zoological Journal of the Linnean SocieQ (1993), 108: 179-196. With 9 figures Femoral ontogeny and locomotor biomechanics of Dryosaurus lettowvorbecki (Dinosauria, Iguanodontia) RONALD E. HEINRICH, CHRISTOPHER B. R U F F AND DAVID B. WEISHAMPEL Department of Cell Biology and Anatomy, Johns Hopkins University, School of Medicine, Baltimore, Maryland 21205 Received October 1991, revised manuscript accepted for publication J U !1992 ~ Femoral ontogeny in the iguanodontian dinosaur DTyosaurus lettowvorbccki (Late Jurassic, Tanzania) is analysed biornechanically using principles of beam theory. Statistically significant differences in cross-sectional properties are found between animals of differing size, reflecting alterations to both the relative amount and distribution of cortical bone during growth. Two explanations are suggested to account for these modifications in bone architecture: ( 1 ) increasing mechanical loads related to increasing body size, and (2) changes in the orientation of these loadings associated with a caudal shift of the centre of gravity. It is argued that D. lettowvorbecki hatchlings were not bipedal as generally presumed but obligate quadrupeds. Based on avian growth rates, the transition from quadrupedality to habitual bipedalisrn is estimated to have occurred within several months of hatching. The biomechanical approach employed here contributes new insight into ontogeny of locomotion in D. lettowuorbccki and provides additional ways of analysing ontogenetic processes among extinct and living species. ADDITIONAL KEY WORDS:-Cross-sectional geometry - development - bipedalism CONTENTS Introduction . . . . . . . . Methods and materials . . . . . Results . . . . . . . . . Weight estimates . . . . . . Cross-sectional properties . . . . Discussion . . . . . . . . Size transitions . . . . . . Femoral ontogeny and locomotion . A quadrupedal-bipedal locomotor shift Summary. . . . . . . . . Acknowledgements . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179 182 185 185 186 189 190 190 193 195 195 195 INTRODUCTION Bones involved in load-bearing are capable of responding to changes in mechanical strain in a precise and systematic manner (Alexander et al., 1984; Lanyon, 1984). As a result, the basic form of limb elements can be profoundly 0024-4082/93/060179+ 18 S08.00/0 179 0 1993 The Linnean Society of London I80 R. E. HEINRICH E l AL. Figure 1. Cladogram depicting the phylogenetic relationships of Dryosaurur to other ornithopodan taxa. altered when subjected to abnormal loadings (Goodship, Lanyon & McFie, 1979; Lanyon et al., 1982; Rubin & Lanyon, 1985). Less well investigated are the changes in skeletal morphology that accompany normal postnatal ontogeny. If long bone growth proceeded isometrically, body mass would increase more rapidly than bone cross-sectional area (A a M2/3)resulting in greater strain magnitude and reduced safety factors with increasing size. This, however, does not occur. Analysing the tibiotarsus of chicks, Biewener, Swartz & Bertram (1986, see also Goodship et al., 1979) demonstrated that strain magnitude at homologous positions along the bone remained constant throughout ontogeny. This uniformity of strain magnitude can be attributed to alterations in the mechanical properties of bone that occur during the course of normal development (Brear, Currey & Pond, 1990 and references cited therein). Mechanical properties may be modified in one or both of the following ways: (1) by altering the material properties of bone (e.g. altering mineral content) or (2) by changing the amount and distribution of bone. The latter constitutes changes in size and shape or, more generally, changes in bone architecture. While changes in material properties of bone certainly occur throughout ontogeny (Brear et al., 1990),it is the changes in bone architecture which are thought to be the primary mechanism by which bone responds to increased loadings (Woo et al., 1981; Lanyon & Rubin, 1985). An ontogenetic sample of femora attributed to the basal iguanodontian dinosaur Dryosaurus lettowvorbecki was analysed to assess quantitatively the importance of architectural changes during development of this particular species. Iguanodontians were ornithopods (Fig. 1; Sereno, 1986; Sues & Norman, 1990; Weishampel & Heinrich, 1992), the group of dinosaurs which became the dominant herbivores in the latter half of the Cretaceous Period (97-65 million years ago). Two species of Dryosaurus are currently recognized, D.altus from the Morrison Formation of the western interior of the United LOCOMOTOR BIOMECHANICS OF DRTOSAURUS 181 States, and D.lettowvorbecki from the Tendaguru Beds of Tanzania (Janensch, 1955; Galton, 1981; Sues & Norman, 1990). These deposits date to roughly 150 million years ago. Both species reached lengths of 6.5 m, over half of which comprised the tail, and fully mature animals are estimated to have weighed approximately 70 kg. (Weight estimates are discussed in detail in Methods and materials.) Hindlimb proportions indicate cursorial specializations in these bipedal animals (Galton, 1971, 1974; Coombs, 1978), and Thulborn ( 1982) estimated their top running speed to be in excess of 40 km h-', the fastest of all ornithischian dinosaurs analysed to date. Overall similarities to Hypsilophodon (Galton, 1974), one of the best studied of all small euornithopodan taxa, suggests that Dryosaurus was a habitual biped. The position of the centre of gravity and the kinematics of locomotion have obvious and important implications for understanding the magnitude and orientation of mechanical loadings acting on the femur. Comparison of grounddwelling bird kinematics with what can be inferred of locomotion in adult Dryosaurus illustrates this point. The centre of gravity in all bipedal dinosaurs is reconstructed as near the hip joint, with the long, heavy tail acting to counterbalance the presacral region of the body during locomotion (Alexander, 1985). Gatesy (1990) has shown that parasagittal limb motion in theropod dinosaurs involved considerable retraction of the femur and movement of the feet symmetrically beneath the pelvis. This was presumably the case in bipedal ornithopods such as Dryosaurus as well (Galton, 1974). In contrast, the parasagittal gait kinematics of birds differs substantially from that of non-avian archosaurs, owing in large part to the loss of the counterbalancing tail. T o support a centre of gravity in front of the hip, birds have adopted a subhorizontal femoral orientation which positions the foot directly under the centre of gravity. Rather than retracting the femur, as do bipedal dinosaurs and other non-avian archosaurs, ground-dwelling birds rely predominantly on flexion at the knee to move the foot under the centre of gravity (Gatesy, 1990). Along with the kinematic differences found in these two groups of bipeds there are structural differences in the hindlimb; bipedal dinosaurs have proportions more characteristic of quadrupedal cursorial mammals than of birds (Coombs, 1978; Anderson et al., 1985). The level of strain, or strain magnitude, acting on a load-bearing element is the product of (1) the mechanical properties of bone, as defined above, and (2) the size and orientation of the mechanical loadings acting on it (Lanyon & Rubin, 1985). The resultant cross-sectional morphology of a load-bearing element reflects this unique interaction. Cross-sectional geometries, then, can be used to reconstruct the mechanical loadings that acted on long bones in vivo (Ruff & Hayes, 1983), and an ontogenetic series of cross-sections provides a means of reconstructing the pattern of loadings that accompanied growth and development. If young animals maintain similar weight distributions and utilize the same locomotory gaits as mature animals, then we might expect crosssectional properties of bone to demonstrate unidirectional changes as body weight increases. Although rates of change may vary during different stages of growth, the direction of change, if any, should remain constant-an assumption supported by data collected from the South American bipedal ratite, Rhea americana (Table 1). Conversely, if the pattern of change in cross-sectional morphology deviates significantly from unidirectional trends, factors other than R. E. HEINRICH ET AL. 182 TABLE 1. Cross-sectional properties of an ontogenetic sequence ot' Rhea amen'cana femora. See Methods and materials section for a discussion of the cross-sectional properties listed in the Table. Abbreviation: AMNH, American Museum of Natural History. Specimen AMNH 1474 was collected from Southern Brazil, specimens AMNH 6470 and 6471 from Uruguay Specimen No. Length yo Cortical area AMNH 6471 AMNH 6470 AMNH 1474 I25 195 210 55.1 50.6 54.2 I,,/I,, 1.499 1.579 1.858 increasing size are probably affecting femoral loadings and modifications to bone architecture. METHODS AND MATERIALS Long bone diaphyses modelled as hollow beams can be analysed using standard engineering beam theory principles (Timoshenko & Gere, 1972; Ruff & Hayes, 1983 and references cited therein). Stresses resulting from the loading of hollow beams are expressed by the equations where and 6, is compressive stress; Fa is axial loading; and A is cross-sectional area; Mbc bb= - I where ob is compressive or tensile stress; M b is the externally applied bending moment about a given axis; c is the distance from neutral plane of bending to the beam's surface; and I is the second moment of area about the neutral axis in the plane of bending. The amount and distribution of bone a t a given cross-section is reflected in the cross-sectional area (A) and the second moment of area ( I ) respectively. To compare the relative amount of cortical bone in cross-sections of varying size, percentage cortical area (cortical area/total subperiosteal area) was calculated. Comparisons of bone distribution involved calculations of the second moments of area about the craniocaudal (I,)and mediolateral ( I y )axes of the femur, as well second moments of area of the as the maximum (Imax)and minimum (Imin) section. The angle 8,measured counterclockwise from the M-L axis of greatest bending strength, provides information on the general orientation of bone distribution. These cross-sectional properties are presented in Fig. 2. Cross-sectional shape can be compared using the ratios I,/Z or Imax/Zmin which in turn provide information about relative resistance to bending loadings in different planes of the femur. Body mass is obviously an important parameter in any analysis of mechanical loading of limb elements. For purposes here weight estimates of D . lettowvorbecki at different stages of ontogeny are of particular interest. Estimates of body mass in dinosaurs have generally applied a technique of volume displacement using LOCOMOTOR BIOMECHANICS OF DRYOSAURUS I83 Subperiosteal area = 395 mm2 Cortical area = 273 mm2 l x = 13345 mm4 ly = 9937 mm' I,, = 13484 mm' lmin= 9798mm' e = 79.00 Figure 2. Proximal cross-section of medium-sized femur (UT 1495/ 14) demonstrating the crosssectional properties; total area, cortical area, 0, and the second moment of areas I,, Iy,I,,,, I,,,$", analysed in this study. Y-Axis indicates craniocaudal orientation, x-axis mediolateral. isometrically scaled models of specific species and density estimates obtained from living reptiles (Colbert, 1962; Alexander, 1985). Anderson, Hall-Martin & Russell ( 1985), however, have presented a regression for calculating weight of bipedal dinosaurs from femoral circumference, W = 0.16 G.73 (3) where W is body weight and C,is midshaft circumference. The exponent is based on a regression of 23 species of quadrupedal mammals, and the proportionality constant from mass estimates of Troodon (Stenonychosaurus discussed by Anderson et al., 1985, is a junior synonym of Troodon; Currie, 1987). Anderson et al. (1985) justify the mammalian exponent on the basis of ( 1 ) similarities in limb proportions between quadrupedal mammals and bipedal dinosaurs, and (2) similarities between the 2.73 exponent calculated for mammals and exponents previously calculated for femoral circumference of three bipedal dinosaurs. Since it could be argued that a quadrupedal based regression might overestimate body mass in bipeds, weight estimates for D. lettowvorbecki are calculated using both equation (3) and the equation, W = 1.08 c.28 (4) 184 R. E. HEINRICH E l AL. calculated by Anderson et al. (1985) from 72 species of birds. In this way we can be reasonably confident that we have bracketed the actual weight of D. lettowvorbecki, assuming that the modern analogues on which the regression equations are based accurately reflect allometric proportions of cross-sectional area to body weight in the femora of small bipedal dinosaurs. All remains referred to D . lettowvorbecki come from a single locality in the Tendaguru Hills, Quarry IG, c. 75 km west of the Tanzanian port city of Lind (Janensch, 1914). Quarry IG represents a monospecific mass death assemblage where animals of all sizes are thought to have perished during widespread drought conditions (Russell et al., 1980). All material is highly disarticulated and disassociated and most elements are fragmentary (Janensch, 1914; G. Maier, personal communication). Among the available specimens, however, are five complete and undistorted femora ranging in length from 120 to 276 mm. These were used to calculate regressions of greater trochanter and medial condyle length, and total condylar width against total length (Fig. 3, Table 2). Lengths of incomplete femora were then estimated using these regressions. If both medial condyle length and condylar width could be measured for a given specimen, the regression of medial condyle against length was used because of its higher correlation coefficient (Table 2). Using this method the resulting range of femoral lengths was calculated to be 109 to 330 mm. For purposes of this study, femora were assigned to one of three sizes classes based on length: small (109-146 mm), medium (180-234 mm), and large (282-330 mm). These classes presumably reflect relative age groups. Geometric data were obtained from 27 femora with naturally broken crosssections. The location of these cross-sections was between c. 50% and 30% of femoral length from the distal end of the bone (all percentages are given from the distal end of the bone). Break position was categorized as either proximal (those breaks occurring between 41% and 50%) or distal (those between 30% and 39%). Since inaccurate length estimates may affect the accuracy of the break location, comparisons of external femoral morphology provided a means of verifying the categorization of each cross-section. O n the basis of femoral length and location of break, each specimen was assigned to one of six groups: smalldistal, small-proximal, medium-distal, medium-proximal, large-distal and largeproximal. The cross-sections at these breaks were photographed and the endosteal and periosteal perimeters manually digitized. Cross-sectional area, and 8, were calculated using a modified version of cortical area, I,, I,, I,,,,,,Imin the program SLICE (Nagurka & Hayes, 1980). These dimensions were then ratios. Statistical used to calculate percentage cortical area and I,/Z, and Imax/Imin analyses were carried out using the multiple comparison Tukey test (Systat version 5.1 ; Wilkinson, 1988). TABLE 2. Regression equations for estimating femoral length (mm) in D . lettowvorbecki. Abbreviations: N , sample size; ?, coefficient of determination; SE, standard error of estimate Character N Slope y-int. ? SE Greater trochanter Medial condyle Condylar width 5 5 5 6.015 4.077 3.810 -7.86 28.55 17.60 0.977 0.992 0.982 3.415 5.624 8.130 LOCOMOTOR BIOMECHANICS OF DRYOSAURUS r I85 Cranial t Medial Medial 4 Cranial D B A Figure 3. Schematic diagram of a D. leftozuvorbecki femur in (A) cranial, (B) lateral, (C) proximal and (D) distal views. Length estimates are based on regressions derived from the three measurements shown; greater trochanter, medial condyle, and condyle width. Abbreviations: L, length; GT, greater trochanter; MD, medial condyle; CW, condylar width. Scale bar = 4 cm. RESULTS Weight estimates Eight weight estimates calculated from femora spanning the ontogenetic size range of our sample are given in Table 3. Since none of the 100 or so fragmentary femora available for study is recognizably larger than no. 8, this specimen probably represents a mature animal. Using the regression derived for quadrupedal mammals as suggested by Anderson et al. (1985), this animal is estimated to have had a mass of 70 kg. T h e only other estimate of body weight for Dryosaurus of which we are aware is 120 kg, given by Thulborn (1982). His method of calculation, however, is not given and therefore we are unable to critically compare Thulborn's estimate with those derived here. TABLE 3. Weight estimates (in kg) for D . lettowvorbecki based on femoral midshaft circumferences (Circ. in mm). ~ Specimen Length Circ. Weightb Weight' 109" 120 143 175" 204 276 299" 3304 37.0 46.0 55.0 69.5 76.0 108.0 111.5 116.5 4.1 6.7 10.0 17.1 21.0 46.7 50.3 55.6 3.1 5.5 9.0 17.1 21.8 56.9 62.1 70.0 "Femoral length is estimated; 'weight estimates based on the bipedal regression equation W = 1.08GzBof Anderson et al. (1985, see text for discussion); 'weight estimates based on the quadrupedal regresion equation W = 0.16G 73 of Anderson el al. (1985, see text for discussion) R. E. HEINRICH E l AL. I86 7o 70 t- $ld A no A "t ".- 30 0 0 O I I I Figure 4. Relationship between percentage cortical area of a cross-section and femoral length (given in mm). Femoral size: 0 , small; A, medium; 0 , large. TABLE 4. Comparisons of percentage cortical area: data for proximal and distal sections of D. lettowvorbecki femora pooled. Abbreviations: N , sample size; SD, standard deviation; SE, standard error of estimate; not significantly different at P = 0.10 ~ ~~ Size Small Medium Large ~ ~~ Jv Mean SD SE Range 11 52.30 5.295 1.596 38.940.6 9 68.34 2.954 0.985 61.7-71.7 8 67.09 3.295 1.165 60.9-71.2 Sign. o.ool ~ Cross-sectional properties Cross-sectional data collected for D. lettowvorbecki femora are summarized in Figs 4-7 and accompanying Tables 4-7. Since we are interested in the architectural differences that exist between successive stages of ontogeny, the statistical results of Tukey test comparisons between small-medium and mediumlarge femora are also presented in Tables 4-7. Percentage cortical area Comparisons of proximal and distal locations within size classes revealed no statistically significant differences in percentage cortical area between the two regions. Therefore, these data were combined to provide more robust comparisons between size classes. Statistically significant differences in percentage cortical area do occur between small and medium sized animals ( P < 0.001) but not between medium and large animals (Fig. 4, Table 4). Ratio of I, lo Iu Statistically significant differences in the second moment of area ratio, I x / I y , LOCOMOTOR BIOMECHANICS OF DRYOSAURUS A 0 0 A A 0 0 O 0 0.4 I 100 187 I I 200 300 Femoral length I ., Figure 5. Relationship between the second moment of area ratio, Ix/Iy,and femoral length (given in mm). Femoral size and break position noted by symbols: 0 , small-proximal; 0 , small-distal; A, medium-proximal; A,medium-distal; 0 , large-proximal; large-distal. TABLE 5. Comparisons of I x / I yvalues between size classes and at two locations along the femoral shaft. Abbreviations: as in Table 4. Location Distal Proximal Jv Mean SD SE Small 7 0.826 0.080 0.030 0.705-0.921 Medium 6 0.780 0.027 0.01 I 0.750-0.821 Large 5 0.794 0.117 0.052 0.597-0.890 Small 4 1.208 0.1 10 0.055 1.0941.324 Medium 3 1.307 0.039 0.022 1.266-1.307 Large 3 1.016 0.216 0.125 0.771-1.180 Size Range Sign -~ ~ = o.082 are found only at P < 0.10 (Fig. 5 , Table 5 ) . Proximal ZJZY ratios of medium and large femora (discussed further under 0) differ at P = 0.083. Ratio of I,,, to I,,, At both proximal and distal femoral locations, differences in the ratio of Zmax/Zmin are statistically significant at the P < 0.10 level. I n each case increases from small to medium-sized femora are followed by decreases from medium to large (Fig. 6, Table 6). This is particularly noticeable proximally where, although large Zmax/Zmin ranges exist, no overlap in Zmax/Zminratios occurs between either small-medium or medium-large groups (Table 7). Although small sample sizes limit the degree of statistical significance, distinct differences in relative maximum/minimum bending strength between femora of differing size are strongly suggested. R. E. HEINRICH ET AL. 188 . . .. I A A A A A A4 A 0 0 0 . 1.0I 100 I 200 Femoral length Figure 6. Relationship between the second moment of area ratio I,,/I,, in mm). Symbols for femoral size and break position as in Fig. 5. I 300 and femoral length (given TABLE 6. Comparisons of Imax/Imin values between D.lettowvorbecki size classes and at two locations along the femoral shaft. Abbreviations: as in Table 4. Location Distal Proximal Size N Mean SD SE Range Small 7 1.276 0.129 0.049 1.09&1.450 = o.087 Medium 6 1.47 I 0.148 0.060 1.326-1.678 __ Large 5 I .402 0.188 0.084 1.1841.690 Small 4 1.269 0.102 0.051 1.127-1.369 p = o.072 Medium 3 1.444 0.081 0.047 1,3761,534 Large 3 1.241 0.059 0.034 1.181-1.299 Sign. = o,053 The angle 0 Distally, 8 is oriented almost mediolaterally, i.e. approximately 180", in animals of all sizes, while proximally the orientation is closer to being craniocaudal, i.e. 90" ( = 270"). The regional differences in the orientation of the axis of maximum bending strength are statistically significant in all proximodistal comparisons within a size class, while no significant differences are found between proximal or distal locations across size classes (Fig. 7, Table 7). The 0 value of one large-proximal cross-section is much smaller than is found in proximal specimens of any size class (Fig. 7). This specimen also has an IJI,, ratio significantly smaller than both of the other large-proximal specimens ratio of the three femora (Fig. 6). All of these (Fig. 5) and the largest Imax/lmin cross-sectional properties suggest that this particular specimen may be better interpreted as a distal rather than proximal cross-section, although the calculated position of the break is 45%. (The other two large-proximal sections LOCOMOTOR BIOMECHANICS OF DRYOSAURUS I89 A A 0 240 A 0 220 . rn A 0. 180 A 160 A ' rn A 200 100 300 Femoral length Figure 7 . Relationship of the axis of greatest bending strength in proximal and distal cross-sections for femora of differing length. Symbols for femoral size and break position as in Fig. 5. TABLE 7 . Comparisons of the orientation of axes of greatest bending strength (degrees) in D.lettowvorbecki femoral cross sections. Abbreviations: as in Table 4. Location Size N Mean SD SE Range Sign. Small 7 171.9 13.87 5.24 I 159.6- 198.3 Medium 6 162.7 17.03 6.953 149.0-195.4 Large 5 182.8 24.65 11.02 151.9-215.7 Small 4 251.8 8.638 4.319 240.7-260.1 __ Medium 3 250.4 8.502 4.908 242.0-259.0 __ Large 3 228.8 ~~ Distal ~ Proximal 43.45 25.09 181.7-267.3 are 48y0.)Two possible explanations may account for this: either the break location has been inaccurately estimated, or the morphology of large femora simply differs appreciably from that of small and medium sized elements. Since all shape properties (second moment of area ratios) of the proximal cross-sections are smaller in large than in medium sized femora (Figs 5, 6), the ZJZY and 8 deviations of the one large specimen may not be as incongruent as they first appear. These differences are, however, larger than would be predicted on the basis of the remainder of the data and it may be that the cross-sectional break is less than the 45% position calculated, although it is unlikely to be distal (30-39y0) as defined here. DISCUSSION Femoral growth in Dryosaurus lettowvorbecki is accompanied by adjustments in both the relative amount and distribution of bone. While it is known that 190 R. E. HEINRICH ET AL. material properties of bone change during development, particularly early ontogeny (Torzilli et al., 1981; Currey & Pond, 1989; Brear el al., 1990), architectural modifications in load-bearing elements are generally attributed to a process of remodelling in response to strain (Lanyon et al., 1982; Lanyon, 1984; Rubin, 1984). Consequently, cross-sectional geometry is a reflection of, and can be used to reconstruct, the mechanical loadings which acted on the load-bearing element in vivo (Ruff & Hayes, 1983). By analysing the crosssectional morphology at various stages of ontogeny (small-medium-large) in D. lettowvorbecki, a pattern of mechanical loadings accompanying normal growth and development of the femur is reconstructed. This pattern is then used to make locomotor inferences. Size transitions The cross-sectional properties of medium sized femora differ most noticeably from those of small femora in two ways. First, they possess a relatively greater amount of cortical bone (Fig. 4, Table 4), and second, they are characterized by a larger Zmax/Zmin ratio (Fig. 6, Table 6). The roughly 33% increase in percentage cortical area corresponds to femoral lengthening from c. 150 to 180 mm and an increase in weight of roughly 7 or 8 kg (Table 3, and discussed further below). Interestingly this increase is not preceded by a recognizable trend towards increasing percent cortical area among small femora (Fig. 4). The second difference found between small and medium sized femora, both proximally and distally, is in relative maximum to minimum bending strength. Again, small femora do not demonstrate any clear trend towards increasing relative maximum bending strength over their size range (Fig. 6). The angle 8 remains the same at both locations over the small-medium size transition (Table 7). Differences in cross-sectional properties between femora of medium and large size are restricted to proximal cross-sections and, in general, are less distinctive than those between small and medium femora. Percentage cortical area remains constant (Fig. 4, Table 4) over almost a doubling in femoral length and an approximate 50 kg increase in body weight (Table 3). The IJZY and ratios of large femora, however, are distinctly smaller than particularily Imax/Zmin those of medium femora, indicating a relative reduction in maximum to minimum bending stress (Tables 5 and 6). Except for the changes in percentage cortical area, the morphology of distal cross-sections is relatively similar in femora of all sizes. More significant architectural changes occur proximally, as demonstrated in Fig. 8 using representative cross-sections of each size class. Increases in both relative amount and distribution of bone occur over the first size transition, while only changes in the distribution of bone occur over the medium-large transition. In the light of these alterations in bone distribution it is interesting that the axis of maximum bending strength remains (with the one large specimen exception discussed previously) constant both proximally and distally. The result of these architectural differences is reflected in cross-sectional shape, which is more elliptical in femora of medium size than in either small or large femora. Femoral ontogeny and locomotion Over the size range exhibited by our sample the most conspicuous alterations in cross-sectional morphology are manifested during what appears to be a very LOCOMOTOR BIOMECHANICS OF DRYOSAURUS 191 1 Cranial Medial SMALL MEDIUM LARGE Figure 8. Proximal cross-sections representing each of the three size classes: small (uncatalogued), medium (U?' 1495/14), and large (WJ 10000). Scale bar = I cm. short period of time, a period arbitrarily defined here as the small-medium size transition. Increases in both the relative amount and the maximum bending strength of bone over this period suggest that loadings on the femur had increased significantly. If increasing body mass alone accounted for these changes in architecture, then we would expect the trends documented over the first size transition to continue, albeit possibly at modified rates, as animals got larger (Table 1 : ZmaX/lmin).The predicted unidirectional trends, however, are not found in any of the comparisons made. Percentage cortical area increases dramatically in relatively small animals, while subsequent increases of over 3000/, in body mass (20-70 kg, Table 3) produce no increases in the relative amount of bone. Proximal and distal Zmax/Zminand proximal Zx/Iyratios exhibit a more parabolic than linear pattern of change (Fig. 6), suggesting that relative bending loads do not continue to increase with increasing size. Instead an apparent initial increase in bending stresses is followed by a relative reduction. Increasing body size alone, therefore, seems an unlikely explanation for the observed architectural changes in cross-sectional morphology. If increasing body size is insufficient to account for the ontogenetic pattern documented, then differences in the orientation of those loadings must also be important. Recent fossil evidence supports this conjecture. Horner & Weishampel (1988) described embryonic and juvenile remains of two euorni thopodans, Orodromeus makelai and Maiasaura peeblesorum. These animals had large heads and small tails relative to body size, proportions unlike those of fully grown individuals. Although no embryonic material of D . lettowvorbecki exists, a large head-to-body size ratio can be inferred for D . lettowvorbecki hatchlings given that the condition occurs in 0. makelai (a hypsilophodontid) and M . peeblesorum (a hadrosaurid), two taxa which phylogenetically bracket Dryosaurus (Fig. 1: Sereno, 1986; Weishampel & Heinrich, 1992). O n this basis we hypothesize that the body centre of gravity in D . lettowvorbecki hatchlings was cranial to the hip, and that this position shifted caudally during development. What effect would a centre of gravity in front of the hip have on femoral loadings and hatchling locomotion? This problem is illustrated schematically in Fig. 9 and can be quantified by the expression M=FxD (5) 192 R. E. HEINRICH E T AL. \ Moment = F x D \ Figure 9. A schematic representation of D.lcttowvorbccki demonstrating the biomechanical effect of the position of the centre of gravity on the bending stresses incurred by the femur. Given that the forces exerted by muscles are identical, a centre of gravity well cranial to the hip (left) would produce a larger moment acting about the femoral midshaft, resulting in larger bending stresses than a centre of gravity positioned near the hip (right). where M is the moment being generated about the femur; F is the mechanical load approximated here by body weight; and D is the perpendicular distance between the femoral midshaft and the mechanical load acting at the centre of gravity. Although calculating actual bending moments about the femur would require knowledge of the forces exerted by the appropriate muscles, Fig. 9 demonstrates the biomechanical importance of the position of the centre of gravity for an animal locomoting on two legs. For a bipedal dinosaur then, the greater the distance from the hip joint to the centre of gravity, the larger the moment acting about the femur and the greater the bending stress on the element. Since the centre of gravity was most cranial in the smallest animals, i.e. hatchlings, and assuming they were bipedal, bending loads should have been relatively greater in these animals than at any other stage of ontogeny. However, none of the cross-sectional properties analysed here support this supposition. The ratios found in femora of increased percentage cortical area and larger Zmax/lmin medium size both suggest that the relatively greatest femoral bending stresses occurred in animals associated with medium size. How might this be accounted for? A centre of gravity positioned well in front of the hip may have made habitual bipedality difficult, if not impossible for hatchlings. We suggest instead that D. lettowvorbecki hatchlings were constrained by the position of their centre of gravity to a quadrupedal stance during early neonatal life. It is postulated that, as these animals grew to some threshold size, the combination of increasing body and tail size effectively shifted the centre of gravity to a position more near the LOCOMOTOR BIOMECHANICS OF DRYOSAURUS I93 hip. That is, the tail became an adequate counter balance. At that time a habitual bipedal posture would have been assumed. Shifting from quadrupedal locomotion, where the body centre of gravity was supported by four limbs, to bipedality, where body mass was distributed over only two, would essentially double the mechanical loadings on the hindlimb. Bending and compressional stresses would increase significantly, requiring structural compensation to maintain acceptable safety factors. Over the smallmedium size transition, distinct and relatively dramatic architectural modifications have been identified. These are ( 1 ) an increase in relative bone mass and (2) an increase in the relative magnitude of greatest bending strength. Since neither of these cross-sectional parameters increases with increasing length among small femora, the abruptness of these changes in cross-sectional properties fits a hypothesis of novel mechanical loadings acting on the femur at a specific body size. Modifications in femoral architecture that occur after the small-medium size transition involve some changes in the distribution but not the relative amount of bone. We suggest that given a quadrupedal-bipedal transition and a shifting centre of gravity, craniocaudal bending stresses in femora of medium size were relatively larger than at any other time during the life of a D . lettowvorbecki individual. As these animals continued to increase in body size and the centre of gravity moved closer to the hip, a relative reduction in the bending moments acting about the femur resulted. Consequently relative craniocaudal bending stress was reduced in the largest animals. A quadrupedal-bipedal locomotor shqt Assuming that D . lettowvorbecki hatchlings were obligate quadrupeds, it is of some interest to decipher the amount of time that elapsed between hatching and the onset of bipedalism. Estimating age from femoral length requires ascertaining a rate of development in these animals, a problem made more difficult because of the uncertainty as to whether dinosaurs were warm or coldblooded. We have attempted to encompass the range of possible growth rates for D . lettowvorbecki by utilizing rates compiled for modern ecto- and endothermic animals (Case, 1978). For phylogenetic reasons, growth rates of reptiles, precocial and altricial birds have been considered. T o determine a speciesspecific growth rate for D . lettowvorbecki from the regressions presented by Case (1978), adult body weight estimates of 56 and 70 kg were used (Table 3). The growth rates calculated in this manner are provided in Table 8. If the femur used to calculate these body weights is not that of a mature animal, or if the method of calculation underestimates actual adult body weight, then the growth rates presented in Table 8 constitute minimal values. Since the architectural modifications of the small-medium size transition coincide with femoral lengthening from c. 150 to 180 mm, estimating body weights at these lengths enables age to be assessed using the formula: Age = Postnatal growth Growth rate where age is in days; postnatal growth is the difference between body weight at a given femoral length minus weight at hatching (estimated to have been c. 0.5 kg I94 R. E. HEINRICH ET AL. TABLE8. Growth rates (gms day-') and associated weight (kg)-age (days) estimates for D.lettowvorbecki. Abbreviation: Sm-Med Trans, small-medium transitional period (days) Growth rate Weight 150 mm Age 150 m m Weight 180 m m Age 180 m m Sm-Med Trans Reptilian" Precocial Bird" Altricial Bird" 6.8 79.2 552.3 12.4 12.4 12.4 I750 150 22 19.8 19.8 19.8 2838 244 35 1088 94 13 Reptilianb Precocial Birdb Altricial Birdb 7.9 91.4 648.9 12.3 12.3 12.3 1494 I29 18 21.4 21.4 21.4 2709 234 33 1215 105 15 Model "Growth rate based on estimated adult body weight of 56.6 kg; bgrowthrate based on estimated adult body weight of 70.0 kg. on the basis of comparisons to similar sized modern iguanas; Weishampel, unpublished data); and growth rate values are those determined from the regressions of Case (1978) as discussed above. Regression equations of weight against femoral length obtained from the data of Table 3 were used to estimate body weight of individuals whose femora were 150 and 180 mm in length (Table 8). Finally, to obtain some idea of the amount of time required to cross the small-medium transition, the age at femoral length 150 mm is subtracted from that at 180 mm (Table 8). Several generalizations can be made on the basis of these calculations. Clearly, unless D.lettowvorbecki possessed growth rates comparable to those of modern avian endotherms, any postulated transition from quadrupedality to bipedality would have been preceded by several years of postnatal development (Table 8). If D.lettowvorbecki growth did follow a n ectothermic pattern, the apparent 'abrupt' increase in percentage cortical area is merely an artefact of our size classification scheme. Likewise, changes to the second moment of area properties analysed would have occurred over a considerably greater period of time than has been suggested here. Interpreting the biomechanical evidence in terms of an ectothermic growth rate, however, is equivocal. We can think of no reason for cross-sectional properties to undergo substantial alterations after several years of growth. A locomotory shift in the second or third year of life is inconsistent with functional and behavioural interpretations of D.lettowvorbecki made to date. Given that dryosaurids were adapted for speed (Coombs, 1978; Galton, 1974, 1981; Thulborn, 1982), it seems unlikely that slower moving quadrupedal individuals would have been accommodated by the swifter moving adults, yet the taphonomic evidence at Tendaguru indicates that individuals of all sizes were living in close approximation, if not herding (Russell et al., 1980). If the architectural modifications found in the femora of Dryosaurus are related to a locomotory shift from quadrupedality to bipedality, then growth rates must have exceeded those of modern reptiles. If the growth rate of D . lettowvorbecki fell somewhere between modern precocial and altricial birds, then the initiation of bipedality (femoral length of c. 150 mm) is estimated to have occurred within five months of hatching and potentially as early as two and a half weeks. Devoting energy towards growth during early ontogeny may well have been of selective advantage for a species that appears to have travelled in large groups. This would also imply that posthatching parental LOCOMOTOR BIOMECHANICS OF DRYOSAURUS 195 care was important during at least early ontogeny of D . lettowvorbecki, as has been suggested for both hadrosaurids (Horner, 1984, Horner & Weishampel, 1988) and hypsilophodontids (Winkler & Murry, 1989). O n the basis of the evidence presented here, we speculate that growth rates in Ornithopoda were more similar to modern endotherms than ectotherms and that all femora categorized in this study as small or medium represent animals in their first year of growth. Those femora classified as large are thought to represent adult or subadult members of the species. SUMMARY Ontogenetic changes in femoral morphology and locomotion were analysed in the iguanodontian dinosaur Dryosaurus lettowvorbecki using cross-sectional data and applying principles of beam theory. The results presented here suggest that locomotor ontogeny in D . lettowvorbecki was more complicated than has generally been recognized. The percentage cortical area (a measure of the relative amount of bone) increases abruptly over a relatively short period during early ontogeny and then remains uniform during subsequent increases in body size. Modifications in cross-sectional shape also occur with increasing size, as demonstrated by differences in second moment of area ratios. The patterns of change in these properties indicate that the orientation of mechanical loadings acting on the femur of D . lettowvorbecki differed at various stages of growth and development. It is suggested that the alterations in femoral architecture described here reflect a shift from quadrupedality to bipedality early in the ontogeny of this animal. ACKNOWLEDGEMENTS We gratefully acknowledge Frank Westphal (Institut f i r Geologie und Palaontologie, Universitat Tubingen, Germany) for permission to study the Dryosaurus specimens and George Barrowclough and Allison Andors (American Museum of Natural History) for access to rhea femora. L. Witmer and an anonymous reviewer provided useful comments on a n earlier version of this manuscript. Funding was provided in part by a Natural Science and Engineering Research Council of Canada Postgraduate Scholarship and Lucille B. Markey Scholarship to REH, and NSF grant support (EAR-8719878, EAR-9004458) and a Visiting Professorship (through Sonderforschungsbereich 230) at the Universitat Tubingen, Germany, to DBW. REFERENCES Alexander RMcN, Brandwood A, Currey JD, Jayeo AS. 1984. Symmetry and precision of control of strength in limb bones of birds. Journal of .Zoology, London 203: 135-143. Alexander RMcN. 1985. Mechanics of posture and gait of some large dinosaurs. .Zoological Journal of d c Linnean Sociey 83: 1-25. Anderson JF, Hall-Martia A, Russell DA. 1985. Long-bone circumference and weight in mammals, birds and dinosaurs. Journal of <oologv, London 207: 53-61. Biewener AA, Swart2 SM, BertJEA. 1986. Bone remodeling during growth: dynamic strain equilibrium in the chick tibiotarsus. CalciJicd Tissue Infernational 39: 390-395. Breu K, Currey JD, Pond CM. 1990. Ontogenetic changes in the mechanical properties of the femur of the polar bear Ursus maritimus. Journal of .Zoology, London 2221 49-58. I96 R. E. HEINRICH E l A L . Case TJ. 1978. Speculations on the growth rate and reproduction of some dinosaurs. Paleobiology 4: 320-328. Colbert EH. 1962. The weights of dinosaurs. Amm’can Museum NoNouitafes,No. 2076: 1-16. Coombs WP.Jr 1978.Theoretical aspects of cursorial adaptations in dinosaurs. Quarterh Review of Biology 53: 393468. JD,Pond CM. 1989. Mechanical properties of very young bone in the axis deer (Axis axis) and humans. Journal of xoology, London 218: 5947. C u m c PJ. 1987. Bird-like characteristics of the jaws and teeth of troodontid theropods (Dinosauria, Saurischia). Journal of Vertebrate Paleontology 7: 72-81. Galton PM. 1971. The mode of life of Hypsilophodon, the supposedly arboreal ornithopod dinosaur. Lethaia 4: Currey 453465. Galton PM. 1974. The ornithischian dinosaur Hypsilophodon from the Wealden of the Isle of Wight. Bulletin of the British Museum (Natural Hisfory) Geology 24: 1-125. Galton PM. 1981. Dryosaurus, a hypsilophodontid dinosaur from the Upper Jurassic of North America and Africa postcranial skeleton. Palaontologisches zcitschriif 55: 27 1-3 1I. Cotemy SM. 1990. Caudofemoral musculature and the evolution of theropod locomotion. Paleobiology 16: 170-186. Cooaship AF., h y o n LE,McFic H. 1979. Functional adaptation of bone to increased stress. Journal of Bone and Joint Surgery 61L 539-546. Homer JR.1984. The nesting behavior of dinosaurs. ScimtzJic American, 2501 130-137. Horncr JR,Weishampel DB. 1988. A comparative embryological study of two ornithischian dinosaurs. Nature 332: 256-257. Janmsch W. 1914. Die Gliederung der Tendaguru-Schichten im Tendaguru-Gebiet und die entstehung der Saurier Lagerstatten. Archiv fur Biontologie 3: 227-262. Janmmch W. 1955. Der ornithopode Dysalotosaurur der Tendaguruschichten. Palaeontographica, Supplmmt VII: 3: 105-176. Lanyon LE. 1984. Functional strains as a determinant for bone remodeling. Calcified Tissue International 36: S56-S6 I. Lanyon LE,Rubin CT. 1985. Functional adaptation in skeletal structures. In: Hildebrand M, Bramble DM, Liem KF, Wake DB, eds. Funcfional vertebrate morphology. Cambridge: Harvard University Press, 1-25. h y o n LE, Cood.hip AE, Pyc E, McFic H. 1982. Mechanically adaptive bone remodelling: a quantitative study on functional adaptation in the radius following ulna osteotomy in sheep. journal of Biomechanics, 15: 141-154. Nagurka ML,Hayes WC.1980. An interactive graphics package for calculating cross-sectional properties of complex shapes. Journal of Biomechanics 13: 419-45 1. Rubin CT. 1984. Skeletal strain and the functional significance of bone architecture. CalciJied Tissue International 36: S 1 I-S 18. Rubin CT, h y o n LE. 1985. Regulation of bone mass by mechanical strain magnitude. CalciJicd Tissue International 37: 41 1 4 1 7. RUB CB, Hayes WC. 1983. Cross-sectional geometry of Pecos Pueblo femora and tibiae-a biomechanical investigation: I. Method and general patterns of variation. American Journal of Physical Anthropology 60: 359-38 I. Rumsell D, B C h d P, McIntomh JS. 1980. Paleoecology of the dinosaurs of Tendaguru (Tanzania).Mmoirc de la Socicte gkologique dc France N o . 139: 169-175. Sereno PC. 1986. Phylogeny of the bird-hipped dinosaurs (Order Ornithischia). National Geographic Research 2: 234-256. Sues HD,Nor~xumD. 1990. Hypsilophodontidae, Tenontosaurus, Dryosauridae. In Weishampel DB, Dodson P, Osmblska H, eds. The Dinosauria. Berkeley: University of California Press, 498-509. Timomhdo SP, Gerc JM.1972. Mechanics of Materials. New York: Van Nostrand Reinhold. Thulborn RA. 1982. Speeds and gaits of dinosaurs. Palacogeography, Palacoclimatology, Palaeoecology 38: 227-256. Torailli PA, Tllrebe K, Burst& AH, Hdple KG. 1981. Structural properties of immature canine bone. Journal of Biomechanical Engineering 103: 232-238. Weimhampel DB, Heinrich RE. 1992. Systematics of Hypsilophodontidae and basal Iguanodontidae (Dinosauria Ornithopcda). Historical Biology 6: 15S184. Wilkinson L. 1988. S T S T A T : The Systemfor ~latistics.Evanston, I L SYSTAT, Inc. Winklcr DA,Murry PA. 1989. Paleoecology and hypsilophodontid behavior at the Proctor Lake dinosaur locality (Early Cretaceous), Texas. In Farlow JO, Ed. Paleobiology qf the dinosaurs. Boulder, Colorado: Geological Socie& of America Special Paper, 238, 5 5 4 I. Woo SLY, Kuci SC, Amiel D, Comcz MA, Hayes WC, White FC, Akcmon WH. 1981. The effect of prolonged physical training on the properties of long bones: a study of Wolff s Law. journal ofBonc andJoinf Surgery 63A: 780-787.
© Copyright 2026 Paperzz