Chapter 3 DINOSAUR PEDAL BIOMECHANICS

Chapter 3. Dinosaur pedal biomechanics

3.1. Morphological changes in pedal phalanges through ornithopod dinosaur evolution: A biomechanical approach

[Moreno K, Carrano MT, Snyder R. Journal of Morphology. Accepted for publication]

3.1.1. Abstract The evolution of ornithopod dinosaurs illustrates a remarkable example of the transition from digitigrady to subunguligrady. During this transition, the pes was drastically altered from the plesiomorphic dinosaurian morphology (four digits, claw-shaped unguals, strongly concavo-convex joints, phalanges longer –proximodistal– than wide – mediolateral–, excavated collateral ligament fossae, presence of sagittal ridge and prominent processes for the attachment of tendons) to a more derived condition (tridactyly, modification of the unguals into hooves, phalanges wider –mediolateral– and thinner – anteroposterior– than long –proximodistal–, lack of collateral ligament fossae, loss of sagittal ridge and processes for tendon attachment, relatively flattened articular surfaces). Our studies examined four non-avian dinosaurs and two birds, documenting the range of pedal morphologies relevant to this particular transition. Our analyses of the external morphology, two-dimensional models (using Finite Element Analysis), and internal bone structure demonstrate that this evolutionary shift generated a loss of digit mobility and flexibility. In addition, pedal posture was modified to align it with the main direction of the ground reaction force, thus becoming better adapted to support high loads. These conclusions can be applied to other, parallel evolutionary changes (in both dinosaurs and mammals) that involved similar transitions to a subunguligrade posture.

3-2

Chapter 3. Dinosaur pedal biomechanics

3.1.2. Introduction Among the most significant changes to have occurred during ornithopod dinosaur evolution are the many modifications of pedal morphology and posture associated with the acquisition of quadrupedalism (Fig. 3.1-1). The transition to a quadrupedal posture occurs in parallel within numerous dinosaur lineages (stegosaurs, ceratopsians, and sauropods; Carrano, 2001), and the acquisition of a subunguligrade pes presents a striking example of convergence, not only with these other dinosaurs, but also with mammals such as rhinocerotids and proboscideans.

Figure 3.1-1. Evolution within Ornithopoda. Changes of stance, posture, and size are evident from the skeletal reconstructions; enlargements highlight changes in digit III pedal morphology. The presence of a softtissue pad (black outline beneath pedal phalanges) has been inferred from trackway morphologies. Skeletal reconstructions are approximately to scale (Lesothosaurus, 1 m total length; Camptosaurus, 8 m; Corythosaurus, 10 m; modified from Paul, 1987; phylogeny simplified from Carrano, 1999).

3-3

Chapter 3. Dinosaur pedal biomechanics

Studies of mammals indicate that as body size increases, limb posture becomes progressively more upright, and limb motion is restricted to a predominantly parasagittal plane (Hildebrand, 1988; Biewener, 1989; Christiansen, 2002). These changes are in response to a reduction in the mass-specific amount of force required to counteract moments about the limb joints (Biewener, 1989). It is reasonable to hypothesize that such changes are valid for quadrupedal dinosaurs as well, although this has not been studied in depth yet. Nonetheless, the physics behind the anatomical features of the pes demonstrates its ability to support high compressive loads (see section 3.2). During ornithopod evolution (Fig. 3.1-1), a change of posture occurred from bipedalism to facultative quadrupedalism, presumably with a consequent forward shift of the centre of mass. Although some uncertainty still exists about how specialized (or obligatory) this quadrupedalism was, abundant evidence is available from different-sized manual/pedal sets of footprints (e.g. Lockley and Hunt, 1995). In addition, a comparative anatomical study of the hadrosaurid ornithopod Maiasaura (Dilkes, 2001) showed that although the forelimb was gracile, it was adapted to carry a portion of body weight, especially in adults. The more primitive ornithopod Dryosaurus may represent an intermediate stage, in which juveniles engaged in facultative quadrupedalism but adults were nearly exclusively bipedal (Heinrich et al., 1993).

3-4

Chapter 3. Dinosaur pedal biomechanics

Figure 3.1-2. Anatomical terms and measurements used in this study. A) Mediolateral view; B) Anterior view.

The pedal morphology of basal ornithopods is shared with theropods, and is therefore plesiomorphic for Dinosauria. This morphology consists of: three primary weight-bearing digits, a phalangeal formula of 2-3-4-5-0, claw-shaped unguals, phalanges that are proximodistally longer than mediolaterally wider, well-developed collateral ligament fossae (Fig. 3.1-2A), prominent proximal processes for the attachment of extensor and flexor tendons, and pronounced development of saddle-shaped (or ginglymoid) joints formed by a sagittal ridge on the proximal phalangeal articular surface (and a complementary furrow on the distal surface) (Fig. 3.1-2B). These features are evident in most basal ornithopods, including Lesothosaurus and “Hypsilophodonts” such as Hypsilophodon, Thescelosaurus, and Tenontosaurus. Basal ankylopollexians such as Camptosaurus show some forelimb modifications that suggest intermittent use in support, but the pes remains quite primitive and retains three claw-like unguals. In all these forms, the pes is distinctly digitigrade, and footprints attributed to primitive ornithopods show discrete digit marks with no indication of a fleshy metatarsal pad. However, more highly derived ornithopods (i.e. iguanodontians) modified this

3-5

Chapter 3. Dinosaur pedal biomechanics

digitigrade pes into a more subunguligade structure. At the same time, the manus became more highly modified for weight bearing, and quadrupedalism was presumably more frequently used. These ornithopods eliminated the first pedal digit (resulting in a phalangeal formula of 0-3-4-5-0) and modified the unguals into hooves. The phalanges are mediolaterally wider and anteroposteriorly thinner than proximodistally long, lack collateral ligament fossae, and exhibit neither a sagittal ridge, a process for tendon attachment, nor excavation of the joint surfaces; all articular surfaces are relatively flattened. These marked postural and morphological differences imply associated differences in pedal kinematics, and therefore in the distribution of forces along the pes. In the present work, our goal was to evaluate the responses of different ornithopod phalangeal morphologies to external loading. Specifically, we examined differences in pedal digit morphology and mobility within and among taxa by studying the external characteristics of the joints. In order to test whether these differences may have been correlated with loading conditions during locomotion, we modeled the distribution of loading in isolated phalanges of various proportions using finite element analysis (FEA). Because trabecular architecture is known to correlate with principal longitudinal stresses and the magnitude of shear stresses (Wolff, 1870; Hayes et al., 1982; Gefen and Seliktar, 2004), we observed the internal trabecular structure using computed tomography (CT) scanned images, and made comparisons to the modeled loading patterns. Finally, we interpreted the results in light of several anatomical “zones” within the phalanx, as a means to understand the consequences of changes in the external and internal morphologies of the distal pedal bones.

3-6

Chapter 3. Dinosaur pedal biomechanics

Figure 3.1-3. Photographs of specimens used in this study in dorsal view. Ornithopods are enclosed in a frame. A) Camptosaurus dispar, left pes, USNM 5473; B) Camptosaurus dispar, right pes, USNM 4277; C) Camptosaurus dispar, left pes, USNM 4697; D) Corythosaurus casuarius, right pes, USNM 15578; E) Saurolophus sp., right pes, AMNH 5271; F) Allosaurus fragilis, left pes, USNM 8424; G) Dromaius novaehollandiae, left pes, P. Kroehler, pers. coll.; and H) Rhea americana, right pes, P. Kroehler, pers. coll. Scale bars = 10 cm.

3-7

Chapter 3. Dinosaur pedal biomechanics

3.1.3.

Materials and methods For this study, we focused on the non-ungual phalanges from nearly complete

specimens of ornithopod feet. We studied three specimens of the ankylopollexian Camptosaurus dispar (USNM 5473, USNM 4277, USNM 4697) and two hadrosaurids, Corythosaurus casuarius (USNM 15578) and Saurolophus sp. (AMNH 5271; mounted in plaster). For comparative purposes, we also examined the pedes of the non-avian theropod Allosaurus fragilis (USNM 8324) and the extant ratites Dromaius novaehollandiae and Rhea americana (Fig. 3.1-3).

3.1.3.1. External morphology

We quantified the distal joint curvature (Fig. 3.1-4A) and sagittal ridge height (Fig. 3.1-5A) of the pedal phalanges in order to describe the basic changes that occurred during ornithopod pedal evolution. Joint curvature reflects the maximal amount of dorsoventral flexion and extension; more flattened phalanges were presumably capable of less dorsoventral movement. The sagittal ridge is a narrow ridge that crosses the proximal joint surface dorsoventrally, dividing it into two concave facets and creating a ginglymoid morphology (Fig. 3.1-2B). This morphology at least partly serves to constrain mediolateral and rotational movements, acting to enforce dorsoventral motion when the joint is under torsional loads. Measurements of the sagittal ridge and joint curvature on these specimens were obtained from the digitised distal joint surfaces using a MicroScribe G2XL (Immersion Corporation M.R.) in combination with the software Rhinoceros 3D (version 3.0, educational license for the Applied Morphometrics Laboratory, National Museum of Natural History, Smithsonian Institution). For practical reasons, the sagittal ridge data were collected from the distal joint surface (i.e.; the sagittal furrow) of the preceding phalanx (Fig. 3.1-2B). Although the sagittal ridge itself is present on the proximal articulation, the contacting phalangeal surfaces conform nearly perfectly, permitting accurate measurements to be taken from either surface.

3-8

Chapter 3. Dinosaur pedal biomechanics

3.1.3.2. Modelling

We created two-dimensional (2D) models of individual phalanges with the software Fempro (Algor V16, Design Check). We constructed basic phalangeal models using two basic geometric shapes, the circle and truncated cone. Either shape (or both) could be modified in order to create variations in the overall shape of the phalanx. Left-right symmetry was constantly maintained (Fig. 3.1-2A) in our models. Although real phalanges often show some degree of asymmetry, especially in certain animals with particular locomotory behaviours (e.g. asymmetrical gait or step, fossorial, etc.), the objective of our model is to show general patterns in phalangeal morphology that can be easily be compared between different animals regardless of individual or specific characteristics. The different degrees of asymmetry should be treated as an additional (untested) variable that may further modify the stress distributions; certainly this aspect is worthy of further study. The scale of the model is assumed to have little effect on the stress distribution. Other studies have shown that for many mammals, stresses are maintained at relatively constant levels over a wide size range without disproportionate changes in the cross-sectional areas of bones (Alexander et al., 1979). Therefore, the results of the present model can be generally compared with pedal bones of different absolute sizes. To study the influence of shaft length (Fig. 3.1-2A), we increased the height of the truncated cone, but not the dimensions of its top, bottom, sides, or of the circle. We used the same method to vary the proximal surface width, by changing the length of the topside of the truncated cone and adjusting the lateral sides for the new shape. For the proximal joint concavity, the portion enclosed by a curve inscribed within the truncated cone was subtracted from the total surface. An equivalent technique was used for the distal convexity: we subtracted the distalmost area formed by a curve inscribed in the circle.

3-9

Chapter 3. Dinosaur pedal biomechanics

3.1.3.3. Finite Element Analysis

We performed a finite element analysis (FEA) in order to examine stress distributions within the model phalanges. The FEA was performed by the software Fempro (Algor V16, Design Check), under the following conditions: (1) Meshing was set to triangular shape at a density of 400–450 triangles per area unit, with optimum meshing symmetry. This construction geometry acts as a braced framework (truss), preventing bending loads in the structure, and is only subjected to longitudinal loading (tension and compression, but not shear or bending). These characteristics are compatible with cancellous bone behaviour (Currey, 2002; Levin, 2002). (2) Material properties were taken from published values (Reilly and Burstein, 1975) for fast-growing haversian bovine bone (isotropic): density = 1895 kg/m3 (Currey, 2002); elastic modulus (E) = 10 GPa; Poisson’s ratio = 0.4; shear modulus (G) = 3.66 GPa. Bovine histology seems to resemble that of dinosaurs (Reid, 1996), at least enough to allow the assumption of broadly similar material properties and minimum strength capability estimates for the models. (3) Loading conditions were set to 1000 N/m2 of pressure and located at the proximal joint surface. This is an approximate value based upon the weights of Camptosaurus and Allosaurus, calculated based on femoral circumferences ~ 300 mm (USNM 6061, USNM 5959, and USNM 8424 (Anderson et al., 1985). Nevertheless, we tested for variations in stress distribution due to loading conditions in advance, using different loading values. The resulting stress magnitudes showed changes proportional to changes in loading input. Therefore loading magnitude was assumed to have little effect on stress distribution. Loading was applied normal to the joint surface, under the assumption that the cartilage and synovial capsule would transmit most normalized forces, as demonstrated in theoretical models (Matthewson, 1982; Dar and Aspden, 2003). (4) Mobility was constrained by fixing the distal joint in all directions (Fig. 3.1-2A), in order to simulate limitations on movement due to the presence of a subsequent phalanx, the ground reaction force (GRF), and the collateral ligaments when the foot was in contact with the ground.

3-10

Chapter 3. Dinosaur pedal biomechanics

(5) For all FEA results, colour range settings were standardized based on the highest and lowest values found after all the model calculations (267–3383 N/m2). This provides a visual means for comparing stress distributions between the models. These loading and constraint conditions correspond to a static analysis of the stance phase in locomotion. As the phalanges were modeled as isolated elements, and loading magnitude and direction have no significant effect, our model will reveal general stress distributions for any phalanx at a given instant when the foot is in contact with the ground. This means that our modeled phalanges could correspond to any location in the foot (different phalanges, digits, or sides), as well as different foot postures during weightbearing walking phases (e.g., early stance, weight bearing, kick off), but always (and only) under static conditions.

3.1.3.4. Internal morphology

We obtained images of the internal structure of the bones by computed tomography (CT) scanning, using a Somatom AR.SP scanner (Siemens Corporation). We scanned the first pedal phalanges of four taxa: Allosaurus, Camptosaurus, a juvenile Corythosaurus, and Saurolophus. The first three were scanned using a consistent field of view (FOV) of 116 mm2 (512 x 512 voxels) and 1 mm slice thickness. The single specimen of Saurolophus (AMNH 5271) required a larger FOV of 350 mm2 (512 x 512 voxels) and 2 mm slice thickness because it was permanently mounted on a plaster base. These images revealed the general trabecular orientation and density of the internal structure of the bones, which are known to correlate with the orientation of principal compressive or tensile stresses, as well as the magnitude of shear stresses (Wolff, 1870; Hayes et al., 1982; Gefen and Seliktar, 2004). We tested these observations against the results from the Finite Element Analysis of two-dimensional models (see above). In these images, the greyscale is an indicator of material density, with white as the highest value and black as the lowest. However, it is important to note that high density (white) does not always indicate dense fossilized bone, but can also correspond to the presence of a harder element such as quartz, as an infilling material (e.g., partial filling of

3-11

Chapter 3. Dinosaur pedal biomechanics

the medullary cavity), or to a transition between a surrounding low-density material (air) to the higher-density fossil (often creating a white outline). 3.1.4. Results 3.1.4.1. External morphology

In general terms, the distal joint curvature is symmetrical in lateral view, and the collateral ligament fossa is located close to the centre of rotation (Fig. 3.1-4A). However, the penultimate phalanges are an exception; they achieve a larger ventral joint surface by means of reducing the dorsal width and increasing ventral width. Here, if the collateral fossa is present (e.g., Camptosaurus, Allosaurus, and ratites) and displaced dorsally, it generates an angle with respect to the ligament that attaches to the ungual.

Figure 3.1-4. Distal joint curvature in pedal phalanges. A) Measurements taken in lateral view (curve, chord), showing the curvature calculation. B) Graph of the average curvature for each digit in the studied specimens.

3-12

Chapter 3. Dinosaur pedal biomechanics

The depth of the collateral ligament fossa can be greater on either the lateral (Camptosaurus) or medial side (Allosaurus), or both may be symmetrical (ratites). On the other hand, Corythosaurus and Saurolophus lack such fossae, and the distal joint remains symmetrical, although the outer sides are better developed in digits II and IV. The average curvature (Fig. 3.1-4B) of Corythosaurus phalanges is nearly constant and close to zero for all three digits. In contrast, digits II and III of Camptosaurus are more curved than those of Allosaurus, but in general curvature remains constant between and within digits (i.e.; a low standard deviation). Allosaurus has one of the largest digit IV curvature, even more than the average in birds, whereas birds have the largest distal joint curvatures in digits II and III. Dromaius and Rhea plotted together and within the same standard deviation.

Figure 3.1-5. Joint surface topography in pedal phalanges. A) Measurements taken in dorsal view (curve, chord), distal joint surface (sagittal furrow), showing the curvature calculation. B) Graph of the average sagittal furrow depth.

3-13

Chapter 3. Dinosaur pedal biomechanics

In most cases, lateral digits have a wider dispersion than the central one (Fig. 3.1-5B), showing a better-developed sagittal ridge. This is especially the case for digit IV, where we found the greatest depth values. Nevertheless, Corythosaurus and Saurolophus retain low values for all digits, which is indicative of their derived hadrosaurid condition (flattened joint surfaces). With the exception of Camptosaurus, the third digit presents a low sagittal ridge, and all digit III phalanges plot together. To summarize, the highest phalangeal curvatures are found in the theropods (Allosaurus and the birds), and the lowest in the hadrosaurids. Camptosaurus is intermediate between the two. Curvature is consistent between the digits in ornithopods, whereas it shows greater variability in theropods.

3.1.4.2. Finite element analysis 2D

In general, our 2D FEA models exhibit their highest stress zones at the contact between the truncated cone and circle (i.e.; the phalanx neck; Fig. 3.1-6), where the extremes of the constraints are placed. The lowest stresses are distributed toward the distal joint. Logically, these stress patterns are highly influenced by the geometric properties of the phalanx. The cone acts as a funnel, concentrating loads toward the outside edges of the narrow section, and the circle efficiently focuses the loads near the loading point, which is evident in the model results (Fig. 3.1-6). Even when these geometric figures are fused, with smooth edges, the stress distribution pattern remains fairly constant (Figs. 3.1-7.1B, D).

3-14

Chapter 3. Dinosaur pedal biomechanics

Figure 3.1-6. Stress von Mises distribution in the basic geometrical shapes used to construct our 2D phalanx model. A) Truncated cone; B) Circle. Grey arrows show the application of stress; grey triangles indicate where the structure was fixed against motion. Internal stress levels are shown from highest (red) to lowest (blue). Note that the truncated cone concentrates stress toward the distal part, whereas the circle effectively concentrates it near the point of application (black arrows).

3-15

Chapter 3. Dinosaur pedal biomechanics

Figure 3.1-7. 2D model of a phalanx in FEA, showing effects of morphological variation. Row 1 varies shaft length and “neck” presence: 1A) original shape, considered a “long shaft”; 1B) removal of “neck” from original shape; 1C) shortened shaft; 1D) shortened shaft without “neck”. Row 2 (A-D) varies the depth/curvature of the proximal concavity. Row 3 (A-D) demonstrates an increase in the proximal joint concavity. Row 4 (A-D) demonstrates a decrease in the distal joint convexity. Row 5 shows the effect of combining narrow proximal width (3D) with flattened distal and proximal joints (4D). Column A represents the same model in each row, for ease comparison.

3-16

Chapter 3. Dinosaur pedal biomechanics

Shortening the shaft length (Figs. 3.1-7.1A–D) increases the maximal stress (compare between figures 3.1-7.1A, C and 3.1-7.1B, D), but tends to lower the minimal stress. Therefore, the range of stress values is wider, and high loads are concentrated in the small area of the phalanx neck (Fig. 3.1-2); this region exhibits the highest stress of all the experimental models (Fig. 3.1-7.1C). Small increases in the concavity of the proximal joint surface (Figs. 3.1-7.2A–D) generate a small decrease in maximal stress (Figs. 3.1-7.2B, C), but a deeper concavity with sharp edges (Fig. 3.1-7.2D) produces higher stress along this surface. Hence, the benefits (i.e. lower stress range) of increasing the proximal concavity have some limitations. Minimal stress remains constant and distally located, which means that changes to the morphology of the proximal concavity have no effects on the distal joint surface, differing from the latter example. These models also show larger stresses distributed along the walls, leaving an internal region with low stress; this was not observed in any of the other experiments. The reduction of the width in the proximal joint surface causes a general decrease in stress (Figs. 3.1-7.3A-D), which becomes more evenly distributed along the shaft. This experiment caused the greatest overall reduction in stress. If we reduce the convexity of the distal joint surface (Figs. 3.1-7.4A-D), minimal stress is raised and maximal stress is slightly decreased. Therefore, a flat structure produces higher but well-distributed stresses. When the two last characteristics are combined, decreasing the proximal concavity and the distal convexity (Fig. 3.1-7.5), the resulting stress has a very low range. This variation exhibited the lowest maximal stress of all the experiments.

3-17

Chapter 3. Dinosaur pedal biomechanics

Figure 3.1-8. Internal structure in pedal phalanges in dorsoventral (A) and sagittal (B) sections. 1) Allosaurus, second phalanx, digit III; 2) Camptosaurus, first phalanx, digit III; 3) Corythosaurus, first phalanx, digit III; and 4) Saurolophus, first phalanx, digit III. mc = medullary cavity. Not to scale.

3-18

Chapter 3. Dinosaur pedal biomechanics

3.1.4.3. Internal morphology

From proximal to distal, the sagittal sections of Allosaurus (Fig. 3.1-8.1A) and Camptosaurus (Fig. 3.1-8.2A) phalanges show a conical structure with thick walls of compact bone, internally filled by a gradient of small (proximal) to large (distal) trabeculae, which end at a large medullary cavity (mc; Figs. 3.1-8.1A,B, 8.2A,B). Both distal joints reveal evenly distributed trabeculae. Bone layering can be observed within either dorsoventral section of the midshafts (Figs. 3.1-8.1B, 8.2B). The only difference is the shape of the dorsoventral cross-section, which is kidney-shaped in Camptosaurus and rounded in Allosaurus. It is important to note that Allosaurus presents no major visible structures in CT images (Fig. 3.1-8.1). However, it was possible to examine its internal bone architecture by direct observation of the broken parts as isolated elements. Note also, that the specimen was longitudinally drilled to insert a metallic support. Hence it was highly fractured. All these features altered CT resulting image, in which the conical gradient of the proximal trabeculae is only slightly noticeable. Yet direct evidence shows an internal structure that is similar to Camptosaurus. Corythosaurus (Fig. 3.1-8.3A) has an internal structure with more evenly distributed trabeculae than Allosaurus and Camptosaurus. Although the trabeculae also form a smooth conical gradient, the medullary cavity and compact bone are absent. Large trabeculae lie near the edges of the bone (Fig. 3.1-8.3B), and no layering can be identified. Saurolophus (Fig. 3.1-8.4A-B) reveals much more evenly distributed trabeculae than the other specimens, with no conical gradient. The sagittal section (Fig. 3.1-8.4A) shows a small zone in the internal part of the midshaft that contains slightly larger trabeculae, but a clear medullary cavity is lacking. In the dorsoventral cut, one “layer” passes parallel to the shaft edge and disappears near either joint (Fig. 3.1-8.4B). This layer may be a scanning artefact: the plaster mounting adds another material interface, which could have produced additional X-ray reflection/refraction. However, observations of broken hadrosaurid phalanges (e.g. phalanx III-3, Edmontosaurus sp., UCM 42353) indicate that this dense CT “layer” could correspond to the boundary of a zone with smaller (and therefore denser)

3-19

Chapter 3. Dinosaur pedal biomechanics

trabeculae located along the medial and lateral perimeter. The other narrow dense zone, which runs from the dorsal edge toward the centre of the phalanx (Fig. 3.1-8.4B), could also be an artefact. Regardless, the salient observation here is that CT images confirm the absence of the medullary cavity and more homogeneous trabecular size compared to the other dinosaurs. 3.1.5. Discussion 3.1.5.1. Pedal mobility and flexibility

Interpreting the data in terms of the flexion-extension ability of the joints (distal joint curvature; Fig. 3.1-4B) and their ability to resist torsional loads (depth of the sagittal ridge; Fig. 3.1-5B), we note that Corythosaurus (juvenile) and Saurolophus have little flexibility in all three digits. This imparts a low resistance to torsion, even in the lateral digits where all the other animals studied have some development of a sagittal ridge. This is typical for a derived ornithopod, and can be explained as a consequence of the more upright position of the pes (Fig. 3.1-1). This reorientation aligns the pes closer to the main direction of the GRF, with a resultant loss of digit mobility. On the other hand, Camptosaurus retains high digital flexibility, being fairly similar to Allosaurus, and hence retaining more plesiomorphic characters as a primitive ornithopod (Fig. 3.1-9). Accordingly, Camptosaurus has a well-developed interlocking phalangeal morphology that counteracts the torsional loads generated by this flexibility. This morphology can be observed even in the third digit, whereas the rest of the animals studied (including birds) show a very shallow sagittal ridge. One possible consequence is an asymmetrical step, in which the animal walked with the pes angled or rotated inward (digits III and IV have deeper sagittal ridges than digit II).

3-20

Chapter 3. Dinosaur pedal biomechanics

Figure 3.1-9. Reconstructions of pedal mobility using 3D element scans. The pedal phalanges of Camptosaurus (A, B) and Allosaurus (C, D) were placed in maximum flexion (A, C) and maximum extension

3-21

Chapter 3. Dinosaur pedal biomechanics

(B, D). Note that in each case, digit II shows the least mobility, but that the two taxa differ with regard to whether digit III or IV is the most mobile.

In Allosaurus, digital flexibility seems quite conservative in general. Here, digits II and III are less flexible than in the ratites, but digit IV has a remarkable mobility (Fig. 3.19C, D), exceeding that seen in the birds. The interlocking phalangeal morphology is also well developed in the lateral digits and, in contrast to Camptosaurus, Allosaurus might have had a more symmetrical step, with the third digit as the primary weight bearer. Dromaius and Rhea have a similar morphology (Fig. 3.1-2), and therefore their pedal function is approximately equivalent, as is evident from their similar locomotory habits (Abourachid and Renous, 2000). The fourth digit has reduced mobility and shortened phalanges, presumably working as a stiffer element in comparison with the other digits. The presence of collateral ligament fossae and their position relative to the distal joint, together with the symmetry of the distal curvature in lateral view, also provides some information about flexion-extension ability. Phalanges are usually built to contribute equally to extension and flexion (via their symmetrical distal curvature), and the collateral ligaments constrain and guide their range of motion. Therefore, in asymmetrical phalanges, with a better-developed ventral side and more dorsally placed collateral ligament, movement is restricted dorsally and greater flexion is supported (but extension is reduced). This is the case for the penultimate phalanges of Camptosaurus, Allosaurus, and the ratites. Their unguals are probably only slightly neutrally extended, with the proximal joint resting high on a footpad, while only a portion of the ungual contacts the ground. The absence of these characteristics in Corythosaurus and Saurolophus indicates that the ungual is more closely aligned with the preceding phalanx, and therefore more extended neutrally. This suggests a different function of the footpad in hadrosaurids, which evenly supports the entire pes.

3.1.5.2. Phalangeal morphology as a consequence of pedal loading

Notice that in most of the animals studied, digit III has a very low sagittal ridge (and shallow corresponding furrow) in each of the phalanges (Fig. 3.1-5B). The sagittal ridge works to reinforce the dorsoventral rotation of the joint. Its weak development (or absence)

3-22

Chapter 3. Dinosaur pedal biomechanics

indicates that digit III might be under low torsional loads, and therefore must be mediolaterally aligned with the GRF. This is confirmed by experiments carried out in starlings and quail (Middleton, 2003), which showed that the centre of pressure is in fact located along digit III, and the lateral digits act later during the walking phase. Variations of pedal anatomy induced minor changes in the position of the centre of pressure, but it always maintained its place within the middle digit. On the other hand, most of the animals studied showed prominent sagittal ridges in digits II and IV (Fig. 3.1-5B). This indicates that the outer digits required more resistance to torsional loads, which probably reflects their main function as stabilizers during standing and walking, as well as their secondary weight-bearing function. Moreover, digit IV appears to have an even more accentuated sagittal ridge, which suggests that this digit, in particular, is usually under higher torsional loads than digit II. Thus the contribution of the digits to locomotion and support is unequal, and specifically is biased toward the midline of the pes in the animals studied. The FEA study is consistent with observations of internal bone structure in these taxa. Compact bone lines the walls of convexo-concave phalanges (Figs. 3.1-8.1, 8.2), where the zones of highest stresses are found in the respective FEA models (Figs. 3.1-7.2B–D). Cancellous bone with small trabeculae is present close beneath the proximal joint surface (Figs. 3.1-8.1, 8.2), which is the same relative position as the elevated stresses (Figs. 3.17.2B–D). These small trabecular zones are underlain by a zone of larger trabeculae and a substantial medullary cavity (Figs. 3.1-7.2B-D), matching the location of a significant lowstress zone in the FEA model. In flattened phalanges, the trabecular architecture is more homogeneous, and compact bone is largely absent internally (Figs. 3.1-7.3, 7.4, 8.3, 8.4). These patterns closely match the loading distributions generated by our models. Taken together, the alignment of the foot, trabecular structure, and loading distribution support the inference that the flattened phalanges of hadrosaurids reflect a modification of foot function from the primitive ornithopod condition. This transition involved changes in posture, flexion-extension limitations, and bone loading.

3-23

Chapter 3. Dinosaur pedal biomechanics

3.1.5.3. Function and Evolution of the Ornithopod Pes

Phalanges with highly convexo-concave joints surfaces are beneficial for bipedal animals. As they have only two supporting limbs, bipedal animals face greater structural requirements for the contact area of the pes, especially if it is to provide equilibrium for the entire body. It is possible to achieve this with long digits that exhibit high, controlled flexibility, which form a “stabilizer platform”. By contrast, quadrupedal animals are more stable (Muller and Verhagen, 2002), and therefore are capable of reducing the contact area of the foot. It is not a coincidence, perhaps, that no bipedal animal exhibits an unguligrade stance. Flattened phalanges with narrow proximal joint surfaces, in combination with restricted mobility and upright posture, create a “columnar support,” which represents the best structure for longitudinal (axial) loadings. This is also the best arrangement for bone resistance to compressive loading (Currey, 2002); thus this morphology would be able to support higher loads and be beneficial for larger-bodied animals. It seems logical then, that dinosaurs such as sauropods, ceratopsians, and stegosaurs, along with mammals such as rhinocerotids and proboscideans, share this phalangeal morphology and also reached large sizes. Another important feature for phalangeal function is the “neck” (Fig. 3.1-5). This structure is a stress concentrator but also reduces the stress in other zones, especially the distal joint surface (compare between models in Fig. 3.1-7.1). Consequently, the distal joint is free to have a high range of motion, because it avoids serving as the main weight-bearing structure. In light of these results, small weight-bearing phalanges with a narrow neck seem improbable, because this small bone would have to support more loading per unit volume than a larger one, and the stress concentration in the phalanx neck could easily exceed bone resistance. Note that a small phalanx with a pronounced neck (Fig. 3.1-7.1C) exhibits the highest stress values of all the experiments. This might explain the reduction of the “neck” in derived pedal phalanges of ornithopods and other large-bodied animals.

3-24

Chapter 3. Dinosaur pedal biomechanics

As stated in “Wolff’s law” (Wolff, 1870), the structure of the body is essentially a blueprint of the forces applied to its structures. When looking at the changes that occurred in ornithopod evolution, it is clear that on the line to Hadrosauridae, ornithopods drastically altered posture, changing the pes from a “stabilizer platform” to a “sub-columnar” structure. The associated soft tissues also altered their function. For digitigrade bipeds, pads act as energy-absorbing cushions and are typically digit-specific, but for subunguligrade quadrupeds, these pads became united beneath the pes and acquired an additional role in support. 3.1.6. Conclusions Hadrosaurids possessed a derived pedal morphology, including flattened phalanges, absence of collateral ligament fossae, and loss of processes for the attachment of flexor and extensor tendons. This morphology indicates that hadrosaurids had a more upright pedal posture than basal ornithopods, which was aligned to the main direction of the GRF during standing. Furthermore, the lack of a stress concentrator (phalanx neck) and the reduction of joint curvature allowed the hadrosaurid pes to support high loads more effectively than the primitive ornithopod pes. In contrast, Camptosaurus, Allosaurus, and the ratites (Rhea and Dromaius) exhibited many of the primitive (basal) conditions of the foot (similar to Allosaurus): concavoconvex joints, marked processes for the attachment of the flexor and extensor tendons, excavated collateral fossae, a dorsal position of these fossae in the penultimate phalanges, a neutral extended claw, and weak sagittal ridges in the lateral digits. In Camptosaurus, the presence of well-developed sagittal ridges in digit III (middle one) suggests that there was high flexibility in all digits, and possibly a mediolaterally asymmetric step. The weak development of the sagittal ridge in digit III of Allosaurus may indicate a more symmetrical step, but digit IV had higher flexibility, as evidenced from its strong joint curvature and marked sagittal ridge. Both ratites also show high joint curvature and development of the sagittal ridge in lateral digits, revealing a particularly elevated flexibility in digits II and III, and a symmetrical gait, despite the stiffer digit IV.

3-25

Chapter 3. Dinosaur pedal biomechanics

In general, digit III is the main weight bearer; lateral digits (II and IV) are under higher torsional loads, with digit IV being more strongly affected. Taken together, this suggests that locomotion in the specimens studied might have been biased toward the midline of the pes. FEA models show loading patterns that are consistent with trabecular structure, allowing us to assume that: primitive pedes (e.g. Allosaurus and Camptosaurus) are adapted to work as “stabilizer platforms”, which require high, controlled flexibility. On the contrary, derived pedes (i.e. hadrosaurids) are well adapted to “columnar support” and resistance to high loads. Although this research was focused on certain dinosaurs, the functional morphological issues discussed here can be applied to other terrestrial tetrapods. Nonungual phalanges are morphologically simple, but at the same time are directly relevant to numerous aspects of animal posture and behaviour. This makes future comparisons between vertebrates especially promising with regard to problems of stability versus functionality, increasing body size, and postural changes during evolution.

3-26

Chapter 3. Dinosaur pedal biomechanics

3.2. Functional Morphology and Evolution of Subunguligrady in Ornithopod Dinosaurs [Moreno K, Carrano M, Snyder R. In preparation]

3.2.1. Abstract The parallel evolution of subunguligrady from a digitigrade-plantigrade posture has occurred at least transiently in different mammals as well as dinosaurs. In ornithopod dinosaurs this transition was gradual, including modifications of pedal and limb posture and increase in size. Our study shows that changes of the external and internal anatomy of the pes also took place, from a plesiomorphic condition of short phalanges 1 to 3, great dorsoventral height, and high development of the interlocking system; to a derived condition in hadrosaurids of longer first phalanx than second and third, reduction of dorsoventral height and reduction of interlocking features. Finite Element Analysis in two dimensions was used to test the influence of external morphology on the stress distribution along the pes. These analyses suggest that the hadrosaurid pes had a remarkable capability of supporting high compressive loads in comparison with basal ornithopods. These changes are evident in the internal structure of pedal phalanges as well, which exhibit high stress concentration zones in basal ornithopods (compact bone and associated medullary cavity) and a more homogeneous internal structure in hadrosaurids. Our findings support the inference that primitive ornithopods were adapted for characteristically high stress digitigrade posture. In contrast, derived ornithopods (hadrosaurids) are more compatible with a subunguligrade posture (at least in adult stages), which typically homogenizes and minimizes compressive stresses. Together, all these modifications served to strengthen the pes for longitudinal loads, and permitted increase in body size but restricted digital mobility. Similarities with other subunguligrade vertebrates indicate that this morphology appears to be a common physical solution that increases resistance to high loads.

3-27

Chapter 3. Dinosaur pedal biomechanics

3.2.2.

Introduction The evolution of a subunguligrade posture has occurred several times in parallel

within quadrupedal terrestrial tetrapods (Fig 3.2-1). The transition is evident in multiple lineages of both mammals (e.g. rhinocerotoids, proboscideans, notoungulates) and dinosaurs (thyreophorans, ceratopsians and sauropods), where it is frequently accompanied by a substantial increase in body size. For most mammal clades in which the transition from plantigrady to sub-unguligrady has occurred, few studies have focused on the morphological and functional changes in the manus and pes. Exceptions include horses (Thomason, 1985) and camelids (Janis et al., 2002), but both groups acquired highly specialized features – monodactyl unguligrady and a dominant pacing gait respectively – that led researchers to focus on relatively derived taxa. As a result, the primitive postures for both clades have received much less attention. Generally, the fossil record of this transition is poorly understood. Archaic ungulates (“Condylarths”) from the Early Eocene, such as the dog-sized arctocyonid Chriacus, retained flexible, plantigrade, and pentadactyl feet with claws. More derived, Late Eocene ungulates, such as the larger Phenacodus and those closely related to more primitive horses, appear to have rapidly evolved an unguligrade pes; few intermediate postural forms are known (Archibald, 1998). So far, equid pedal evolution remain the best-studied postural transition, but one that is not very informative for other groups. Equids had a fairly parasagittal pedal posture right from the beginning. Basal forms such as Hyracotherium (Eocene) and Mesohippus (Oligocene), were probably already subunguligrade (Camp and Smith, 1942; Thomason, 1985), and maintained a basic morphological design of digit III, the main weight-bearing structure, similar to the derived unguligrade Equus. Few modifications of morphology or the proportions of phalanges, plus only subtle variation in muscle, tendon and ligament arrangement are present in equids (Camp and Smith, 1942). Their basic pattern was already established in the primitive group Condylarthra, especially for the hindlimb (Simpson, 1951), however, subunguligrady has appeared at least transiently in numerous other ungulate groups. In which little is known about the details of the transition.

3-28

Chapter 3. Dinosaur pedal biomechanics

Figure 3.2-1. Parallel evolution of subunguligrady (enclosed in a box) in mammals and dinosaurs. (modified from Springer et al., 2003; Carrano, 1999).

3-29

Chapter 3. Dinosaur pedal biomechanics

Within dinosaurs, recent work on posture and locomotion has focused primarily on sauropods (e.g.; Carrano and Wilson, 2001; Bonnan, 2003), and specifically on determining the timing and sequence of certain morphological changes. Fewer studies have addressed the functional implications of these changes (e.g. Coombs, 1978), and none have considered the evolution of subunguligrady in quadrupedal ornithischian dinosaurs. Ornithopods have been essentially disregarded, because workers have generally assumed that their pedal posture was digitigrade despite evidence from footprints, which shows that the pes was supported by a fleshy pad (Currie and Sarjeant, 1979). However, recent morphological work supports inferences from ichnology that derived ornithopods achieved a subunguligrade posture (see section 3.1). In fact, ornithopod dinosaur evolution represents a potentially rich area for the investigation of transition in pedal posture, because their relatively abundant fossil record reveals marked differences in the morphology of digit III (the main weight-bearing structure, as in perissodactyls) that are associated with a postural change from digitigrady (a more horizontal support of the pes, similar to plantigrady) to sub-unguligrady. Basal ornithopods such as Lesothosaurus; “Hypsilophodonts” such as Hypsilophodon and Thescelosaurus; and even ankylopollexians such as Camptosaurus and Tenontosaurus possessed four pedal digits with claw-shaped unguals, phalanges proximodistally longer than mediolaterally wide, well-developed collateral ligament fossae, prominent proximal processes for the attachment of extensor and flexor tendons, and pronounced development of saddle-shaped (or ginglymoid) joints (Fig. 3.2-1). This morphology is also associated with bipedalism and smaller relative body sizes plesiomorphically. More derived iguanodontians modified this digitigrade pes into a more subunguligrade structure, eliminating the first pedal digit and modifying the unguals into hooves. These ornithopods could have developed quadrupedalism in at least the adult stages (Norman, 1980). Finally, the more specialized hadrosaurid phalanges become mediolaterally wider and anteroporteriorly thinner than proximodistally long, and lack structures like the collateral ligament fossae, sagittal ridge, processes for tendon attachment, and excavation of the joint surfaces. Hadrosaurids also changed their posture, probably achieving quadrupedality in

3-30

Chapter 3. Dinosaur pedal biomechanics

adult forms (Dilkes, 2001). These dinosaurs could exceed 10 m long, which is a record size magnitude for all Ornithopoda. As has been shown in previous studies on isolated phalanges of ornithopods (see section 3.1), this evolutionary shift generated a loss of digit mobility and flexibility. In addition, pedal posture was modified to align the pes with the main direction of the ground reaction force, thus becoming better adapted to support higher loads. Associated morphological changes appear to have responded to different loading conditions, as well as kinematics. In the present work we employ three distinct approaches in order to study the effect of the transition to subunguligrady within Ornithopoda on the different morphologies and postures in the pes (with emphasis on digit III): (1) description of the external anatomy and characterization of changes in pedal bone morphology; (2) Finite Element Analysis (FEA), in which we test theoretical predictions of shape-stress relationships; and (3) analysis of bone internal structure from CT scanned images, in order to estimate principal compressive and tensile stresses, as well as the magnitude of shear stresses. We then relate patterns in these data to other features, such as body weight and size increase.

3.2.3. Materials and methods We obtained data from digit III in other ornithopods and theropods (avian and noavian), some of which has been published elsewhere (see section 3.1). These data provide a framework for the morphological changes that occurred during the transition from digitigrady to subunguligrady within Ornithopoda (Table 3.2-1 and 2), and permit comparisons to be made between ornithopods and outgroups like birds and non-avian theropods. Digit III was chosen because it is presumed to have been aligned with the ground reaction force (GRF) within the mesaxonic ornithopod foot. Digit III also shows greater development than the two outer digits, further suggesting that it might have been the principal weight-bearing structure.

3-31

Chapter 3. Dinosaur pedal biomechanics

Table 3.2-1. Data used for the calculations used in figures 3.2-3 to 5. Sign Specimen Rh Rhea americana

Dr

Ale

All1

All2

Ca

Co

Sa

Dromaius novahollandiae

Alectrosaurus olseni

Allosaurus fragilis

Allosaurus fragilis

Camptosaurus dispar

Corythosaurus casuarius

Saurolophus indet.

Coll. Num. Pers. Coll.

Pers. Coll.

6554 AMNH

680 AMNH

8424 USNM

5473 USNM

15578 USNM

5271 AMNH

Phalanx III-1 III-2 III-3

III-1 III-2 III-3 III-1 III-2 III-3 III-1 III-2 III-3 III-1 III-2 III-3 III-1 III-2 III-3 III-1 III-2 III-3 III-1 III-2 III-3

L 43.0 29.6 15.0

58.2 37.3 21.7 110 77 63 152 97 73 107 77.7 72 86 59 43 57.7 20.1 15.8 110 37 32

MLW 12.1 12.4 13.3

14.4 12.9 13.6 31 26 26 70 60 51 42 35 32.4 52 53 48 45 53.7 45.5 72 88 85

DVH 9.3 10 9.5

11.3 7.7 6.2 20 17 14 46 36 33 35 32 25.3 27 30 30 24.3 26.5 23.2 47 37 34.5

GEOM 16.9 15.4 12.4 14.8

FMC

76

21.2 15.5 12.2 15.9

85

40.9 32.4 28.4 33.5

229

78.8 59.4 49.7 61.5

381

54.0 44.3 38.9 45.3

299

49.4 45.4 39.6 44.6

289

39.8 30.6 25.5 31.4

212

71.9 49.4 45.4 54.5

440

Coll. Num. = collection number; L = proximodistal length; MLW = mediolateral width; DVH = dorsoventral height; GEOM = geometrical mean (LxMLWxDVH)1/3, average values shown in bold; and FMC = femoral midshaft circumference. *Juveniles. Values in mm.

3-32

Chapter 3. Dinosaur pedal biomechanics

Specimens measured were housed in the following museums: United States of America: American Museum of Natural History, New York (AMNH); Museum of Comparative Zoology, Harvard University, Cambridge (MCZ); Museum of the Rockies, Bozeman (MOR); National Museum of Natural History, Washington DC (USNM); Peabody Museum, Yale University, New Haven (YPM). Canada: Royal Ontario Museum, Toronto (ROM); Royal Tyrrell Museum of Palaeontology, Drumheller (RTMP); National Museum of Canada, Ottawa (NMC). Argentina: Fundación Miguel Lillo, San Miguel de Tucumán (PVL); Museo de la Universidad del Comahue, Neuquén (MUCPv). England: British Museum of Natural History, London (BMNH); France: Muséum National d’Histoire Naturelle, Paris (MNHN CNJ); Belgium: Institute Royale des Sciences Naturelles de Belgique, Brussels (IRSNB); South Africa: South African Museum, Cape Town (SAM); China: Institute of Vertebrate Palaeontology and Palaeoanthropology, Beijing (IVPP).

3.2.3.1. External morphology

We measured the proximodistal length of the digit III non-ungual pedal elements (metatarsals and phalanges), and compared each length with that of the fourth phalanx. Our data were taken from 39 ornithopod specimens: 8 basal ornithopods, 4 Camptosaurus, 7 iguanodontians, and 20 hadrosaurids. In addition, we also studied 9 theropod specimens: 2 allosaurids, 2 coelurosaurs, 2 ornithomimids, 1 tyrannosaurid, and 2 ratites (Table 3.1-1). When possible, we obtained three proxies for body size: dorsoventral height on the midshaft of the phalanges (DVH); geometric mean (GEOM), which was calculated from the multiplication of three measurements: phalangeal length, mediolateral width and dorsoventral height in the shaft, and then taking the cubic root to the total value; and the femoral midshaft circumference (FMC)(Table 3.1-2). Body size data were used to examine potentially size-related patterns of change within the ornithopod pes.

3-33

Chapter 3. Dinosaur pedal biomechanics

Table 3.2-2. Data used for the calculation of pedal bone length/third phalanx ratio (in parenthesis). Specimen Theropoda Aves Ratites Rhea americana Dromaius novahollandiae Tyrannosauridae Alectrosaurus olseni Ornithomimidae Dromiceiomimus brevitertius Struthiomimus altus Coelurosauria Ornitholestes hermani Compsognathus longipes Allosauridae Allosaurus fragilis Allosaurus fragilis Ornithischia Pisanosaurus mertii Heterodontosaurus tucki Orodromeus makelai Laosaurus consors Xiaosaurus dashampensis Othnielia rex Parksosaurus warreni Gasparinisaura cincosaltensis Camptosauridae Camptosaurus dispar Camptosaurus dispar Camptosaurus dispar Camptosaurus dispar Iguanodontidae Iguanodon bernissartensis Iguanodon bernissartensis Iguanodon bernissartensis Iguanodon bernissartensis Iguanodon bernissartensis Iguanodon atherfieldensis Iguanodon mantelli

Coll. Num.

MT-III

III-1

III-2

III-3

Pers. Coll. Pers. Coll.

325 (21.7) 362 (16.7)

43 (2.9) 58 (2.7)

30 (2.0) 37 (1.7)

15 (1.0) 22 (1.0)

AMNH 6554

529 (8.4)

110 (1.7)

77 (1.2)

63 (1.0)

ROM 797 TMP 85.8.3

296 (7.1) 365 (9.2)

61 (1.5) 81 (2.0)

97 (1.3) 58 (1.5)

42 (1.0) 40 (1.0)

TMP 66.2.3 MNHN CNJ 79

111 (6.6) 81 (4.7)

37 (2.2) 25 (1.4)

28 (1.7) 18 (1.1)

17 (1.0) 18 (1.0)

AMNH 680 USNM 8424

420 (5.8) 345 (4.8)

152 (2.1) 107 (1.1)

97 (1.3) 78 (1.1)

73 (1.0) 72 (1.0)

PVL 2577 SAM MOR 294 YPM1882 IVPP V63701 MCZ 4454 ROM 804 MUCPv 208

96 (6.2) 74 (4.9) 54 (6.4) 125 (5.1) 70 (5.0) 76 (5.8) 156 (4.6) 56 (1.2)

23 (1.5) 22 (1.5) 19 (2.3) 43 (1.8) 18 (1.3) 25 (1.9) 60 (1.8) 49 (1.1)

18 (1.2) 17 (1.1) 13 (1.6) 33 (1.4) 16 (1.1) 17 (1.3) 39 (1.2) 45 (1.0)

15 (1.0) 15 (1.0) 9 (1.0) 24 (1.0) 14 (1.0) 13 (1.0) 34 (1.0) 47 (1.0)

USNM 5473 USNM 4697 USNM 4277 YPM 6801

236 (5.5) 226 (6.1) 229 (5.3) 208 (7.2)

86 (2.0) 90 (2.4) 95 (2.2) 60 (2.1)

59 (1.4) 46 (1.2) 51 (1.2) 35 (1.2)

43 (1.0) 37 (1.0) 43 (1.0) 29 (1.0)

IRSNB 1536 IRSNB G1714 IRSNB R1639 IRSNB 1562 IRSNB Q1534 BMNH R11521 IRSNB 1551

338 (9.5) 358 (8.3) 370 (9.3) 340 (8.9) 350 (10.1) 221 (2.9) 290 (10.0)

121 (3.4) 115 (2.7) 114 (2.9) 132 (3.5) 125 (3.6) 78 (1.0) 105 (3.7)

45(1.3) 48 (1.1) 49 (1.2) 47 (1.2) 40 (1.1) 89 (1.1) 34 (1.2)

36 (1.0) 43 (1.0) 40 (1.0) 38 (1.0) 35 (1.0) 77 (1.0) 29 (1.0) Continued…

3-34

Chapter 3. Dinosaur pedal biomechanics

Specimen Hadrosauridae Hadrosaurus indet. Hadrosaurus indet.* Edmontosaurus edmontoni Edmontosaurus edmontoni Edmontosaurus annectens Edmontosaurus sp. Gryposaurus incurrimanus Gryposaurus incurrimanus Corythosaurus casuarius* Corythosaurus casuarius Corythosaurus casuarius Hypacrosaurus altispinus Hypacrosaurus altispinus Lambeosaurus lambei Lambeosaurus lambei Lambeosaurus lambei Saurolophus indet. Saurolophus osborni Prosaurolophus maximus Prosaurolophus maximus *Juveniles

Coll. Num. FMNH (mounted) MOR 471 NMC 8399 ROM 867 USNM 2414 ROM 5167 ROM 764 TMP 80.22.1 USNM 15578 AMNH 5240 ROM 845 TMP 82.10.1 NMC 8501 TMP-L 82.38.1 TMP-R 82.38.1 ROM 6474 AMNH 5271 AMNH 5220 RO M-L 7871 ROM-R 7871

MT-III 376 (15.0) 227 (15.4) 351 (11.1) 330 (10.8) 340 (8.5) 412 (10.1) 312 (10.2) 316 (11.6) 245 (15.5) 380 (15.2) 391 (13.0) 285 (11.4) 364 (11.7) 442 (15.4) 399 (11.1) 365 (11.6) 274 (8.6) 370 (9.3) 320 (10.4) 307 (9.4)

III-1 128 (5.1) 64 (4.4) 111 (3.5) 109 (3.6) 120 (3.0) 141 (3.4) 109 (3.6) 121 (4.5) 58 (3.7) 135 (5.4) 117 (3.9) 111 (4.4) 109 (3.5) 148 (5.2) 133 (3.7) 123 (3.9) 110 (3.4) 130 (3.3) 124 (4.1) 114 (3.5)

III-2

III-3

38 (1.5) 17 (1.2) 35 (1.1) 62 (2.0) 50 (1.3) 53 (1.3) 33 (1.1) 36 (1.3) 20 (1.3) 40 (1.6) 34 (1.1) 31 (1.3) 42 (1.4) 38 (1.3) 40 (1.1) 40 (1.3) 37 (1.2) 40 (1.0) 31 (1.0) 36 (1.1)

25 (1.0) 15 (1.0) 32 (1.0) 31 (1.0) 40 (1.0) 41 (1.0) 31 (1.0) 27 (1.0) 16 (1.0) 25 (1.0) 30 (1.0) 25 (1.0) 31 (1.0) 29 (1.0) 36 (1.0) 31 (1.0) 32 (1.0) 40 (1.0) 31 (1.0) 33 (1.0)

3.2.3.2. Modelling

Two-dimensional models were made with the software Fempro (Algor V17.1 personal license). They consist of four geometrical objects shaped as phalanges I to IV, in which the respective nodes at the distal and proximal surface of the joints were manually interconnected by lines, generating a truss like mesh. All this structure rests on a fifth object which fits its ventral profile and the zone intended to be surface of contact with the ground was modified according to the postures studied (Fig. 3.2-2). These models were constructed to permit Finite Element Analysis (FEA) of stress distributions and magnitudes within phalanges of different shapes, sizes, and pedal posture (see below). Initial results were reported in section 3.1.

3-35

Chapter 3. Dinosaur pedal biomechanics

Two primary variables are tested in this paper, phalangeal length and pedal posture, which were accommodated by the following models:

(1) Phalangeal length variations: We created three models of the ornithopod digit III, each consisting of three phalanges and one ungual (Fig. 3.2-2), in 2D sagittal cross sections. In these models we arbitrarily varied the proportions of the first phalanx only. These variations were iterative, with either short (length proportions 1.71:1:1 for phalanges 1:2:3), medium (3.1:1.2:1) or long first phalanx models (3.3:1:1), which will be referred to as Short, Medium and Long models hereafter.

Figure 3.2-2. Scheme of FE modelling, loading, constraints and materials: 1) objects shaped as phalanges 1 to 4, bovine bone material; 2) manual line connection of joint nodes, human cartilage; 3) supportive object, human fat pad; 4) edge modeled for minimal pad condition; 5) 1000 N/m2 surface pressure load for all models, including Dig-KO; 6) 1000 N/m2 surface pressure load for Dig-FTD posture; 7) 1000 N/m2 surface pressure load force Sub-FTD; 8) constraint fixed in all directions, located in proximal joint of the first phalanx.

3-36

Chapter 3. Dinosaur pedal biomechanics

Neither ungual morphology, contact surface of the fat pad nor the cartilage area were modified. In this way, any resulting variation of stress would correspond only to differences in the phalangeal proportions. To test how the stress distribution of these different models are affected by changes in posture, each set was also set in three different postures:

a. Pedal bones arbitrarily elevated 45 degrees above the horizontal, with complete fat pad support of the pes, and full touchdown (FTD) posture, which correspond to the static stance (resting position). This experiment resembles the subunguligrade condition (Sub-FTD). b. Pedal bones arbitrarily elevated 45 degrees above the horizontal, with minimum fat pad support of the pes. This posture simulates the digitigrade condition at static kick-off posture (Dig-KO). It is important to remark here that, although the kickoff posture is part of a dynamic gait cycle, here it is only considered as a frozen instant at the final phase of ground contact. c. Pedal bones aligned with the horizontal, with minimum fat pad support of the pes. This posture represents full touchdown posture (Dig-FTD) of a digitigrade in static stance (resting position).

Therefore the overall total corresponds to nine models. The experiment corresponding to Medium in this test was also used for the Saurolophus model in the next FEA experiment (see below). Therefore, resultant values can be compared between them.

(2) Pedal posture variations (digitigrady vs. subunguligrady): We constructed two basic models of the ornithopod digit III. These corresponded to the pedal proportions of Camptosaurus and Saurolophus in sagittal section with the same postural set as in the last experiment (see above; Sub-FTD, Dig-KO, and Dig-FTD). This resulted in an overall total of six models.

3-37

Chapter 3. Dinosaur pedal biomechanics

Ungual morphology, cartilage area and surface of contact of the fat pad were made equivalent in all models; thus the analysis for both taxa can be contrasted directly. In addition, any stress variations between equivalent postures should mainly result from differences in phalangeal proportions and morphology.

3.2.3.3. Finite Element Analysis

Finite element analysis (FEA) was performed on the models under the following conditions (Fig. 3.2-2): (1) Meshing was set to a triangular shape at a density of 400 triangles per unit area, optimized for the best meshing symmetry. (2) Material properties were taken from Reilly and Burstein (1975) for fast-growing haversian bovine bone (isotropic): density = 1895 kg/m3; elastic modulus (E) = 10 GPa; Poisson’s ratio = 0.4; shear modulus (G) = 3.66 GPa. Bovine histology seems to generally resemble that of dinosaurs, allowing us to assume broadly similar material properties and provide a minimum strength capability for the models. In addition, it is known that the mechanical properties of bone are relatively invariant between adults of different terrestrial tetrapod species, and are not important factors in explaining the maintenance of skeletal stresses in animals of different sizes (Biewener, 1990; Currey, 2002). Values for cartilage were taken from Chen et al. (2001) for human feet: density = 1200 Kg/m3; E = 10 MPa; and Poisson’s ratio = 0.4. Values for fat pad material properties were also taken from Chen et al. (2001): density = 1000 Kg/m3 (water); E = 1.15 MPa; and Poisson’s ratio = 0.49. The function of articular cartilage and fat pad is to support and distribute loads, and in the case of cartilage, to provide lubrication in the diarthrodial joints (Boshetti et al., 2004). Softtissue stiffness was possibly higher for dinosaurs, however the values chosen from the literature represent general conditions for the synovial capsule and fat pad in humans. This probably gives an underestimation of the soft tissue performance (e.g. Gefen, 2003; Cheung et al., 2005), but also might highlight bone performance. (3) Loading conditions arbitrarily were set to 1000 N/m2 of surface pressure, and located at the “surface of contact with the ground”, to recreate the Ground Reaction Force

3-38

Chapter 3. Dinosaur pedal biomechanics

(GRF). Hence the forces were all vertical (perpendicular to the ground surface), as befits a model that is only static. (4) Constraints were located at the proximal joint of phalanx 1 and fixed in all directions, in order to simulate translational displacement limitations due to the presence of the rest of the body mass. (5) Linear static analysis. This element formulation propagates forces only through nodes at the mesh corners, ignoring the interaction of the adjacent matrix. This simplifies the calculations and speeds processing, but reduces model accuracy, underestimating its behaviour. Several important limitations should be considered. First, the model corresponds to a 2D sagittal cross section, and therefore forces generated by the 3D nature of the pes, as well as the effects of muscles, tendons, ligaments, fasciae, etc. are not considered. Although the lack of many such important features means that our model is far from being realistic, nonetheless it reveals information about certain specific variables (phalangeal length and pedal posture variations) whose particular contribution would otherwise be difficult to isolate. Second, the model replicates a static posture, and extrapolations to the dynamic condition are very restricted. Studies have shown that the dynamics of gait as well as passive dynamics of energy stored in bone, tendon and other tissues, exert a major influence on structural performance (Cavanagh et al., 1997; Full et al., 2002). Therefore, our study does not address questions concerning the dynamic of pedal function in ornithopods. Third, the material properties, constraints and loading condition applied to these models, along with the limitations discussed above, tended to underestimate pedal structure behaviour. Therefore, data provided cannot be used for further calculations, such as the safety factor or be directly compared with stress magnitudes obtained by experimentation with modern taxa. Stress magnitudes are used to identify general loading patterns that provide some constraints on how pedal biomechanics might have worked, providing a starting point for further studies.

3-39

Chapter 3. Dinosaur pedal biomechanics

3.2.3.4. Internal morphology

We obtained CT images of three taxa: Camptosaurus, a juvenile Corythosaurus, and Saurolophus, in order to visualize general trabecular orientation and density of the internal structure of the bones. Settings and details of the CT analysis are described in section 3.1. In these images (Fig. 3.2-10), the material density is shown as a greyscale code in which white represents the highest value and black the lowest. Some of the high density (white) zones do not necessarily indicate denser fossilized bone, but instead correspond to X-ray reflection of a harder element such as quartz or pyrite, which could be present within the infilling material gained during the fossilization process. Also, drastic transitions in material density (air-rock) usually generate a white outline. These images were utilized to compare FEA results with the main orientation of principal compressive or tensile stresses, as well as the magnitude of shear stresses experienced by the bone, and are evident in its internal architecture (Wolff, 1870; Hayes et al., 1982; Gefen and Seliktar, 2004).

3-40

Chapter 3. Dinosaur pedal biomechanics

Figure 3.2-3. Comparison of phalangeal geometric mean (GEOM = (LxMLWxDVH)1/3; L = proximodistal length, MLW = mediolateral width, DVH = dorsoventral height) versus femoral midshaft circumference (FMC). Dinosaurs: Ale = Alectrosaurus olseni AMNH 6554; All1 = Allosaurus fragilis AMNH 680; All2 = Allosaurus fragilis USNM 8424; Ca = Camptosaurus dispar USNM 5473; Co = Corythosaurus casuarius 15578 USNM; Sa = Saurolophus sp. AMNH 5271. Ratite birds: Dr = Dromaius novaehollandiae; Rh = Rhea americana. (A) regression line when excluding Sauroplophus; y = 0.1275x + 5.65; R2 = 0.94; (B) regression line considering all data; y= 0.146x + 9.6; R2 = 0.88. Data in Table 3.2-2, values in mm.

3.2.4. Results 3.2.4.1. External morphology

Comparing the results of the geometric mean from phalangeal dimensions (GEOM = (LxMLWxDVH)1/3; L = proximodistal length, MLW = mediolateral width, DVH = dorsoventral height) against the femur midshaft circumference (FMC), it can be seen that most of the values are linearly related (Fig. 3.2-3). The linear regression that includes Saurolophus shows a smaller predictability (y= 0.146x + 9.6; R2 = 0.88; Fig. 3.2-3-B) than when Saurolophus is excluded (y = 0.1275x + 5.65; R2 = 0.94; Fig. 3.2-3-A). Although the

3-41

Chapter 3. Dinosaur pedal biomechanics

amount of data available is reduced, the dimensions of the pedal phalanges in Saurolophus suggest a lower body mass in comparison with Allosaurus (AMNH 680). This conflicts with the expected differences because of the much slender construction of the latter. In contrast, FMC appears to give a better proxy for body mass relative to other species (Saurolophus has the largest FMC) than the phalangeal geometric mean. Modifications of pedal scaling seem to be source of this disparity, including an increase in metatarsal and phalanx III-1 length (Fig. 3.2-4) and a reduction of the dorsoventral height (DVH; Fig. 3.25), which appear not to follow isometric scaling.

Figure 3.2-4. Pedal bone length/third phalanx length ratio in digit III. Upper graph shows proportional differences in proximodistal length between phalanx 1 and respective metatarsal, whereas the lower graph shows the proportional differences in proximodistal length between phalanx 2 and 1 (n = 48; data in table 3.21).

3-42

Chapter 3. Dinosaur pedal biomechanics

Figure 3.2-5. Graph showing the differences on relative dorsoventral height in phalanges of digit III. DVH = Phalangeal dorsoventral height; FMC = femoral midshaft circumference. Dinosaurs: All2 = Allosaurus fragilis USNM 8424; Ca = Camptosaurus dispar USNM 5473; Co = Corythosaurus casuarius 15578 USNM; Sa = Saurolophus sp. AMNH 5271. Ratite birds: Dr = Dromaius novaehollandiae; Rh = Rhea americana. Data on Table 3.2-2.

Third metatarsal length/non-ungual third phalanx ratio in digit III shows distinctive scaling patterns (Fig. 3.2-3). In ratite birds, metatarsal III is almost 20 times longer than the third phalanx, but phalanges 1 and 2 are only slightly longer (2.8 and 2.0 times). In basal ornithopods, theropods and camptosaurids each pedal bone maintains the lowest length ratio differences in comparison with the third phalanx: the metatarsals are up to six times larger, the first phalanges less than two times larger and the second phalanges are almost equal. Camptosaurus exhibit a slightly larger first phalanx. Iguanodontids reveal a marked increase of the first phalanx and metatarsal in comparison with the third phalanx (8 and 3 times respectively), retaining the proportions of the second phalanx seen in theropods, basal ornithopods and camptosaurids. Finally, hadrosaurids achieve the greatest relative length of the first phalanx (10 times larger than the third phalanx), including birds, and the largest

3-43

Chapter 3. Dinosaur pedal biomechanics

metatarsal of all non-avian dinosaurs (~4 times), but still retain the similarity between the second and third phalanges. Phalangeal DVH/FMC ratio, expressed as a percentage shows (Fig. 3.2-5) that Saurolophus has remarkably thinner first to third phalanges. But in Corythosaurus, the first phalanx appears to be more proportionally similar to camptosaurids, although the values are also low. Once again Camptosaurus reveals closer proportional similarities with Allosaurus in all phalanges. And birds show greater dorsoventral height of the first phalanx, although the rest are fairly close to the ones of Allosaurus and Camptosaurus.

3-44

Chapter 3. Dinosaur pedal biomechanics

Figure 3.2-6. Stress distributions of two-dimensional finite element analysis for phalangeal length variations: Long, Medium and Short first phalanx models. Sub-FTD = Subunguligrade in full touch down posture; DigKO = Digitigrade in kick-off posture; Dig-FTD = Digitigrade in full touch down posture.

3-45

Chapter 3. Dinosaur pedal biomechanics

3.2.4.2. 2D Finite element analysis

(1) Phalangeal length variations: Comparing the general stress patterns (Fig. 3.2-6), maximum stress (in von Mises) is concentrated at the first joint between the first and second phalanges in all models. It is then distributed proximally and distally through the edges of the bones and cartilage. Little stress reaches the third phalanx and even less reaches the fat pad. The stress concentration on the ventral sides of the bones and cartilage corresponds to tension (i.e.; maximum principal stress), and that on the dorsal side to compression (i.e.; minimum principal stress). For all three postures studied (Sub-FTD, DigKO, and Dig-FTD), the Short model presents maximum von Mises stress from 27% to 58% higher than in the Long model.

Figure 3.2-7. Comparison of tension and compression stresses for the phalangeal length variations in three different postures (same as described in Fig. 3.2-6). Open signs = absolute value of maximum tension; closed signs = absolute value of maximum compression.

3-46

Chapter 3. Dinosaur pedal biomechanics

Several differences are found between tension and compression stress for the different postures (Fig. 3.2-7). The highest values in all experiments are found in Dig-FTD posture and the lowest are present in Sub-FTD for the Long and Medium models, but in Dig-KO for the Short model. Tension and compression patterns are proportional in all models. For Sub-FTD posture the tension is 7% lower than compression, whereas in DigKO the tension is slightly lower (~2%) and in Dig-FTD posture it is somewhat higher (~1%).

Figure 3.2-8. Stress distributions of two-dimensional finite element analysis for Camptosaurus pedal model versus Saurolophus pedal model in three different postures.

3-47

Chapter 3. Dinosaur pedal biomechanics

(2) Pedal posture variation: The general stress distribution in the Camptosaurus model is quite different from that in the Saurolophus model (Medium model in previous test) in all the experiments (Fig. 3.2-8). In Camptosaurus, the maximum stress is usually located all along the entire dorsal and ventral edges of the phalanges, including the ungual, leaving a low-stress region in the centre of each bone and cartilage structure. In addition, the fat pad shows higher stress zones below the joint, which is especially notable in the Dig-FTD posture (Fig. 3.2-8). In contrast, in Saurolophus the stress is concentrated at the first joint and then disperses toward the edges, with no central lower or higher stress in the fat pad. In both models, tension and compression are concentrated on the ventral and dorsal sides respectively. For all the experiments the Camptosaurus model presented stresses 164% to 208% higher than the Saurolophus models (Fig. 3.2-9). Camptosaurus exhibited the lowest tension and compression values in the Dig-KO posture. In the Sub-FTD posture, tension was slightly lower (~1%) than compression, but in Dig-KO tension it was nearly 14% higher and in Dig-FTD is was elevated about 35%. In Saurolophus, the tension and compression pattern is somewhat reversed. In the Sub-FTD and Dig-KO postures, respectively, tension was 7% and 2% lower than compression, but in the Dig-FTD posture tension was slightly higher (~1%).

3-48

Chapter 3. Dinosaur pedal biomechanics

Figure 3.2-9. Comparison of tension and compression stresses between Camptosaurus and Saurolophus models in three postures. Sub-FTD = Subunguligrade in full touch down posture; Dig-KO = Digitigrade in kick-off posture; Dig-FTD = Digitigrade in full touch down posture. Open signs= absolute value of maximum tension; close signs = absolute value of maximum compression.

3-49

Chapter 3. Dinosaur pedal biomechanics

Figure 3.2-10. CT images of the digit III non-ungual phalanges of ornithopods: A = Camptosaurus dispar USNM 4277; B = Juvenile Corythosaurus casuarius USNM 15578; C = Saurolophus sp. AMNH 5271. Views: Di = Distal; Do = Dorsal; Me = Medial; La = Lateral; Po = Proximal; Ve = Ventral.

3-50

Chapter 3. Dinosaur pedal biomechanics

3.2.4.3. Internal morphology (Fig. 10)

Different internal structures and external morphologies are evident between the CT images of digit III in Camptosaurus, juvenile Corythosaurus and Saurolophus digit III (Fig. 3.2-10). In Camptosaurus the first phalanx exhibits in the proximal part a trabecular conical gradient, oriented toward a large medullary cavity, which in turn is surrounded by compact bone that reinforces the “neck” of the phalanx. In the distal part, larger and more homogeneous trabeculae are present. The second and third phalanges present compact bone in the flexor and extensor processes for the attachment of tendons, and a reduced medullary cavity appears in the centre of the phalangeal neck (Fig. 3.2-10A.2). In the coronal view (Fig. 3.2-10A.1), the sagittal ridges and associated grooves that form their ginglymoid (saddle-shaped) joints are apparent. In transverse view (Fig. 3.2-10A.3-5) the medullary cavities are well defined for all three phalanges. Also, in the second and third phalanges, a pattern of larger trabeculae running from the cavity toward the dorsolateral and ventromedial corners is displayed. The juvenile Corythosaurus presents a more homogeneous trabecular structure, lacking a medullary cavity but still showing a conical gradient in the first phalanx (Fig. 3.210B1-2). This conical gradient appears to be more dorsally displaced, as can be noticed in the sagittal section (Fig. 3.2-10B.2). In all three phalanges, a denser bone layer can be seen along the lateral, dorsal and medial edges, leaving a less dense zone on the ventral side (Fig 10B.3-5). Slight development of the ginglymoid joints is evident (Fig. 10B.1). In Saurolophus, the third digit appears to have an even more homogeneous structure than in Corythosaurus (Fig. 3.2-10C). The internal structure of all phalanges is quite similar; all of them seem to have a denser bone wedge running from the dorsal to the central part, parallel to the lateral, dorsal, and medial edges. In addition, they all lack a medullary cavity and a conical gradient of trabeculae (Fig. 3.2-10C.3-5). The joint morphology is fairly simple with no development of sagittal ridges, and the mediolateral width is reduced in the distal phalanges (Fig. 3.2-10C.1). In both hadrosaurids

3-51

Chapter 3. Dinosaur pedal biomechanics

(Saurolophus and Corythosaurus) the cross-sectional shape is ellipsoidal (i.e.; wider in mediolateral direction) in comparison with a more rounded one in Camptosaurus.

3.2.5. Discussion 3.2.5.1. Stress distribution in the ornithopod pes through evolution

The evolution of the pes in Ornithopoda is marked by changes in pedal morphology that had consequences for stress distribution within the foot. From the data obtained in this study, we can state that the pes became more asymmetrical along the line to hadrosaurids (Fig. 3.2-3 and 11), maintaining basal proportions (i.e.; Theropoda, basal ornithopods and Ankylopollexia) in the second phalanx, but the length of metatarsals and first phalanges systematically increased (Fig. 3.2-4), while reducing dorsoventral height (Fig. 3.2-5). Using Camptosaurus as example of the basal condition, the maximum stress in each phalanx is concentrated along the edges and in the phalangeal neck, as shown in section 3.1. When the whole pes is considered the maximum stresses are distributed all along the pes, in which each bone experiences significant magnitudes of stress, probably higher than the more derived hadrosaurids (values are at least 160% higher in our models; Fig. 3.2-8). Some of this high stress might be the result of a comparatively shorter first phalanx, although this morphology might perform slightly better in bending (Fig. 3.2-7). However, most of the stress might be a consequence of the greater joint curvature and its associated wider range of movement of pedal elements (see section 3.1). For Camptosaurus, as well as more primitive ornithopods, the soft tissues have a bigger role in distributing loads. Specifically, higher-stress zones correspond to tendon attachments, and some of the loads appear to be supported even by the fat pad that lies adjacent to the ventral surface of the joint (Fig. 3.2-6). This corresponds to the position of digital pad prints in basal ornithopods and theropods (e.g. Thulborn and Wade, 1984). In conclusion, the primitive ornithopod pes was under high stress. In order to relieve these stresses, the soft-tissue acted as a specific support system that permitted a wide range

3-52

Chapter 3. Dinosaur pedal biomechanics

of digit mobility whilst extracting bone overloads. This condition probably contributed to constrain their maximum body weight. In hadrosaurids, the stress pattern is reverted. Stress is lower within each phalanx and has a more homogeneous distribution than in basal taxa (section 3.1). But looking at the general stress distribution along the pes, although lower overall regardless of pedal posture, stresses are concentrated mainly in the first joint (III-1, 2; Fig. 3.2-8). Thus, it is evident that hadrosaurid pedal morphology probably was under more homogeneous loads (i.e.; similar values of tension and compression; Fig. 3.2-9), and stress magnitude was decreased via first phalanx lengthening (Fig. 3.2-6). As a result, general performance of the pedal structure was significantly improved in hadrosaurids, which helped to avoid stress concentration zones and support higher loads. It is well known that phalangeal morphology is a useful tool for habitat predictability. For ungulate mammals there are environmental factors that influence pedal morphology, which has revealed in African bovids a distinction between taxa that frequent the open and heavy cover from those that frequent the forest and light cover taxa (e.g.; DeGusta and Vrba, 2005). Open/heavy cover ungulates have shorter phalanges relative to width and greater dorsoventral height, and forest/light cover taxa present the inverse condition. Although it has not been studied which differences in the mechanical properties of the environment would have lead to such a distinction, it is interesting to note a somewhat analogous pattern in the ecological and evolutionary context of ornithopods. Primitive ornithopods from the Jurassic lived in a more humid climate, probably more similar to a heavy cover habitat (Bakker, 1980; Spicer, 2003) and had proximodistally short, but mediolaterally wide and anteroposteriorly higher phalanges. Toward the end of the Cretaceous, hadrosaurids experienced dryer conditions and abundance of light cover areas (Bakker, 1980; Spicer, 2003). In concert, they evolved proximodistally longer and mediolaterally wider but anteroposteriorly lower pedal bones, especially in first phalanges. These characteristics gave them different stress distributions (see above), which are related to the wider mobility range and more horizontal position of the pedal bones in primitive ornithopods; as well as limited movement and a more columnar support of the pes in

3-53

Chapter 3. Dinosaur pedal biomechanics

hadrosaurids. These similarities between bovids and dinosaurs suggests that environmental conditions are important factors for pedal morphology in both groups. If we extrapolate our results on pedal stress distribution, we could hypothesize that a forest/light cover area would promote pedal strengthening, in order to support higher loads and an open or heavy cover environment would favour pedal mobility against high loading capability. Further studies are necessary to better test this hypothesis.

3.2.5.2. Pedal posture

In general, we can deduce from the FE models that a subunguligrade posture would generate higher compression in the dorsal part of the pes and lower tension on the ventral side of the phalanges. In contrast, a digitigrade posture would generate higher tension in the ventral side than the dorsal one, whether in full touchdown or kick-off postures. The presence of a fat pad greatly reduces overall loads in the Sub-FTD posture, whereas the stress became larger due to a greater contact surface in the Dig-FTD posture for both Camptosaurus and Saurolophus. However, we found some differences in stress behaviour due to different postures (Fig. 3.2-9). The pes of Camptosaurus experienced modifications in tension and compression, while the Saurolophus pes appears to have fairly consistent stresses in all postures (i.e.; similar tension and compression values in each case). In the Camptosaurus FEA the absolute value of maximum compression is much lower than the maximum tension in both digitigrade postures (Dig-KO and Dig-FTD), but it is slightly higher in the subunguligrade one (Sub-FTD), this means that a subunguligrade posture in Camptosaurus would have been comparatively less adapted to compressive loads than a digitigrade one. Digitigrade models of Camptosaurus reveal high tension stresses, which were likely to be controlled and minimized by soft tissue such as tendons, ligaments, and/or aponeurosis (e.g. Roberts et al., 1997; Solomonow, 1997; Khaw et al., 2005; Robinson et al., 2005). Evidence of the presence of some of those tensional structures in Camptosaurus is the great development of collateral fossae and tendon attachments.

3-54

Chapter 3. Dinosaur pedal biomechanics

Saurolophus has little differentiation of tension and compression in the different postures, which varied less than 2% in each posture. The low maximum stress values in our model correlate with the low development of the soft-tissue supporting system in hadrosaurids. The muscle extensor digitorium brevis in all likelihood was absent (Dilkes, 2000). Also, flexor muscle scar areas and collateral ligament fossae were highly reduced (see section 3.1). A similar condition has been found in other large subunguligrades, in which fewer elastic structures are present (e.g. elephant; Weissengruber and Forstenpointner, 2004). Therefore, in Saurolophus the main supporting and dissipative structures might have been mainly the pedal elements and fat pad. In summary, primitive ornithopods were unlikely to have had a subunguligrade posture, due to the high compressive loads that this posture generates, for which they are not well adapted. Instead, their anatomy is focused on the extraction of tension loads, mainly by high development of tendons and ligament attachments, which are more compatible with digitigrady. In contrast, hadrosaurids are essentially built to minimize compressive stresses, which are typical of a subunguligrade stance (Fig. 3.2-11).

Figure 3.2-11. Evolution of the pes in Ornithopoda. Some of the morphological and biomechanical changes are illustrated here: presence of hooves (Hadrosaurids) or claws (other ornithopods); posture; proportion of phalangeal length; magnitude of stresses in pedal bones; and body size.

3-55

Chapter 3. Dinosaur pedal biomechanics

3.2.5.3. Changes in the pes due to increase in size

The wide range exhibited by ornithopods through ontogeny and phylogeny ranges from about 30 cm to more than 10 m body length, which would demand substantial adaptations by the skeleton (Fig. 3.2-11). Such adaptation comprises reduction of the distal limb, restriction of mobility, and restructuring of the internal bone architecture and pedal posture (see above). In addition, body size seems to correlate with body posture. Alignment of the leg bones in the parasagittal plane will decrease torsion and shear at the joints, and transmits the energy more directly (i.e.; Biewener, 1990). In ornithopods we observe a similar pattern. The small primitive ornithopods had a more reduced joint angle (i.e. crouching posture; Heinrich et al., 1993), and derived ornithopods, the hadrosaurids, had a more parasagittal posture, which was likely more pronounced in adulthood. Although comparison between mammals and dinosaurs must be done carefully, similarities in limb posture have been established between the two groups (Carrano, 2001). Like all subunguligrade-unguligrade placental mammals, in which this posture evolved separately (e.g. rhinocerotids, proboscideans, equids, etc.; Lovegrove and Haines, 2004), hadrosaurids developed large body sizes. Subunguligrady therefore seems to be a common solution to the problem of large body size. It is interesting to note that pedal morphology is more derived in adult hadrosaurids than in juveniles. In the juvenile Corythosaurus pes studied, the first phalanx morphology seems to be somewhat closer to basal ornithopods. This is evident in the greater dorsoventral height, the presence of the conical gradient of trabeculae, incipient ginglymoid joints, and greater joint curvature (see also section 3.1). It is possible that these morphological characteristics correspond to ontogenetic postural changes from digitigrady (in the juvenile) to subunguligrady (in the adult). This is in general agreement with analyses of appendicular myology on an ontogenetic series of the hadrosaurid Maiasaura (Dilkes, 2001). Furthermore, FEA suggests that higher tension load in the ventral side could be generated by a more digitigrade posture (Fig. 3.2-9) and might have favoured a greater development of the ventral bone structure in the first phalanx of Corythosaurus (Fig. 3.2-

3-56

Chapter 3. Dinosaur pedal biomechanics

10B). This is due to the ellipsoidal cross section of the first phalanx, in which the internal bone structure would have served to prevent bending loads in the dorsoventral direction. This structure is absent in Camptosaurus, probably due to the rounder cross-section and the presence of compact bone in the phalangeal neck, which would have redistributed loads. However, Corythosaurus distal phalanges (III-2, 3) are more similar to those of Saurolophus, and so, were likely to have been used in a subunguligrade posture. Still, their incipient interlocking system suggests that torsional loads were present about the joints. Therefore, digit posture and/or mechanics might have been different from the adults. It is reasonable to suggest that the first phalanx might have been held at a lower angle, more closely aligned with the horizontal plane (i.e.; ground), and/or had a greater range of movement in juveniles. Other studies had noticed anatomical changes in the hadrosaurid Maiasaura (Dilkes, 2001) which indicated that adults had a more slender pes than juveniles. This observation was attributed to fat pad stress dissipation capabilities. Our study indicates that these changes might correspond to a more complex scenario where different morphologies, kinematics and stress distributions all contributed to minimize stress loads. It has been found that artiodactyls (Skedros et al., 2003) experienced a mass reduction of the distal limb related to the metabolic cost of functional use. The distal end experiences a greater range of velocity changes in swing phase than the proximal end, and therefore reduction of the distal limb mass would be advantageous, in terms of conservation of angular momentum and mechanical work during locomotion. Although this observation has been made in cursorial animals, it might be equally valuable for non-cursorial animals that reached large body size. Such a condition generates high shock energy in the distal limbs even at low speeds, therefore necessitating a great deal of energy conservation. Minimization of distal limb mass and surface of contact, as well as a more parasagittal posture are necessary concomitant changes.

3-57

Chapter 3. Dinosaur pedal biomechanics

3.2.6. Conclusion Many morphological changes occurred during the evolution of the pes in Ornithopoda. In digit III the plesiomorphic condition for all the phalanges is: 1 to 3 are short; great dorsoventral height; high development of the interlocking system by means of sagittal ridges and correspondent grooves; great joint curvature; and deep collateral ligament fossae. The derived condition in hadrosaurids is: lengthening of the first phalanx over second and third; reduction of dorsoventral height, joint curvature and lack of collateral ligament fossae. Bone distribution changed accordingly, by reducing concentration zones such as compact bone around the phalanx neck and associated medullary cavity to form a more homogeneous internal structure. Together, all these modifications served to strengthen the pes for longitudinal loads, but restricted digit mobility. Resemblance with the African bovid phalangeal morphology and the degree of habitat openness, suggest that these could have been initiated by climate change. However, this hypothesis requires further investigation. Anatomical features of primitive ornithopods were adapted for digitigrade postures, via soft-tissue extraction of the characteristically high tension loads; and derived ornithopods (hadrosaurids) were more compatible with a subunguligrade posture, which homogenized and minimized compressive stresses. However, juvenile hadrosaurids were likely to have a more digitigrade posture, thus the fully derived condition (subunguligrady) was achieved only in the adult stage. These forms evolved large body sizes, which appear to be a common evolutionary characteristic of all subunguligrade vertebrates.

3-58

Chapter 3. Dinosaur pedal biomechanics

3.3. References Abourachid A, Renous S. 2000. Bipedal locomotion in ratites (Paleognatiform): examples of cursorial birds. Ibis 142:538-549. Alexander RM, Jayes AS, Maloiy GMO, Wathuta EM. 1979. Allometry of the limb bones of mammals from shrews (Sorex) to elephant (Loxodonta). Journal of Zoology 189:305-314. Anderson JF, Hallmartin A, Russell DA. 1985. Long-bone circumference and weight in mammals, birds and dinosaurs. Journal of Zoology 207:53-61. Archibald DJ. 1998. Archaic ungulates ("Condylarthra"). In: Janis CM, Scott KM, Jacobs LL, editors. Evolution of Tertiary Mammals of North America: Volume 1, Terrestrial Carnivores, Ungulates, and Ungulate like Mammals. Cambridge: Cambridge University Press. p 292-325. Bakker RT. 1980. Dinosaur heresy - dinosaur renaissance: Why we need endothermic archosaurs for a comprehensive theory of bioenergetic evolution. In: Thomas RDK, Olsen EC, editors. A Cold Look at Warm-Blooded Dinosaurs. Washington, DC: AAAS. p 351-462. Biewener AA. 1989. Scaling body support in mammals: Limb posture and muscle mechanics. Science 245:45-48. Biewener AA. 1990. Biomechanics of mammalian terrestrial locomotion. Science 250:1097-1103. Bonnan MF. 2003. The evolution of manus shape in sauropod dinosaurs: Implications for functional morphology, forelimb orientation, and phylogeny. Journal of Vertebrate Paleontology 23:595-613. Boshetti F, Pennati G, Gervaso F, Peretti GM, Dibini G. 2004. Biomechanical properties of human articular cartilage under compressive loads. Biorheology 41:159-166. Camp CL, Smith N. 1942. Phylogeny and functions of the digital ligaments of the horse. Memories of the University of California. Berkeley: University of California. p 173.

3-59

Chapter 3. Dinosaur pedal biomechanics

Carrano MT. 1999. What, if anything, is a cursor? Categories versus continua for determining locomotor habit in mammals and dinosaurs. Journal of Zoology 247:29-42. Carrano MT. 2001. Implications of limb bone scaling, curvature and eccentricity in mammals and non-avian dinosaurs. Journal of Zoology 254:41-55. Carrano MT, Wilson JA. 2001. Taxon distributions and the tetrapod track record. Paleobiology 27:564-582. Cavanagh PR, Morag E, Boulton AJM, Young MJ, Deffner KT, Pammert SE. 1997. The relationship of static foot structure to dynamic foot function. Journal of Biomechanics 30:243-250. Chen W-P, Tang F-T, Ju C-W. 2001. Stress distribution of the foot during mid-stance to push-off in barefoot gait: a 3-D finite element analysis. Clinical Biomechanics 16:614-620. Cheung JT, Zhang M, Leun AK, Fan Y. 2005. Three-dimensional finite element analysis of the foot during standing - a material sensitivity study. Journal of Biomechanics 38:1045-1054. Christiansen P. 2002. Locomotion in terrestrial mammals: the influence of body mass, limb length and bone proportions on speed. Zoological Journal of the Linnean Society 136:685-714. Coombs WP. 1978. Theoretical aspects of cursorial adaptation in dinosaurs. The Quarterly Review of Biology 53:393-417. Currey JD. 2002. Bones: Structure and Mechanics. Princeton and Oxford: Princeton University Press. 436 p. Currie PJ, Sarjeant WAS. 1979. Lower Cretaceous dinosaur footprints from the Peace River Canyon, British Columbia, Canada. Palaeogeography Palaeoclimatology Palaeoecology 28:103-115. Dar FH, Aspden RM. 2003. A finite element model of an idealized diarthrodial joint to investigate the effects of variation in the mechanical properties of the tissues.

3-60

Chapter 3. Dinosaur pedal biomechanics

Proceedings of the Institution of Mechanical Engineers Part H- Journal of Engineering in Medicine 217:341-348. DeGusta D, Vrba E. 2005. Methods for inferring paleohabitats from the functional morphology of bovid phalanges. Journal of Archaeological Science 32:1099-1113. Dilkes DW. 2000. Appendicular myology of the hadrosaurian dinosaur Maiasaura peeblesorum from the Late Cretaceous (Campanian) of Montana. Transactions of the Royal Society of Edinburgh - Earth Sciences 90:87-125. Dilkes DW. 2001. An ontogenetic perspective on locomotion in the Late Cretaceous dinosaur Maiasaura peeblesorum (Ornithischia: Hadrosauridae). Canadian Journal of Earth Sciences 38:1205-1227. Full RJ, Kubow T, Schmitt J, Holmes P, Koditscheek D. 2002. Quantifying dynamic stability and maneuverability in legged locomotion. Integrative and Comparative Biology 42:149-157. Gefen A. 2003. Plantar soft tissue loading under the medial metatarsals in the standing diabetic foot. Medical Engineering & Physics 25:491-499. Gefen A, Seliktar R. 2004. Comparison of the trabecular architecture and the isostatic stress flow in the human calcaneus. Medical Engineering & Physics 26:119-129. Hayes WC, Snyder BM, Levine BM, Ramaswamy S. 1982. Stress-morphology relationships in trabecular bone of the patella. In: Galagher RH, editor. Finite Elements in Biomechanics. New York: Wiley. p 223-267. Heinrich RE, Ruff CB, Weishampel DB. 1993. Femoral ontogeny and locomotor biomechanics of Dryosaurus lettowvorbecki (Dinosauria, Iguanodontia). Zoological Journal of the Linnean Society 108:179-196. Hildebrand M. 1988. Analysis of vertebrate structure. New York: John Wiley & Sons. Janis CM, Theodor JM, Boisvert B. 2002. Locomotor evolution in camels revisited: A quantitative analysis of pedal anatomy and the acquisition of the pacing gait. Journal of Vertebrate Paleontology 22:110-121. Khaw F, Mak P, Johnson GR, Briggs PJ. 2005. Distal ligamentous restraints of the first metatarsal. An in-vitro biomechanical study. Clinical Biomechanics 20:653-658.

3-61

Chapter 3. Dinosaur pedal biomechanics

Levin SM. 2002. The tensegrity-truss as a model for spine mechanics: Biotensegrity. Journal of Mechanics in Medicine and Biology 2:375-387. Lockley MG, Hunt AP. 1995. Dinosaur tracks and other fossil footprints of the western United States. New York: Columbia University Press. 338 p. Lovegrove BG, Haines L. 2004. The evolution of placental mammal body sizes: evolutionary history, form, and function. Oecologia 138:13-27. Matthewson MJ. 1982. The effect of a thin compliant protective coating on hertzian contact stresses. Journal of Physics D-Applied Physics 15:237-249. Muller M, Verhagen JHG. 2002. Optimisation of the mechanical performance of a two-duct semicircular duct system - Part 3: The positioning of the ducts in the head. Journal of Theoretical Biology 216:443-459. Norman DB. 1980. On the ornithischian dinosaur Iguanodon bernissartensis from the Lower Cretaceous of Bernissat (Belgium). Institut Royal des Sciences Naturelles de Belgique, Memoires 178:1-103. Paul GS. 1987. The science and art of restoring the life appearance of dinosaurs and their relatives. In: Czerkas SJ, Olson EC, editors. Dinosaurs Past and Present. Los Angeles: Natural History Museum. p 5-49. Reid REH. 1996. Bone histology of the Cleveland-Lloyd dinosaurs and of dinosaurs in general, part I: Introduction to bone tissues. Brigham Young University Geology Studies 41:25-71. Reilly DT, Burstein AH. 1975. Elastic and ultimate properties of compact bone tissue. Journal of Biomechanics 8:393-396. Roberts TJ, Marsh RL, Weyand PG, Taylor CR. 1997. Muscular force in running turkeys: the economy of minimizing work. Science 275:1113-1115. Robinson JR, Bull AM, Amis AA. 2005. Structural properties of the medial collateral ligament complex of the human knee. Journal of Biomechanics 38:1067-1074. Simpson GG. 1951. Horses. New York: Oxford University Press. 247 p.

3-62

Chapter 3. Dinosaur pedal biomechanics

Skedros JG, Sybrowsky CL, Parry TR, Bloebaum RD. 2003. Regional differences in cortical bone organization and microdamage prevalence in Rocky Mountain mule deer. The Anatomical Record 274:837-850. Solomonow M. 1997. Comment on "role of muscle belly and tendon of soleus, gastrocnemius and plantaris in mechanical energy absorption and generation during cat locomotion". Journal of Biomechanics 30:307. Spicer RA. 2003. Changing climate and biota. In: Skelton PW, Spicer RA, Kelley SP, Gilmour I, editors. The Cretaceous World. Cambridge: Cambridge University Press. p 85-162. Springer MS, Murphy WJ, Eizirik E, O'Brien SJ. 2003. Placental mammal diversification and the Cretaceous-Tertiary boundary. Proceedings of the National Academy of Sciences of the United States of America 100:1056-1061. Thomason JJ. 1985. The relationship of trabecular architecture to inferred loading patterns in the third metacarpals of the extinct equids Merychippus and Mesohippus. Paleobiology 11:323-335. Thulborn RA, Wade M. 1984. Dinosaur trackways in the Winton Formation (MidCretaceous) of Queensland. Memoirs of the Queensland Museum 21:413-517. Weissengruber GE, Forstenpointner G. 2004. Musculature of the crus and pes of the African elephant (Loxodonta africana): insight into semiplantigrade limb architecture. Anatomical Embryology 208:451-461. Wolff J. 1870. Über die innere Architektur der Knochen und ihre Bedeurung für die Frage von Knochenwachstum. Virchows Archive on Pathological Anatomy 50:389-450.

3-63