• biomechanics;
  • bone;
  • evolution;
  • material properties;
  • fish;
  • amphibians;
  • reptiles;
  • mammals


  1. Top of page
  2. Abstract
  6. Acknowledgements

The biomechanical performance of long bones is dictated by four key factors: element size, element shape, loading conditions, and material properties. Our understanding of the latter of these has been mostly limited to eutherian mammals and birds, which show similarity. Whether their possession of comparable material properties reflects common ancestry or independent evolution is uncertain. In the present analysis, we tested the bending strength, modulus, and failure strains of the femur and its pterygiophore homolog in actinpterygian fish. Sixty-nine specimens representing basal character states in seven major vertebrate crown clades were tested. These data were then coupled with avian and mammalian data from the literature and analyzed in an evolutionary context using phylogenetic character analysis. Mean values of 188 MPa for yield strength, 22.4 GPa for Young's modulus, and 8,437 μ∈ for yield strain were obtained for the long bones. Analysis of variance (ANOVA) revealed comparable values between clades that span a 30,000-fold range of body mass. We conclude that material properties of the first long bones 475 million years ago were conserved throughout evolution. Major locomotory challenges to femora during vertebrate evolution were almost solely accomplished by modifications of element size and shape. Anat Rec 268:115–124, 2002. © 2002 Wiley-Liss, Inc.

In vertebrates, the appendicular skeleton provides leverage for locomotion and support on land (Alexander, 1994). The major skeletal elements serving in these roles are collectively known as “long bones” because their longitudinal dimensions typically exceed the transverse ones (Marieb and Mallatt, 1992). These bones include the femur, tibia, fibula, metatarsi, and phalanges of the hindlimb, and the humerus, radius, ulna, metacarpi, and phalanges of the forelimb.

The biomechanical performance of long bones is dictated by four key factors: element size, element shape, loading conditions, and material properties (Beaupré and Carter, 1992). The latter of these, material properties, are quantified measures of the physical performance of tissues expressed in the absence of geometric and scaling influences (Keaveny and Hayes, 1993). These include commonly assessed parameters such as strength (maximum stress [force/area] prior to failure), Young's modulus (longitudinal material stiffness), and failure strain (amount of deformation before the onset of failure). Collectively these provide an effective indication of how osseous tissues behave when subjected to loads (Cochran, 1982). They can also be useful for making comparisons between materials.

Vertebrate structures such as long bones, antlers, tooth enamel, skull bones, dermal scutes, and auditory bulla are composed of osseous tissue consisting of hydroxyapatite (mineral), collagen (protein), water, and trace amounts of other proteins. Hydroxyapatite is the stiffest, most brittle constituent, with a Young's modulus of approximately 110 gigapascals (GPa = 109 N/m2) (Currey, 1984). On the other hand, collagen is tough and pliant, with a Young's modulus of only 350 megapascals (MPa = 106 N/m2) (Catanese et al., 1999) (Fig. 1). The proportion of these constituents gives each osseous composite a unique mechanical signature relative to the other tissues (Fig. 1). Collectively the constituent tissue properties emergently influence whole-element functional capacities, such as bearing weight, absorbing impacts, resisting wear, resonating sound waves, etc. (Currey, 1984).

thumbnail image

Figure 1. Range for the Young's modulus, among osseous tissues in vertebrates. Measures for long bones representing a diversity of avian and eutherian long bones show a comparable mechanical signature. The specific range for each osseous tissue type is largely determined by the proportion of the two major components: brittle and stiff hydroxyapatite mineral, and pliant but tough collagen protein. The modulus values for these components by themselves are also denoted.

Download figure to PowerPoint

Because variance in material properties is possible among osseous tissues, these attributes are a potential avenue by which natural selection can modify the biomechanical capacities of the skeleton. Of the commonly assessed material properties (strength, failure strain, and modulus), modulus is considered by many to be the most important from both the biomechanical and the evolutionary standpoint (e.g., Currey, 1984; Fyhrie and Vashishth, 2000). This property conveys rigidity for bones to function while maintaining strain levels below catastrophic failure and fatigue threshold levels (Pattin et al., 1996). Thus, modulus ensures functionality during the reproductive lifespan of an animal. In theory, natural selection can modify modulus and directly influence functionality through a strain-sensitive cellular mechanism that controls bone size and form during life (Fyhrie and Vashishth, 2000). Functional stress and strains are indirectly set by this process and are maintained well below failure thresholds that are usually never experienced by a bone. It is not clear whether failure levels are truly a target of natural selection; nevertheless, they are informative for assessing in vitro mechanical differences between materials and are necessary in assessments of modulus from bending tests (Popov, 1968).

The material properties of avian and eutherian mammal long bones are well established (Burr, 1980), but little is known about the elements of taxa belonging to outgroup clades. The literature contains data on only a few derived reptiles and one marsupial (Peterson and Zernicke, 1986, 1987; Currey, 1987) (Table 1). Virtually nothing has been reported on the material properties of fish, amphibian, and montotreme clades that represent half of the world's 43,000 living osteichthyan species (Fig. 2).

Table 1. Material properties of long bones for a diversity of Amniote taxa
TaxonElementUltimate bending strength (MPa)Young's modulus (GPa)Yield strain in bendinga (μ∈)Reference
  • a

    Ultimate strength ÷ Young's modulus.

  • b

    Value based on a combination of values for ultimate strength and Young's modulus from Engesaeter et al. (1979) and Turner and Burr (1993).

 Alligator (American Alligator)Femur17411.914,600Currey (1987)
 Geochelone (tortoise)Fibula13310.712,400Currey (1987)
 Dipsosaurus (lizard)Tibia316Peterson and Zernicke (1987)
 Anolis (lizard)Femur316Peterson and Zernicke (1986)
 Phoenicopterus (flamingo)Tarsometatarsus25928.89000Currey (1987)
 Spheniscus (penguin)Various2131712,500Currey (1987)
 Anser (goose)Femur26319.613,400McAlister and Moyle (1983)
 Exacalfatoria (quail)Femur311Biewener (1982)
 Larus (gull)Femur160Carrier and Leon (1990)
 Gallus (chicken)Femur96Yamada (1970)
 Colinus (bobwhite)Femur193Biewener (1982)
 Protemnodon (wallaby)Tibia16.8Currey (1987)
 Rattus (rat)Femur182/––/29.46,200bEngesaeter et al. (1979); Turner and Burr (1993)
 Halichoerus (seal)Tibia16816.710,100Currey (1987)
 Felis (cat)Femur145Yamada (1970)
 Ovis (sheep)Femur90Deloffre et al. (1995)
 Sus (boar)Femur151Yamada (1970)
 Equus (horse)Femur24719.912,400Schryver (1978)
 Bos (cow)Femur17918.79,000Currey (1987)
 Homo (human)Femur20814.814,100Currey and Butler (1975)
thumbnail image

Figure 2. Phylogenetic hypothesis for osteichthyan vertebrates depicting clades for which material tests in bending have been conducted on the long bones. Note the paucity of data for basal members (branch points) of the major vertebrate clades (in black) from which the evolution of the similar material property suite of birds (Aves) and eutherian mammals can be deduced.

Download figure to PowerPoint

Sharma (1944) was the first to contrast the material properties of avian and mammalian long bones and note their mechanical similarity. Subsequent investigations have confirmed this shared mechanical signature and established the range of variance among taxa (Table 1, Fig. 1). As determined from bending tests, the richest data set from which interspecific comparisons can be made (Table 1), ultimate bending strength typically ranges from 100–300 MPa, Young's modulus from 15–29 GPa, and failure strain from 6,000–14,000 μ ∈ (microstrain = mm/mm × 106). Variations within each range are posited to reflect minor intracladal modifications for specific mechanical functions related to locomotion (Currey, 1987).

Sharma (1944) inferred that the mechanical equivalence between avian and eutherian long bone tissue stems from their shared amniote ancestry dating back approximately 340 million years (Fig. 3A). This theory has gone largely unchallenged. Nevertheless, in one of the few analyses that included outgroups to the Eutheria and Aves, slightly lower values for Young's modulus were found (Currey, 1987). Long bones from two Galapagos tortoise specimens averaged 12.6 GPa, and values for two specimens of American alligator averaged 12.1 GPa. The potential implications of these findings with regard to how birds and eutherian mammals came to have similar material properties are considerable. Although based on very small sample sizes, Currey (1987) raised the possibility that these values represent the plesiomorphic character state for the Amniota. Thus, the similarity in mechanical signatures between avian and mammalian long bones would have evolved homoplastically (Fig. 3B), rather than being a primitive (plesiomorphic) amniote character state, as Sharma originally contended.

thumbnail image

Figure 3. Phylogenetic hypothesis for the Amniota depicting the two competing scenarios of how eutherian mammals and birds acquired comparable long bone material properties. A: Hypothesis suggested by Sharma (1944), whereby the material property suite of birds and mammals was plesiomorphic for the Amniota approximately 340 million years ago, and was simply retained through the cladogenesis of the Aves and Eutheria. B: Hypothesis favored by Currey (1987), implying that low modulus values for chelonians and crocodylians signify the primitive character state in the Amniota, and that birds and mammals independently evolved similar material property suites.

Download figure to PowerPoint

In the present investigation, we sought to reconstruct the biomechanical evolution of the femur from the establishment of its precursors nearly 475 million years ago in basal osteichthyan fish through its cladogenesis in extant crown clades. In addition to providing the first evolutionary insight into osseous material property evolution on a broad phylogenetic scale, these data would serve to address the controversy surrounding the genesis of the shared mechanical signatures of birds and eutherian mammals.

We used the paleontological literature to identify robust phylogenetic hypotheses for the Ostyeichthyes (bony fish and their ancestors, including birds and mammals) and the changes in long bone form and function that occurred throughout evolution. We then used the comparative anatomical literature to identify seven extant taxa that collectively retain these attributes, and subsequently used them as functional analogs to model long bone evolution. Finally, we assessed the material properties of femoral homologs in 69 specimens and statistically analyzed the results with reference to phylogeny.


  1. Top of page
  2. Abstract
  6. Acknowledgements

Properties and Elements Selected for Analysis

Because material properties are quantifiable measures of mechanical performance, they can be standardized for use in interspecific comparisons and in statistically based phylogenetic character analyses. In the present study, the material properties, Young's modulus, ultimate bending strength, and yield strain were chosen for examination. As noted above, these parameters provide ample indication of material mechanical performance and are a potential means by which natural selection can alter such performance. The long bone elements selected for investigation were the tetrapod femur and the homologous basal pterygiophore element in basal osteichthyan fishes, the metapterygium (Gegenbaur, 1865; Edwards, 1989; Shubin, 1995). These elements are relatively large (even in small animals) and have elongate cylindrical diaphyseal regions that are free of muscle attachments. These features make them suitable for whole-bone bending tests (see below) and the application of engineering beam theory, for which specimens must have high aspect ratios (length to diameter (Sedlin and Hirsch, 1966; Turner and Burr, 1993)) and cross-sectional shapes approximating circles or ellipses.

Specimen Selection

Long bones from the specimens examined in this study were structurally homologous to extinct representatives of extant groups. A restriction on the use of such functional analogs (Radinsky, 1987; Edwards, 1989) for extinct taxa is that living vertebrates are not exactly like those that represented crucial stages in the history of vertebrate locomotion (Witmer, 1995). Nevertheless, Witmer (1995) has demonstrated that functional and biomechanical attributes can often be determined for ancestral taxa if the attributes 1) correlate with osteological structure, and 2) can be inferred decisively using phylogenetic character analysis of extant taxa (“extant phylogenetic bracketing,” sensu Witmer (1995)). The biomechanical attributes considered in the present examination (see below) conform to these criteria, and thus the use of these functional analogs provide a reasonable reconstruction of the biomechanical attributes of long bones throughout evolution.

The long bone attributes considered in the present study were gross morphology, and the modes and media of locomotion that influence loading conditions. Collectively these attributes each play a major role in the performance of long bones (Currey, 1959; Carter et al., 1976; Rubin and Lanyon, 1984; Lanyon and Rubin, 1985), and therefore they were considered important to characterize in ancestral taxa so that biomechanically appropriate functional analogs could be chosen. We identified the ancestral locomotory attributes (hereafter collectively referred to as locomotory grades) and their evolutionary sequence of appearance from two sources: 1) paleontological studies of the locomotion by extinct taxa, and 2) comparative anatomical studies of extant taxa that retain basal anatomical and locomotory attributes within their respective clades (Table 2).

Table 2. Phylogenetic affiliations, locomotory attributes, and analogous taxa
CladeAppearance (mybp)MediumBasal anatomical and locomotory attributesExtant functional analogsTesting
  1. Information from: Matthew (1937); Simpson (1941); Enlow and Brown (1956, 1957, 1958); Schmalhausen (1968); Jenkins (1971); Moy-Thomas and Miles (1971); Jarvik (1972); Jenkins and Parrington (1976); Edwards (1977); Holmes and Carroll (1977); Jenkins and Wejs (1979); Pearson and Westoll (1979); Patterson (1982); Rose (1982); Cifelli (1983); de Ricqles et al. (1983); Jenkins and Krause (1983); Holmes (1984); Krause (1984); Bemis et al. (1986) Parrish (1986); Ahlberg (1989); Coates and Clack (1990); Francillon-Vieillot et al. (1990); Ahlberg and Milner (1994); Coates (1994); Walker and Liem (1994); Shubin (1995); Coates (1996); Hu et al. (1997).

Osteichthyes (bony fish)∼475AquaticRigid fins with small, semi-mobile pterygiophores within body wall; gliding plane locomotionPolypterus (bichir)This study
Sarcopteryia (lobe-fin fish)∼450AquaticMobile lobed fins with large pterygiophores external to body wall; appendicular fin propulsion at slow speedsNeoceratodus (lungfish)Untestable-pterigiophores are secondarily cartilaginous
Stem-Tetrapoda (aquatic tetrapod outgroups)∼350AquaticLimbs; benthic scullingCryptobranchus (hellbender)This study
Tetrapoda (amphibians)∼345AmphibiousLimbs; terrestrial sprawlingAmbystoma (salamander)This study
Amniota (terrestrial vertebrates)∼340TerrestrialLimbs; terrestrial sprawlingVaranus (varanid lizard)This study
Archosauria (semi- and erect-postured reptiles, incl. birds)∼225TerrestrialLimbs; erect posture; parasagital locomotionAves (birds)Data from the literature (see text)
Cynodontia (early mammals)∼225TerrestrialLimbs; semi-erect posture; scansorialityDidelphis (opossum) Tachyglossus (echidna)This study
Derived Eutheria (erect-postured mammals)∼60TerrestrialLimbs; erect posture; parasagital locomotionBos (cow)Canis (dog)Data from the literature (see text)

Eighty-three adult individuals representing seven genera (Table 3) were acquired from field collection, colleagues, and the commercial pet trade. It was our intention to garner material property data for 12 specimens of each taxon. Useable information was obtained for 69 of the 83 specimens (see below). Only a single specimen of the monotreme Tachyglossus was obtained in this effort (Table 3). All specimens originally had been captured in the wild, except the Rattus specimens, which were raised in the laboratory.

Table 3. Specimens used to assess material properties of long bones
TaxonPhylogenetic affiliationLocomotory meansNo. tested with useable dataMean body mass/S.D. (kg)Mean snout-vent length/S.D. (cm)
  • a

    Pterygiophores from four of the twelve specimens were too small for material testing and were not used.

  • na, not available.

Polypterus sp.ActinopterygiaGliding-plane fins8a0.27/.2325.7/8.7
Cryptobranchus alleganiensisAmphibiaBenthic walking limbs120.59/0.2333.0/2.9
Ambystoma tigrinumAmphibiaAmphibious terrestrial sprawling120.049/9.712.7/.82
Varanus xanthematicusReptilia (Amniota)Terrestrial sprawling120.43/0.2127.7/4.9
Tachyglossus aculeatusMonotremata (Amniota)Scansorial locomotion1nana
Didelphis marsupialisMarsupialis (Amniota)Scansorial locomotion122.66/1.233.0/7.5
Rattus norvegicusEutheria (Amniota)Scansorial locomotion120.31/0.0819.2/2.9

Live specimens were euthanized using a chemical overdose (immersion in 1% MS 222 for fish and amphibians, intramuscular injection of nembutal [120 mg/kg body weight] for reptiles, and intraperitoneal injection of pentobarbital sodium euthanasia solution [1 ml/4.54 kg body weight] for mammals). The specimens were kept frozen at −20°C before testing (Sedlin and Hirsch, 1966). One to two days before testing, the specimens were thawed to room temperature and the paired femora and pterygiophores were removed. The bones were delicately stripped of flesh so as not to introduce stress concentrations by damaging the diaphyseal surfaces of the cortices (Currey, 1962). The cleaned bones were immediately placed in physiological Ringer's solution (Humason, 1979) and refrigerated at 7°C. The bones were allowed to equilibrate in a hydrated state to an ambient temperature of 22°C prior to testing.

Material Testing

The elements were tested to failure in a three-point bend configuration (Popov, 1968; Turner and Burr, 1993) using a model 858, MTS Mini Bionix servo-hydraulic material loading frame (Material Testing Systems, Inc., Eden Prairie, MN). Specifically, the bones were placed in a cradle in which axial rotation was possible. This eliminated the potential for torsional shear loading that is introduced when whole bones, with their elliptical cross-sections, are secured at the ends. The slightly curved bones were positioned so that the dorsal (convex) aspects loaded in tension. This configuration helped facilitate failure primarily in tension rather than in longitudinal shear, which is inherently present in any three-point loading test (Turner and Burr, 1993). Loading was carried to failure at a strain rate of 0.05% ∈/sec, which is a rate typical of slow to rapid locomotion (McElheney, 1966; Carter and Hayes, 1976; Lanyon et al. 1975; Lanyon and Baggott, 1976). Because drying alters material properties from their physiological norm, the bones were irrigated with water throughout testing (Sedlin and Hirsch, 1966; Pelker et al., 1984). An 1112 N load cell was used to monitor load during the experiments. The load range was varied between 222 and 1112-N to accommodate interspecific variability in specimen size. Force and displacement readings during testing were recorded using TestStar TSII software (Material Testing Systems, Inc.) (Fig. 4). Mid-diaphyseal area moments of inertia were determined by capturing magnified images of the bones with a Sony CCD DXC-970MD camera (Sony, Inc., Tokyo, Japan), and then digitizing and analyzing the images using NIH Image graphics software (National Institutes of Health, Bethesda, MD) on a Macintosh 7100 power computer (Apple Computer, Inc., Cupertino, CA). The ultimate strength of the specimens was determined using the maximum load values from each test (Fig. 4) and the three-point bend beam theory equations of Popov (1968) and Turner and Burr (1993): ultimate bending strength = MC/I (M = the bending moment of the beam [M = PL/4, where P = the peak force, L = beam length], C = the distance from the neutral plane of bending to the tensile surface of the bone cortex, and I = the second moment of area). It should be noted that this standard for estimating material properties assumes that the tissue being loaded behaves with linear elasticity up to failure. Although bone approximately conforms to these criteria, its microstructural heterogeneity leads to non-brittle fracturing owing to localized failure prior to rupture. Furthermore, progressive failure beginning with the outermost surface in-line with the application of force is the norm in three-point bending experiments with bone. This occurs because bending stresses are differentially higher at this location, and bone is weaker in tension vs. compression (Reilly and Burstein, 1975). For these reasons, our strength data are best viewed as very close approximations of the actual material properties of these tissues.

thumbnail image

Figure 4. Load displacement curve monitored during a whole-bone, three-point bending test on a specimen of Varanus. The elastic region, maximal load, and rupture point are noted.

Download figure to PowerPoint

Estimates of Young's modulus for entire bones derived by coupling bend tests with beam theory assumptions are typically in error. For example, the values for Young's modulus in machined specimens of femora measured in bending or tension are typically two to four times greater than the values for whole femora from the same taxa tested in bending (Maeda, 1964; Yamada, 1970; Cowin, 1988). Beam theory violations and system compliance promote underestimation such as these. Specifically, beam theory equations are only valid if length : width aspect ratios are in excess of 16:1 (Turner and Burr, 1993). Very few long bones conform to these specifications, and technically they are not beams. As such, when loads are applied in a mechanical testing frame, they act more like rigid bodies and deflect less easily. This structural rigidity facilitates system compliance (displacement between the various fixtures and materials in the testing apparatus) before the bone actually undergoes bending itself. This leads to an overestimation of the actual displacement of the bone being tested and underestimation of modulus up to fivefold (Turner and Burr, personal observations). Furthermore, since long bone cross-sections are elliptical, they often show axial rotation during bending. This increases the amount of measured displacement and also promotes underestimation of modulus.

To avoid these complications, we used a protocol similar to that used by Keller et al. (1986), Keller and Spengler (1989), and Turner and Burr (1993) to assess modulus in rat bones. Single-element foil strain gauges (FBY-06.60-11-005LE [Tokyo Soki Kenkyo, Ltd., Tokyo, Japan] and FCB-2-11 [Measurements Group, Inc., Raleigh, NC]) were trimmed to their smallest working size. The gauges were placed on the mid-diaphyseal dorsal surfaces of the contralateral long bones to those tested in the loading frame. The strain-gauged bones were then placed in the loading cradle. Calibrated weights were hung from the elements at midshaft and the strains generated for each increment of weight (five increments in total) were recorded five times each. Mean strain per unit mass was determined. Young's modulus was ascertained by dividing this value into a stress reading from each dynamic load test. The particular stress value used in this calculation was taken from the pre-yield region, where increases in load are a linear function of deformation (Cochran, 1982) (Fig. 4).

Useable material property data were obtained for 69 of the 83 specimens. Specimens were discarded owing to slippage within the testing cradle, operator error, or data acquisition complications. The four smallest specimens (all from Polypterus) were also discarded because they were found to be too small for appropriate testing.

Data Interpretation

The mean material property values of each taxon's long bones, and femoral data from the literature were plotted with reference to a phylogeny of the Osteichthyes based on Rosen et al. (1981) and Gauthier et al. (1988). Evolutionary trends were inferred by standard methods of comparative biology (Harvey and Pagel, 1991; Brooks and McLennan, 1991). The first round of this analysis entailed determining whether a basal taxon's material properties were outside the typical range for birds and eutherian mammals. It was considered a priori that if all values for the amniote taxa were found to be within this range, it would provide strong support for common ancestry. To assess for gradual orthoclinal patterns or stepped cladogenic events leading to the establishment of the avian and eutherian mammal material property signature (McKinney and McNamara, 1991) statistical testing was employed. An analysis of variance (ANOVA) with a Bonferroni/Dunn post-hoc paired t-test using StatView software (SAS Institute, Inc., Cary, NC) was conducted. Tests of whether the material property values were statistically distinguishable at a 0.05 level of significance between consecutive basal evolutionary locomotory grades (e.g., benthic-walking limbs vs. amphibious sprawlers, amphibious vs. terrestrial sprawlers (Table 2)) were made.


  1. Top of page
  2. Abstract
  6. Acknowledgements

The mean values from the long bones tested in this study were all found to be within the bounds of the range occupied by birds and eutherian mammals (Table 4). There appeared to be no clear major pattern of increases or decreases in values leading up to the cladogenesis of the birds and mammals (Table 4). Statistical analyses bore this out by showing no significant differences between consecutive locomotory grades in nearly all instances. Specifically, the mean values for the ultimate strength of each long bone in the present study ranged from 149 to 222 MPa in the taxa for which multiple individuals were tested (Table 4). The single Tachyglossus long bone had an ultimate strength of 271 MPa (Table 4). All values fell within the range for birds and eutherian mammals (100–300 MPa). Statistical analysis of these data indicated that the long bone strength values for consecutive evolutionary locomotory grades were indistinguishable at the 0.05 level of significance, with the lone exception of the comparison between bones from the benthic walking Cryptobranchus and the terrestrial salamander Ambystoma.

Table 4. Material properties of metapterygia and femora for a diversity of vertebrates
TaxonElementsNumber testedMean ultimate bending strength/S.D. (MPa)Mean Young's modulus/S.D. (GPa)Mean yield strain in bending/S.D. (MPa)a
  • a

    Ultimate strength/Young's modulus.

  • na, not applicable.

Polypterus sp.Pelvic metapterygia8155/44.517.6/7.88807/5993
Cryptobranchus alleganiensisFemora12207/34/422.3/7.79282/3229
Ambystoma tigrinumFemora12149/50.218.1/5.28232/1931
Varanus xanthematicusFemora12188/35.722.8/5.68245/1492
Tachygossus aculeatusFemur1271/na24.8/na10927/na
Didelphis marsupialisFemora12222/42.723.9/7.89289/2236
Rattus norvegicusFemora12176/30.026.0/7.36769/1610

Mean Young's modulus values ranged from 17.6 to 26.0 GPa (Table 4). The single specimen of Tachyglossus had a long bone modulus of 24.8 GPa (Table 4). All values occurred within the range typical of birds and eutherian mammals (15–29 GPa). Statistical analysis of these data demonstrated that the values for consecutive evolutionary locomotory grades were indistinguishable at a 0.05 level of significance.

Mean yield strain values ranged between 6,769 μ ∈ and 9,289 μ ∈ (Table 4). The single specimen of Tachyglossus had a value of 10,927 μ ∈ (Table 4). All values fell within the range of avian and mammalian taxa (6,000–14,000 μ ∈). A statistical analysis of these data showed that the values for consecutive evolutionary locomotory grades are indistinguishable at a 0.05 level of significance.

A cladogram of the Osteichthyes plotted with the present Young's modulus, yield strain, and strength values are depicted in Figure 5 along with data for femora from the literature (Table 1).

thumbnail image

Figure 5. A cladogram for the osteichthyan vertebrates depicting the evolution of the material properties of femora and their homologs. These data are from the present study and the literature (Table 1). The results suggest that the most basal pterygiophores of primitive osteichthyan fish possessed a suite of mechanical material properties within the range exhibited by birds and eutherian mammals. This primitive suite of material property attributes (the mechanical signature) went largely unchanged with the establishment of the major sarcopterygian vertebrate clades (Amphibia, Sauria, Monotremata, Marsupialia, and Eutheria). Properties for extinct stem taxa and the major crown clade, Actinistia, can be inferred. Only the Young's modulus values from the literature for tortoises (Chelonia) and aligators (Crocodylia) fall outside the avian and eutherian range. This may reflect an actual evolutionary signal, although sampling error cannot be ruled out at this time.

Download figure to PowerPoint


  1. Top of page
  2. Abstract
  6. Acknowledgements

An evolutionary understanding of skeletal material properties has not been previously obtainable for a number of reasons, including: 1) a general lack of testing outside of the Eutheria and Aves, 2) small, statistically limiting sample sizes for the few taxa that have been examined, 3) a lack of focus on any one homologous element, and 4) difficulties with testing the small bones that are found in the majority of osteichthyan taxa. In the present study we overcame these shortcomings by utilizing engineering protocols designed for small-element testing, and focusing our efforts on numerous specimens of the same element in a diversity of non-avian, non-eutherian taxa. The results allow us to infer biomechanical changes (or a lack thereof) in the material properties of the femur spanning 475 million years of evolution, using phylogenetic character analysis for the first time.

Interesting evolutionary insight can be garnered from phylogenetic character analysis of these material property data, as depicted in Figure 5. Particularly intriguing is that the data for the basal osteichthyan analog, Polypterus (E = 17.6 GPa, ultimate strength = 155 MPa, yield strain = 8,807 μ ∈), falls well within the range typical of birds and eutherian mammals. In fact, with respect to the values from the literature (Table 1), they bear similarity to cow femora (E = 18.7 GPa, strength = 179 MPa, yield strain = 9,600 μ ∈) (Currey, 1987), the “model” values used in many biomechanical studies of the skeleton (Burr, 1980). This suggests that the first “long bone” tissues that evolved nearly half a billion years ago were mechanically very much like extant mammalian bones despite their diminutive size, limited mobility within the body wall, and unique function (supporting shark-like planing fins).

The phylogenetic character analysis (Fig. 5) further revealed that a primitive suite of material property attributes established in these fish remained largely unchanged, showing only minor, statistically indiscernible increases and decreases within the bounds seen in today's birds and eutherian mammals. Such material property stasis occurred despite major changes in morphology and scale, as well as in modes and media of locomotion.

With regard to the disparity of opinions regarding how the Aves and Eutheria came to possess similar material properties, these data strongly support the interpretation that this mechanical signature was the primitive condition within the Amniota (Sharma, 1944). Evidence for this includes comparable “avian/mammalian-like” material properties in amniote outgroups (Actinopterygian and Amphibia) and in amniote reptiles that retain the primitive terrestrial sprawling mode of locomotion (current results; Fig. 5) (Peterson and Zernicke, 1986, 1987).

These findings are at odds with Currey's (1987) suggestion that the basal amniote condition was typified by long bones with lower modulus values than those found in birds and mammals. Phylogenetically they suggest that the low modulus values of crocodylians and chelonians are independent derivations in these derived amniotes (Fig. 5). Nevertheless, owing to the small sample sizes for the crocodilians (n = 2) and tortoises (n = 2) tested by Currey, a strong case cannot be made that the crocodylian and chelonian material properties are actually aberrant. In the present study, all four of the values for femoral modulus in these reptiles fall within a 95% confidence interval for osteichthyan long bones, and one of the four values (16.1 GPa) is a typical avian/eutherian value. The testing of further specimens will be an interesting line of future investigations to address this matter. Along these lines, one of the present authors (G.M.E.) has begun assessing Young's modulus using a nano-indentation testing protocol (Rho et al., 1997). Data for crocodilians (Alligator mississipiensis: 24 GPa, n = 3) and a chelonian (Macrochlemystemmincki: 22 GPa, n = 1) have so far shown properties similar to those found in other osteichthyan clades.

A question that arises from this analysis is why femora and their homologs have maintained their material properties within restricted bounds, showing only minor locomotory adaptations (Currey, 1987), or effects from genetic drift (Futuyma, 1986; Ridley, 1996) (Fig. 5). We believe that the evidence strongly points to some form of constraint (Futuyma, 1986; Ridley, 1996), the nature of which is not currently determinable. Nevertheless, it appears to have had a profound influence of the outcome of vertebrate morphological evolution. For instance, with respect to the skeleton, this would explain why major locomotory challenges such as the evolutionary transition to land did not involve major changes in femoral material properties. Instead, changes in whole-element size and shape were the primary avenues by which natural selection acted in such instances.

A useful outcome from the present study is that material properties for the femur can be inferred from the character analysis for extinct or untested extant taxa. For example, although extant coelacanths (Actinistia) or do not retain osseous long bone elements for which we can assess material properties (Table 2), we can infer these attributes for their Mesozoic counterparts that did. The most parsimonious conclusion based on outgroup analysis (Brooks and McLennan, 1991; Harvey and Pagel, 1991) and phylogenetic bracketing (Witmer, 1995) is that the femoral homologs in these animals were mechanically similar to those of the other vertebrates (Fig. 5).

In a previous study Biewener (1982) pointed out that the similarities in avian and eutherian mammal long bone strength extend over a 14,000-fold range of body mass (0.05–700 kg). In the present study similar values were found in animals as small as 0.023 kg (a specimen of Polypterus), thus extending this range to 30,000-fold. Furthermore, this generalization can now be extended to include both Young's modulus and yield strain, the former of which is considered by far the most important with regard to element form, function, and evolution (Currey, 1984; Fyhrie and Vashishth, 2000) (see above).

In summation, this study suggests that the material properties of the femur have not substantially varied during evolution, regardless of phylogeny, changes in locomotory modes and media, and gross morphology. The shared avian and mammalian material property suite is a reflection of common ancestry. Thus the biomechanical performance of our own skeletons should be viewed in light of our aquatic ancestry, and not solely as a product of adaptation to our particular terrestrial niche (Gould and Verba, 1982). In the absence of major material property changes during evolution, it is clear that the majority of locomotory challenges to the femur during vertebrate evolution were accomplished by modifications of other key determinants of biomechanical performance, namely element size and shape. Further substantiation of the patterns revealed in this investigation, their extension to other long bone elements (particularly phalangeal elements that qualitatively appear to have atypically low whole-element stiffness in bats (Papadimitriou et al., 1996)), and the potential causes of material property evolutionary stasis will serve as interesting foci in future investigations of skeletal material property evolution.


  1. Top of page
  2. Abstract
  6. Acknowledgements

We thank the many people who made this research possible, including: Marvalee Wake, Dennis Carter, Sharon Swartz, Kevin Padian, Mimi Koehl, David Wake, Dale Denardo, Tim Cooper, Steve Deban, Eric Nauman, David Lindberg, Marjolein Van Der Meulin, Dan Leiberman, Brian Simonson, Allen Collins, Chris Meyer, Rob Guralnik, Pang Chan, Jennifer Warnock, Jeninifer Lomis, Mike O'Donnell, Sam Cunningham, Marc Olin, Josh Butler, Mike Shapiro, Yvette Justice, Josh Whorley, Dan Mulcahey, Mark Goodwin, Pat Holyroyd, Mark Levenston. Rick Staub, Alan Woodward, Paul Moler, Bob Jones, Chris Schneider, Diette Walker, Mike Fagan, Roy Caldwell, Randall Voss, John Richmond, Gary Chun, Patrick Couper, Mario Garcia-Paris, Jennifer Clack, Isalee Smith, Jim O'Reilly, Kris Lappin, Chris Rose, Dave Varricchio. We also thank the University of California Museums of Vertebrate Zoology and Paleontology. This research was carried out under Animal Care and Use Safety Protocol R055-1097, UC–Berkeley.


  1. Top of page
  2. Abstract
  6. Acknowledgements
  • Ahlberg PE. 1989. Paired fin skeletons and relationships of the fossil group Porolepiformes (Osteichthyes; Sarcopterygii). Zool J Linn Soc 96: 119166.
  • Ahlberg PE, Milner A. 1994. The origin and diversification of tetrapods. Nature 368: 507514.
  • Alexander R. 1994. Bones: the unity of form and function. London: Weidenfeld and Nicholson. 223 p.
  • Beaupré GS, Carter DR. 1992. Finite element analysis in biomechanics. In: BiewenerAA, editor. Biomechanics: structures and systems. Oxford: Oxford University Press. p 149174.
  • Bemis WE, Burggren WW, Kemp NE. 1986. The biology and evolution of lungfishes. J Morphol 190 (Suppl 1): 1383.
  • Biewener AA. 1982. Bone strength in small mammals and bipedal birds: do safety factors change with body size? J Exp Biol 98: 289301.
  • Brooks DR, McLennan DA. 1991. Phylogeny, ecology, and behavior. Chicago: University of Chicago Press. 434 p.
  • Burr DB. 1980. The relationships among physical, geometrical and mechanical properties of bone, with a note on the properties of non-human primate bone. Yearb Phys Anthropol 23: 109146.
  • Carrier D, Leon LR. 1990. Skeletal growth and function in the California gull (Larus californicus). J Zool Lond 222: 375389.
  • Carter DR, Hayes WC. 1976. Bone compressive strength: the influence of density and strain rate. Science 194: 11741176.
  • Carter DR, Hayes WC, Schurman DJ. 1976. Fatigue life of compact bone. II. Effects of microstructure and density. J Biomech 9: 211218.
  • Catanese J, Iverson EP, Ng RK, Keaveny TM. 1999. Heterogeneity of the mechanical properties of demineralized bone. J Biomech 32: 13651369.
  • Cifelli RL. 1983. The origin and affinities of the South American Condylartha and early Tertiary Litopterna (Mammalia). Am Mus Novit 2772: 149.
  • Coates MI, Clack JA. 1990. Polydactyly in the earliest known tetrapod limbs. Nature 347: 6669.
  • Coates MI. 1994. The origin of vertebrate limbs. Dev Suppl 121: 169180.
  • Coates MI. 1996. The Devonian tetrapod Acanthostega gunnari Jarvik: postcranial anatomy, basal tetrapod interrelationships and patterns of skeletal evolution. Trans R Soc Edinb Earth Sci 87: 363421.
  • Cochran GVB. 1982. A primer of orthopaedic biomechanics. New York: Churchill Livingstone. 413 p.
  • Cowin SC. 1988. Bone biomechanics. Boca Raton: CRC Press. 345 p.
  • Currey JD. 1959. Differences in tensile strength of bone of different histological types. J Anat 93: 8795.
  • Currey JD. 1962. Stress concentrations in bone. Q J Microsc Sci 103: 111133.
  • Currey JD, Butler G. 1975. The mechanical properties of bone tissue in children. J Bone Joint Surg 57A: 810814.
  • Currey JD. 1979. Mechanical properties of bone with greatly differing functions. J Biomech 12: 313319.
  • Currey JD. 1984. The mechanical adaptations of bones. Princeton: Princeton University Press. 294 p.
  • Currey JD. 1987. The evolution of the mechanical properties of amniote bone. J Biomech 20: 10351044.
  • Deloffre P, Hans D, Rumelhart C, Mitton D, Tsouderos Y, Meunier PJ. 1995. Comparison between bone density and bone strength in glucocorticoid-treated aged ewes. Bone 17 (Suppl): 409S414S.
  • de Ricqles A, Meurnier FJ, Castanet J, Francillon-Viellot H. 1983. Comparative microstructure of bone. In: HallBK, editor. Bone. Vol. III. Bone matrix and bone specific products. Boca Raton: CRC Press. p 177.
  • Edwards JL. 1977. The evolution of terrestrial locomotion. In: HechtMK, GoodyPC, HechtBM, editors. Major patterns in vertebrate evolution. New York: Plenum Press. p 553577.
  • Edwards JL. 1989. Two perspectives on the evolution of the tetrapod limb. Am Zool 29: 235254.
  • Engesaeter LB, Ekland A, Langeland N. 1979. Methods for testing the mechanical properties of the rat femur. Acta Orthop Scand 49: 512518.
  • Enlow DH, Brown SO. 1956. A comparative histological study of fossil and recent bone tissues. Part 1. Texas J Sci 8: 403443.
  • Enlow DH, Brown SO. 1957. A comparative histological study of fossil and recent bone tissues. Part 2. Tex J Sci 9: 185214.
  • Enlow DH, Brown SO. 1958. A comparative histological study of fossil and recent bone tissues. Part 3. Tex J Sci 10: 187230.
  • Francillon-Vieillot H, de Buffrenil V, Castanet J, Geraudie J, Meunier FJ, Sire JY, Zylberberg L, de Ricqles A. 1990. Skeletal biomineralization: patterns, processes and evolutionary trends. In: CarterJG, editor. Microstructure and mineralization of vertebrate skeletal tissues. New York: Van Nostrand. p 471548.
  • Futuyma DJ. 1986. Evolutionary biology. Sunderland: Sinauer Associates. 600 p.
  • Fyhrie DP, Vashishth D. 2000. Bone stiffness predicts strength similarity for human vertebral cancellous bone in compression and for cortical bone in tension. Bone 26: 169173.
  • Gauthier J, Kluge AG, Rowe T. 1988. Amniote phylogeny and the importance of fossils. Cladistics 4: 105209.
  • Gegenbaur C. 1865. Untersuchungen zur vergleichenden Anatomie der Wirbeltiere. Vol. II. Leipzig: Wilhelm Engelmann. 72 p.
  • Gould SJ, Verba ES. 1982. Exaptation: a missing term in the science of form. Paleobiology 8: 415.
  • Harvey PH, Pagel MD. 1991. The comparative method in evolutionary biology. Oxford: Oxford University Press. 239 p.
  • Holmes R, Carroll RL. 1977. A temnospondyl amphibian from the Mississippian of Scotland. Bull Mus Comp Zool 147: 489511.
  • Holmes R. 1984. The Carboniferous amphibian Proterogyrinus scheelei Romer, and the early evolution of tetrapods. Phil Trans R Soc Lond B 306: 431527.
  • Hu Y, Wang Y, Luo Z, Li C. 1997. A new symmetrodont mammal from China and its implications for mammalian evolution. Nature 390: 137142.
  • Humason GL. 1979. Animal tissue techniques. 4th ed. San Francisco: W.H. Freeman and Company. 661 p.
  • Jarvik E. 1972. Middle and Upper Devonian Porolepiformes from East Greenland with special reference to Glyptolepis groenlandica n. sp. Meddelelser om Gronland 187: 1178.
  • Jenkins FA. 1971. The postcranial skeleton of African cynodonts. Bull Peabody Mus Nat Hist 36: 1216.
  • Jenkins FA, Parrington FR. 1976. The postcranial skeletons of the Triassic mammals Eozostrodon, Megazostrodon, and Erythrotherium. Phil Trans R Soc Lond B Biol Sci 273: 379410.
  • Jenkins FA, Weijs WA. 1979. The functional anatomy of the shoulder in the Virginia opossum (Didelphis virginiana). J Zool 188: 379410.
  • Jenkins FA, Krause DW. 1983. Adaptations for climbing in North American multituberculates (Mammalia). Science 220: 712715.
  • Keaveny TM, Hayes WC. 1993. Mechanical properties of cortical and trabecular bone. In: HallBK, editor. Bone. Vol. VII. Bone growth, part B. Boca Raton: CRC Press. p 285344.
  • Keller TS, Spengler DM, Carter DR. 1986. Geometric, elastic, and structural properties of maturing rat femora. J Orthop Res 4: 5767.
  • Keller TS, Spengler DM. 1989. In vivo strain gauge implantation in rats. J Biomech 15: 911917.
  • Krause DW. 1984. Mammalian evolution in the Paleocene: beginning of an era. In: GingerichPD, BadgleyCE, editors. Mammals: notes for a short course. Publication EO1-1040-005-91. Knoxville: University of Tennessee. p 87109.
  • Lanyon LE, Hampson WGJ, Goodship AE, Shah JS. 1975. Bone deformation recorded from in vivo strain gauges attached in the human tibial shaft. Acta Orthop Scand 46: 256268.
  • Lanyon LE, Baggott DG. 1976. Mechanical function as an influence on the structure and form of bone. J Bone Joint Surg 58-B: 436443.
  • Lanyon LE, Rubin CT. 1985. Functional adaptation in skeletal structures. In: HidebrandM, BrambleDM, LiemKF, WakeDB, editors. Functional vertebrate morphology. Cambridge: Harvard University Press. p 125.
  • Maeda M. 1964. Study on the bending strength of long limb bones in large animals. J Kyoto Pref Med Univ 73: 233257.
  • Marieb EN, Mallatt J. 1992. Human anatomy. Redwood City: Benjamin/Cummings Publishing Co. 731 p.
  • Matthew WD. 1937. Paleocene faunas of the San Juan Basin, New Mexico. Trans Am Phil Soc 30: 1510.
  • McAlister GB, Moyle DD. 1983. Some mechanical properties of goose femoral cortical bone. J Biomech 16: 577589.
  • McElheney JH. 1966. Dynamic response of bone and muscle tissue. J Appl Physiol 21: 12311236.
  • McKinney ML, McNamara KL. 1991. Heterochrony: the evolution of ontogeny. New York: Plenum Press. 437 p.
  • Moy-Thomas JA, Miles RS. 1971. Paleozoic fishes. 2nd ed. London: Chapman Hall. 149 p.
  • Papadimitriou HM, Swartz SM, Kunz TH. 1996. Ontogenetic and anatomical variation in mineralization of the wing skeleton of the Mexican free-tailed bat, Tadarida brasiliensis. J Zool Lond 240: 411426.
  • Parrish JM. 1986. Locomotory adaptations in the hindlimb and pelvis of the thecodontia. Hunteria 12: 335.
  • Patterson C. 1982. Morphology and interrelationships of actinopterygian fishes. Am Zool 22: 241259.
  • Pattin CA, Caler WE, Carter DR. 1996. Cyclical mechanical property degradation during fatigue loading of cortical bone. J Biomech 29: 6979.
  • Pearson DM, Westoll TS. 1979. The Devonian actinopterygian, Cheirolepis agassiz. Trans R Soc Edinb 70: 337399.
  • Pelker RR, Friedlaender GE, Markham TC, Panjabi MM, Moen CJ. 1984. J Orthop Res Soc 1: 405411.
  • Peterson JA, Zernicke RF. 1986. The mechanical properties of limb bones in the lizard Anolis equestris [Abstract]. Am Zool 26: 133A.
  • Peterson JA, Zernicke RF. 1987. The geometric and mechanical properties of limb bones in the lizard, Dipsosaurus dorsalis [Abstract]. J Biomech 20: 902.
  • Popov EP. 1968. Introduction to mechanics of solids. Englewood Cliffs, NJ: Prentice Hall, Inc. 571 p.
  • Radinsky LB. 1987. The evolution of vertebrate design. Chicago: University of Chicago Press. 188 p.
  • Reilly DJ, Burstein AH. 1976. The elastic and ultimate properties of compact bone tissue. J Biomech 8: 393405.
  • Rho RY, Tsui TY, Pharr GM. 1997. Elastic properties of human cortical and trabecular lamellar bone measured by nanoindentation. Biomaterials 18: 13251330.
  • Ridley M. 1996. Evolution. Cambridge: Blackwell Science. 719 p.
  • Rose KD. 1982. Skeleton of Diacodexis, the oldest known artiodactyl. Science 216: 621623.
  • Rosen DE, Forey PL, Gardiner BG, Patterson C. 1981. Lungfishes, tetrapods, paleontology, and plesiomorphy. Bull Am Mus Nat Hist 167: 163275.
  • Rubin CT, Lanyon LE. 1984. Dynamic strain similarity in vertebrates: an alternative to allometric limb bone scaling. J Theor Biol 107: 321327.
  • Schmalhausen II. 1968. The origin of terrestrial vertebrates. New York: Academic Press. 314 p.
  • Schryver H. 1978. Bending properties of cortical bone of the horse. Am J Vet Res 39: 2528.
  • Sedlin ED, Hirsch C. 1966. Factors affecting the determination of the physical properties of femoral cortical bone. Acta Orthop Scand 37: 2948.
  • Sharma DN. 1944. Note on the comparison of rat and chicken bone. Ind J Med Res 51: 686687.
  • Shubin N. 1995. The evolution of paired fins and the origin of tetrapod limbs. In: HechtMK, editor. Evolutionary biology. New York: Plenum Press. p 3986.
  • Simpson GG. 1941. Mounted skeleton and restoration of an early Paleocene mammal. Am Mus Novit 1155: 15.
  • Turner CH, Burr DB. 1993. Basic biomechanical measurements of bone: a tutorial. Bone 14: 595608.
  • Walker WF, Liem KF. 1994. Functional anatomy of the vertebrates. Fort Worth: Saunders College Publishing. 788 p.
  • Willems G, Celis JP, Lambrechts P, Braem M, Vanherle G. 1993. Hardness and Young's modulus determined by nanoindentation technique of filler particles of dental restorative materials compared with human enamel. J Biomed Mater Res 27: 747755.
  • Witmer L. 1995. The extant phylogenetic bracket and the importance of reconstructing soft tissues in fossils. In: ThomasonJJ, editor. Functional morphology in vertebrate paleontology. Cambridge: Cambridge University Press. p 1933.
  • Yamada H. 1970. Strength of biological materials. EvansFG, editor. Baltimore: Williams and Wilkins Co. 297 p.
  • Zioupos P, Currey JD, Casinos A, de Buffrenil V. 1997. Mechanical properties of the rostrum of the whale Mesoplodon densirostris, a remarkably dense bony tissue. J Zool Lond 241: 725736.