone) is then refilled with new bone that is deposited in concentric lamellae. This sequence occurs in a spatially and temporally tightly coordinated manner, and normally without significantly affecting overall bone mass or markedly impairing the biomechanical properties of the bone undergoing remodeling.

The seasonal variation in bone structure observed in the pedicles of the older bucks deviates in a number of aspects from the above picture. Our findings indicate a strong seasonal rhythm of bone formation and resorption, with formation probably considerably lagging behind resorption. In consequence, it may take some months until the bone regains a dense structure. The large cavities resulting from the extensive resorption activity are filled in with complex secondary lamellar systems that possess multiple vascular canals. These systems are referred to as complex secondary osteons to indicate their difference from smaller and more typically formed secondary osteons. The seasonal changes in the roe deer pedicle may be classified as an unusual case of remodeling, thereby corroborating the statement by Hall (2005) that regional bone remodeling can occur in very different ways.

In the yearling roe buck, no remnants of mineralized cartilage were observed in the inner pedicle zone. Cartilage remnants were also absent from the pedicles of the older bucks, although in these cases it could be argued that they had been removed in the course of the repeated cycles of extensive resorption and infilling. The increase in pedicle diameter occurred via sub-periosteal bone apposition, i.e. by direct (intramembranous) ossification. Our findings are thus consistent with the view expressed by Gruber (1937) that pedicle formation in roe deer occurs via direct (intramembranous) ossification. It should, however, be mentioned that in other species (fallow deer, red deer) initial pedicle growth (not secondary growth in thickness) has been reported to involve endochondral ossification (Kierdorf et al. 1994; Li & Suttie, 1994).

Antler bone

Antler bone formation in roe deer can be summarized as follows. Initially a trabecular framework of woven bone with more or less cylindrical pores is formed that, in a process of endochondral ossification, largely replaces the preexisting trabeculae of mineralized cartilage. Resorption of the cartilaginous trabeculae is performed by chondroclasts, whose occurrence and activity has been demonstrated in previous studies (Banks, 1974; Kierdorf et al. 1995a; Szuwart et al. 2002). In the process of cartilage replacement, woven bone is initially laid down onto a scaffold of mineralized cartilage (Banks, 1974; Kierdorf et al. 1995a). The presence of remnants of this mineralized cartilage within the woven bone trabeculae of hard antlers of roe deer demonstrates that the replacement of cartilage by bone during antler growth is incomplete, as was previously shown also to be the case in red deer (Landete-Castillejos et al. 2012a). The above description pertains to the formation of the main cortex and the cancellous core of the roe deer antler. In contrast to these antler zones, the outer cortex lacks remnants of mineralized cartilage, thereby indicating that it is not formed by endochondral ossification, but by sub-periosteal direct (intramembranous) ossification.

The soft tissue spaces lined by the trabecular scaffold of woven bone are filled by endosteal deposition of lamellar bone in the form of primary osteons oriented in the long axis of the antler. In this paper, the entire tubular lamellar system formed within an intertrabecular cavity is regarded as one primary osteon. Such primary osteons may contain one or several vascular canals, and are often morphologically more variable than secondary osteons. Our use of the term primary osteon in antler bone corresponds to that of Krauss et al. (2011) in their study of red deer antlers. This concept of the primary osteon is the basis for the latter authors' statement that the number of lumina within the trabecular framework of developing antlers corresponds to the number of primary osteons present in mature antler bone.

Whereas in the center of the antler shaft the lamellar infilling is incomplete, leading to the formation of the cancellous antler core, in more peripheral areas the intertrabecular spaces are completely filled in, causing a compaction of the bone and the formation of a dense antler cortex. This process somewhat resembles the so-called lamellar compaction of fine cancellous bone occurring in the metaphysis of long bones, which was described as involving ‘endosteal lamellar deposition on the surface of pre-existing endochondral trabeculae [..] containing spicules of calcified cartilage matrix’ (Enlow, 1962a, p. 82). Contrary to the situation in antlers, the appearance of the lamellar bone formed in the long bone metaphyses is rather irregular (Enlow, 1962a). This can be attributed to the fact that in the antlers the distinctly tubular nature of the trabecular framework determines the shape of the lamellar bone filling in the intertrabecular cavities.

Our findings indicate that in the main cortex and the cancellous antler core, the infilling of the intertrabecular voids by primary osteons was in places apparently preceded by bone resorption on the trabecular scaffold. A corresponding observation was previously reported by Gomez et al. (2013) for red deer in a study using fluochrome labeling of forming antler bone. Using BSE imaging, we observed the presence of bright (relatively hypermineralized) thin scalloped lines around primary osteons. On the basis of their morphological characteristics and the observation of signs of resorption on the trabecular scaffold, these lines were interpreted as cement lines formed at reversal sites. Skedros et al. (1995) previously reported the occurrence of highly mineralized seams around primary osteons in mature antler bone. They termed these structures hypermineralized lamellae and assumed that they function as analogs of cement lines in secondary bone. Krauss et al. (2011) likewise observed hypermineralized lines around primary osteons in mature antler bone of red deer, which they interpreted as resting lines that were laid down before the start of osteon formation. Based on the findings of the present and our previous study (Gomez et al. 2013), we suppose that the hypermineralized lamellae of Skedros et al. (1995) and the resting lines described by Krauss et al. (2011) may likewise represent cement lines formed at reversal sites following resorption on the trabecular scaffold. Previous studies on antler growth demonstrated early osteoclastic resorption of the bone that had been laid down on the scaffold of the mineralized cartilage, so that chondroclastic, osteoblastic and osteoclastic processes could be observed side by side in a single section (Banks, 1974; Kierdorf et al. 1995a). This means that right from the beginning of the formation of the main cortex and the cancellous core, bone resorption occurs in the antler growth process.

The finding that not only secondary osteons but, according to our interpretation, sometimes also primary osteons were delimited by cement lines can complicate the distinction between them. Our study also suggests that secondary osteons in antler bone, like those in pedicle bone, may not always conform to the typical morphological picture of relatively small circular lamellar systems with a single central vascular canal. A recent fluorochrome labeling study of cortical bone formation in red deer antlers confirmed the formation of secondary osteons during antler osteogenesis also in this species (Gomez et al. 2013). It was demonstrated that the formative period of secondary osteons overlapped with that of primary ones. As these secondary osteons were formed in bone that was still undergoing modeling, they were referred to as modeling osteons (Gomez et al. 2013).

Our findings corroborate the view that secondary osteons are scarce in antlers and that most of the osteons present in antler bone are primary ones (Chen et al. 2009; Launey et al. 2010; Krauss et al. 2011; Gomez et al. 2013). In consequence, the forming antler can be said to show very limited remodeling activity during its short lifespan. Signs of ongoing (limited) remodeling activity were observed in the permanently viable antlers of castrated fallow bucks (Kierdorf et al. 2004a).

The present study confirmed the observation previously made for red deer antlers (Gomez et al. 2013) that during antler growth bone resorption at trabecular endosteal surfaces is followed by bone formation at the same sites. Generally, modeling in a specific bone location is distinguished from remodeling in that osteoblastic bone formation is not preceded by osteoclastic bone resorption at modeling sites (Erben, 1996; Seeman, 2008). Recently, however, Maggiano (2012) suggested that the term modeling should be applied to all processes occurring at periosteal or endosteal surfaces during bone development and growth, including those where bone deposition follows resorption at a given site. As shown, primary osteon formation in growing antlers may involve resorption on the endosteal trabecular surface prior to lamellar infilling of the intertrabecular cavities. Nevertheless, the process can be characterized as a case of bone modeling in the sense defined by Maggiano (2012).

The outer antler cortex, which was referred to as the sub-velvet zone by Rolf & Enderle (1999), has thus far received little attention. We show that where it is thin, the outer cortex consists of woven bone. In areas where the antler exhibits protuberances (pearls) and the outer cortex is broader, the woven bone forms a trabecular framework whose voids are incompletely filled with primary osteons. These observations suggest a less mature nature of the outer compared with the main cortex. In contrast to the main cortex, resorption on the trabecular scaffold does not occur prior to the infilling of the intertrabecular spaces.

It has been shown that at sites of bone mineralization, proteoglycans of the osteoid are partially degraded and lost, and that there is a decrease in the glycosaminoglycan content of bone with tissue age (Pugliarello et al. 1970; Baylink et al. 1972; Grynpas & Hunter, 1988). The more intense autofluorescence observed in the outer cortex of the roe deer antlers is therefore seen as indicative of a higher proteoglycan content and thus a younger age of this zone compared with the main cortex. In line with our findings, a higher degree of autofluorescence of the bone matrix of newly formed trabecular excrescences compared with the established (older) matrix was recently reported by Taylor et al. (2012). The above view of a younger age of the antler's outer cortex compared with the main cortex is in line with the results of a recent fluorochrome labeling study in red deer (Gomez et al. 2013).

On the basis of our findings we conclude that the outer cortex is formed late during antler growth by intramembranous ossification as a consequence of periosteal activity occurring in the more proximal antler portions. This is in line with the view of Bubenik (1966), who stated that the pearls of the antler develop secondarily during late antler growth. Late formation by intramembranous ossification explains the lack of mineralized cartilage remnants and signs of bone resorption in the outer cortex that are both prominent features of the endochondrally forming central antler portions.

The antlers of castrated deer tend to develop conspicuous bony protuberances (von Raesfeld, 1919; Bubenik, 1966; Goss, 1983; Kierdorf et al. 1995b). Histological investigations in castrated fallow bucks have shown that these protuberances, which develop in proximal and distal antler portions, are formed by periosteal bone apposition onto the original antler surface (Kierdorf et al. 2004a). These protuberances can therefore be characterized as a hypertrophic form of the structures that form during normal antler growth, due to (seasonally renewed) activation of the antler periosteum in the castrates.

During the short lifespan of the antlers, a large amount of bone has to be mineralized. Thus, in adult red deer stags (≥ 3 years old), dry antler weight accounted for 28% of total skeletal weight (Gomez et al. 2012). The resulting high calcium and phosphorus demand for antler mineralization is partly met by mobilization from the skeleton (Meister, 1956; Banks et al. 1968; Baxter et al. 1999). The more recently formed bone in the transition zone and the cancellous core of the roe deer antlers showed signs of hypomineralization. This could indicate that during peak mineralizing activity the mineral demands of the forming antler bone could not be met in full. If this was the case, the findings would indicate an impairment of the mineralization process caused by physiological exhaustion, especially regarding calcium supply (Landete-Castillejos et al. 2012b). An alternative explanation would be that velvet shedding, which is caused by a rise in circulating androgens levels, sets an abrupt end to the antler mineralization process regardless of the stage of antler maturation. This was observed in an experimental situation in which, due to anti-androgen treatment, out-of-season antler casting had been induced that was followed by regeneration of a new set of antlers. These antlers were cleaned of velvet at a very immature stage of bone development after a short period of growth (Kierdorf et al. 1993). In the roe deer, the most recently formed antler portions, namely the outer cortex and the youngest bone formed in the cancellous and transition zones, showed signs of incomplete development and hypomineralization. This suggests that also during normal antler development, velvet shedding can be triggered before full maturation of the antler bone has been accomplished.

Conclusion

Our study showed that the pedicles of roe deer exhibit a pronounced seasonal variation in bone structure, alternating between a stage of high porosity around antler casting and a stage of very compact structure during the hard antler period. The observations suggest that the mechanical properties of the pedicle vary markedly between different periods of the antler cycle, and biomechanical studies investigating this seasonal variation are therefore encouraged. Due to the seasonally occurring extensive resorption and formation processes, affecting large volumes of bone at a time and involving large numbers of osteoclasts and osteoblasts, the pedicle bone of older deer is a heavily remodeled structure. The pedicle may therefore be a useful model system to study the mechanisms controlling osteoclast differentiation and activity as well as the effects of certain drugs on bone formation and resorption processes. Understanding the mechanisms that underlie the annually repeated formation of a dense and presumably mechanically very strong bone like the pedicle may also provide clues on how to increase the density and biomechanical competence of osteoporotic human bones.

Antler bone is basically composed of a trabecular framework, whose voids are filled with primary osteons. Notwithstanding the fact that antlers are composed largely of primary bone, there is also evidence of resorptive activity during their formation. Osteoclastic resorption followed by bone deposition at the same trabecular endosteal surfaces is considered to be the cause for the occurrence of cement (reversal) lines around primary osteons from the main cortex and in the cancellous core of the antlers. The outer cortex that is present in the proximal antler portions is formed late during the growth process by periosteal apposition and lacks signs of resorption.

From a biomechanical point of view, pedicles and antlers must be viewed as an entity. By studying antlers and pedicles, a more complete picture of the biomechanics of cranial appendages in deer can be achieved than is possible by investigating antlers alone.

Acknowledgements

The authors thank Detlev Klosa for his skillful help with BSE imaging. The very helpful comments of two anonymous reviewers are gratefully acknowledged.

Ancillary