1983

The Journal of Experimental Biology 215, 1983-1993

© 2012. Published by The Company of Biologists Ltd
doi:10.1242/jeb.064790

RESEARCH ARTICLE
Comparison of structural, architectural and mechanical aspects of cellular and
acellular bone in two teleost fish
Liat Cohen1, Mason Dean3, Anna Shipov1, Ayelet Atkins1, Efrat Monsonego-Ornan2 and Ron Shahar1,*
1
Koret School of Veterinary Medicine and 2School of Biochemistry and Nutrition, The Robert H. Smith Faculty of Agriculture, Food
and Environment, The Hebrew University of Jerusalem, Israel and 3Max Planck Institute of Colloids and Interfaces, Department of
Biomaterials, Am Mühlenberg 1, 14424 Potsdam, Germany
*Author for correspondence (rons@savion.huji.ac.il)

Accepted 13 February 2012

SUMMARY
The histological diversity of the skeletal tissues of fishes is impressive compared with that of other vertebrate groups, yet our
understanding of the functional consequences of this diversity is limited. In particular, although it has been known since the mid-
1800s that a large number of fish species possess acellular bones, the mechanical advantages and consequences of this
structural characteristic – and therefore the nature of the evolution of this feature – remain unclear. Although several studies have
examined the material properties of fish bone, these have used a variety of techniques and there have been no direct contrasts of
acellular and cellular bone. We report on a comparison of the structural and mechanical properties of the ribs and opercula
between two freshwater fish – the common carp Cyprinus carpio (a fish with cellular bone) and the tilapia Oreochromis aureus (a
fish with acellular bone). We used light microscopy to show that the bones in both fish species exhibit poor blood supply and
possess discrete tissue zones, with visible layering suggesting differences in the underlying collagen architecture. We performed
identical micromechanical testing protocols on samples of the two bone types to determine the mechanical properties of the bone
material of opercula and ribs. Our data support the consensus of literature values, indicating that Youngʼs moduli of cellular and
acellular bones are in the same range, and lower than Youngʼs moduli of the bones of mammals and birds. Despite these
similarities in mechanical properties between the bone tissues of the fish species tested here, cellular bone had significantly
lower mineral content than acellular bone; furthermore, the percentage ash content and bone mineral density values (derived from
micro-CT scans) show that the bone of these fishes is less mineralized than amniote bone. Although we cannot generalize from
our data to the numerous remaining teleost species, the results presented here suggest that while cellular and acellular fish bone
may perform similarly from a mechanical standpoint, there are previously unappreciated differences in the structure and
composition of these bone types.
Key words: acellular bone, teleost fish, mechanical properties.

INTRODUCTION The bone matrix is permeated by an interconnected network of

Bone forms a tough and protective load-bearing framework for the
bodies of vertebrates. The magnitude and direction of the principal
loads to which bones are exposed can vary. Furthermore, bone is
loaded repeatedly and cyclically during the animal’s life; therefore,
fatigue damage in the form of microcracks is bound to accumulate
within the bone material. Most vertebrate bones are able to cope
with a changing mechanical environment and damage accumulation
through adaptation of their morphology and self-repair (termed
modeling and remodeling, respectively). Both processes depend on
the ability of bone to sense local changes in its mechanical
environment, such as increasing strain or strain energy density levels,
which result from the accumulation of microcracks or changing
loads. This mechanosensory ability is widely thought to be provided
by bone cells – the osteocytes (Adachi et al., 2009; Bonewald, 2011).
Osteocytes arise from the bone-forming cells (osteoblasts) (Franz-
Odendaal et al., 2006). In the bones of mammals, some osteoblasts
become embedded within the extracellular matrix they deposit
(osteoid), and this matrix then becomes mineralized. Once embedded
and trapped within the matrix, the osteoblasts become known as
osteocytes, residing within small spaces named osteocytic lacunae.
microscopic channels (canaliculi), which contain cellular processes
of osteocytes, and thus embedded osteocytes are able to
communicate with their neighbors.
In the middle of the 19th century it was observed that in the great
majority of teleost fish species (and thus a large proportion of
vertebrates), bone is acellular and does not contain osteocytes. The
ubiquity of osteocytes throughout vertebrate animals, including
mammals, birds, amphibians and reptiles, and their proposed
mechanosensory function make this observation surprising.
Interestingly, the bone of basal teleosts is cellular; therefore, the
loss of osteocytes is a derived property (Parenti, 1986; Kranenbarg
et al., 2005a); as the main function of bone is mechanical, this
suggests that the evolution of acellularity confers a mechanical
advantage to the skeleton. However, based on current knowledge
of mammalian bone biology, lack of cells would hinder the
functionality of fish bone, restricting its ability to respond to
changing loading conditions and to repair microdamage. And yet,
various reports suggest that acellular fish bone functions
mechanically in a manner similar to its cellular counterpart. For
example, several authors have reported that the acellular bone of

THE JOURNAL OF EXPERIMENTAL BIOLOGY

1984 L. Cohen and others
neoteleosts – the eight most-derived superorders of bony fishes,
which include the Percomorpha, the largest and most diverse group
of fishes – responds to mechanical loads by the process of modeling,
as a cellular bone would (Huysseune et al., 1994; Hegrenes, 2001;
Kilhara et al., 2002; Kranenbarg et al., 2005a; Kranenbarg et al.,
2005b; Nemoto et al., 2007; Kitamura et al., 2010; Totland et al.,
2011). Furthermore, although osteocytic lacunae are believed to act
as stress concentrators (e.g. Nicolella et al., 2006), and therefore
their absence could conceivably decrease the rate of microdamage
accumulation, it is reasonable to assume that acellular bone still
accumulates some fatigue damage from repeated cyclic mechanical
loads, and therefore would benefit from the ability to replace
damaged areas. However, the manner in which acellular fish bone
is able to undergo modeling and remodeling without the strain-
sensing osteocytes is currently unknown, although osteoblasts and
bone lining cells are reasonable candidates for mediators of these
processes (Witten and Huysseune, 2009).
The structure and mechanical properties of bones from a large
variety of vertebrates from different taxonomic levels (mammals,
amphibians, reptiles and birds) have been extensively studied. It is
thus somewhat surprising that so little is known about the structure
and mechanical properties of the bones of modern teleost fish, as
the number of teleost fish species likely represents at least 50% of
the rest of the vertebrate species combined (Currey, 2010). Very
few studies have examined the material properties of fish bone, with
cellular fish bone being the most studied (Roy et al., 2000; Rho et
al., 2001; Erickson et al., 2002; Wang et al., 2002; Zhang et al.,
2002; Paxton et al., 2006) and, to the best of our knowledge, with
only one study providing values for acellular fish bone (Horton and
Summers, 2009). However, none of these previous studies tested
and compared the two bone types, and valid comparisons among
the studies are difficult because they employed a wide diversity of
methods and focused on disparate species and different skeletal
elements. The dearth of data regarding the mechanics of fish bone
tissue most likely stems from the fact that fish bones are typically
small and their shape is extremely irregular, making sample
preparation and testing very difficult (Currey, 2010). In order to
properly compare the mechanical properties of cellular and acellular
bone, it is imperative that samples of the two bone types be tested
by exactly the same method using multiple replicates.
In the study reported here, we examined in detail the structure
and mechanics of two bones (operculum and rib) obtained from fish
with acellular bones [the tilapia Oreochromis aureus (Steindachner,
1864); Acanthopterygii] and from fish with cellular bones (the
common carp, Cyprinus carpio L.; Ostariophysi). We used various
techniques to describe their morphological features, mean mineral
density and microstructural features. These analyses are paired with
standardized mechanical testing methods to determine and compare
the mechanical properties of cellular and acellular fish bone tissue.
MATERIALS AND METHODS
Specimens and sample preparation
Sixteen freshly killed, healthy 18month old carp weighing
672–1281g, and 15 freshly killed, healthy 14month old tilapia,
weighing 356–576g, were obtained from a commercial fishery
(Madan, Maagan Michael, Israel).
The operculum and fifth rib were then harvested from all fish by
manual dissection within 1h. Samples of all bones were collected
for histological study and immediately fixed in 4%
paraformaldehyde solution (Sigma, St Louis, MO, USA). Samples
were also collected for light microscopy, micro-computed
tomography (micro-CT) scanning and mechanical testing. These
THE JOURNAL OF EXPERIMENTAL BIOLOGY
samples were wrapped in saline-soaked gauze and stored at –20°C
until further testing.
Histology and light microscopy
Rib and operculum bone samples, taken from four carp and four
tilapia, were fixed overnight in 4% paraformaldehyde in PBS at
4°C. They were then decalcified in 0.5moll–1 EDTA, pH7.4. Once
decalcified, they were dehydrated in graded ethanol solutions and
Histo-clear© (National Diagnostics, Atlanta, GA, USA). They were
then embedded in Paraplast (SPI supplies, West Chester, PA, USA),
cut by microtome into 5m sections and mounted on glass slides
(Superfrost, Thermo Scientific, Barrington, IL, USA). The
embedded sections were deparaffinized in xylene, washed twice with
ethanol, rehydrated through a graded series of ethanol solutions and
stained with hematoxylin and eosin (H&E).
Five histological sections from each of the four carp opercula
were used to obtain estimates of osteocyte density within the bone.
To determine osteocyte density, a large rectangle was outlined
digitally over a high-resolution image of each sample such that it
covered as much of the section as possible, the area of the rectangle
was determined and all osteocytes within this region were counted
manually. Osteocyte density was determined as the number of
osteocytes within the rectangle divided by its area.
In addition, several unstained, fresh samples of ribs and opercula
from four carp and four tilapia were prepared for light microscopy.
Transverse and axial rib slices were sectioned from the proximal
region of the fifth rib, and both axial and dorso-ventral strips were
prepared from the operculum. The regions from which the samples
in the different bones were obtained are schematically shown in
Fig.1. Bone samples were prepared using a slow-speed water-cooled
diamond-blade saw (Isomet® low speed saw, Buehler, Lake Bluff,
IL, USA), and their surfaces ground and polished using a
grinding–polishing device (Minimet®1000, Buehler). The prepared
surfaces were examined by a reflected-light microscope (Olympus®
BX 51 microscope, Olympus, Tokyo, Japan) at different
magnifications, and images were captured using a high-resolution
camera (Olympus® DP 71 digital camera).
Water, organic and ash content
Opercula from eight mature carp and eight mature tilapia were used
for determination of ash content. Small pieces (800–1200mg) of
each operculum were carefully dissected and cleaned of all soft
tissue, then placed in acetone solution for 12h to remove all lipids.
Samples were weighed with a semi-analytical scale (±0.05mg) to
determine lipid-free wet mass, then placed in a ceramic cup and
heated in an oven (Tuttnauer, Jerusalem, Israel) to 100°C for 3h to
remove all unbound water. After heating, the samples were
immediately weighed to determine their dry mass (organic and
mineral content) and placed (within their ceramic cups) in an oven
at 500°C for 10h, so that all organic material was burned. The
remaining ash was weighed immediately, so that the dry bone ash
content (%) of each sample could be determined. Water (wet mass
– dry mass), organic (dry mass – ash mass) and ash content were
also calculated by dividing by wet mass to determine a percentage.
Mechanical testing
Mechanical testing was performed with the samples immersed in
saline, using a custom-built micromechanical testing device (see
Fig.2). All samples were thawed immediately before testing for 24h
at 4°C, then for 1h at room temperature. The operculum beams
were tested in three-point bending; rib samples were tested in axial
compression because three-point bending tests require long, straight

Dorso-ventral
Axial
Fig.1. Schematic drawing of the samples obtained for mechanical testing.
Two kinds of rectangular beam were excised from the operculum. From
each fish, one beam was cut such that its long axis was in the axial
(cranio-caudal) direction and a second beam (from the contralateral
operculum) was cut such that its long axis was in the dorso-ventral
direction. The rib sample was removed as a short (~2.5mm long) section
obtained from the proximal, straight and uniform part of the fifth rib.
samples with a consistent cross-section, and these conditions could
not be adequately met for ribs.
Operculum samples
Two beams were prepared from opercula from each of eight carp
and seven tilapia: beams from the left operculum were oriented along
the axial orientation of the operculum, while beams prepared from
the right operculum were oriented in a perpendicular (dorso-ventral)
orientation (see Fig.1). Thirty beams of roughly similar size were
prepared in all (14 from tilapia opercula and 16 from carp opercula).
The length of the axial beams, for both types of fish, ranged between
11.4 and 21.93mm, and that of the dorso-ventral beams ranged
between 12.31 and 23.73mm. The width of the dorso-ventral beams
was 4.94±0.42mm (mean ± s.d., here and for all subsequent values
unless otherwise stated) for the carp and 5.03±0.20mm for the
tilapia. The width of the axial beams was 5.25±0.30mm for the carp
and 5.11±0.07mm for the tilapia. The thickness of the beams was
the native thickness of the operculum, and was measured for each
beam at several locations. Thickness ranged between 0.60 and
1.04mm in the carp samples and between 0.48 and 0.84mm in the
tilapia samples; however, there was never more than 5% variation
in depth across the length of any given beam.
Rib samples
Eight samples of the left fifth rib of carp and seven samples of the
left fifth rib of tilapia were prepared. A tubular segment (carp
segment length 2.66±0.04mm; tilapia segment length
2.65±0.19mm) was obtained from the proximal, reasonably straight
part of each rib, using the Isomet® slow-speed water-cooled
diamond-blade saw (see Fig.1). In compression testing (see below),
stress calculation requires knowledge of the cross-sectional area of
the sample; as biological samples do not have a perfectly uniform
cross-sectional area (CSA), we chose samples with relatively small
CSA variation (<10%) and set the effective transverse bone CSA
THE JOURNAL OF EXPERIMENTAL BIOLOGY
Structure and mechanics of acellular bone 1985
A
Fig.2. Schematic drawing of the micromechanical testing device with close-
up photographs demonstrating the loading scenario of (A) 3-point bending
of operculum beams and (B) compression testing of rib sections. Note in B
the platen is in contact with the rounded end of the anvil, allowing the
platen to compensate for slight deviations from parallelism that may occur
between the two cut edges of the rib section.
for each sample (mm2) as an average of the CSA of the proximal
and distal ends, as determined from digital images (ImageJ, NIH,
Bethesda, MD, USA, version 1.44I).
Three-point bending of operculum beams
The operculum beams were tested by three-point bending. The
beams were placed in the custom-built micromechanical testing
device within a saline-containing test chamber (see Fig.2A). A
stationary anvil was attached to the wall of the test chamber. This
anvil consisted of two supports, which were 8mm apart. A
moving anvil with a single prong was attached to a load cell (model
31, Honeywell, Sensotec, Columbus, OH, USA), which was
attached in turn to a high-precision linear motor [Physik
Instrumente (PI) GmbH, Karlsruhe, Germany]. The tip of the
moving prong was centered between the bottom supports (see
Fig.2A). All parts of the testing system (including the chamber
itself) were made of stainless steel. The parts contacting the sample
(bottom supports and upper prong) had rounded profiles (1mm
diameter) to reduce stress concentration. The upper prong was
brought into contact with the tested beams at a predetermined
preload (2N), the chamber was filled with physiological saline
solution at room temperature until the sample was fully immersed,
and bending tests were conducted under displacement control at
a rate of 500mm–1 up to a final displacement of 1500m (at this
displacement the beams were in the plastic, post-yield zone, but
had not fractured yet). Force–displacement data were collected by
custom-written software (LabView, National Instruments, Austin,
TX, USA) at 50Hz. The resulting load–displacement curves were
used to estimate the Young’s modulus of the beam material. It
should be noted that as shear deformation was not taken into
consideration, estimated values of the Young’s modulus are
slightly underestimated. However, as the mean span-to-depth ratio
of all tested beams was 12.06±3.52, this effect is relatively minor
(Spatz et al., 1996; Draper and Goodship, 2003). Young’s modulus
1986 L. Cohen and others
was calculated based on the following relationship obtained from
beam theory:
S × L3
E= , (1)
48 I
where E is the Young’s modulus of the bone material in the direction
coinciding with the long axis of the beam (Nmm–2), S is the slope
of the linear part of the force–displacement curve (Nmm–1), L is
the distance between the stationary supports (mm) and I is the
moment of inertia (mm4) of each operculum beam. The moment of
inertia I of a rectangular beam is calculated as follows:
b × d3
I= , (2)
12
where b is the beam width (mm) and d is the beam depth (mm).
Compression testing of rib samples
All rib samples were tested by axial compression under displacement
control in the same micromechanical testing device (Fig.2B). Each
sample was placed between two anvils, one of which was stationary
and attached to the wall of the test chamber, while the other was
movable and attached to a load cell, which was attached in turn to
a high-precision linear motor. The moving anvil had a well-
lubricated ball-like edge of 10mm diameter upon which a stainless
steel platen with a corresponding lubricated half-sphere cavity was
fitted, such that the platen could rotate to adjust for slight
misalignment of the sample’s edges (see Fig.2B). One end of the
sample was placed against the platen and the movable anvil was
moved carefully until it contacted the opposite end of the sample,
and a desired preload was obtained (2N). The sample was then fully
immersed in saline and loading proceeded at a constant rate of
60mmin–1, up to a total displacement of 60m.
Force–displacement data were collected by custom-written software
(LabView). The resulting load–displacement curves were analyzed
to calculate the Young’s modulus of the rib sections tested.
The modulus was calculated as the ratio of stress to strain, where
the stress was determined by:
F each species of fish. The quantitative data were compared between
σ= , (3)
A level of significance was set at P≤0.05.
where  is stress (Nmm–2), F is the applied force (N) and A is the
average bony cross-sectional area of the rib section (mm2)
determined as described above. The strain was determined by:
ΔL
ε= , (4)
L However, as the mechanical behavior of bone is strongly linked to
where  is the strain (dimensionless), L is the shortening of the
compressed rib segment under load (mm) and L is its original length
(mm). Thus the modulus could be extracted from the results of the
experiment as:
FL
E= . (5)
AΔL
As the rib segments were quite stiff, part of the movement of the
anvil was due to compression of the load cell. Load cell compliance
was determined by compressing the two stainless anvils against each
other, and evaluating the resulting load–deformation curve. Thus
for a given load, sample compression could be determined by
subtracting the load cell compression displacement from the total
sample-load cell compression displacement. Because for the same
load the deflection of the load cell was found to be about 50% of
that of the sample-load cell unit, sample stiffness was similar to that
THE JOURNAL OF EXPERIMENTAL BIOLOGY
of the load cell, making the correction valid (if the load cell was
much more compliant than the sample, such a correction would be
inaccurate).
In order to validate the results obtained by this experimental setup,
first a hollow circular tube was prepared from Delrin®
(polyoxymethylene). This material has known mechanical properties
(the Young’s modulus of Delrin® is 3.2GPa, quite similar to that
of the bones tested here). The tube was 2.5mm long, with a 1.3mm
o.d. and 0.95mm i.d., and was thus similar in dimensions to the
tested rib sections. The tube was compressed 4 times and the
Young’s modulus for Delrin® determined by these experiments was
2.92±0.096GPa, within an acceptable margin of error of the
material’s actual stiffness.
Micro-CT scanning
The tubular rib sections and opercular beams were scanned using
a micro-CT scanner (1174 scanner, SkyScan®, Kontich, Belgium).
The X-ray source was set at 50kVp (peak kilovoltage) and 800A.
In total, 450 projections were acquired over an angular range of
180deg. The operculum samples of both carp and tilapia were
scanned with an isotropic voxel size of 27.6m and an integration
time of 3900ms, while the carp and tilapia rib samples were scanned
at an isotropic voxel size of 11.1 and 8.6m, with integration times
of 3600 and 3500ms, respectively. All scans were performed with
a 0.25mm aluminium filter to decrease beam-hardening effects.
Scans were reconstructed and analysis was performed using
commercial software (NRecon® SkyScan® software, version 1.6.1.2
and CT analyser® SkyScan® software, version 1.9.3.2, respectively).
Bone mineral density (BMD) was determined based on calibration
with two phantoms of known mineral density (0.25 and
0.75mgcm–3) supplied by SkyScan®, which were scanned under
exactly the same conditions as the operculum and rib specimens.
Statistics
Statistical analyses of quantitative data (BMD, ash content and
mechanical testing results) were made to find differences between
carp and tilapia groups, as well as between ribs and opercula within
and within groups by the non-parametric Mann–Whitney test. The
RESULTS
Histology and light microscopy
The current study focused primarily on the comparison of the
material properties of the bones of these two teleost species.
its structure, we also provide a brief description of some features
discerned from a light microscopy study of stained and unstained
bone samples.
Fish opercula are bony plates, thin and slightly curved in the
medio-lateral direction. Both tilapia and carp opercula exhibited
three distinct tissue zones, observable under reflected light
microscopy in either coronal sections (i.e. opercula divided into
rostral and caudal sections) or transverse sections (i.e. divided into
dorsal and ventral sections). The lateral (outer) and medial (inner)
thirds of the thickness of the bone showed clear lamellar organization
(see area bracketed in Fig.3B), sandwiching a middle, less-ordered
(isotropic) tissue (see area bracketed in Fig.3A), similar in
appearance to ‘woven’ bone (Currey, 2008). The blood supply to
all bones in both types of fish is quite sparse, but blood vessels
tended to be localized in the disordered woven zone. The lateral
and medial lamellar zones appeared thicker and more compact in

A B
100µm
C D
25µm
E F
25µm
G H
50µm
I J
50µm
tilapia than in carp. A similar layering of tissue zones was also
observed in frontal (i.e. perpendicular to the long axis) and
longitudinal cross-sections of ribs of both species, where a less-
organized inner zone (sometimes perforated by the central lumina
of the ribs) was surrounded by an outer zone of concentrically
ordered layers (Fig.3F).
When samples of opercula of both species were polished on their
medial or lateral faces (i.e. perpendicular to the sections mentioned
above), no distinct tissue layering or zonal arrangement was visible
(Fig.4). However, on these polished surfaces, we consistently saw
dense collections of what appeared to be small voids in the tissue,
similar in size and density in the two bone types and typically smaller
THE JOURNAL OF EXPERIMENTAL BIOLOGY
Structure and mechanics of acellular bone 1987
Fig.3. Light microscopy and histology images of operculum
and rib samples from carp (left column) and tilapia (right
column). (A)Low magnification cross-section of carp
operculum. White bracket indicates the extent of the
disorganized (ʻwovenʼ bone) mid-region. (B)Low magnification
cross-section of tilapia operculum. Black bracket indicates the
extent of the layered region. (C)High magnification cross-
section of carp operculum. Black arrows point to osteocyte
lacunae. Note the woven bone region on the right of the
image. (D)High magnification cross-section of tilapia
100µm operculum. (E)High magnification flat medial surface of the
carp operculum, showing osteocyte lacunae (black arrow) and
their network of canaliculi (black arrowheads). (F)Axial cross-
section of a rib from a tilapia, demonstrating outer layered
zones, sandwiching a woven interior zone. (G)Cross-section
of carp operculum stained with H&E, with osteocyte nuclei
visible within lacunae (black arrows). (H)Cross-section of
tilapia operculum stained with H&E. Note the absence of cells.
The structural layering arrangement can be clearly seen.
(I)Cross-section of carp rib stained with H&E. Osteocyte
nuclei can be clearly seen within lacunae (black arrows).
25µm
(J)Cross-section of tilapia rib stained with H&E. Note the
layered structure and absence of cells.
100µm
50µm
50µm
and with a much higher density than the osteocytic lacunae in the
cellular carp bone. The voids appeared as small ‘pores’ (2–4m
wide) and/or wavy and occasionally branched ‘lines’ (~15–40m
long) (Fig.4); we took these to be the same structural features, but
viewed in different orientations, and therefore interpret the structures
as tapered or blind-ended tubules. As tubules were only visible in
median (and not coronal or transverse) sections of opercula, we
verified them to be consistent structural features and not preparation
artifacts by examining multiple additional tissue sections from
individuals of both species, prepared using a variety of methods
(embedded vs non-embedded samples, polished vs unpolished, etc.),
but the appearance described above remained consistent. Although
1988 L. Cohen and others
A B
200µm
C D
10µm 200µm 10µm
opercula polished in the median plane showed no discrete tissue
zones, tubules appeared to clump spatially according to their
orientation, with clusters of pores and clusters of wavy lines
appearing to localize separately (Fig.4A). These clusters did not
seem to be linked to the medio-lateral position in the operculum
(e.g. ‘pores’ did not appear to be localized only in the outer, lamellar
layer). Similar tubules were observed in rib samples as well but
with a lower density than seen in opercula.
Osteocyte distribution and density
Osteocytes can be seen within lacunae scattered within the sections
obtained from the carp operculum and rib (black arrows in Fig.3G,I).
Osteocyte lacunae showing clearly stained nuclei were counted in
several randomly selected cross-sections of four different carp
and yielded a mean (±standard deviation) density of
424±33osteocytesmm–2. There are no adequate data for lacunar
shapes and sizes for fishes, but if we assume that they are similar
to those seen in mammals [i.e. elliptical in shape with major and
minor diameters of 10 and 5m, respectively (McCreadie et al.,
2004; Hannah et al., 2010)] the area fraction taken by the osteocytic
lacunae in carp bones is 1.6%. The osteocytic lacunae (black arrows
in Fig.3C,E) in the bones of the carp show distinct canalicular
networks interconnecting adjacent lacunae (black arrowhead in
Fig.3E); however, their distribution is less dense than in mammals
(R.S., personal observation) (Skedros, 2005; Li et al., 1991).
BMD
The BMD of carp opercula, determined by micro-CT, was
significantly (P0.001) lower than the BMD of the tilapia opercula
(median 766mgcm–3, range 735–824mgcm–3 and median
912mgcm–3, range 883–963mgcm–3, respectively, see Fig.5).
Similarly, the BMD of carp rib was significantly (P0.001)
lower than the BMD of tilapia rib (median 785mgcm–3,
range 737–837mgcm–3 and median 1066mgcm–3, range
1038–1102mgcm–3, respectively, see Fig.5). In tilapia, the BMD
of the operculum was significantly (P0.002) lower than the BMD
of the rib (Fig.5), while no significant difference was observed
THE JOURNAL OF EXPERIMENTAL BIOLOGY
Fig.4. Tubules seen in the paramedian plane of
the opercular bone of carp (A,B) and tilapia (C,D),
viewed under reflected light microscopy
(unstained). Arrows in each image indicate
individual tubules; however, there are many
tubules visible in each panel. All images show
predominantly end-on views of tubules except D,
where some tubules are visible in a more
longitudinal view (the sample is somewhat
transparent and so structures are visible beneath
the surface). Higher magnification views of
morphologies shown in C and D are presented in
20µm the insets in those images. All of these
morphologies were seen on opercular samples
polished from both lateral (A,D) and medial (B,C)
sides and also in ribs. Note the different zones of
tissue orientation/organization in A, and also how
tubule openings follow the tissue contours around
the vessel in the center of B, supporting our
conclusion that these are structural features of
both bone types and not artifacts.
20µm
between the BMD of the operculum and rib of carp (Fig.5), although
there was a tendency towards a lower BMD in the operculum
(P0.09).
Youngʼs modulus
The computed Young’s moduli of the opercula and ribs of both
types of fish (tilapia and carp) are presented in Fig.6. The mean
value of Young’s moduli obtained from the operculum samples
reveals that the stiffness of the acellular bones of tilapia is similar
to that of the cellular bones of the carp, with values ranging between
3.5 and 10.87GPa. In the carp, the Young’s moduli of the operculum
bone in the axial and dorso-ventral directions were not statistically
different (medians of 4.1 and 5.7GPa, respectively, and ranges of
3.6–10.0 and 3.9–6.8GPa, respectively, P0.29), suggesting that this
bone is transversely isotropic. In the tilapia opercula, the median
Young’s modulus in the axial direction (7.7GPa, range
5.3–10.9GPa) was approximately 40% higher than in the dorso-
ventral direction (5.8 GPa, range 5.1–6.0GPa).
The rib sections were loaded in compression along the long axis
of the rib. The median values of Young’s modulus obtained
(8.6GPa for tilapia, range 4.1–11.0GPa, and 7.5GPa for carp, range
3.6–14.5GPa) were approximately 40% higher than those found for
the dorso-ventral orientation of the operculum.
Water, organic and ash content
The dry bone ash content of opercula in tilapia (64.49%) is higher
than that of carp (57.58%), in agreement with the values reported
above for bone mineral density (Fig.7A). Water, ash and organic
content relative to lipid-free wet mass were 14.2%, 49.4% and
36.4%, respectively, for carp, and 14.0%, 55.5% and 30.5%,
respectively, for tilapia (Fig.7B).
DISCUSSION
In this study we compared the material composition (BMD; ash,
water and organic content), structure and mechanical properties
between the acellular opercula and ribs of a neoteleost fish (O.
aureus) and those of a more basal teleost (the carp C. carpio) whose

1200
Rib
1100
BMD (mg cm–3)
1000 Operculum
900
Rib Operculum
800
700
600
Carp Tilapia
Fish species
Fig.5. Box plot of bone mineral density (BMD) values in carp and tilapia
opercula and ribs.
bones are cellular. Whereas material composition was shown to vary
significantly between cellular and acellular bone in the species tested
here, and gross structure to vary to some degree, the mechanical
properties were shown to be largely similar (with the exception of
the orientation-dependent moduli of tilapia opercula).
Structurally, the tissues are easily visibly distinguished from one
another by the presence and absence of osteocytes, but are more
similar to each other in gross, histological morphology than they
are to the more-studied cellular bone types of amniotes. In particular,
differences in osteocyte presence and associated canalicular density
are striking, even for the cellular boned carp where osteocytes are
much less densely packed than those of birds and mammals (424
vs 700–1800cellsmm–2) (Li et al., 1991; Skedros, 2005). The low
osteocyte density, coupled with the apparent relative scarcity of
blood vessels in both tilapia and carp bone, make both fish bone
types observed here appear, at lower magnifications, as more solid
(less porous) blocks of tissue compared with amniote bone, also
suggesting they may have a lower metabolic rate. Our data and
published work on medaka (Ekanayake and Hall, 1988) point to a
relative lack of blood vessels in teleost bone; however, Moss reported
extremely vascular regions in both acellular and cellular bone (Moss,
1963) and so it remains unclear whether fish bone as a rule is less
vascular than amniote bone.
Although lacking the rich cellular patterning of most amniote
bones, carp and tilapia bones still exhibited pronounced structural
features. The obvious tissue zones, viewed in coronal/transverse
cross-sections of opercula and ribs, suggest differences in underlying
collagen organization. In particular, the disorganized central (woven)
region may be the remains of the earliest primary bone formed during
development, whereas the laminated lateral and medial layers point
to a later process of layered tissue deposition in both types of bone.
This finding agrees with those of Moss (Moss, 1960; Moss, 1961),
who described fish bone (both cellular and acellular) as a mixture
of lamellar and woven bone, and is further supported by the common
use of opercular annuli to age post-hatch individuals of many species
of bony fishes (e.g. Frost and Kipling, 1959; Vilizzi and Walker,
1999; Khan and Khan, 2009). The observed histological differences
therefore likely reflect underlying structural patterning, as collagen
organization has been shown to play a role in mineral platelet
packing, size and distribution in vertebrate bones (Weiner et al.,
1999; Currey, 2008; Bigi et al., 2000). It is unclear to what degree
fish endoskeletal bones exhibit the hierarchical structural
organization of amniote bony tissues, but one study of endoskeletal
THE JOURNAL OF EXPERIMENTAL BIOLOGY
Structure and mechanics of acellular bone 1989
12
A
10
8
6
4
2
Young’s modulus (GPa)
0
Carp Carp Tilapia Tilapia
axial DV axial DV
Fish species and beam orientation
16
B
14
12
10
8
6
4
2
0
Carp Tilapia
Fish species
Fig.6. Youngʼs moduli of (A) carp and tilapia operculum bone in the dorso-
ventral (DV) and axial directions (from bending tests), and of (B) carp and
tilapia fifth rib bone (from compression tests).
bone in carp (Bigi et al., 2000) and studies of intramuscular fish
bone (mineralized myoseptal tendons) (e.g. Lee and Glimcher, 1991;
Zhou et al., 2007) and scales (made of acellular bony tissue in both
cellular and acellular boned species, yet perhaps dental in
evolutionary origin) (e.g. Meunier, 1987; Bigi et al., 2000; Sire and
Huysseune, 2003) point to structural systems in fish bone types that
are similar to but simpler than amniote bone, where tissue layers
differ in their collagen orientation (Lee and Glimcher, 1991; Currey,
2008). The visible evidence of tissue organization in the ribs and
opercula of the two species in our study provides a roadmap for
future biological and material science examinations into fundamental
questions of growth, mineralization and organization in fish bones.
The observed interspecific differences in tissue layers (e.g. the
thicker woven region in carp opercula relative to that of tilapia) also
offer a valuable opportunity for testing and modeling basic tissue-
specific structure–function relationships.
The dense network of narrow tubules (2–4m wide, ~15–40m
long), a major structural feature observed under reflected light
microscopy in ground and polished sections, remains more difficult
to interpret (Fig.4). These structures do not appear to be artifacts,
as we observed them in rib and opercular bones of both species
(albeit with a higher density in opercular bone) and under multiple
forms of sample preparation. However, their structural interpretation
and the extent to which they perforate the thickness of the skeletal
element are not clear. The tubules are apparently not associated with
osteocytes or vascular tissue, and bear strong resemblance to the
canaliculi of Williamson [~2–4m wide canals with an apparent
nutritive function in the cellular dermal skeletons and endoskeletons
of basal bony fishes (Sire and Meunier, 1994)] and endoskeletal
tubules (~2–8m wide) observed in the bones of several teleost
1990 L. Cohen and others

68 A B
66

% Ash content of dry bone

64

62

60

58

56
Carp Tilapia

Fig.7. (A)Ash content (percentage by mass) of the operculum of carp and tilapia. (B)Ternary diagram of the constituents of various vertebrate skeletal
elements, compiled and modified from Currey (Currey, 2002) by Macesic and Summers (Macesic and Summers, 2012), and including data from our study.
The diagram shows bone from terrestrial vertebrates (gray circles), teleost fish (black circles) and calcified cartilage (white circles) from the propterygia
(pelvic skeletal elements composed of an unmineralized cartilage core covered by a mineralized rind) and vertebral column of elasmobranch fishes (sharks,
rays and relatives). Relative proportions were determined by percentage wet mass. Comparative data were obtained from previous publications (Currey,
2002; Macesic and Summers, 2012; Porter et al., 2006; Toppe et al., 2007; Zioupos et al., 2000).

species and containing thin Sharpey’s-like fibers, probably periosteal
in origin (Kölliker, 1857) [figs11,12 in Moss (Moss, 1961); figs4–6
in Hughes et al. (Hughes et al., 1994)]. It is possible also that the
features we observed are the bundles of ‘uncalcified collagen fibers’
described by Moss (Moss, 1960; Moss, 1963) [which may also be
the ‘T-fibers’ of Hughes et al. (Hughes et al., 1994)], anchored
directly in the matrix but not housed in tubes; however, the pored
end-on perspective of these structures argues they are tubular in
nature (e.g. Fig.4B,C). We could not verify the contents or detail
the structure and distribution of these tubules; however, their
presence in both an acanthopterygian fish with acellular bones
(tilapia) and an ostariophysian fish with cellular bones (carp), and
the reported presence of similar structures in bones/scales of basal
cellular bony fishes and several other species of acellular
acanthopterygian fishes, suggests these are extremely widespread
structural features. A three-dimensional analysis of tissue patterning
would provide fundamental insight into the material organization
of fish bone, and the simple shape of the opercula and ribs make
them ideal skeletal elements for examination. Studies are currently
underway to examine these features in three dimensions using high-
resolution X-ray imaging and scanning electron microscopy.
Despite gross structural similarities between the bone types here
(aside from cell presence or absence), the acellular operculum and
ribs of tilapia have a much higher mineral density (by as much as
25%) and ash content (by about 10%) than the cellular bones of the
carp. The lower mineral density of cellular bone could be attributed
to its higher porosity due to the presence of lacunae (Currey, 2008);
however, the low percentage volume occupied by the osteocytic
lacunae in the carp [estimated in our study to be around 1.6%,
compared with ~5–9% in amniotes (Currey, 1988)] makes such an
explanation unlikely. A literature review of percentage ash values
for acellular and cellular fish bone (N>20 species for each bone
type; M.D., J. W. C. Dunlop and R.S., in preparation) suggests that
a lower mineral content is a consistent feature of cellular bone (~50%
ash) relative to acellular bone (~58% ash); these differences are also
evident in Toppe et al.’s (Toppe et al., 2007) comparison of several
acellular and cellular species. Furthermore, compared with amniote
bones, which have ~61–74% dry ash content (Currey, 2008; Szpak,
2011), fish bones are in general less well mineralized (Biltz and
Pellegrino, 1969; Lees, 1987a; Szpak, 2011; Toppe et al., 2007)
(Fig.7B). This supports theoretical conjectures that mineral is less
tightly packed among collagen fibrils in fish bone relative to
mammalian bone (Lees et al., 1984; Lees, 1987a; Lees, 1987b), but
the reasons for the differences among fish species are unclear; a
much wider array of species must be studied to understand this
feature in a phylogenetic context and studies correlating bulk tissue
mineral density and nanoscale mineral packing would be very
instructive.
As the BMD of cellular fish bone is significantly lower than that
of acellular fish bone, we expected Young’s modulus to also be
much lower. The Young’s modulus of bone is determined by its
composition (mostly mineral density and porosity) and by its
microarchitecture. However, counter to our predictions, values of
Young’s moduli of bone material obtained from samples of ribs
and opercula from the two fish types tested here were quite similar:
for both cellular and acellular fish bone, median Young’s modulus
values were found to be in the range 4.1–8.5GPa. The reason for
the similarity of material stiffness in bones with such a large
difference in mineral density is not clear, but surely relates to some
degree to the organization and interactions of mineral and organic
components in the tissues (Fig.7B).
There are very few studies reporting the material properties of
fish bones; however, the values for Young’s moduli observed in
this study are within the range reported (3.7–8.4GPa) recently for
the acellular rib bone of another fish species, the great sculpin,
Myoxocephalus polyacanthocephalus (Horton and Summers, 2009)
(Fig.8A). Such values are lower than those reported for the cortical
bones of birds and mammals. For instance, Young’s modulus of
equine cortical bone was reported to be between 17.4 and 23.5GPa
in the axial direction, and between 9.3 and 13.9GPa in the transverse
direction (Shahar et al., 2007). Currey reported on the Young’s
modulus of a wide range of amniote bones (but no fish), which were
all tested by a similar protocol, and with the exception of antlers
(which are particularly compliant because of their exceptionally low
mineral content), the range of values reported fell between 14.2 and
28.2GPa (Currey, 1999). Even in 4week old mice, which could be
expected to have low moduli as a result of their young age and
consequent low mineral content, the axial Young’s modulus of

THE JOURNAL OF EXPERIMENTAL BIOLOGY

 Structure and mechanics of acellular bone 1991

Kranenbarg et al., 2005a

A Cellular Acellular
Species with material data

Polypteriformes
17.6 GPaB 1
P. senegalensis
Acipenseriformes
15.5 GPaI 2
Lepisosteiformes C. harengus

Amiiformes 8.51 GPaB,C I 3,4

Actinopterygii
C. carpio
Osteoglossomorpha
8.50 GPaI 5,6
Elopomorpha
D. rerio

Teleostei 16.68 GPaH 7

Clupeomorpha
O. mykiss
B Unpublished
tilapia data Ostariophysi 14.30 GPaH 7
14

S. trutta
12 Protacanthopterygii
Modulus (GPa)

10
7.20 GPaB,C 3
Inc.
bichir Basal neoteleosts
data
8 O. aureus

6
Paracanthopterygii
Without Euteleostei 6.48 GPaB 8
bichir data
4
Acanthopterygii
Comp./ Hard./
bend. indent. Neoteleostei M. polyacanthocephalus

Fig.8. (A)Phylogeny of actinopterygian fishes detailing the current state of knowledge of the phylogenetic distribution and material properties of cellular (red)
and acellular (green) bone. Pie charts at the tips of the phylogeny represent the relative proportions of cellular and acellular boned species sampled for each
group, as reported from a literature consensus by Kranenbarg et al. (Kranenbarg et al., 2005a). Pie chart size is scaled according to the number of species
with available data, not the overall number of species in each group (e.g. Kranenbarg et al. collected data for 281 acanthopterygian fishes but that group
contains >10,000 species); however, the differences in pie chart size among groups are roughly representative of group size (e.g. there are many more
acanthopterygian species than polypteriform species). The eight species for which material properties data have been reported in the literature are listed in
the right-hand column, colored according to bone type and with blue dashed arrows indicating the phylogenetic groups to which they belong. Mean moduli
are presented inside each fish icon with superscript letters representing test method (B, bending; C, compression; I, indentation; H, hardness) and numbers
representing literature sources (see below). Mean moduli per species have been, in many cases, averaged from a variety of skeletal elements (e.g. ribs,
opercula, vertebrae, intramuscular bones) and tissue orientations, and from data collected by non-equivalent methods (e.g. nanoindentation and bending);
values are therefore provided only to show the approximate range and order of magnitude of existing data for fish bone. We strongly urge readers to consult
the original works and not cite these values directly. (B)Means and standard deviations for bone moduli versus testing method (comp./bend.,
compression/bending; hard./indent., hardness/indentation) for the eight examined cellular and acellular boned species. Many of the cellular bone samples
were tested dry and by hardness/indentation, therefore likely over-estimating the average modulus for cellular bone relative to acellular bone. This inflation
of modulus is emphasized by nanoindentation data for tilapia opercula (shown; A.S., unpublished data), which are approximately two times larger than the
modulus we report in three-point bending. Three-point bending data for bichir (Polypteriformes) pectoral girdle (cellular) report a particularly high Youngʼs
modulus (17.6GPa) and standard deviation (7.8GPa), suggesting either abnormal properties relative to cellular bone from other fishes or that three-point
bending was an improper test for a skeletal element with such a non-uniform shape; we therefore show the effect of removing this value from the cellular
bone data for bending/compression. Pie charts and phylogeny modified from fig.1 of Kranenbarg et al. (Kranenbarg et al., 2005a), with tree topology
adjusted to reflect the current hypothesis of sister group relationship between Clupeomorpha and Ostariophysi (e.g. Zaragüeta-Bagils et al., 2002).
References for material property data: 1(Erickson et al., 2002); 2(Rho et al., 2001); 3this study; 4(Roy et al., 2000); 5(Zhang et al., 2002); 6(Wang et al.,
2002); 7(Paxton et al., 2006) – we estimate Youngʼs modulus from their microhardness data using an equation from Rapoff et al. (Rapoff et al., 2008);
8
(Horton and Summers, 2009).

cortical bone was reported to be 8.9±1.1GPa (Lev-Tov Chattah et
al., 2009).
The relatively low stiffness of fish bone is likely related to its
comparatively low mineral content, but whether the lower stiffness
in fact confers a functional advantage on fishes (e.g. whether more
compliant bone is better suited to fish ecologies) is unknown.
Fish bone stiffness is probably also related to some unique features
of its microarchitecture. In the carp (cellular bone), the moduli
in orthogonal directions within the median plane of the operculum
were similar, and substantially lower than the axial modulus
measured in the rib. In tilapia (acellular bone), the axial modulus
in the rib and the modulus in the antero-posterior orientation in
the transverse plane of the operculum were similar, and much
higher (by about 50%) than the modulus in the dorso-ventral
orientation in the transverse plane of the operculum. These
findings suggest that there may be an inherent difference between
the hierarchical architecture of acellular and cellular bone, at least
in the bones of these two fishes, beyond the presence/absence of
osteocytes and their lacunae. We believe this will be clarified
through analyses of the ontogeny of tissue mineralization and of

THE JOURNAL OF EXPERIMENTAL BIOLOGY

1992 L. Cohen and others
crystal orientation in species with cellular and acellular bone
types. It would be of particular interest to examine how tissue
material properties and mineral microstructure change in fish
bones that show large-scale ontogenetic changes or spatial grades
in the degree of cellularity (Moss, 1961; Moss, 1963; Witten et
al., 2001; Witten and Huysseune, 2009).
It is important to note that material property data for bone are
highly dependent on sample preparation and testing techniques, and
therefore comparisons among studies that used different
methodologies may not be valid. To the best of our knowledge,
only eight studies (including ours) have performed mechanical tests
aimed at determining material properties of fish bone, and only two
of these [ours and that of Horton and Summers (Horton and
Summers, 2009)] have focused on acellular bone (see Fig.8).
Ignoring for a moment the differences in sample orientation (i.e.
axial vs transverse tests), the skeletal element examined and
phylogeny (Fig.8A), the combined data from these studies suggest
that cellular bone, on average, is stiffer than acellular bone (11.4±2.6
vs 7.0±1.3GPa) (Fig.8B). However, the majority of studies
examining cellular bone tested dried samples using nanoindentation,
a technique for measuring nanoscale material moduli. Not only is
the indentation modulus often higher for dry than for hydrated
biological samples (Guidoni et al., 2006; Enders et al., 2004) but
also it is unclear how it can be compared with Young’s moduli
obtained by common techniques used to test materials at the
micrometer scale, such as bending, tension and compression tests
(Hengsberger et al., 2003). When all available fish material property
data are plotted according to testing method, the values generated
by nanoindentation (i.e. the majority of data for fish cellular bone)
are higher and show a wider range of variation (Fig.8B). This effect
of testing method is underlined by nanoindentation data of carp (Roy
et al., 2000), which provided an indentation modulus more than twice
our reported Young’s modulus in compression from axial tests, and
by nanoindentation data of the same tilapia opercula used in this
study but tested dry (A.S., unpublished data), which provided an
indentation modulus of 14.5±1.5GPa (Fig.8B), approximately twice
as high as the Young’s modulus we report from three-point bending
of the same tissue. Based on the likelihood that existing data for
fish bone material properties are test dependent and on our
comparison of tilapia and carp bone, we posit that cellular and
acellular bone are, in general, equivalent from the standpoint of
material stiffness.
Our study presents the first comparison of acellular and cellular
fish bone materials, performed by uniform methodology on samples
of standardized size. Yet, although these species provide valuable
reference points for acellular and cellular bone, they are quite
phylogenetically disparate (Fig.8A), with one a derived perciform
neoteleost (tilapia: Acanthopterygii) and the other a more basal
teleost (carp: Ostariophysi). In fact, the scant data on fish bone
material properties have been drawn from a very wide phylogenetic
sampling, made up of one polypteriform fish [Polypterus
senegalensis (Erickson et al., 2002)] and one to two representatives
each from four superorders of Teleostei: Clupeomorpha [Clupea
harengus (Rho et al., 2001)], Ostariophysi [Danio rerio (Wang et
al., 2002; Zhang et al., 2002); Cyprinus carpio (this study) (Roy et
al., 2000)], Protacanthopterygii [Oncorhynchus mykiss, Salmo trutta
(Paxton et al., 2006)] and Acanthopterygii [Myoxocephalus
polyacanthocephalus (Horton and Summers, 2009); Oreochromis
aureus (this study)]. All of the studied species are freshwater fishes
except M. polyacanthocephalus and C. harengus, and the two
acanthoptherygian species are the only acellular boned fish that have
been examined (Fig.8).
THE JOURNAL OF EXPERIMENTAL BIOLOGY
Our data suggest that fish bone has fundamental structural,
compositional and material performance differences relative to other
bony tissues; yet it remains unclear what selective pressures might
have led to the loss of osteocytes in some fish groups. We are
currently exploring the microstructure of cellular and acellular fish
bone in multiple species. These analyses will allow us to test the
hypothesis that differences in fibril structure and fibril-array
arrangement are responsible for the differences found between fish
bone types (i.e. mineral density and isotropy), and between fish bone
and that of other vertebrates (e.g. low stiffness and cellularity). To
provide a useful phylogenetic perspective and allow accurate
comparisons of mechanical property data for acellular and cellular
fish bone, it would be valuable for future work on fish bone to
standardize material testing methodology and, if possible, to choose
test species according to those phylogenetic groups for which there
are existing data. It would be particularly useful to see comparisons
within the Perciformes, which have many species of convenient size
for testing and, although largely acellular boned, also contain several
reportedly cellular-boned species within the Scombrinae (Kölliker,
1857; Moss, 1961). The Protocanthopterygii also represent a
particularly interesting group for study as they possess nearly equal
numbers of cellular- and acellular-boned species (Moss, 1961;
Kranenbarg et al., 2005a; Kranenbarg et al., 2005b).
In the comparison of carp and tilapia bone, the lack of cells does
not seem to confer an obvious material performance advantage
relative to cellular fish bones; as such, the cause of the evolutionary
development towards acellularity in neoteleost fish bone remains
unclear. An explanation previously offered was that acellularity
could increase the stiffness of fish bone, which is low relative to
that of mammalian bone (Horton and Summers, 2009). Our data
support Horton and Summers’ rejection of this hypothesis because
the bones of the more basal teleost, which are cellular, were shown
here to have a similar stiffness to that of acellular bones from a
neoteleost, despite the presence of osteocytic lacunae. Alternatively,
the primary function of osteocytes may be osteocytic osteolysis
(Clark et al., 2005; Bonucci, 2009; Witten and Huysseune, 2009;
Bonewald, 2011). As fish are unlikely to face calcium deficiency
given the high availability of minerals in their natural habitat (both
marine and fresh water), selective pressure would have favored the
loss of osteocytes, as the high metabolic cost of their maintenance
would outweigh the benefit. The loss of osteocytes in fishes then
implies that the other functions typically performed by these cells
are either of lesser importance to fish bone or are addressed through
as yet unidentified mechanisms.
ACKNOWLEDGEMENTS
The authors wish to thank Prof. John Currey for critically reviewing the manuscript
and making many useful comments, Dr Ofer Ashulin and Prof. Berta Sivan for
help with obtaining samples, Dr Joshua Milgram for help with sample preparation
and Dr Gilad Segev for help with statistical analysis.
FUNDING
This work was supported in part by an Alexander von Humboldt Post-doctoral
Fellowship to M.D.
REFERENCES
Adachi, T., Aonuma, Y., Ito, S.-I., Tanaka, M., Hojo, M., Takano-Yamamoto, T. and
Kamioka, H. (2009). Osteocyte calcium signaling response to bone matrix
deformation. J. Biomech. 42, 2507-2512.
Bigi, A., Koch, M., Panzavolta, S., Roveri, N. and Rubini, K. (2000). Structural
aspects of the calcification process of lower vertebrate collagen. Conn. Tiss. Res.
41, 37-43.
Biltz, R. M. and Pellegrino, E. D. (1969). The chemical anatomy of bone: I. A
comparative study of bone composition in sixteen vertebrates. J. Bone Joint Surg.
51, 456-466.
Bonewald, L. (2011). The amazing osteocyte. J. Bone. Miner. Res. 26, 229-238.

Bonucci, E. (2009). The osteocyte: the underestimated conductor of the bone
orchestra. Rend. Lincei Sci. Fis. Nat. 20, 237-254.
Clark, W. D., Smith. E. L., Linn. K. A., Paul-Murphy, J. R., Muir, P. and Cook, M. E.
(2005). Osteocyte apoptosis and osteoclast presence in chicken radii 0-4 days
following osteotomy. Calcif. Tissue Int. 77, 327-336.
Currey, J. (1988). The effect of porosity and mineral content on the Youngʼs modulus
of elasticity of compact bone. J. Biomech. 21, 131-139.
Currey, J. D. (1999). The design of mineralized hard tissues for their mechanical
functions. J. Exp. Biol. 202, 3285-3294.
Currey, J. D. (2002). Bones, pp. 436. Princeton: Princeton University Press.
Currey, J. D. (2008). Collagen and the mechanical properties of bone and calcified
cartilage. In Collagen: Structure and Mechanics (ed. P. Fratzl), pp. 397-420. New
York: Springer.
Currey, J. D. (2010). Mechanical properties and adaptations of some less familiar
bony tissues. J. Mech. Behav. Biomed. Mater. 3, 357-372.
Draper, E. R. C. and Goodship, A. E. (2003). A novel technique for four-point bending
of small bone samples with semi-automatic analysis. J. Biomech. 36, 1497-1502.
Ekanayake, S. and Hall, B. K. (1988). Ultrastructure of the osteogenesis of acellular
vertebral bone in the Japanese Medaka, Oryzias latipes (TeIeostei, Cyprinidontidae).
Am. J. Anat. 182, 241-249.
Enders, S., Barbakadse, N., Gorb, S. and Arzt, E. (2004). Exploring biological
surfaces by nanoindentation. J. Mater. Res. 19, 880-887.
Erickson, G. M., Catanese, J., 3rd and Keaveny, T. M. (2002). Evolution of the
biomechanical properties of the femur. Anat. Rec. 268, 115-124.
Franz-Odendaal, T. A., Hall, B. K. and Witten, P. E. (2006). Buried alive: how
osteoblasts become osteocytes. Dev. Dyn. 235, 176-190.
Frost, W. E. and Kipling, C. (1959). The determination of the age and growth of pike
(Esox lucius L.) from scales and opercular bones. ICES J. Mar. Sci. 24, 314-341.
Guidoni, G., Denkmayr, J., Schöberl, T. and Jager, I. (2006). Nanoindentation in
teeth: influence of experimental conditions on local mechanical properties. Philos.
Mag. 86, 5705-5714.
Hannah, K. M., Thomas, C. D. L., Clement, J. G., De Carlo, F. and Peele, A. G.
(2010). Bimodal distribution of osteocyte lacunar size in the human femoral cortex as
revealed by micro-CT. Bone 47, 866-871.
Hegrenes, S. (2001). Diet-induced phenotypic plasticity of feeding morphology in the
orangespotted sunfish, Lepomis humilis. Ecol. Freshwater Fish. 10, 35-42.
Hengsberger, S., Enstroem, J., Peyrin, F. and Zysset, P. (2003). How is the
indentation modulus of bone tissue related to its macroscopic elastic response? A
validation study. J. Biomech. 36, 1503-1509.
Horton, J. M. and Summers, A. P. (2009). The material properties of acellular bone in
a teleost fish. J. Exp. Biol. 212, 1413-1420.
Hughes, D. R., Bassett, J. R. and Moffat, L. A. (1994). Histological identification of
osteocytes in the allegedly acellular bone of the sea breams Acanthopagrus
australis, Pagrus auratus and Rhabdosargus sarba (Sparidae, Perciformes,
Teleostei). Anat. Embryol. 190, 163-179.
Huysseune, A., Sire, J. Y. and Meunier, F. J. (1994). Comparative study of lower
pharyngeal jaw structure in two phenotypes of Astatoreochromis alluaudi (Teleostei:
Cichlidae). J. Morphol. 221, 25-43.
Khan, M. A. and Khan, S. (2009). Comparison of age estimates from scale, opercular
bone, otolith, vertebrae and dorsal fin ray in Labeo rohita (Hamilton), Catla catla
(Hamilton) and Channa marulius (Hamilton). Fish. Res. 100, 255-259.
Kilhara, M., Ogata, S., Kawano, N., Kubota, I. and Yamaguchi, R. (2002). Lordosis
induction in juvenile red sea bream, Pagrus major, by high swimming activity.
Aquaculture 212, 149-158.
Kitamura, K., Suzuki, N., Sato, Y., Nemoto, T., Ikegame, M., Shimizu, N., Kondo,
T., Furusawa, Y., Shigehito, W. and Hattori, A. (2010). Osteoblast activity in the
goldfish scale responds sensitively to mechanical stress. Comp. Biochem. Physiol.
156A, 357-363.
Kölliker, A. (1857). On the different types in the microscopic structure of the skeleton
of osseous fishes. Proc. R. Soc. Lond. B 9, 656-668.
Kranenbarg, S., van Cleynenbreugel, T., Schipper, H. and van Leeuwen, J. L.
(2005a). Adaptive bone formation in acellular vertebrae of sea bass (Dicentrarchus
labrax L.). J. Exp. Biol. 208, 3493-3502.
Kranenbarg, S., Waarsing, J. H., Muller, M., Weinans, H. and van Leeuwen, J. L.
(2005b). Lordotic vertebrae in sea bass (Dicentrarchus labrax L.) are adapted to
increased loads. J. Biomech. 38, 1239-1246.
Lee, D. D. and Glimcher, M. J. (1991). Three-dimensional spatial relationship
between the collagen fibrils and the inorganic calcium phosphate crystals of pickerel
(Americanus americanus) and herring (Clupea harengus) bone. J. Mol. Biol. 217,
487-501.
Lees, S. (1987a). Considerations regarding the structure of the mammalian
mineralized osteoid from viewpoint of the generalized packing model. Conn. Tiss.
Res. 16, 281-303.
Lees, S. (1987b). Possible effect between the molecular packing of collagen and the
composition of bony tissues. Int. J. Biol. Macromol. 9, 321-326.
Lees, S., Bonar, L. and Mook, H. (1984). A study of dense mineralized tissue by
neutron-diffraction. Int. J. Biol. Macromol. 6, 321-326.
Lev-Tov Chattah, N., Sharir, A., Weiner, S. and Shahar, R. (2009). Determining the
elastic modulus of mouse cortical bone using electronic speckle pattern
interferometry (ESPI) and micro computed tomography: a new approach for
characterizing small-bone material properties. J. Biomech. 45, 84-90.
Li, K., Zernicker, R. F., Barnard, R. J. and Li, A. F. Y. (1991). Differential response
of rat limb bones to strenuous exercise. J. Appl. Phys. 70, 554-560.
THE JOURNAL OF EXPERIMENTAL BIOLOGY
Structure and mechanics of acellular bone 1993
Macesic, L. J. and Summers, A. P. (2012). Flexural stiffness and composition of the
batoid propterygium as predictors of punting ability. J. Exp. Biol. 215 (in press).
McCreadie, B. R., Hollister, S. J., Schaffler, M. B. and Goldstein, S. A. (2004).
Osteocyte lacuna size and shape in women with and without osteoporotic fracture. J.
Biomech. 37, 563-572.
Meunier, F. J. (1987). Nouvelles données sur lʼorganisation spatiale des fibres de
collagène de la plaque basale des écailles des Téléostéens. Ann. Sci. Nat. Zool. 9,
113-121.
Moss, M. L. (1960). Osteogenesis and repair of acellular teleost bone. Anat. Rec. 136,
246-247.
Moss, M. L. (1961). Studies of the acellular bone of teleost fish: 1. Morphological and
systematic variations. Acta Anat. 46, 343-462.
Moss, M. L. (1963). The biology of acellular teleost bone. Ann. N. Y. Acad. Sci. 109,
337-350.
Nemoto, Y., Higuchi, K., Baba, O., Kudo, A. and Takano, Y. (2007). Multinucleate
osteoclasts in medaka as evidence of active bone remodeling. Bone 40, 399-408.
Nicolella, D. P., Moravits, D. E., Gale, A. M., Bonewald, L. F. and Lankford, J.
(2006). Osteocyte lacunae tissue strain in cortical bone. J. Biomech. 39, 1735-1743.
Parenti, L. R. (1986). The phylogenetic significance of bone types in euteleost fishes.
Zool. J. Linn. Soc. 87, 37-51.
Paxton, H., Bonser, R. H. C. and Winwood, K. (2006). Regional variation in the
microhardness and mineralization of vertebrae from brown and rainbow trout. J. Fish
Biol. 68, 481-487.
Porter, M. E., Beltrán, J. L., Koob, T. J. and Summers, A. P. (2006). Material
properties and biochemical composition of mineralized vertebral cartilage in seven
elasmobranch species (Chondrichthyes). J. Exp. Biol. 209, 2920-2928.
Rapoff, A. J., Rinaldi, R. G., Hotzman, J. L. and Daegling, D. J. (2008). Elastic
modulus variation in mandibular bone: a microindentation study of Macaca
fascicularis. Am. J. Phys. Anthropol. 135, 100-109.
Rho, J. Y., Mishra, S. R., Chung, K., Bai, J. and Pharr, G. M. (2001). Relationship
between ultrastructure and nanoindentation properties of intramuscular herring
bones. Ann. Biomed. Eng. 29, 1082-1088.
Roy, M. E., Nishimoto, S. K., Rho, J. Y., Bhattacharya, S. K., Lin, J. S. and Pharr,
G. M. (2000). Correlations between osteocalcin content, degree of mineralization and
mechanical properties of C. carpio rib bone. J. Biomed. Mater. Res. 54, 547-553.
Shahar, R., Zaslansky, P., Barak. M., Friesem, A. A., Currey, J. D. and Weiner, S.
(2007). Anisotropic Poissonʼs ratio and compression modulus of cortical bone
determined by speckle interferometry. J. Biomech. 40, 252-264.
Sire, J.-Y. and Huysseune, A. (2003). Formation of dermal skeletal and dental tissues
in fish: a comparative and evolutionary approach. Biol. Rev. 78, 219-249.
Sire, J.-Y. and Meunier, F. J. (1994). The canaliculi of Williamson in holostean bone
(Osteichthyes, Actinopterygii): a structural and ultrastructural study. Acta Zool. 75,
235-247.
Skedros, J. G. (2005). Osteocyte lacuna population densities in sheep, elk and horse
calcanei. Cells Tissues Organs 181, 23-37.
Spatz, H. C., OʼLeary, E. J. and Vincent, F. V. (1996). Youngʼs moduli and shear
moduli in cortical bone. Proc. R. Soc. Lond. B 263, 287-294.
Szpak, P. (2011). Fish bone chemistry and ultrastructure: implications for taphonomy
and stable isotope analysis. J. Archaeol. Sci. 38, 3358-3372.
Toppe, J., Albrektsen, S., Hope, B. and Aksnes, A. (2007). Chemical composition,
mineral content and amino acid and lipid profiles in bones from various fish species.
Comp. Biochem. Physiol. 146B, 395-401.
Totland, G. K., Fjelldal, P. G., Kryvi, H., Lokka, G., Wargelius, A., Sagstad, A.,
Hansen, T. and Grotmol, S. (2011). Sustained swimming increases the mineral
content and osteocyte density of salmon vertebral bone. J. Anat. 219, 490-501.
Vilizzi, L. and Walker, K. F. (1999). Age and growth of the common carp, Cyprinus
carpio, in the River Murray, Australia: validation, consistency of age interpretation,
and growth models. Environ. Biol. Fish. 54, 77-106.
Wang, X. M., Cui, F. Z., Ge, J., Zhang, Y. and Ma, C. (2002). Variation of
nanomechanical properties of bone gene mutation in the zebrafish. Biomat. 23,
4557-4563.
Weiner, S., Traub, W. and Wagner, H. D. (1999). Lamellar bone: structure-function
relations. J. Struct. Biol. 126, 241-255.
Witten, P. E and Huysseune, A. (2009). A comparative view on mechanisms and
functions of skeletal remodeling in teleost fish, with special emphasis on osteoclasts
and their function. Biol. Rev. 84, 315-346.
Witten, P. E, Hansen, A. and Hall, B. K. (2001). Features of mono- and
multinucleated bone resorbing cells of the zebrafish Danio rerio and their contribution
to skeletal development, remodeling, and growth. J. Morphol. 250, 197-207.
Zaragüeta-Bagils, R., Lavoué, S., Tillier, A., Bonillo, C. and Lecointre, G. (2002).
Assessment of otocephalan and protacanthopterygian concepts in the light of
multiple molecular phylogenies. C. R. Biol. 325, 1191-1207.
Zhang, Y., Cui, F. Z., Wang, X. M., Feng, Q. L. and Zhu, X. D. (2002). Mechanical
properties of skeletal bone in gene-mutated Stöpseldtl28d and wild-type zebrafish
(Danio rerio) measured by atomic force microscopy-based nanoindentation. Bone
30, 541-546.
Zhou, H., Burger, C., Sics, I., Hsiao, B. S., Chu, B., Graham, L. and Glimcher, M.
J. (2007). Small-angle X-ray study of the three-dimensional collagen mineral
superstructure in intramuscular fish bone. J. Appl. Crystallogr. 40, S666-S668.
Zioupos, P., Currey, J. D. and Casinos, A. (2000). Exploring the effects of
hypermineralisation in bone tissue by using an extreme biological example. Conn.
Tiss. Res. 41, 229-248.