Histological evidence for muscle insertion in extant amniote femora: implications for muscle reconstruction in fossils

Abstract

Since the 19th century, identification of muscle attachment sites on bones has been important for muscle reconstructions, especially in fossil tetrapods, and therefore has been the subject of numerous biological and paleontological studies. At the microscopic level, in histological thin sections, the only features that can be used reliably for identifying tendon–bone or muscle–tendon-bone interactions are Sharpey's fibers. Muscles, however, do not only attach to the bone indirectly with tendons, but also directly. Previous studies failed to provide new indicators for muscle attachment, or to address the question of whether muscles with direct attachment can be identified histologically. However, histological identification of direct muscle attachments is important because these attachments do not leave visible marks (e.g. scars and rugosities) on the bone surface. We dissected the right hind limb and mapped the muscle attachment sites on the femur of one rabbit (Oryctolagus cuniculus), one Alligator mississippiensis, and one turkey (Meleagris cuniculus). We then extracted the femur and prepared four histological thin sections for the rabbit and the turkey and five histological thin sections for the alligator. Sharpey's fibers, vascular canal orientation, and a frayed periosteal margin can be indicators for indirect but also direct muscle attachment. Sharpey's fibers can be oriented to the cutting plane of the thin section at high angles, and two Sharpey's fibers orientations can occur in one area, possibly indicating a secondary force axis. However, only about 60% of mapped muscle attachment sites could be detected in thin sections, and frequently histological features suggestive of muscle attachment occurred outside mapped sites. While these insights should improve our ability to successfully identify and reconstruct muscles in extinct species, they also show the limitations of this approach.

Introduction

Muscle reconstruction in extinct tetrapods

From the early days of paleontology, reconstructions of animals were used to illustrate the conclusions made from fossil bones, such as the reconstruction of the dinosaur Iguanodon by Gideon Mantell. By the mid-19th century, life reconstructions were on display and in books, e.g. the Crystal Palace sculptures of various extinct animal groups or the painting in the Sedgwick Museum in Cambridge, England, depicting various marine reptiles.

Studies of muscles of extinct animals were necessary for creating accurate life reconstructions. The majority of research in the fields of myology and muscle homologies began in the late 19th and the first half of the 20th century. Based on already existing myological research on extant animals (e.g. Gadow, 1882; Gadow & Selenka, 1891), further studies aiming at homologizing muscle groups among different extant tetrapods were conducted (e.g. Romer, 1923a, 1927a, 1942; Appleton, 1928a,b; Haines, 1935; Davis, 1936; Hudson, 1937; Howell, 1938a,b). Simultaneously, the results of these studies were used for muscle reconstructions in extinct tetrapods with a focus on fossil synapsids (e.g. Gregory & Camp, 1918) and dinosaurs (e.g. Romer, 1923b, 1927b). However, these studies were based on morphological observations only, i.e. dissections of recent material and embryological studies.

W. Gross (1934) was one of the first to study bone histology on extinct tetrapods, although he did not apply the results to muscle reconstruction. In 1944, C. M. Goss was among the first to look for muscle attachments at the microscopic level and classified several direct attachments. Research on muscle–bone interactions, primarily focusing on Sharpey's fibers, picked up pace (e.g. Knese & Biermann, 1958; Hoyte & Enlow, 1966; Barton & Keenan, 1967; Jones & Boyde, 1974; Garant & Cho, 1979; Hurov, 1986) in the late 1950s to late 1970s, but studies on muscle reconstructions in extinct tetrapods dropped to almost zero, one exception being Coombs (1979).

Although myological studies on extant animals (e.g. Ghetie, 1976; Jones, 1979; McGowan, 1979; Rowe, 1986; Raikow, 1990; and Baumel, 1993) increased again starting in the late 1970s, it was not until the 1990s that these new results were incorporated into studies on extinct tetrapods, e.g. Bryant and Seymour (1990) and Gatesy (1990, 1995). More histological research (e.g. Francillon-Vieillot et al. 1990; Frost, 1990a-d) and the extant phylogenetic bracket approach (EPB) by Witmer (1995) has sparked more research on muscle reconstructions over the last 10 years. Again, reconstructing muscles in extinct taxa was based on macroscopic features and focused on biomechanical applications (e.g. for limb muscles, Goss, 1944; Hutchinson, 2000, 2001a,b, 2002, 2004, 2006; Hutchinson & Gatesy, 2000; Carrano & Hutchinson, 2002; Gatesy, 2002; Jasinoski et al., 2006; Remes, 2007; Otero & Vizcaino, 2008; Bonnan, 2010; Holliday et al., 2010; Shaw, 2010; Maidment and Barrett. 2011; and Persons & Currie, 2011). At the same time, new insights were gained into muscle–tendon–bone interactions at the microscopic level for extant tetrapods (e.g. Benjamin et al. 2002; Hieronymus, 2002, 2006; McNeill Alexander, 2002; Suzuki et al. 2002, 2003; Trotter, 2002; Suzuki & Hayashi, 2008, 2010; and Thomopoulos et al. 2010) as well as extinct tetrapods (Organ & Adams, 2005; Zhou et al. 2010; Klein et al. 2012). However, the distinction between the two modes of muscle attachment, direct and indirect, was discussed neither in macroscopic nor in microscopic studies, and only Suzuki studied the muscle–bone and muscle–tendon–bone interactions microscopically (Suzuki et al. 2002, 2003; Suzuki, 2003; Suzuki & Hayashi, 2008, 2010). To clarify terminology, direct attachment means that the muscle fibers attach to the bone via fibrocartilage (Dolgo-Saburoff, 1929), whereas tendinous or indirect attachment refers to muscle attachment via tendons on the periosteum with fibers penetrating the periosteum (Woo et al. 1988). Much more needs to be understood about applying bone histology to muscle reconstructions since little work has been done on the topic.

Previous histological studies

Features of bone, seen in thin sections under a microscope, were first observed by Havers (1691) and later confirmed, for example by Albinus (1756). The method of bone histology for scientific purposes was then established, which included most of the characters we observe in thin sections today. In the 19th century, Sharpey (1848) described structures we now call Sharpey's fibers that are believed to be indicative of (tendinous) muscular attachment, e.g. Sharpey (1848), Suzuki et al. (2002, 2003), Suzuki (2003) and Hall (2005). Although bone histology has been established for some 300 years and used scientifically, little work has been done on the evaluation of muscle attachment on a bone histological level.

One of the few exceptions is the previously mentioned work of Suzuki. In the course of his studies, he identified several attachment types for extant taxa (Suzuki, 2003). Two of his attachment types, the unmediated and the fibrocartilage mediated (Suzuki et al. 2002, 2003; Suzuki, 2003), are indirect attachments, whereas his fleshy type is a direct attachment (Suzuki et al. 2002, 2003; Suzuki, 2003). Nevertheless, Suzuki's focus was on reptiles and classical histological sections in which the muscles were still attached to the bone, and the bone was decalcified, not on petrographic thin sections.

The extant phylogenetic bracket

Results from studies on extant animals are applied to their extinct relatives based on the extant phylogenetic bracket (EPB) as introduced by Witmer (1995; see also Bryant & Russell, 1992) The most frequently used pair of brackets is crocodiles (e.g. Hutchinson & Gatesy, 2000) and birds (e.g. Hutchinson, 2001a,b) for reconstructing the biology of dinosaurs. Earlier researchers applied similar approaches implicitly when reconstructing soft tissues of synapsids (Gregory & Camp, 1918) and dinosaurs (Romer, 1922, 1923b, 1927b).

Research questions and approach

According to Hall (2005), Sharpey's fibers previously could be associated only with fibrous entheses (an enthesis being ‘the site of attachment of a tendon or ligament’, http://medcyclopaedia.com, accessed on 7 December 2011). As tendons are not necessarily associated with muscles, e.g. the meniscus tendons, the existence of Sharpey's fibers in histological thin sections is not an unequivocal indicator of muscular attachment. Furthermore, so far only Suzuki et al. (2002, 2003) and Suzuki (2003) had shown muscle–tendon–bone interaction on a histological level. Therefore, we wanted to test three hypotheses:

1. Muscle insertions reliably leave a trace in the cortical histology of long bones.

2. Commonly used histological indicators such as Sharpey's fibers are reliably reflect the insertion of muscles or connective tissue.

3. Sites of direct muscle attachment differ histologically from sites of indirect muscle attachment.

The background for this research, of course, is provided by the paleobiological research program of reconstructing muscles in extinct tetrapods. We thus wanted to test whether histology can be used for this purpose.

Our approach was rather straightforward and consisted of first studying the hind limb musculature and then sectioning the femur and studying its histology for one individual each of three extant model organisms: the domestic rabbit, the American alligator and the domestic turkey.

Materials and methods

Study animals

The study animals were chosen for the purpose of representing a wide variety of phylogenetic groups among Amniota and thus much of the EPB of fossil amniotes (Fig. 1). Furthermore, one representative each of the clades Crocodylia and Aves was chosen to provide the EPB for non-avian dinosaurs and non-crocodilian crurotarsi. Additionally, the choice of study animals was influenced by their metabolic rate, with two specimens with a high basal metabolic rate (the rabbit and the turkey) and one with a low basal metabolic rate (the alligator). Lastly, the chosen animals represent three different modes of life: one terrestrial quadruped specializing in running and jumping (the rabbit); one semi-aquatic quadruped (the alligator); and one bipedal runner and flier (the turkey).

Fig. 1.

Phylogenetic relationships of each of the three studied species. (A) The common rabbit, Oryctolagus cuniculus (cladogram modified from Westheide & Rieger, 2009). (B) Alligator mississippiensis (cladogram modified from Holliday & Witmer, 2009). (C) The common turkey, Meleagris gallopavo (cladogram modified from Westheide & Rieger, 2009).

Oryctolagus cuniculus f. domestica Linnaeus, 1758

Extensive myological work has been done on the domestic rabbit, Oryctolagus cuniculus f. domestica (e.g. Bensley, 1948; Popesko et al. 1992), providing a basis for this study. The study specimen was obtained from a local butcher's shop. Its weight without skin, the head, and internal organs was 2164 g; the feet had already been cut off as well. The femur length measured 99 mm. The epiphyseal plate had not closed yet, indicating that the specimen was a juvenile.

Alligator mississippiensis Daudin, 1802

We obtained a complete specimen of Alligator mississippiensis that had died of burns and had been brought to the Reptilauffangstation NRW reptile sanctuary by a private owner. The specimen weighed 6860 g with internal organs and was a juvenile female, approximately 5 years old. Its femur measured 86 mm in length.

Meleagris gallopavo f. domestica, Linnaeus 1758

Meleagris gallopavo f. domestica, the domesticated turkey, was chosen because of its easy availability and similar body mass as the two other model species. Meleagris represents the avian side of the EPB for extinct dinosaurs. The study specimen was obtained from a local butcher's shop and weighed 8660 g without internal organs, head, feet, and feathers. Its femur was 129 mm long.

Dissection and documentation

Dissections were necessary for two reasons. First, we wanted to map the muscular attachments on the bone and familiarize ourselves with the myology of the study animals, thereby getting a better understanding for proposed homologies and terminology in the literature. And secondly, dissection was required to fulfill the aim of this study, to combine macroscopical and microscopical observations of muscular attachment in a single specimen.

Fixation and preservation

First, the pelvic region along with the hind limbs was detached anteriorly of the first sacral vertebra. In the case of Alligator mississippiensis, the tail was cut off after the 3rd caudal vertebra; the rest of the animal was skeletonized. Afterwards, the material was placed into a fixation solution consisting mostly of formaldehyde for a week, thereby sealing the tissue permanently. For preservation, the specimens were put into a solution of 0.5% formaldehyde and fungicides. A good example to differentiate between fixation and preservation is the mummification of the dead in Ancient Egypt: with embalming of the corpse being the fixation, and storing of the mummy in a dry and cool tomb being the preservation. Before dissecting, the preserved specimens had to be in the preservation solution for 3 days to remove the remaining formaldehyde.

Dissections

The dissection of the hind limbs was performed following Bensley (1948), McLaughlin (1980), and Popesko et al. (1992) for Oryctolagus; Romer (1923a), Chiasson (1962), and Brinkmann (2010) for Alligator; Gadow & Selenka (1891), Ghetie (1976), Rowe (1986), and VandenBerge and Zweers (1993) for Meleagris. For conflicting nomenclature for muscles in Alligator, Romer (1923a) was given priority over the others. In Meleagris, priority was given to Rowe (1986).

Documentation of muscles and attachment sites

The muscle attachment sites on the femur were mapped in an anterior and posterior view (Figs 3–5) to visualize all of them. Color-coded pins were used to prevent confusion between the individual muscles.

Fig. 3.

Muscle attachment sites (cross-hatched areas) as detected by dissection and major anatomical features mapped onto the right femur of Oryctolagus cuniculus in anterior (top) and posterior view (bottom). Muscles with direct attachment are colored in green, those with indirect attachment are colored in orange. For muscles with both attachment types, the dominant one is shown.

Fig. 5.

Muscle attachment sites (cross-hatched areas) as detected by dissection and major anatomical features mapped onto the right femur of Meleagris gallopavo in anterior (top) and posterior view (bottom). Muscles with direct attachment are colored in green, those with indirect attachment are colored in orange. For muscles with both attachment types, the dominant one is shown.

Sketches were digitally redrafted in Adobe illustrator cs3.

Thin sectioning

Thin sectioning protocol

The femora of the three study specimens were sectioned along four or five transverse planes, resulting in a total of 14 thin sections for the three individuals. One or two sections were located on the proximal epiphysis and/or metaphysis, one in the midshaft region, and one or two on the distal metaphysis and/or epiphysis (Fig. 2). The thin sections were labeled according to their position from proximal to distal.

Fig. 2.

Position and code number of the thin sections cut from the Oryctolagus cuniculus (left), Alligator mississippiensis (middle), and Meleagris gallopavo (right) right femora. Sections were numbered from proximal to distal.

The thin sectioning followed the standard procedure used at the bone and petrographic thin section laboratory of the Division of Paleontology of the University of Bonn. First, the bones were embedded in epoxy resin (Araldite 2020). After being cut with a rock saw with a thin saw blade, the bones were ground with F400, F600 and F800 powder and then cleaned and glued onto glass slides using Araldite 2020. Next, the glued block was cut again, and the remaining embedded bone slice was ground and polished with F600 and F800. When the slides had been ground down to the required thickness of 30–50 μm, they were covered with a coverslip using UV glue type Verifix LV 740.

For cutting thin sections from fresh bone, it is generally recommended to degrease bones with acetone prior to embedding. This is to improve adhesion and minimize the risk of detachment of the bone from the glass slide. Recent bone detaches more readily than fossil bone. Furthermore, it does so unpredictably, i.e. after the initial gluing of the bone block to the glass slide, and possibly during every consecutive step until the final embedding. Fossil bone, on the other hand, usually does not detach once it is ground down thin enough (after cutting the block with the rock cutting saw). However, we did not follow the general degreasing recommendations, but degreased the samples after embedding and cutting based on the rationale that the resulting increase in surface would make the process more efficient. We encountered a problem with this method: the epoxy resin became brittle and fractured after 2½ days in acetone, but without exacerbating the interpretation of the ready thin sections. This phenomenon was most likely caused by weak zones that formed when heat, created by the curing of the resin, sequestered the grease. We therefore suggest degreasing the specimens before embedding, after all.

Imaging of the thin sections

The histological thin sections were drawn with a camera lucida mounted on a Wild M5 dissecting microscope, and the drawings were scanned and used for mapping of the histological features in the thin sections. Thin sections were permanently coverslipped and imaged in conventional transmitted light (LM) and polarized transmitted light (PL) using a Leica DMLP Polarizing Microscope configured with a 360° rotating stage and PL filters and using Leica N PLAN 40/0.65, N PLAN POL 10/0.25, and N PLAN POL 2.5/0.07 objective lenses. LM and PL images were acquired with a Leica DFC420 color camera using the Leica imageaccess easylab 7 software (Leica, Wetzlar, Germany).

Choice of thin section position

The positions of the thin sections can be seen in Fig. 2. The planes of section were chosen to optimally cover the attachment sites on the proximal and distal end of the femur as well as on the midshaft. All sections represent transverse planes. The position of the thin sections was chosen to equally represent indirect and direct muscle attachments.

Terminology

For the description of the histological thin sections we follow the terminology of Francillon-Vieillot et al. (1990). We will use the term Sharpey's fibers for fibrous structures that are visible under polarized and normal light. They can be short or long, straight or curved. Sharpey's fibers (SF) are associated with fibrous entheses of tendons attaching to bone (Sharpey, 1848; Hall, 2005). Sharpey's fibers are collagenous fibers that occur as fiber bundles (Sharpey, 1848) and penetrate the periosteum at high angles (Francillon-Vieillot et al. 1990; Hall, 2005). We found a high angle of orientation to the periosteum to be an unnecessary condition for Sharpey's fibers. Additionally, we use the term ‘cross pattern’ for a combination of two SF orientations in one spot. Furthermore, when deposition of lamellar bone in the vascular canals has just begun or has not yet been completed, the osteons are called ‘incipient’.

Results

Description of muscles inserting or originating on the femur

In general, the muscles observed during the dissections showed no significant difference to the literature. Exceptions from this occur in Meleagris, where we were unable to find M. gemellus, although a muscle fitting to its origination and position was observed. The muscle differed in that it was attached by a short tendon instead of a direct attachment. In another instance, we detected muscle tissue directly attached to the bone, but it could not be homologized with any known muscle. Therefore, we could not reconstruct whether this muscle originated at or inserted on this spot.

Attachment sites

The muscles, which were mapped macroscopically on the femur (Figs 3–5), showed macroscopic and microscopic features in their attachments. These characteristics are described below. The histological features are described in more detail, starting with the most unambiguous feature. Histological features together with the muscles are summarized in Table 1.

Table 1.

Summary of the muscles shown in Figs 7–9, listing their histological features, the completeness of the histological evidence or the absence of evidence

Muscle	Thin sections	Type of attachment	Histological features
M. iliacus	Or-1	direct	not featured
Mm. gemelli cranialis and caudalis and obturator internus	Or-1	indirect	Sharpey's fibers
M. quadratus femoris	Or-1	direct	frayed margin
M. gluteus superficialis	Or-1	indirect	frayed margin
M. pectineus	Or-2	indirect	not featured
M. vastus intermedius 2	Or-2, Or-3	direct	(steeply inclined) Sharpey's fibers, incompletely
M. adductor longus	Or-2, Or-3	direct	Sharpey's fibers
M. adductor magnus	Or-3	direct	Sharpey's fibers
M. gastrocnemius caput laterale	Or-4	indirect	not featured
M. gastrocnemius caput mediale	Or-4	indirect	not featured
M. puboischiofemoralis internus 2	Al-1	indirect	Sharpey's fibers
M. puboischiofemoralis externus 1	Al-1	indirect	not featured
M. puboischiofemoralis externus 2	Al-1	direct	not featured
M. femorotibialis internus	Al-2, Al-3, Al-4	direct	Sharpey's fibers; incomplete cross pattern, incomplete vascular canal orientation, incomplete
M. iliofemoralis	Al-2	direct	not featured
M. femorotibialis externus	Al-2, Al-3	direct	Sharpey's fibers, incomplete vascular canal orientation, incomplete
M. coccygeofemoralis longus	Al-2	direct	Sharpey's fibers, incomplete
M. coccygeofemoralis brevis	Al-2	direct	Sharpey's fibers, incomplete
M. adductor femoris 1	Al-3	direct	not featured
M. adductor femoris 2	Al-3	direct	not featured
M. gastrocnemius externus superficialis	Al-4, Al-5	indirect	not featured
M. vastus medialis	Me-1, Me-2, Me-3, Me-4	direct	Sharpey's fibers, only Me-4 vascular canal orientation, only Me-4
unidentified	Me-1	direct	frayed margin, incomplete
M. quadratus femoris or M. gluteus profundus	Me-1	indirect	frayed margin
M. obturator internus	Me-1	indirect	cross pattern
M. crureus	Me-2, Me-3	direct	vascular canal orientation, incomplete, only Me-3
M. gluteus minimus	Me-2	indirect	not featured
M. vastus externus	Me-2, Me-3	direct	not featured
Mm. gluteus superficialis and obturator externus	Me-2	indirect	Sharpey's fibers, incompletely
M. ischiofemoralis	Me-2	direct	Sharpey's fibers, incomplete
M. adductor magnus	Me-3, Me-4	direct	Sharpey's fibers, incomplete vascular canal orientation frayed margin, incomplete
M. adductor longus	Me-3	direct	not featured
M. subcrureus	Me-4	direct	Sharpey's fibers
Mm. flexor digitorum longum and peroneus profundus	Me-4	indirect	Sharpey's fibers vascular canal orientation
M. flexor hallucis longus	Me-4	indirect	Sharpey's fibers, incomplete vascular canal orientation
M. tibialis cranialis	Me-4	indirect	vascular canal orientation, incomplete

Fig. 4.

Muscle attachment sites (cross-hatched areas) as detected by dissection and major anatomical features mapped onto the right femur of Alligator mississippiensis in anterior (top) and posterior view (bottom). Muscles with direct attachment are colored in green, those with indirect attachment are colored in orange.

Muscle attachment sites on a macroscopical scale

Macroscopically, we observed that muscles with tendinous attachment show rugosities and scars on the bone. These attachment sites have a well-developed margin and are generally smaller than those of direct attachment sites. In the observed specimens, tendinous attachment is restricted exclusively to the proximal and distal epiphyses. In contrast, direct attachment occurs regularly on the shaft and to the proximal and distal epiphyses as well. Sites of direct attachment leave a smooth surface on the femur, although enlarged pores may occur, e.g. at the site of M. vastus intermedius in Oryctolagus. These pores are likely to match the exit points of transverse vascular canals in histological thin sections. Generally speaking, directly attaching muscles have the tendency to leave no macroscopical evidence on the bone.

Attachment sites in thin sections

The most often occurring evidence for muscle attachment are SF (Fig. 6A). They do not seem to be restricted to indirect attachment, but frequently occur at direct muscle attachment sites as well.

Fig. 6.

Histological correlates of muscle attachment sizes. (A) Sharpey's fibers in section Al-4. (B) Vascular canal orientation in section Me-3. The vascular canals in the thin section are oriented longitudinally (1), whereas the vascular canals in the sharp ridge are oriented radially (2). (C,D) Steeply inclined Sharpey's fibers in section Or-3. (C) Arrow points to a cone-like structure representing steeply inclined Sharpey's fibers under lower magnifications. (D) Arrow points to V-like fibrous structures under higher magnification. (E) Cross pattern representing two Sharpey's fiber orientations in section Al-4. The arrows indicate the two orientations. (F) Arrow points to the frayed bone margin in section Or-1.

V-shaped, short fibrous structures, each touching an osteocyte lacuna, represented SF penetrating the bone at steep angles relative to the sectioning plane (Fig. 6C). They are visible as such at high magnification, while appearing cone-like at lower magnifications (Fig. 6D). When focusing at specific depths of the field at high magnifications, the structures appear cone-like as well when the fibrous structures are out of focus. The ‘cross pattern’ proved to be two different SF orientations in one spot (Fig. 6E,F). This feature is best seen in polarized light (Fig. 6E) and with the λ-filter (Fig. 6F), regardless of the magnification. Fiber orientations can be distributed equally, although one orientation usually dominates. The fiber distribution may change within one location.

Furthermore, the orientation of vascular canals may be indicative of muscle attachment, especially when radial vascular canals occur in parts of thin sections that are dominated by longitudinal vascular canals (VC) (Fig. 6B). Important for this feature is that the differently oriented vascular canals ‘disturb’ the homogeneity of the vascular canal alignment (Fig. 6B). Another criterion for muscle attachment may be a frayed periosteal margin, indicating a higher amount of blood vessels penetrating the periosteum from inside the bone for muscle nutrition (Fig. 6F). These histological criteria for muscle attachment can occur individually and combined, although most combinations involve SF.

Description of thin sections

Each thin section is described individually, and according to its position on the femur from proximal to distal. After a brief description of the general properties of the bone in the thin section, detailed descriptions of the histological features that are connected to muscle attachment sites follow. Muscles showing no histological features are mentioned briefly with their hypothetical positions, as are histological features that have no corresponding muscle.

Oryctolagus cuniculus

Section Or-1

This is a section through the femur head and the greater trochanter. The outer surface is rough and appears frayed along the two lobes (Fig. 7A). The cortex consists of woven bone, developing into fibrolamellar bone (FLB). The VC feature incipient secondary osteons and are oriented longitudinally. SF penetrate the periosteum along the medial margin of the lateral lobe (Fig. 7A). These SF can be related to the tendon of Mm. gemelli cranialis and caudalis and obturator internus. Laterally, the margin is frayed extensively, corresponding to the attachment of Mm. gluteus superficialis and quadratus femoris, although the frayed portion exceeds the possible attachment site of both muscles. M. iliacus, which should be visible on the anteromedial portion of the section, is not evident (Fig. 7A). It seems as if the margin in this portion is damaged to some extent, possibly explaining the lack of evidence of attachment of this muscle.

Fig. 7.

Camera lucida drawings of the transverse sections of the right femur of Oryctolagus cuniculus from proximal to distal, comparing muscle attachment sites and histological correlates. (A) Or-1, section through the femur head and greater trochanter showing strong Sharpey's fiber bundles corresponding to attachment of the tendon for Mm. gemelli and M. obturator internus, as well as areas of a frayed bone margin with some correspondence to the attachment of Mm. gluteus superficialis and quadratus femoris. Attachment of M. iliacus is possibly obscured by damage of the cortex along the anteromedial side. (B) Or-2, section through the proximal metaphysis with Sharpey's fibers corresponding to the attachments of Mm. vastus intermedius 2 and adductor longus and some Sharpey's fibers in the deeper cortex which correspond well to the attachment of M. quadratus femoris in an ontogenetically earlier stage. (C) Or-3, the midshaft section with steeply inclined Sharpey's fibers that correspond to the attachment of M. vastus intermedius 2 and Sharpey's fibers corresponding to the attachments of Mm. adductores longus and magnus. (D) Or-4, section through the distal epiphysis with thick Sharpey's fibers bundles corresponding to the insertion of the lateral meniscus tendon. Note that the epiphyseal plate is not yet closed, indicating osteological immaturity of the specimen. Muscles with direct attachment are colored in green, those with indirect attachment are colored in orange. The black bar in (D) represents the meniscus tendon, a structure that is not associated with muscle attachment.

Section Or-2

This is a section of the proximal metaphysis. The cortex shows a cavity located medially to posteromedially; this could be a nutritional foramen (Fig. 7B). The matrix in the cortex is parallel-fibered to woven bone, making up FLB. The VC are oriented longitudinally. SF occur along the posterolateral margin, penetrating the periosteum at low angles (39°–53°). These fibers can be associated with M. adductor longus (Fig. 7B). The deep cortex features SF along the posterolateral portion (Fig. 7B). These SF move closer to the periosteum posteriorly and penetrate it at the posterior tip. The deeper fibers may be evidence for the ontogenetically earlier position of M. quadratus femoris, whereas the outermost SF cannot be associated with any muscle (Fig. 7B). Along the anterior margin, the cortex features SF most likely from M. vastus intermedius 2. M. pectineus, expected on the posterior medial part of the section, cannot be correlated with any histological feature (Fig. 7B).

Section Or-3

This section covers the midshaft. The highly vascularized cortex is made of parallel-fibered bone, grading into FLB. Vascular canals are oriented longitudinally, but a reorientation towards a circumferential orientation occurs primarily in the anterior portion. One line of arrested growth (LAG) occurs, which is unusual for an animal bred for meat production but can be explained by the animal being acquired in early spring, and therefore having lived through the preceding winter. The lateral tip contains posteriorly positioned SF (Fig. 7C). In addition, the VC appear to be oriented differently from the rest of the cortex. Steeply inclined SF relative to the cutting plane occur almost along the entire anterior margin (Fig. 6C,D; Fig. 7C). They represent M. vastus intermedius 2. On the medial posterior side, SF occur along a small portion, evidencing M. adductor magnus.

Section Or-4

Or-4 represents a section through the epiphyseal region on or near the condyles. A cortex is not clearly identifiable. The matrix consists of parallel-fibered bone. The section consists of three areas – an anterior, center, and posterior area – showing that the epiphyseal plate was not yet closed. A strong anterior concavity is present in which dense SF occur along the lateral and central part and along a short portion of the medial part of the concavity. They do not correspond to any muscle attachment but reflect the attachment of the meniscus tendon. Attachment correlates for Mm. gastrocnemii capitis laterale and mediale are not evident at their expected positions on the lateral and medial anterior margins of the section, respectively.

Alligator mississippiensis

Section Al-1

This section covers the proximal metaphysis. The cortex consists of parallel-fibered bone tissue. Vascular canals are oriented radially and longitudinally. In the anterior region, SF can be related to the tendon of M. puboischiofemoralis internus 2 (Fig. 8A). Along the lateral side, anteriorly positioned SF occur that do not correspond to any muscle attachment. Mm. puboischiofemorales externi 1 and 2 are not evident in the posterolateral corner of the thin section, possibly because this region as been ground too thin (Fig. 8A).

Fig. 8.

Camera lucida drawings of the transverse cross-sections of the right femur of Alligator mississippiensis from proximal to distal end, comparing muscle attachment sites and histological correlates. (A) Al-1, section through the proximal metaphysis showing M. puboischiofemoralis externus. The cortex was ground too thin where Mm. puboischiofemoralis interni 1 and 2 were expected. (B) Al-2, section though the proximal end of the shaft with incomplete evidence for Mm. femorotibiales internus and externus and adductores femoris 1 and 2, as well as M. iliofemoralis. (C) Al-3, the midshaft section with incomplete evidence for the attachment of Mm. femorotibiales internus and externus. (D) Al-4, section cut through the shaft distally with extensive evidence for M. femorotibialis internus. Note the cross pattern along the anterior margin representing two Sharpey's fiber orientations, one for the muscle and one corresponding to the knee tendon. (E) Al-5, section through the transition from distal metaphysis to distal epiphysis showing no congruence between histological features and muscle attachments. Muscles with direct attachment are colored in green, those with indirect attachment in orange.

Section Al-2

This is a section through the proximal shaft. Parallel-fibered matrix forms bone in the cortex. The VC are oriented longitudinally and radially. The lateral side of the section shows SF (Fig. 8B). These fibers can be associated with M. femorotibialis externus. SF occur along the medial anterior side, partly covering the attachment site of M. femorotibialis internus (Fig. 8B). Posteriorly, along the medial side, SF occur in a small concavity (Fig. 8B). Here, they correspond to M. coccygeofemoralis longus, M. coccygeofemoralis brevis, or both. The posterior side bears SF deeper in the cortex; they could be evidence for the former position of M. puboischiofemoralis externus 1 or 2 (Fig. 8B). M. femorotibialis was expected along the posterior side and M. iliofemoralis along the anterior side; neither of them was observed. The anterior margin came out too thin after the grinding, obscuring any possible attachment of M. iliofemoralis (Fig. 8B).

Section Al-3

This section covers the midshaft. The cortex consists of parallel-fibered bone. Some of the VC are irregular but the majority are longitudinally oriented. This results in a different VC orientation in an area that also features SF (Fig. 8C). This area corresponds to the attachment of M. femorotibialis, although attachment of this muscle should be visible along the entire lateral margin. More SF occur along the medial side, covering attachment of M. femorotibialis internus to about 50% (Fig. 8C). Attachment of Mm. adductores femoris 1 and 2 is not evident. These two muscles should be seen along the posterior part of the margin (Fig. 8C). In the deep cortex, SF occur posterolaterally, indicating an ontogenetically earlier position of muscular attachment (Fig. 8C).

Section Al-4

This section covers the distal shaft. The cortical bone consists of parallel-fibered bone. The vascular canals are typically oriented longitudinally. SF occur in the posterior central region, where they are accompanied by a different VC orientation (Fig. 8D). Two small areas on the posteromedial side bear SF as well, and SF occur in combination with a different VC orientation along the medial portion (Fig. 8D). The anterior margin bears SF in two different orientations with high angles (Figs 6E,F and 8D). One set of these SF is oriented medially, which is the dominant orientation on the medial part of the anterior margin. The second set of SF is oriented almost perpendicular to the first, i.e. laterally, and dominates the lateral part of the anterior margin. In between, both orientations briefly occur in equal part (Fig. 8D). One orientation could indicate the knee tendon, the other orientation a muscular attachment. More SF occur anterolaterally. All the aforementioned features correspond to the attachment of M. femorotibialis internus, which is visible for almost its entire extent (Fig. 8D). SF in combination with a different VC orientation along the lateral margin do not correspond to muscular attachment. M. gastrocnemius externus superficialis is not evident in the posterolateral portion of the thin section, although the cortex bears no signs of destruction by grinding here (Fig. 8D).

Section Al-5

This cross-section cuts transversely through the transition from distal metaphysis to epiphysis. The cortex throughout the thin section. All bone tissue is made of a parallel-fibered bone matrix. The VC are oriented longitudinally. Along the posterolateral margin tendinous material is attached to the bone. In this section, VC orientation differs from the dominant orientation, and the margin is partially frayed (Fig. 8E). No muscle corresponds to this feature; instead its position corresponds to the two SF orientations in section Al-4. Weakly expressed SF occur along the anterolateral corner of the section (Fig. 8E). These fibers do not correspond to muscle attachment, to tendon or to ligament insertions. M. gastrocnemius externus superficialis is not visible in thin sections at the observed attachment sites, i.e. along the posterolateral margin and anterolaterally (Fig. 8E).

Meleagris gallopavo

Section Me-1

Me-1 is a section through the femur head and the greater trochanter. The cortical matrix is made of parallel-fibered bone. Vascular canals are generally oriented longitudinally. The thickened cortex along the lateral side (Fig. 9A) has lamellar to parallel-fibered bone. In the most lateral part of the section, SF occur on either side of the thickened cortex, and the more posteriorly situated SF have two orientations (Fig. 9A). The dominant set SF of is oriented posteriorly at a high angle, whereas the second set is oriented anteriorly at a low angle. This site partially corresponds to the tendinous attachment of M. obturator internus, although the site is too large to be associated only with this attachment. SF in the posteriormost portion do not correspond to any observed muscle attachment. Laterally, along the anterior margin of the section, it is frayed (Fig. 9A). Anteromedially, the frayed margin corresponds to either M. quadratus femoris or M. gluteus profundus; which one depends on the exact level of the thin section. The section was intended to be in the distalmost portion of M. quadratus femoris, and a slight deviation of about 1 mm from this plane would have caused it to cut M. gluteus profundus instead. Most of the remaining frayed margin corresponds to the unidentified muscle mentioned above. At the posterior central margin, M. vastus medialis is not visible in thin sections at the observed attachment sites. Along the lateral margin posteriorly as well as anteriorly, the cortex bears signs of having been ground too thin.

Fig. 9.

Camera lucida drawings of the transverse cross-sections of the right femur of Meleagris gallopavo from proximal to distal end, comparing muscle attachment sites and histological correlates. (A) Me-1; cross-section through the femur head and greater trochanter showing two different Sharpey's fiber orientations corresponding to the attachment of the tendon for M. obturator internus. Depending on the exact level of the section, there is evidence of attachment of M. gluteus profundus or M. quadratus femoris as well. (B) Me-2; section through the proximal metaphysis showing the attachment of Mm. gluteus superficialis and obturator externus and, incompletely, M. ischiofemoralis. (C) Me-3; the midshaft section in which only the attachment of Mm. crureus and adductor magnus left any histological trace. (D) Me-4; section through the distal epiphysis showing attachment sites of Mm. vastus medialis, adductor magnus, subcrureus, flexor digitorum longum, peroneus profundus, flexor hallucis longus, and tibialis cranialis. Note the extensive development of Sharpey's fibers without corresponding muscle attachment sites. Muscles with direct attachment are colored in green, those with indirect attachment in orange.

Section Me-2

This section is cut through the proximal metaphysis. The bone matrix consists of parallel-fibered bone. The VC are oriented longitudinally. Along the lateral side, the margin features SF in a small area (Fig. 9B). These SF do not correspond to any muscle. A little more posteriorly, SF occur again accompanied by a different VC orientation (Fig. 9B). Here, the features can be associated with the tendinous attachment of Mm. gluteus superficialis and obturator externus. In the same area, SF are present in the deeper cortex as well, suggesting the ontogenetically earlier attachment of a muscle. Posterior to the attachment site, VC continue to be oriented differently (Fig. 9B). Even further posteriorly, more fibers are observed. These could be SF that are partially aligned almost parallel to the periosteum and become steeper posteriorly (Fig. 9B). They could also be evidence of parallel-fibered bone. They do not correspond to any muscular attachment. Along the posteromedial and medial section, M. ischiofemoralis is indicated by the presence of a medium-sized area of SF, located medially (Fig. 9B). Along the anterior margin, four muscles cannot be associated with histological features (M. vastus medialis medially, M. crureus more laterally, and Mm. gluteus minimus and vastus externus). Along the anterior margin, the cortex seems to have been ground too thin.

Section Me-3

This is the mid-shaft section. The cortex is FLB consisting of a woven bone matrix. Orientation of the VC is longitudinal as well as circumferential. The inner posterior margin features two sharp ridges, one of which shows VC oriented perpendicularly to the dominant VC orientation (Fig. 9C). The VC seem to partially penetrate the periosteum on the outer surface. This is correlated with the attachment of M. adductor magnus. Additionally, bone fibers in the tip of the ridge are oriented in the dominant fiber orientation (Fig. 9C). The outer surface between the two ridges shows SF along a notch (Fig. 9C). They do not correspond to any muscular attachment. In the anteriormost part of the thin section, VC are oriented differently again, matching the anteriormost part of the attachment area of M. crureus (Fig. 9C). Histological features do not provide evidence for most of the attachment of M. crureus. M. vastus medialis on the medial and posteromedial side, M. adductor longus on the posterior side lateral to M. adductor magnus, and M. vastus externus on the posterolateral side are not reflected in bone histology.

Section Me-4

This is a section of the distal epiphysis. The cortical bone has a woven bone matrix forming FLB. The VC are predominantly radially oriented, although longitudinal orientation occurs. In the posterior concavity, SF are accompanied by a difference in VC orientation, although the VC orientation exceeds occurrence of SF, being visible throughout almost the entire posterior concavity (Fig. 9D). The area of SF seen in this section corresponds partially to the tendon of Mm. flexor digitorum longum and peroneus profundus and partially to M. flexor hallucis longus. More medial to the concavity, the margin is frayed and weakly developed SF are seen at the posteromedial tip of the section (Fig. 9D). Both correspond to the attachment of M. adductor magnus. On the anteromedial tip of the section, SF and a different VC orientation correspond to M. vastus medialis. Slightly medial to the anterolateral tip of the section, SF correlate with the attachment of M. subcrureus. More SF occur along the lateral side up to the posterior lateral tip, accompanied by a posterolateral cross pattern, none of which s to a muscular attachment site (Fig. 9D). M. tibialis cranialis should be evidenced by more than the small area of VC orientation on the lateral side of the posterior concavity, but it is not.

Discussion

Identification of muscle attachment sites in thin sections

Sharpey's fibers

According to the literature (e.g. Hall, 2005), SF are the most unambiguous indicator for tendon–bone interactions and, so far, the only indicator used for the identification of muscle attachment. Although the results of this study do not contradict this statement, the absence of histological evidence for several muscles with tendinous attachment (Figs 7D and 8A,D,E,9B) casts doubt on the completeness and accuracy of muscle reconstruction based on SF alone. SF occur in all thin sections, most commonly in Meleagris (16 SF sites), followed by Alligator (14 SF sites) and Oryctolagus (8 SF sites).

SF be used for identifying muscle attachment not only can in the outermost bone layers, but also for ontogenetically older muscle positions (see sections Or-2 and Al-3).

Identification of SF is easiest when they are aligned along, or close to, the sectioning plane. However, the appearance of SF when cut at higher angles could not be gleaned from the literature. The V-shaped structures seen in section Or-3 under higher magnifications could belong to these steeply inclined SF. The V-shaped fibrous structures are associated with conical structures seen under lower magnifications that correspond to the general orientation of SF, i.e. perpendicular to the periosteum (Fig. 6C,D). Therefore, the V-shaped structures/conical structures are SF cut at higher angles.

On the lateral side, posteriorly in section Me-2 (Fig. 9B), several generations of differently oriented SF suggest a changing orientation of tensile force during ontogeny rather than changes in position of the attachment. The change in force orientation could be due to a shift of the relative position of the two attaching muscles earlier during ontogeny. The region of changing SF is associated with the shared tendon of Mm. gluteus superficialis and obturator internus.

Suzuki (Suzuki et al. 2002, 2003; Suzuki, 2003) cites two attachment types for which SF are typical, although he only addresses attachment in reptiles. His ‘unmediated’ insertion type has SF at an angle of about 90° to bone surface, and his ‘fibrocartilage-mediated’ attachment type is typically of a lower, more acute angle (Suzuki et al. 2002, 2003). Our study found SF in both 90° and more acute angles, indicating both insertion types in Alligator, with the ‘unmediated’ type dominating. For Oryctolagus and Meleagris, a distinction between high and low angle attachment is possible, although Suzuki's (2003) and Suzuki et al.'s (2002, 2003) division is not applicable because neither of the two species show a perichondreum. For fossil taxa, Suzuki's (Suzuki et al. 2002, 2003; Suzuki, 2003) division could prove to be less applicable, because a perichondreum is not fossilized. Distinguishing between his two proposed attachment types could be problematic.

The indirect (tendinous) and direct (fleshy) attachment cannot be discerned by different SF patterns or angles. In crocodiles, Suzuki (2003) was able to associate dense bundles of SF with ‘unmediated’ attachment and sparse SF with ‘fibrocartilage-mediated’ attachment. This result could not be repeated for the observed Alligator in general, although SF density appeared to be lower at some spots than at others.

While SF were visible in both cases of attachment, the high angle SF were observed only in one case of direct muscle attachment. Yet, according to the old adage that ‘the absence of evidence is not evidence of absence,’ we cannot exclude the presence of this feature in indirect attachments. Even more, a high angle orientation of SF to a transverse cutting plane seems to be rather rare, as we had only one case of high angle SF in 13 thin sections.

The cross pattern

When more than one SF orientation occurs in one spot of a thin section, a cross pattern is created. Here, one set of fibers can be assigned to a specific muscular attachment, whereas the second fiber orientation belongs to something different. This feature was observed in thin sections Al-4 and Me-1 (Figs 8D and 9A).

In the case of the two orientations along the anterior margin in section Al-4 (Fig. 6C,D), it can be inferred from the dissection, the position of the thin section in the distal metaphysis, and the literature (e.g. Chiasson, 1962) that the second set corresponds to the insertion of the knee tendon. M. femorotibialis inserts on the knee tendon slightly more distal to the thin section position. M. femorotibialis internus is the distalmost muscle contributing to this tendon, but not the only one (M. femorotibialis externus is another example). Action from all contributing muscles could create a second force axis in the distal part of M. femorotibialis internus, causing the second SF set.

In section Me-1, the occurrence of two sets of fibers crossing each other remains unexplained. The area of occurrence corresponds well to the attachment of the M. obturator internus tendon, but this tendon is not split. It is more likely that the cross pattern in this thin section is caused by the interaction between parallel-fibered bone and SF. The secondary orientation aligns with the fiber orientation throughout the area of thickened cortex.

Nevertheless, cross patterns can be generalized as resulting from two simultaneously occurring orientations of tensile force. Similar to the high angle SF, we cannot exclude crossing SF orientations for two muscles or two tendons, which create a tensile force in the same spot. Although we could find this feature only in association with the combination of a direct muscle attachment with tendon action, the rarity of the feature in the studied thin sections prevents further interpretation.

Vascular canal orientation

On their own, VC oriented differently from the dominant VC orientation are not sufficient indicators for muscle attachment. However, they can be used as indicators when combined with outer surface VC and/or SF (Figs 8C,D and 9D). This feature was observed 10 times, in sections Or-1, Al-3 to Al-5, and sections Me-3 and Me-4.

Suzuki et al. (2002, 2003) and Suzuki (2003) discussed vascular canal orientation as indicative for muscle attachment. In section Or-1 and Al-5, VC orientation is accompanied by a frayed margin, in sections Al-3 and Al-4 by SF, and in sections Me-3 and Me-4, different VC orientation has no additional features.

VC that are oriented towards the periosteum in a transverse plane may possibly rearrange the bone fabric into the same orientation to help better to take up tensile stresses caused by muscle attachments. On the other hand, muscles producing strong forces could cause the underlying bone to improve the bone fabric, thereby causing the reorientation of the VC.

Frayed bone margin

The correlation between a frayed periosteal margin and a muscle attachment site may result from blood vessels penetrating the periosteum and entering the attached muscle or tendon, thereby enhancing nutrient supply. A frayed margin was observed in sections Or-1, Al-5, Me-1, and Me-4, accounting for six sites in total, all of which are thin sections through epiphyseal regions (Figs 7A, 8E and 9A,D). Macroscopically, this feature is represented by a rough outer bone surface. Unfortunately, this feature is unreliable, as it is easily confused with bone loss from the process of preparing the thin section because of the very thin cortex in the epiphyseal region. In sections Me-1 for example, a frayed margin occurs in an area of cortex that appears to be ground too thinly anterolaterally to anteriorly (Fig. 9A). A frayed surface in section may also be associated with long-grained bone texture in juvenile dinosaurs (Tumarkin-Deratzian et al. 2006; Brown et al. 2009; Tumarkin-Deratzian, 2009).

Discussion of ambiguous histological evidence for muscle attachment

False negatives: no evidence for muscle in thin section

The lack of macroscopic or histological evidence for some muscle attachment sites is shared by all studied specimens. Indeed, of the 48 possible muscle attachments observed during this study (15 indirect and 33 direct attachments), 19 (40%) are not evident from any of the features mentioned in the previous sections. Of these 19 attachments, six are tendinous and 13 are directly by muscle fibers. This means that 40% of the indirect attachments and 39% of the direct attachments were not evident from histology.

One reason for the absence of signs for muscle attachment may be the improper placement of the plane of section. For example, in section Or-1, (Fig. 7A), M. iliacus may not be observed because the section was cut too proximally. However, in this case Mm. gluteus superficialis and quadratus femoris should not have been seen as well, because they are on the same level as M. iliacus (Fig. 3). Another reason is that the cortex was ground too thinly, as is the case in section Me-2. Potential signs of attachment of Mm. vastus medialis and crureus could not be observed along almost the entire anterior part of the section because of this.

According to Suzuki (Suzuki et al. 2002, 2003; Suzuki, 2003), muscle force load and evidence for attachment are connected in reptiles with fibrocartilage-mediated entheses. A hyaline cover of the periosteum serves as the main absorber for muscle force, so that only the strongest muscles (in terms of tensional stress) leave a fibrous signature in the bone, i.e. SF (Suzuki et al. 2002, 2003; Suzuki, 2003). Interestingly, the thin sections of Alligator appeared to be the most fibrous of all of the three observed specimens, even though it is the specimen with the least amount of evidence for muscle attachment.

Poor correspondence between histological evidence and actual muscle attachment

Section Al-4 has the largest muscle attachment site, with M. femorotibialis internus spanning more than 75% of the entire circumference of the thin section. However, the evidence for its attachment site was incomplete (Fig. 8D). Additionally, SF are sparse along the anteromedial and the anterolateral margins (Fig. 8D). This pattern could be caused by the strength of tensile muscle forces. Similar to Suzuki's (Suzuki et al. 2002, 2003; Suzuki, 2003) explanations for the strength of muscle force in fibrocartilage-mediated entheses in reptiles, SF could be more likely to form where muscle imparts high forces to the bone.

Additionally, histological features can span the attachment sites of two muscles [e.g. the frayed margin correlated with Mm. gluteus superficialis and obturator femoris in section Or-1 (Fig. 7A) or the evidence for Mm. coccygeofemorales brevis and longus in section Al-2 (Fig. 8B)], making it difficult to differentiate between the muscles unless the observer has dissected the animal personally and knows exactly where the thin sections were cut. Suzuki et al. (2002, 2003) and Suzuki (2003) do not mention this problem, and neither do Suzuki & Hayashi (2008, 2010).

False positives: evidence in thin sections but no muscles

Sixteen sites occur where muscle attachment features are seen but with no muscle associated with them. Twelve of these sites show SF, two sites have frayed margins, and three have VC (where one is accompanied by a frayed margin). These occurrences are false positives for muscle attachments at the time of death but are potentially informative about ontogenetically earlier entheses and ligament attachment.

In the anterior concavity of section Or-4, the SF correspond to the meniscus tendon. Therefore, SF could be associated with the attachment of articulation tendons and ligaments instead of muscles. In Meleagris, the surplus of SF to muscle attachment may be due to a more fibrous bone matrix. If the fibers seen in the thin section are not SF but are instead bone fibers, this would explain why there is no muscle corresponding to them. Nevertheless, this explanation does not fit a fast-growing, endothermic animal such as a bird, which usually features fibrolamellar bone instead of lamellar zonal bone.

As members of Crocodylia do not have menisci in the knee, the VC orientation in section Al-5 cannot be interpreted as an attachment of a meniscus tendon. Furthermore, in the course of the dissection of the alligator limb, we did not encounter a tendon attaching to the femur in this region. Unfortunately, the dissection instructions (Romer, 1923a; Chiasson, 1962; and Brinkmann, 2010) did not provide an arthrological overview to shed light on this matter.

The SF found in the posterior part of section Or-2 show up in the deeper cortex first (Fig. 7B). As mentioned in the description of section Or-2, the deeper SF are interpreted as evidence for ontogenetically earlier attachment of M. quadratus femoris. Nevertheless, attachment of M. quadratus femoris is limited to the posterior side of the third trochanter in the adult Oryctolagus cuniculus.

Comparison of the occurrence of histological indicators in the studied specimens

SF and VC orientation are informative in all studied specimens. The V-shaped fibrous structures, interpreted as steeply inclined SF, are only seen in Oryctolagus (Fig. 7C). This is due to the steep angle at which the SF must be oriented to the plane of section, apparently a rare situation in the observed thin sections and a peculiarity of M. vastus intermedius 2 so far. A cross pattern is more common and can be seen one time each in Alligator and Meleagris (Figs 8D and 9A). If one were to cut a cross-section through the distal metaphysis of Oryctolagus, cutting M. vastus intermedius 2, the same cross pattern would be predicted because M. vastus intermedius 2 inserts on the knee tendon as well. Frayed margins could be seen in Oryctolagus and Meleagris, and their absence in Alligator is attributed to the thick perichondreum surrounding the epi- and metaphyses of the bone (e.g. seen in Al-1; Fig. 8A).

Concluding remarks

Both direct and indirect (tendinous) muscle attachment sites can be identified in histological thin sections of a taxonomically diverse sample of extant amniotes (Oryctolagus, Alligator, Meleagris). The most reliable indicator for muscular attachment is Sharpey's fibers. Yet, they are not limited to muscular attachment sites but can rather be associated with the meniscus tendons, for instance. When oriented to the plane of section at high angles, Sharpey's fibers take on a cone-like appearance under lower magnifications and a V-shaped fibrous shape under higher magnifications. Sharpey's fibers that cross each other at two different orientations are indicative of a second muscle or a tendon in the same spot. Finding this feature in a histological thin section, therefore, is more valuable than the other features, as it provides information about two attachments rather than one.

Another good histological indicator of muscle attachment sites is vascular canals arranged in a different orientation than the dominant orientation in the thin section, preferably radial vascular canals oriented towards the periosteum, thereby disturbing the homogeneity of the vascular canal pattern. Finally, a frayed periosteal margin may be indicative for muscle attachment as well, although it is the most ambiguous character.

These features can be used to identify specific muscles in a given thin section of tetrapod bones, as long as the exact position of the thin section is known. The muscles identified with these features do not necessarily leave macroscopic traces on the bone surface. Such macroscopic traces, however, are good indicators of where it might prove worthwhile cutting a thin section. Scars and rugosities are likely to bear SF of some kind (i.e. normal, high angle, and cross pattern) and enlarged pores correspond to vascular canals.

Nevertheless, none of the histological features mentioned above had exclusive association with either direct or indirect sites of muscle attachment. The occurrence of a ‘cross pattern’ or of high-angle Sharpey's fibers was too scarce as to allow a strict association, but further research on those two features should help clarify the matter. A cross pattern is to be expected in all instances where several muscles contribute to one tendon. This pattern usually occurs in the distal metaphysis of the femur. Sharpey's fibers and vascular canal orientation, on the other hand, can be seen as general indicators for muscle attachment, occurring with direct and indirect attachment. The feature of the frayed margin should be handled with care, as it can be a sign of muscular attachment, especially when it occurs at the epiphyses, but it can also be a sign of damage of the cortex. The latter is more likely for extinct animals that underwent fossilization and diagenesis. It is unlikely to find this feature in thin sections of animals with a thick perichondreum surrounding the periosteum.

Overall, only about 60% of the muscle attachment sites known to be intersected by the thin sections are correlated by histological features, yet these were thin sections from extant animals. In extinct animals this percentage is likely to be lower, as it is not possible to differentiate between single muscle attachment sites and attachment sites where two or more muscles or a tendon attach without leaving a gap in histological features. Conversely, histological features such as SF and radial vascular canal do not necessarily correspond to muscle attachment sites at all. However, histological evidence may help phylogenetic bracketing as a location-specific correlate regardless of externally visible correlates. Moreover, it may assist with identification of unusual attachments in fossils without a strict extant homolog. An example for the latter is Shychovski et al.'s (2010) work on ligament attachments in tyrannosaurs, and Klein et al.'s (2012) work on neck ribs in sauropod dinosaurs.

Author contributions

M.S. and H.P. conceived and designed the experiments, H.P. performed the experiments, analyzed the data and wrote the paper, M.S. and H.P. contributed reagents/materials/specimens/analysis tools, M.S. critically improved the manuscript.