Simulated Range of Motion and Hind Foot Posture of the Middle Jurassic Sauropod Rhoetosaurus

Nair Jay (Orcid ID: 0000-0001-9218-8124) Jannel Andréas (Orcid ID: 0000-0002-6625-5693)

70 Jannel “Keep your feet on the ground”: Simulated range of motion and hind foot posture of sauropods — a case study of the Middle Jurassic Rhoetosaurus brownei

LRH: ANDRÉAS JANNEL ET AL.

RRH: BIOMECHANICS OF AN EARLY SAUROPOD PES

Andréas Jannel , Jay P. Nair , Olga Panagiotopoulou , Anthony Romilio and

Steven W. Salisbury

Andréas Jannel. School of Biological Sciences, The University of Queensland, Brisbane,

QLD 4072, Australia. E-mail: [email protected]; [email protected]

Jay P. Nair. School of Biological Sciences, The University of Queensland, Brisbane, QLD

4072, Australia. E-mail: [email protected]; [email protected]

Olga Panagiotopoulou. Monash Biomedicine Discovery Institute, Department of Anatomy and Developmental Biology, Monash University, Clayton, VIC 3800, Australia. E-mail: [email protected]

This is the author manuscript accepted for publication and has undergone full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the Version of Record. Please cite this article as doi: 10.1002/jmor.20989

Anthony Romilio. School of Biological Sciences, The University of Queensland, Brisbane,

QLD 4072, Australia. E-mail: [email protected]

Steven W. Salisbury. School of Biological Sciences, The University of Queensland, Brisbane,

QLD 4072, Australia. E-mail: [email protected]

ABSTRACT The biomechanics of the sauropod dinosaur pes is poorly understood, particularly among the earliest members of the group. To date, reasonably complete and articulated pedes in Early–Middle Jurassic sauropods are rare, limited to a handful of taxa. Of these, Rhoetosaurus brownei, from eastern Australia, is currently the only one from the

Gondwanan Middle Jurassic that preserves an articulated pes. Using Rhoetosaurus brownei as a case exemplar, we assessed its paleobiomechanical capabilities and pedal posture.

Physical and virtual manipulations of the pedal elements were undertaken to evaluate the range of motion between the pedal joints, under both bone-to-bone and cartilaginous scenarios. By employing the results as constraints, virtual reconstructions of all possible pedal postures were generated. We show that Rhoetosaurus brownei was capable of significant digital mobility at the osteological metatarsophalangeal and distal interphalangeal joints. We assume these movements would have been restricted by soft tissue in life but that

their presence would have helped in the support of the animal. Further insights based on

anatomy and theoretical mechanical constraints restricted the skeletal postures to a range

encompassing digitigrade to subunguligrade stances. The approach was extended to

additional sauropodomorph pedes, and some validation was provided via the bone data of an

African elephant pes. Based on the resulting pedal configurations, the in-life plantar surface

of the sauropod pes is inferred to extend caudally from the digits, with a soft tissue pad

supporting the elevated metatarsus. The plantar pad is inferred to play a role in the reduction

of biomechanical stresses, and to aid in support and locomotion. A pedal pad may have been

a key biomechanical innovation in early sauropods, ultimately resulting in a functionally

plantigrade pes, which may have arisen during the Early to Middle Jurassic. Further

mechanical studies are ultimately required to permit validation of this long-standing

hypothesis.

KEYWORDS: Sauropoda; range of motion; posture; pes; biomechanics; Rhoetosaurus brownei

1 INTRODUCTION

Sauropods represent the heaviest terrestrial organisms known to have existed, exemplifying one of the most emblematic groups of dinosaurs. With body masses exceeding 40 tonnes

(Bonaparte & Coria, 1993; Alexander, 1998; Mazzetta, Christiansen & Fariña, 2004; Benson et al., 2014; Lacovara et al., 2014), one might ask how these gigantic animals could move, let alone support themselves, without their bones failing under such heavy loads. Many studies have reconstructed sauropod locomotion using a variety of approaches, including comparative osteology (e.g., Hay, 1910; Christiansen, 1997; Carrano, 2005; Sander et al.,

2011), paleoichnological analysis (e.g., Wilson & Carrano, 1999; Wilson, 2005), biomechanical investigations (e.g., Christian, Heinrich & Golder, 1999; Preuschoft et al.,

2011; Klinkhamer et al., 2018), and/or computational techniques (e.g., Sellers et al., 2013). In sum, the majority of these studies proposed that sauropods likely had fairly restricted limb movements with reduced mobility at the main joints.

These aforementioned studies have nonetheless focused entirely on analysis of the upper and middle long bone segments (the stylopodium and zeugopodium, respectively). For the vast majority of terrestrial vertebrates, however, the lower limb segments (the autopodium), comprising the manus and pes, represent the only skeletomuscular parts that interact with the substrate. Accordingly, the biomechanical properties of the autopodium not only operate for the support of the animal but also facilitate the transmission of forces into the more proximal parts of the limbs during locomotion (Floyd, 2014). So far, however, most locomotor studies have generalized sauropod autopodia to function as singular and static body segments, with a limited capacity for movement at the wrist and/or ankle joints (e.g., Sellers et al., 2013), if at

all. Consequently, little is known about the exact biomechanical abilities of sauropod

autopodia, and the extent to which these complex structures could withstand stresses and

simultaneously propel these terrestrial giants.

As often noted, the manus and pes of sauropods are morphologically divergent structures

(Upchurch, Barrett & Dodson, 2004; Bonnan, 2005). While the manus comprises an elevated

colonnade of metacarpals, the pes is generally proposed to contain more horizontally

positioned metatarsals. However, the attribution of these configurations rests largely on the

interpreter’s opinion of how the autopodial elements articulate, lacking rigorous support from

biomechanical insights. A detailed understanding of the paleobiomechanical abilities of the

autopodia in sauropods has therefore been deficient largely because certain parameters have

been difficult to estimate with precision. These comprise ranges of motion (ROM) between

ossifications, posture and autopodial-level kinematics. Consequently, the posture and

biomechanics of the sauropod autopodia have often been open to subtly variant

interpretations (e.g., Marsh, 1878; Hatcher, 1901; Paul, 1988; Gallup, 1989; McIntosh, 1990;

Bonnan, 2005; Carrano, 2005; Wilson, 2005).

In fact, published observations concerning movements within the sauropod pes remain

limited to a handful of studies, mostly detailing motion at a few joints within the parasagittal

plane. For instance, Hatcher (1901) observed that the unguals of Diplodocus carnegii (CM

94) were capable of substantial ‘vertical’ movements based on their articular morphology.

Gallup (1989: p. 2) noticed that the articulation at the metatarsophalangeal joint of pes

specimen FMNH 241-50 (now referred to Cedarosaurus weiskopfae, D'Emic, 2013) allowed for significant motion in the vertical plane. A similar appraisal was made by Bonnan (2005:

p. 353), observing that the unique articular surface of the unguals at the most distal

interphalangeal joints permit considerable flexion in Plateosaurus engelhardti, an early

sauropodomorph. Whilst these observations provide preliminary indication for the flexibility

at some joints within the sauropod pes, quantification of the associated pedal ROMs remains

lacking (specifically in a three-dimensional space).

Given this backdrop, we sought to investigate the paleobiomechanical capabilities and test

the likelihood of potential postures in early sauropod pedes, using the Middle Jurassic

Rhoetosaurus brownei as an exemplar. The scope of this study is to provide a preliminary

constraint-based framework (Gatesy, Baker & Hutchinson, 2009) for the acquisition of more

robust and uniform paleobiomechanical data in the reconstruction of sauropod autopodia, an

approach fundamentally applicable to other extinct taxa.

To evaluate this overarching aim, a set of hierarchical investigations was performed on the

pes of Rhoetosaurus brownei. First, we measured the flexibility of each pedal joint using

various ROM evaluations in a 3D context (physical or virtual; with cartilage or bone-to-

bone), an aspect so far not yet quantified in the literature of sauropods. Second, each ROM

analysis of Rhoetosaurus brownei was statistically compared between procedural methods

(physical, virtual) and articular assumptions (bone-to-bone, cartilage) with ANOVA in order

to determine the most exhaustive technique for paleobiomechanical analysis in extinct taxa.

Third, identical ROM evaluations were performed with an African bush elephant (Loxodonta

africana) pes to provide corroboration of our methods to modern taxa. As the largest living terrestrial animal, the African elephant provides an ideal analogue because of the comparably high degree of mechanical forces acting upon their feet. This allows for assessment of the

ROM at each osseous pedal joint from a ‘blind anatomical perspective’, whilst yielding pedal

ROM quantification that has been so far surprisingly scarce in elephants. Fourth, we generated the full range of osseous pedal postures that were achievable by the pes of

Rhoetosaurus brownei, grounded from the postural morphotypes attributed to sauropods in the literature. This was achieved via a constraint-based approach to geometrically estimate autopodial postural morphotypes, using the bounds of the ROM results as an initial input.

Postures assumed unlikely in previous works (but which are reconstructed within the constraints of the ROM results) are not discounted a priori, but instead are critically assessed.

The aim of this investigation was to ascertain the most plausible pedal posture(s) adopted naturally in Rhoetosaurus brownei based on anatomy, comparisons with living taxa, and ichnology.

Institutional Abbreviations: CM: Carnegie Museum of Natural History (Pittsburgh,

U.S.A.); FMNH: Field Museum of Natural History (Chicago, U.S.A.); MB.R., see MfN;

MdCL: Musée des Confluences (Lyon, France); MfN, Museum für Naturkunde –

Leibniz−Institut für Evolutions− und Biodiversitätsforschung an der Humboldt−Universität zu Berlin (collection numbers: MB.R.) (Berlin, Germany); POL: Collection of the centre de conservation et d'étude (Lons-le-Saunier, Jura, France); QM: Queensland Museum (Brisbane,

Australia); RVC: The Royal Veterinary College, University of London (Hatfield, U.K.);

YPM: Yale Peabody Museum (New Haven, Connecticut, U.S.A.).

2 MATERIALS & METHODS

2.1 Study specimens

2.1.1 Rhoetosaurus brownei (case study)

This study focuses on the nearly complete right lower hindlimb of the holotypic specimen of

Rhoetosaurus brownei (QM F1659, see Supplementary Figure S1), a sauropod from the

Middle Jurassic of Queensland, Australia (Longman, 1926; Nair & Salisbury, 2012). The tibia, fibula, astragalus and pes were examined firsthand, while physical manipulations were conducted on replicas. As preserved, the tibia and astragalus are conjoined by matrix in a partially disarticulated arrangement (Nair & Salisbury, 2012). Therefore, replicas of these elements have been detached and re-articulated to facilitate postural reconstructions. The pes preserves digits I–IV complete (4 metatarsi, 12 phalanges), but is missing digit V (estimated to be a metatarsal and a single phalanx based on other taxa). When excavated the pes was found articulated, and subsequent preparation freed some of the bones from the matrix.

However, the proximal phalanges of each digit (specifically, phalanges II-1+2, III-1+2+3 and

IV-1+2) remain tightly interconnected by matrix, precluding manipulation between them. A sensitivity analysis was therefore performed to estimate the potential effects at each interconnected interphalangeal joint on the overall ROM of the pes (see Supplementary Data for details). To permit consistent comparisons with our physical analysis, this study focuses on the well-preserved metatarsophalangeal (MTP) and distalmost interphalangeal (DIP) joints of each digit (Fig. 1).

2.1.2 African bush elephant (Loxodonta africana; validation taxon)

In order to validate our approach based on an extant taxon, we applied the same virtual based

procedures to the pes of an African elephant (Loxodonta africana; based on the left pes of a

3500 kg, 19 year old bull, RVC ID 0602151). Computed tomography scans of the specimen

were obtained courtesy of John R. Hutchinson (GE Lightspeed 16-detector unit: voxel size

3.20 mm; voltage 120 kVp; scanned February 2015). The individual bones were digitally

modelled and volume rendered using Mimics by OP (version 17.0; Materialise, Inc., Leuven,

Belgium), with the left pes being subsequently mirrored for comparison to the right pes of

Rhoetosaurus brownei.

2.1.3 Additional sauropods (comparative taxa)

We compared the postural data derived from Rhoetosaurus brownei with the pedes of

additional sauropod and non-sauropod sauropodomorph taxa in order to place our inferences

into a broader evolutionary context (Fig. 2, and see Supplementary Data and Fig. S6–7 for

further details on the phylogeny). We selected the Late Triassic Plateosaurus engelhardti to

represent a non-sauropod (out-group) exemplar, using a complete left pes (POL 70, formerly

Dimodosaurus polignensis). The sauropods pedes compared comprised three Late Jurassic

taxa: (1) the right pes of the North American diplodocoid Diplodocus carnegii (CM 94, mounted cast at MfN); (2) the right pes of the North American macronarian Camarasaurus sp. (mounted cast at MdCL); (3) the right pes of the East African macronarian Giraffatitan brancai (reconstructed mounted pes at MfN).

Three-dimensional digital surface models of these specimens were generated using photogrammetry (Falkingham, 2012; Mallison & Wings, 2012; see Rhoetosaurus

photogrammetry below for specifications). With the exception of POL 70 (Plateosaurus

engelhardti), these comparative specimens were mounted, thus precluding the manipulations

of individual elements. Because the 3D models of the sauropod pedes resulted in a single

interconnected mesh, each pedal element was virtually isolated in order to reconstruct their

‘pseudo-articular’ surfaces. Given this limitation, we choose not to report the pseudo-pedal

ROMs of these specimens. These additional postural reconstructions, nonetheless, provide

relevant comparative and contextual instances of corroborating information. Although outside

of our present focus on Rhoetosaurus brownei, further quantification of the pedal ROMs and

postures of these additional taxa are required.

2.2 Osteological neutral pose of the autopodium

Measurements of ROMs are relative to an osteological neutral pose (ONP), which does not correspond to a natural standing pose of an animal in life (Fig. 1). ONP has been defined, in lateral view, to be the position of two articulated bones where their long axes are parallel to each other (Reiss & Mallison, 2014: p. 3). The ONP provides an objective means for comparative analysis of ROMs, derived from different approaches (i.e., physical versus virtual; and/or between taxa). With regards to intra-autopodial joints, we here consider the

ONP to correspond to an alignment of the long axes of articulated bones being parallel to each other, not only in lateral, but also in dorsal and distal views (Fig. 3). This revised ONP allows for ROM measurements between autopodial joints, in a 3D space, to be objectively compared. These positions were initially assessed virtually (in Autodesk Maya 2016; www.autodesk.com), and applied to the physical manipulations. Fundamentally, ROMs are measured within three degrees of freedom (DOF), from the centre of rotation of each joint

(Fig. 1B): (1) a x-axis where lateromedial axial rotations occur; (2) a y-axis where

dorsoplantar flexions occur; and (3) a z-axis where abduction/adduction occurs. For

definitions of these terms of motion, see Table 1.

2.3 Range of motion analysis

In this study, we measure the ROM within the pes of Rhoetosaurus brownei considering

bone-to-bone ROM manipulations, in both traditional physical and virtual settings, combined

with an attempt to incorporate the effect of cartilage (Fig. 4). Determining the angular ROM

at appendicular joints is an approach used to infer basic limb postures and functionality. The

method has been often used to evaluate the limits of flexion/extension at joints of extinct

animals, such as in dinosaur forelimbs (e.g., Carpenter, 2002; Senter, 2005; Senter & Parrish,

2005; Senter & Robins, 2005; Senter, 2006; Senter & Parrish, 2006; Bonnan & Senter, 2007;

Langer, de França & Gabriel, 2007). For dinosaurs, ROM estimations have almost always

been limited to bone-to-bone manipulations, which is problematic because the role of

cartilage and other soft tissues has been deemed highly significant in biomechanics (e.g.,

Schwarz-Wings, Wings & Meyer, 2007; Bonnan et al., 2010; Holliday et al., 2010; Hutson &

Hutson, 2012).

Witmer (1995) proposed that the ‘Extant Phylogenetic Bracket’ framework be used when reconstructing soft-tissues of extinct animals. However for sauropods, extant bracketing

relatives (i.e., birds and crocodiles) are morphofunctionally very dissimilar and therefore may

not represent appropriate correlates for soft tissue inferences. Instead, sauropods have been

regularly compared to elephants in functional studies (e.g., Christiansen, 1997; Henderson,

2006); these examples are ‘level III inferences’ of the Extant Phylogenetic Bracket, which

employ correlates beyond an immediate phylogenetic bracket (see Witmer, 1995; Perry &

Prufrock, 2018).

Within the pes of sauropods, the full suite of ligaments and muscles is difficult to reconstruct with precision. While sauropod pedes are occasionally well preserved to indicate muscle/ligament scars (e.g., CM89, CM 94 [Hatcher, 1901]), not all pedal soft tissues in life may have been associated with them. Further, the known myology of extant phylogenetic bracketing taxa might be radically divergent to those of sauropods. Rhoetosaurus brownei itself only shows a limited number of scars on the metatarsi, but not enough to identify corresponding connective tissues with certainty. Thus, for the purposes of this study, we did not take into consideration the effects of most soft tissues (except for pseudo-cartilage), but acknowledge such limitations in the discussion of our results.

2.3.1 Pseudo-cartilage thickness

An arbitrary 20 mm space between all pedal bones was applied, representing ‘pseudo- cartilage’ in our bone-to-cartilage ROM manipulations from an experimental standpoint. This was thought to provide the full sets of ROM movements achievable at a joint, i.e., ranging from a bone-to-bone to thick cartilaginous facets. In this context, it is important to emphasize

that such quantified data is entirely lacking in fossil taxa such as sauropods. The choice for

this cartilaginous thickness is detailed below (also see Supplementary Data).

Among extant animals, cartilage thickness has been demonstrated to be correlated to body

weight in a simple allometric manner (Simon, 1970; Stockwell, 1971). Surprisingly however,

explicit cartilage thicknesses as proportions of associated autopodial bones have yet to be

documented in the literature, unlike at the ends of stylopodial and zeugopodial bones

(Holliday et al., 2010, see Supplemental Table S2). When compared with the elephants, as

the largest living terrestrial organisms, few studies have actually specified the thickness

cartilage in their limbs (Simon, 1970; Stockwell, 1971, see Supplemental Data; Malda et al.,

2012). Variation in reported cartilage thickness (e.g., ranging between 1.0–11.0 mm in Egger

et al., 2008) is likely contingent on multiple factors, including ontogeny, species, size, or specific region of joint, among others (Egger et al., 2008; Malda et al., 2012, see

Supplemental Data).

Given that the vast majority of sauropods would have easily surpassed the masses of modern species of elephants, their respective cartilage thicknesses would have presumably

been greater. Although we lack quantification in sauropods, Holliday et al. (2010) evaluated

the proportion of cartilage thickness in the long limbs of several extant phylogenetic

bracketing taxa, ranging between a maximum of 10.8% of bone length (in Alligator

mississippiensis) and 1.8% (in Coturnix japonicus), which we incorporate to this discourse.

Applying these proportional ranges to the pedal phalanges of Rhoetosaurus brownei yields

hypothetical cartilage thicknesses of 15.3–40.8 mm (mean 34.4 mm) and 2.5–6.8 (mean 5.7

mm) using the Alligator mississippiensis and Coturnix japonicus cartilage proportions,

respectively (Supplementary Table S3).

Because the aforementioned values are correlated from data of the long bones rather than

autopodial elements (considerably shorter), we conducted a sensitivity analysis to estimate

the effect of cartilage thickness on ROM measurements (see Supplementary Data). The

sensitivity analysis demonstrates non-statistically significant differences in ROM values

recorded between large articular thicknesses (i.e., between arbitrary gaps of 20, 30, and 40

mm [Supplementary Text and Figure S10]), an outcome that may be explained by the higher

degree of disarticulation. Given the insensitivity shown at the higher values, we set the value of maximum cartilage thickness to be 20 mm. Additional support for a cartilage thickness being less than 20 mm thick is inferred by the preservation of the study specimen (QM

F1659), in which the matrix-filled spaces between phalanges range from 5–20 mm. We

acknowledge that additional studies are required to further substantiate cartilage thickness

estimates within the pes of sauropods and other fossil taxa.

For Loxodonta africana, observations were made on the generated 3D model. In this

model, the gap between the bones was taken to be indicative of cartilaginous thickness.

Measured, these gaps range from 0.5–1.5 mm. Hence, the thickness of virtual cartilage was

set as 1.5 ± 0.5 mm, corresponding with the thickness evaluated at the femoral cnemial

condyles (taken from Egger et al., 2008).

2.3.2 Physical manipulation (analysis A)

Replica pedal elements of Rhoetosaurus brownei were physically manipulated and examined

for ROM, as bone-to-bone articulations (analysis A1), and as articulations with hypothetical

intervening cartilage (analysis A2). Because cartilage has elastic properties that absorb and

dissipate forces applied to a joint, we used compressive and flexible polyurethane foam,

similar to those used in engineering (Mills, 2007) and biomedical studies (Patel et al., 2008;

Netti, 2014). A rectangular (120 x 200 mm), 20 mm thick piece of closed-cell foam was

placed on the epiphyses of each articular surface to replicate articular cartilage

(Supplementary Figure S8). For each joint, the phalanx was moved from the ONP to an

assumed point of dislocation in each Cartesian plane, and photographed using an iPhone6

camera (8 MP, f/2.2, 29mm). The ROM at each joint was evaluated twice from the ONP

sequentially in each 3D plane (i.e., parasagittal, frontal, transverse planes; Fig. 4). Ten

measurements were taken at each joint, to account for variability (Fig. 4). All ROM

measurements were taken by one of us (AJ) to avoid inter-observer variability. The angle of

ROM was measured in Adobe Photoshop CS6 (13.0.6).

2.3.3 Virtual manipulation (analysis B)

3D surface models of the replica pedal elements were virtually manipulated and examined for

ROM, as bone-to-bone articulations (analysis B1), and as articulations with hypothetical

intervening cartilage (analysis B2). These virtual models were obtained for each element

using photogrammetry (Falkingham, 2012; Mallison & Wings, 2012; Romilio & Salisbury,

2014). Photographs were taken with an iPhone6 camera (8 MP, f/2.2, 29mm) to produce a

spherical coverage of each bone. Approximately 120 photographs were taken to cover the

entire external surface of each element. In Agisoft Photoscan Professional (v1.0.3;

www.agisoft.com) the photographs were aligned, generated into dense clouds, triangulated

into a surface mesh, given a textured map and outputted to Wavefront Object file format

(*.obj). The 3D models were imported into Autodesk Maya 2016 (www.autodesk.com),

where the osteological elements of each digit were positioned into their ONP (Fig. 3) to be

comparable with that used for the physical manipulations (analyses A1 and A2). Virtual

cartilage was represented as a 20 mm gap between bones from their respective ONP, thus

comparable to the thickness of foam used for the physical analyses.

In order to virtually evaluate a ROM objectively for a given joint, a computational procedure was initially required, described as follows (Supplementary Figure S9). Each joint

was assigned a single center of rotation, achieved by positioning a virtual sphere at the distal

end of the proximalmost of the two adjacent bones that comprise the joint. Virtual marker

locators were applied with respect to the center of rotation of each sphere (Fig. 1B, and

Supplementary Figure S9). Rigging joints were then linked to each virtual marker locator, and aligned to the same 3D coordinates. A bone-joint rotation was considered biologically plausible if the virtual movement did not result in extreme disarticulation or collision of virtual surface models. The ROM at each joint was evaluated twice from the ONP sequentially in each 3D plane (i.e., parasagittal, frontal, transverse planes). Ten measurements were taken at each joint (to account for variation) by one of us only (AJ) to eliminate inter-observer variability.

2.3.4 Statistical analysis

All statistical analyses were completed using R (R Development Core Team, 2017) to

perform analyses of variance (ANOVA) between analyses of approach-type and condition of

articular surfaces. First, ANOVAs were performed to observe variations of ROM between

like-analyses from different approaches – i.e., analyses A1 vs. B1 (direct bone-to-bone

ROM), and analyses A2 vs. B2 (ROM of bones with articular cartilage). Subsequently,

ANOVAs were performed to observe variations of ROM between like-analyses of different

conditions at the articular surfaces: analyses A1 vs. A2, and B1 vs. B2. Visual inspection of

residual plots did not reveal any obvious deviations from homoscedasticity or normality. P- values were obtained by multiple comparisons of the associated model.

2.4 Pedal posture reconstruction

In order to determine the likely weight-bearing stance of the pes of Rhoetosaurus brownei,

we applied a constraint-based like exclusion model (sensu Gatesy, Baker & Hutchinson,

2009). This approach works by applying iterative rounds of constraining information to

ultimately infer a likely posture in an extinct animal. Here, we investigated the full array of

postures using virtual manipulations. These postures were then further delimited based on

theoretical biomechanical insights, comparisons with extant taxa, and ichnological

considerations.

2.4.1 Generalized postural morphotypes

Pedal postural morphotypes in tetrapods represent differences in autopodial configurations with respect to the substrate, generally in a parasagittal plane. The most commonly invoked are plantigrady, digitigrady, subunguligrady and unguligrady (Fig. 5A). While aspects of postural classifications vary in the literature, some definition may be acknowledged. A plantigrade posture implies that the entire autopodium is in contact with the substrate

(Carrano, 1997). A digitigrade posture involves the portion of pes distal to, and including, the

MTP joint being in contact with the substrate (Schaller et al., 2011), while a subunguligrade posture is similar, but involves the elevation of the MTP joint above the substrate

(Panagiotopoulou et al., 2012; Panagiotopoulou et al., 2016). Unguligrady is similar to subunguligrady, however only the end of the distal phalanx (often hoof-like unguals) contacts the substrate (McClure, 1993; Carrano, 1997). Previous inferences of pedal posture in sauropods recognised the aforementioned morphotypes at least once (e.g., Marsh, 1878; Paul,

1988; Gallup, 1989; Carrano, 2005), in addition to two other terms, ‘semi-plantigrady’ and

‘semi-digitigrady’ (e.g., Hatcher, 1901; McIntosh, 1990; Bonnan, 2005; Wilson, 2005).

However, these latter two are covered within the definitions of the morphotypes, described above. Despite the commonplace assignment of pedal postures to a categorical morphotype in almost all previous works, the rationale behind specific postural determinations is usually unstated, or ambiguous at best.

2.4.2 Quantification of pedal posture

Intriguingly, all aforementioned postural morphotypes lack a quantitative measure of comparison in the literature, typically being associated with the interpreter’s opinion of how the autopodium interacts with the substrate. In an attempt to further appraise this feature, we introduce a novel geometric scheme of angular dimensions as measures of posture in

Rhoetosaurus brownei, a method that is easily applicable to other tetrapods (Fig. 5B). The

ROM of each joint was assigned as follows: α corresponds to the angle at the MTP joints relative to the substrate, β corresponds to the angle at the IP joints relative to the substrate, and γ corresponds to the angle of the distalmost phalanx relative to the substrate (see Fig. 4B for specifications). This method provides a quantitative tool for precisely comparing postures between animals deemed to exhibit the same general morphotype. It is important to note that these morphotypes are simply arbitrary categories of pedal configuration, and that an in-life autopodial posture may encompass a swathe of subtly different postures.

2.4.3 Pedal posture in Rhoetosaurus brownei

The virtual ROM values with hypothetical cartilage were used in producing virtual postural reconstructions. Each digit was configured according to the different postural morphotypes

(Fig. 6A). The digits were then rearticulated at the proximal ends of the metatarsi, resulting in

This article is protected by copyright. All rights reserved. Jannel 20 distinct arch profiles. Each assembly was then positioned in connection to a vertically aligned tibia and tarsus (i.e., fixed to the tibial long axis) (Fig. 6B).

Five virtual postures in Rhoetosaurus brownei were outputted and selected for assessment in a systematic biomechanical framework. These postures correspond to the aforementioned generalized morphotypes recognized among modern terrestrial tetrapods (i.e., ranging from plantigrady to unguligrady, Fig. 6A), which have been attributed to sauropods at least once in the literature. Although these particular ‘discrete’ postures could be considered as part of a

‘continuous’ range, the description and rationale for each is as follows: o Posture 1 is a plantigrade condition, corresponding to the joints disposed at their ONP in

contact with the substrate (i.e., α1 = β1 = γ1 = 0). This posture is important because: (1) it

represents one of the original diagnostic traits supporting the establishment of Sauropoda

(Marsh, 1878); (2) it is considered as the primitive morphotype in archosaurs (Charig,

1972); (3) it is present in modern crocodilians as an outgroup of the sauropod EPB

(Parrish, 1987); (4) and it is also recorded within some large terrestrial mammals (e.g.,

Ursidae, Davis, 1964). This pose conforms to one boundary of our continuum of

postures, within the present constraint-based approach. o Posture 2 corresponds to a mid-digitigrade condition where 50% of the maximal

dorsiflexion at the MTP (α2 = αmax, β2 = γ2 = 0). For sauropods, this posture may be 1 2 synonymous with previous statements of “semi-plantigrady” (sensu Hatcher, 1901;

Bonnan, 2005), and/or “semi-digitigrady” (sensu McIntosh, 1990; Wilson & Sereno,

1998; Wilson, 2005), terms that have also been applied to elephants (Fowler, 2001;

Ramsay & Henry, 2001).

dorsiflexion recorded at the MTP joints of each digit (i.e., α3 = αmax, β3 = γ3 = 0). This

posture represents the presumed basal dinosauromorph condition that is retained within

the major lineages of dinosaurs (Sereno, 1991; Carrano, 2005; Mallison, 2010).

Particularly, this posture is commonly restored in basal sauropodomorphs (Yates, 2004),

and has sometimes been proposed historically for sauropods (Tornier, 1909; Holland,

1910; Matthew, 1910). Among modern animals, the posture is characteristic of birds as a

second outgroup of the sauropod EPB (e.g., Bonnan, 2016). o Posture 4 corresponds to a prim-subunguligrade condition (Latin primus: ‘finest’, literally ‘first’), reflecting the obligate initial elevation of the IP joints to facilitate a

subunguligrade posture. Because of the high number of incremental and alternative

postures possible, for simplification posture 4 was constrained by applying the maximal

dorsiflexion at the DIP joints of each joint (i.e., α4 =0, β4 = βmax, γ4 = 0). This posture

represents one exemplar of subunguligrady, a morphotype previously proposed for

Sauropoda (Carrano, 2005) and often attributed to elephants (Hutchinson et al., 2011;

Panagiotopoulou et al., 2012). o Posture 5 corresponds to an unguligrade condition, initiated by the elevation of the

proximal end of the ungual with respect to the substrate (α5 = αmax, β5 = βmax, and γ5 > 0).

This posture has been suggested for the sauropod Giraffatitan brancai (Paul, 1988), and

generally attributed to most ungulates (McClure, 1993; Carrano, 1997). This pose

conforms to the second boundary of our continuum of postures, within the present

constraint-based approach.

These five postural morphotypes were also reconstructed for comparisons in the additional aforementioned sauropodomorph taxa. To validate our approach, the same postures were blindly evaluated in our virtual elephant pes. These were then compared to its established in vivo posture obtained from CT data. It should be mentioned that despite being the largest living terrestrial animal, little is actually known about the skeletal foot posture in elephants, perhaps because dissections of their autopodia have been limited to a handful of studies (e.g.,

Ramsay & Henry, 2001; Weissengruber & Forstenpointner, 2004; Weissengruber et al.,

2006; Hutchinson et al., 2011). Consequently, like for sauropods, a wide range of postural morphotypes has been historically attributed to elephants (e.g., Fowler, 2001; Ramsay &

Henry, 2001; Weissengruber & Forstenpointner, 2004; Weissengruber et al., 2006;

Hutchinson et al., 2011; Panagiotopoulou et al., 2012). Ultimately, these reconstructions provide an opportunity to estimate the respective postural morphotype(s) in elephants.

2.4.4 Metapodium arch curvature

Among descriptions of sauropod autopodia, a commonly used character relates to the proximal configuration of the metapodium (i.e., metacarpus or metatarsus) (e.g., Bonnan,

2003; Apesteguía, 2005; Bonnan, 2005; Nair & Salisbury, 2012; Carballido et al., 2017).

While the feature is often used in comparative morphological descriptions, these reconstructions lack quantification. Furthermore, the potential role of metapodial curvature is often overlooked. Morphofunctionally, the arches in the feet of living animals have been proposed to provide a propulsive lever and act as shock absorbers during their locomotion

(Ward, Kimbel & Johanson, 2011). For instance, the development of longitudinal and transverse arches of the foot has been described as a key adaptation for bipedalism in

Analogous arch configurations have also been expounded to provide a morphofunctional benefit in crocodilians, lizards and elephants (Brinkman, 1980a; b; Weissengruber &

Forstenpointner, 2004). To our knowledge, however, no study has yet explicitly quantified variations in autopodial arch configuration in living taxa, making comparisons to sauropods and other extinct taxa problematic.

In an attempt to quantify this feature in a systematic manner, we introduce a geometric measure to estimate the angle of metapodium arch curvature, which can be applicable in both extinct and extant taxa. The metapodium arc angle (Φ) represents a geometric measure of two angles of curvature at the proximally bridged metapodium, resting on both external and internal arc curvatures (see Fig. 7 for its estimation). It provides a means of measuring and comparing architectural differences in the spread of the metapodium (hereafter focusing on metatarsus curvatures). Overall, higher Φ angles denote increasingly convex forms; those closer to zero indicate straighter metatarsal bridges (Table 2). The potential morphofunctional role of this feature in sauropods is further discussed in relation to the pedal posture of

Rhoetosaurus brownei, below.

2.4.5 Hypothetical plantar outline of the pes

A hypothetical outline in plantar perspective was reconstructed for each posture to allow for a preliminary observation of soft tissue inference (Fig. 14, and Supplementary Figure S12).

Such outlines do not reflect the actual in vivo margins of soft tissue of the animal plantarly, nor are they intended to conform exactly to known sauropod track shapes. These virtual outlines are purely employed here as a comparative tool to discuss the implications of

This article is protected by copyright. All rights reserved. Jannel 24 variations between postural reconstructions and actual tracks. A precise life-like outline would be difficult to reconstruct without a comprehensive understanding of most, if not all, associated pedal soft tissues. Virtual outlines for each posture were generated in Autodesk

Maya 2016 (www.autodesk.com; see Supplementary Data).

3 RESULTS

3.1 Rhoetosaurus brownei: ROM with bone-to-bone articulations

3.1.1 Physical manipulation (analysis A)

At the metatarsophalangeal (MTP) joints of Rhoetosaurus brownei, the sagittal plane involves most of the maximal excursion angle (i.e., excursion angle relates to the full swath of motion in a plane; Table 1), with values greater than 75° recorded for all digits (Fig. 8).

These data contrast with the frontal plane (Fig. 8), where the maximal excursion angles do not exceed 25° (as little as 7° in digit I). In the transverse plane, the excursion angles range between 50–60° in the medial digits I-II, and extend up to 95° in digit IV (Fig. 8).

At the distalmost interphalangeal (DIP) joints the largest excursion angles are generally recorded in the transverse plane (95–117°), followed closely by the angles of the sagittal plane (83–104°). In the frontal plane the excursion angles range between 50–70° in digits I,

III and IV, and approach 100° in digit II (see Supplementary Text 8.1.1 for additional details).

3.1.2 Virtual manipulation (analysis B)

For the MTP joints of Rhoetosaurus brownei, the maximal excursion angles generally occur in the sagittal plane, with values greater than 70° recorded for digits I-III, but approaching

40° in digit IV. The frontal plane records the minimal excursion angles, with no value exceeding 25°. The transverse plane records relatively uniform excursion angles of approximately 40° (Fig. 8).

At the DIP joints, the greatest excursion angles are generally recorded in the sagittal plane

(78–110°), followed closely by the angles of the transverse plane (67–98°). In the frontal plane the excursion angles range between 50–60° in digits I and IV, and between nearly 80–

90° in digits II-III (see Supplementary Text 8.1.2 for additional details).

3.2 Rhoetosaurus brownei: ROM with hypothetical intervening cartilage

3.2.1 Physical manipulation (analysis A)

For the MTP joints of Rhoetosaurus brownei, the maximal excursion angles occur in the sagittal plane with the recorded values always exceeding 80°. The frontal plane records the minimal excursion angles, with no value exceeding 40°. In the transverse plane, the excursion angles range between nearly 75° up to 88° (Fig. 8).

At the DIP joints, the greatest excursion angles are generally recorded in the transverse plane, with recorded values greater than 85° in occurring all digits. Although comparatively lower overall, the sagittal (80–110°) and frontal planes (60–95°) record substantial overlap in excursion angles (see Supplementary Text 8.2.1 for additional details).

3.2.2 Virtual manipulation (analysis B)

For the MTP joints of Rhoetosaurus brownei, the maximal excursion angles occur in the sagittal plane, with values greater than 75° recorded for all digits. These data contrast with the frontal plane, where the maximal excursion angles do not exceed 45°. The transverse plane records relatively uniform excursion angles of approximately 42° (Fig. 8).

At the IP joints, the greatest excursion angles are generally recorded in the sagittal plane

(104–116°), followed closely by the angles of the transverse plane (83–102°). In the frontal plane, the measured excursion angles are 65–98° (see Supplementary Text 8.2.2 for additional details).

3.3 ROM in African bush elephant (Loxodonta africana)

3.3.1 With bone-to-bone articulations

For the MTP joints of the elephant, the maximal excursion angles generally occur in the sagittal plane, with values greater than 45° for all digits (Fig. 9). The frontal plane records the minimal excursion angles, with no value exceeding 25°. In the transverse plane, the excursion angles range between nearly 40° and up to 70° (Fig. 9).

At the IP and DIP joints, the greatest excursion angles are generally recorded in the transverse plane (45–75°), followed closely by the angles of the sagittal plane (35–65°). In the frontal plane, the measured excursion angles are 13–36° (see Supplementary Text 8.3.1 for additional details).

3.3.2 Articulations with hypothetical intervening cartilage

For the MTP joints of the elephant, the maximal excursion angles occur in the sagittal plane, with values greater than 55° for all digits. The frontal plane records the minimal excursion angles, with no value exceeding 53°. In the transverse plane, the excursion angles range between nearly 48° up to 75° (Fig. 9).

At the IP and DIP joints, the greatest excursion angles are recorded in the sagittal plane

(66–78°), followed closely by the angles of the transverse plane (44–74°), and those of the frontal plane (28–52°) (see Supplementary Text 8.3.2 for additional details).

3.4 ANOVA: Effects of methodological choices on ROM results

3.4.1 Distinctions between approaches

Despite generally similar trends, statistically significant differences (p<0.05) between

specific ROMs dimensions were recorded between the physical and virtual approaches, as

evaluated by ANOVA (Supplementary Data S4). Overall, the use of a virtual approach in the

sagittal plane decreases the digital ROM by 3% in bone-to-bone articulation, but increases it by 12% with hypothetical cartilage. In the frontal plane, the use of a virtual approach increases the digital ROM by 3% for bone-to-bone, and 5% with hypothetical cartilage articulations. However in the transverse plane, it is decreased by 27% and 23%, for these respective articular scenarios (Supplementary Data S1). This indicates that the application of different approaches influences the overall ROM outcomes evaluated within each DOF. In sum, the high proportion of statistically significant p-values alludes to the dissimilarity between the physical and virtual approaches.

3.4.2 Distinctions between articular scenarios

Likewise, comparison of the ROM results between bone-to-bone and with hypothetical

cartilage assumptions (A1 vs. A2; and B1 vs. B2; Fig. 9) report statistically significant

differences (p<0.05) between specific ROMs dimensions as evaluated by ANOVA

(Supplementary Data S5). Overall, the presence of cartilaginous facets increases the digital

ROM in all planes for both physical and virtual approaches. In the sagittal plane, the

inclusion of cartilage increases the digital ROM by 7.5% for the physical approach, and 25%

for the virtual approach. In the frontal and transverse planes, the presence of cartilage

increases ROM by 34% and 6% for the physical approach, and 36% and 10% for the virtual

approach, respectively. Similar outcomes are recorded in the virtual ROM of the elephant

where the inclusion of cartilage increases the phalangeal ROM by 45% in the sagittal plane,

70% in the frontal plane, and 6% in the transverse plane (see Supplementary Data S1). This

alludes to the significance that soft tissues such as cartilage have on the overall ROM results.

3.4.3 Repeated measurements

Of note, comparisons of the ROM values of repeated intra-observer measurements (n=10)

appear overall to be non-significant (p>0.05), with the percentage of significant p-values

inferior to 50% (Supplementary Data S5). This indicates that repeated measurements do not

necessarily affect the ROM being evaluated (Fig. 10). Nonetheless, irrespective of articular

assumptions, ANOVAs record a percentage of significant p-value greater than 40% for analysis A, and inferior to 15% for analysis B. This suggests that ROM evaluated by traditional physical manipulation of bones generates substantially more variations amongst repeated measurements than ROM evaluated virtually (see Supplementary Data S5).

3.5 Pedal posture in Rhoetosaurus brownei and other sauropodomorphs

3.5.1 Postural morphotypes in Rhoetosaurus brownei

The two boundaries of our constraint-based approach for the pes of Rhoetosaurus brownei

(i.e., posture 1 [plantigrade] and posture 5 [unguligrade]) can be rejected based on the

following rationales. In full plantigrady, the ONP of each digit is aligned with the transverse

plane, resulting in the digits being parallel to the ground surface (Fig. 11). This constrains the

proximal metatarsal bridge to form an almost linear alignment (Fig. 11; Table 2) that can only

articulate with the wedge-shaped cranial face of the astragalus. As a result, the digits are

significantly splayed distally. Such a specific configuration of the pes is, to our knowledge,

unknown amongst graviportal living organisms. While a linear metatarsal proximal arch is

observed in large plantigrade terrestrial tetrapods, such as crocodilians (Brinkman, 1980a)

and some mammals (e.g., bears, wombat; Davis, 1964), the metatarsi in these taxa do not

diverge distally, and appear to be tightly configured with the tarsus. A plantigrade

morphotype would therefore form a rather poor anatomical support for the pes in

comparatively larger sauropods (Fig. 11). In unguligrady, only the distal portion of the ungual

of each digit firmly engages the substrate, resulting in a fully elevated pes (Fig. 11). In this

posture, the proximal articulation of the metatarsus forms a semi-circular arch that can only

engage with the distal astragalar surface. This configuration results in a non-articulating gap

between the metatarsals II and III and the astragalus in Rhoetosaurus brownei. In addition,

the large mediolaterally compressed unguals, combined with their bevelled and concave

articular facets, would not have been anatomically adapted for loading the tips of the unguals

only. Among modern unguligrade tetrapods for instance, the distalmost phalanx include

hoofed and keratinous-covered material that engage the substrate via a flat distal surface (e.g.,

Equidae, among other unguligrade artiodactyls; McClure, 1993; Clifford, 2010). Thus, we

conclude that the unguligrade configuration in Rhoetosaurus brownei is also unlikely given

its anatomy.

Discounting the bounding postural morphotypes, three further intermediate postures can

be considered in Rhoetosaurus brownei: mid-digitigrady, absolute digitigrady and subunguligrady (Fig. 11, postures 2–4). In all three morphotypes, the proximal articulation of the metatarsus forms a curved arch, being semi-circular in absolute digitigrady (posture 3), but less-tightly curved in mid-digitigrady and subunguligrady (postures 2 and 4).

Morphofunctionally, autopodial arches have been proposed to provide a propulsive lever, function as shock absorbers during locomotion, and act as a counterbend, minimizing stresses to which the bones are subjected (Ward, Kimbel & Johanson, 2011). As an analogue, arched construction bridges are ideally suited for evenly distributing centrally applied loading forces

(Mahmoud, 2009). Fixed at each end, an arched bridge can be vertically loaded at its centre more heavily than a straight bridge because forces are transmitted towards the two fixed ends

(i.e., carried laterally along the curve); in contrast, a straight bridge would bend under comparable loads. In these postures, the proximal metatarsal arch articulates with the craniodistal margin of the astragalus in mid-digitigrady, and to the distal astragalar surface in absolute digitigrady and subunguligrady (Fig. 11). Compared to the bounding postural morphotypes (postures 1 and 5), these configurations form a more robust metapodial-tarsal articulation.

Mid- and absolute digitigrady (postures 2–3) differ from the subunguligrade pose (posture

4) in one aspect. In the former, all phalanges fully interact with the substrate, with the caudal margin of the autopodium-ground interface occurring along the MTP joints. In contrast, this corresponding margin is at the DIP joints in subunguligrady. In living taxa, the full contact of phalanges with the substrate usually forms a prerequisite for the adequate support of the feet.

For Rhoetosaurus brownei, the hinge-like and robust MTP joints (Nair & Salisbury, 2012) may have provided greater anatomical integrity for load bearing compared to subunguligrady, which relies on relatively more fragile IP joints. This may offer a mechanical advantage for the mid- to absolute digitigrade range of postures, over other ones. A further anatomical advantage of postures 2–4 concerns the inclination of the long axis of the metatarsi, aligning

(2013) mechanically evaluated that when an element fixed at one end is loaded more parallel to its long axis (as analogous to postures 2–4), the stress relayed is expected to be reduced

(Fig. 12A). This is also consistent with the idea that larger animals tend to present more upright postures to align the limb closer to the ground reaction force in order to reduce the effects of the forces exerted by muscles (Biewener, 1989b; a).

In all three morphotypes, pedal flexibility is attained in all planes (Fig. 13). In the sagittal plane, the ground would have prohibited plantiflexion, although elimination of this barrier produces highly recurved digits, particularly in absolute digitigrady and subunguligrady.

Similarly, maximal hypothetical dorsiflexion ranges are achievable, resulting in a dorsally procurved pes (particularly in mid-digitigrady), although soft tissue could have limited these movements in life. In the frontal plane, the maximal abduction to adduction excursion of the phalanges results in the unguals occupying a wide mediolateral range. In mid-digitigrady, the maximal plantiflexion and maximal adduction produces an osteological ‘overlap’ between unguals II and III (Fig. 13), which would have been impossible in life. In the transverse plane, lateral and medial transverse rotations can reach their maximal ROM values without restriction (Fig. 13).

3.5.2 Virtual pads and pedal outlines

In most morphotypes, the inclusion of a virtual pad beneath the elevated metatarsus increases the plantar surface area of the pes (Fig. 11 and 14). This would likely confer a mechanical advantage because a larger plantar surface acts to better dissipate the pressure exerted on the pes during loading (Fig. 12B). Because pressure depends on the amount of force applied over

a certain area ( = , with F = W, normal component force [N], and A = surface area of 퐹 퐴 contact), a larger휎 surface area results in a much smaller magnitude of pressure being exerted.

This relationship represents a fundamental mechanical principle, and it is also observed

among living terrestrial tetrapods, perhaps being best exemplified in the feet of large-bodied mammals that carry heavy loads such as elephants, rhinoceroses, hippopotamuses, and camels.

Tracks may provide further clues regarding the pedal posture and form of sauropods. Here,

the generation of hypothetical pedal outlines around the pedal morphotypes of Rhoetosaurus

brownei (Fig. 14) can provide a means of comparison with tracks (acting as pseudo-tracks).

Intriguingly, our pseudo-tracks show comparatively similar outlines between morphotypes 1–

4 in neutral, maximal abduction and adduction poses. At the neutral pose, the hypothetical

plantar outline presents four clear, cranially oriented and spread digital contours. The greatest

spread of digits is exhibited in absolute digitigrady, with digit I being cranially projected and

digit IV laterally directed (Fig. 12). The presence of a hypothetical fifth digit and soft tissue

pad results in an undulating lateral margin coupled with a rounded caudal margin. In maximal

abduction, the outlines are similar between all morphotypes, being narrow and compact in

digital spread. The digital contour of digit I projects craniomedially, while digits III and IV

are generally craniolaterally orientated. In maximal adduction, the outlines are also similar

between all morphotypes, being largely orbicular, with few clear digital contours observed.

These include craniolateral and caudolateral projections representing unguals I and IV,

respectively (Fig. 14). These particular outlines are reminiscent of some actual sauropod

tracks (e.g., Farlow, Pittman & Hawthorne, 1989; Thulborn, 1990).

3.5.3 Postural morphotypes in other sauropodomorphs

All common tetrapod postural morphotypes may be evaluated in the pedes of Plateosaurus

engelhardti, Diplodocus carnegii, Camarasaurus sp., and Giraffatitan brancai (Fig. 15; see

Supplementary Figures S13–16, for detailed views of these morphotypes for each taxon). For these morphotypes, each sauropod exhibits similar osseous pedal configurations to the pes of

Rhoetosaurus brownei (Fig. 15). Based on this, we infer that their overarching anatomy and associated mechanical principles would therefore be similar. In light of the models evaluated in Rhoetosaurus brownei, the mid-digitigrade, absolute digitigrade, and subunguligrade morphotypes (i.e., postures 2–4) are deemed to be the anatomically most robust postures in

Diplodocus carnegii, Camarasaurus sp. and Giraffatitan brancai. In the non-sauropod

Plateosaurus engelhardti, all postural morphotypes are achievable and anatomically adequate.

3.6 Pedal postures in the African bush elephant (Loxodonta africana)

A single static pedal posture is typically attributed to elephants. However, based on

osteological ROM flexibility alone, the range of common postural morphotypes can be

reconstructed and accordingly evaluated in the African bush elephant (Fig. 17). In

plantigrady, the proximal metatarsal arch is partially disarticulated, with the calcaneum being

positioned at the ground, resulting in an implausible anatomical configuration. Based on this,

and the CT information, the plantigrade morphotype (posture 1) is rejected. Instead, when

compared to the in vivo CT-scanned posture, the closest virtually reconstructed morphotype

corresponds to a range around subunguligrady and unguligrady (postures 4–5), although

closer to the latter. Nonetheless, the mid- and absolute digitigrade postures cannot be totally

disregarded due to their anatomical plausibility.

4 DISCUSSION

4.1 ROM evaluations

Our measurements show that the MTP and DIP joints of the right pes of Rhoetosaurus brownei demonstrate significant degree of flexibility in all three axes of rotation. The ranges

of dorsiflexion/plantiflexion (parasagittal movements) achieved at the MTP and DIP joints

are notable, generally registering excursion angles greater than 100° and 80° respectively.

These large flexion/extension ranges are consistent with the observations of Gallup (1989)

and Bonnan (2005) who inferred marked parasagittal movements at some pedal joints of

Cedarosaurus weiskopfae (D'Emic, 2013) and Plateosaurus engelhardti. In Rhoetosaurus brownei, the large parasagittal ranges at the MTP joints relate in part to the morphology of the distal regions of the metatarsi (Nair & Salisbury, 2012). Specifically, movements are unrestricted by the presence of shallow intercondylar grooves bordered by transversely broad, roller-like condyles. Similarly, the concave articular facets of the unguals at the DIP joints permit large movements in the parasagittal plane (e.g., Nair & Salisbury, 2012: fig. 12H). The substantial parasagittal osteological movements evaluated for Rhoetosaurus brownei may simply represent an evolutionary holdover from more plesiomorphic sauropod precursors.

These precursors, namely basal sauropodomorphs (Yates, 2012), are inferred to have been

capable of substantial intra-autopodial motions (Mallison, 2010). As exemplified in our

models of Plateosaurus engelhardti, their autopodia have been reconstructed with digits that

are largely individually free (e.g., Weems, 2003), allowing for such motions to be potentially fully usable in life (Fig. 15). It is therefore expected that sauropod digits should exhibit similar patterns to basal sauropodomorphs, considering osteology alone without additional constraints.

The presence of significant excursion angles in the frontal and the transverse planes is

compelling, given all previous works have noted movements, if any, in the parasagittal plane alone. Within the frontal and transverse planes, we recorded excursion angles greater than

20° and 40° (respectively) for the MTP joints, and greater than 80° (both planes) at the DIP joints. The latter results are consistent with earlier studies that described the beveled proximal articulation of sauropod unguals as a contributing factor to their lateral orientation (e.g.,

Upchurch, Barrett & Dodson, 2004; Bonnan, 2005; Nair & Salisbury, 2012). When the compound range of motion is considered wholly, it is significantly greater than any motion in a single plane, as traditionally inferred. This is further supported by the movements estimated

between the interconnected phalanges of digits II–IV in Rhoetosaurus brownei (see

Supplementary Data). When the virtually isolated phalanges are incorporated, the excursion

angles at these digits increases the ROM by 10–30% (see Supplementary Data). This corroborates our aforementioned findings that osteological ROM at the pedal joints within a sauropod pes is significantly greater than previously expected. These results suggest that evaluations of ROM should investigate movements in all planes. Recording only

flexion/extension motion at a joint disregards the capacity for movements to be conferred in other planes, as also advocated by Kambic, Roberts & Gatesy (2014).

Similar results are recorded in the African bush elephant pes. Our evaluations demonstrate

that the pedal joints of an elephant are capable of a significant degree of flexibility in all

DOFs when osteology alone is considered. The ROM values are also further increased with the addition of cartilaginous facets. However, the pedal joints of an elephant are fully embedded in soft tissue, which suggest that such a large range of motion is not entirely employed in life. Interestingly, these results still suggest that elephants retain pedal

osteological movements through the course of their evolution. These ROM capabilities may therefore be related to specific morphofunctional aspects. For instance, Hutchinson et al.

(2011: fig.3) depicted a variation in autopodial flexibility under alternate loading conditions, an observation also previously alluded to by Weissengruber et al. (2006). Hutchinson et al.

(2011) described that the pedal joints, notably the MTP and predigit joints, recorded an increase of dorsiflexion during loading, which they suggested to provide a passive stabilizing support.

Similarly, while we recorded substantial motions in all planes in the pes of Rhoetosaurus brownei, such a range may not have been entirely employed in life, likely because of additional in vivo constraints (e.g., soft tissues, posture) (Fig. 11). As for elephants, one

explanation for the extensive osseous movements in Rhoetosaurus brownei relates to the

safety factor of the assembly (Mattheck, Bethge & Schäfer, 1993). These movements may

have operated as a stress-reducing mechanism in the support and locomotion of the animal.

For instance, during any state of contact with the substrate (stance phase), the pes must remain stable whilst simultaneously accommodating any positional changes within the assembly of bones to avoid biomechanical failure. Although the availability of flexibility

might move stresses onto less resistant materials (e.g., tendons), this risk of failure may be

counteracted by soft tissues, such as the presence of a cushioning pad beneath the elevated

bones (as occurs in elephants; see Weissengruber et al., 2006). This pad may accommodate

the slight osteological movements initiated under loading conditions. Fundamentally, the

capacity for osteological movements coupled with specific soft tissues could provide a means of securing the safety factor under the load carrying capacity of the autopodium.

4.2 Variations in procedural and articular assumptions

Our results show that variations in procedural and articular assumptions produce similar

overall trends of ROM values. However, when subjected to ANOVA the four sets of ROM

results are statistically significantly different (p<0.05). We propose that differences between

physical and virtual manipulations can be explained by distinctions in the objectivity inherent

to each method. In physical manipulations, the free articulation of the bony elements do not

allow for the exact replication of ROM values at a same joint, ensuing that results with this

approach are more likely to be inconsistent. This issue is concomitant with the handling of

fragile fossils and/or the physical constraints associated with manipulation of large specimens

such as sauropod bones. Virtual manipulations, in contrast, are not subject to these problems.

For example, the presence of fixed centers of rotation at each joint ensures the consistent

replication of ROM values. While this technique provides a procedure that is more objective

and easily replicable, it should be acknowledged that a fixed axis might also add an artificial

constraint of the true capacity of movements at a joint.

A fundamental limitation with measuring ROM in extinct taxa, regardless of the approach used, is the determination of the point of disarticulation between bones. Although instances

of extreme dislocation in well-defined joints might be obvious, the joint margins of many

fossil animals, including sauropod autopodia, are not always clearly defined. Therefore, we

recommend conducting sensitivity analyses in the form of repeated measurements to

constrain confidence (yielding standard error) for the ROM being evaluated. Although our

study only accounted for intra-observer variability, inter-observer variability ought to be

assessed in future to estimate its effect on overall ROM measurements, if any. Based on our

outcomes, we propose that the virtual methodology overall provides an alternative and

comparatively more objective approach in the evaluation of ROM. The technique is further

advantageous because it can be extended to compare different taxa, and/or specimens of

differential preservation quality (Fig. 15).

Our results also show variability in statistically significant differences between the

articular assumptions (Fig. 8 and 10; Supplementary Table S5), but generally exhibit greater

ROM values with hypothetical cartilage, both for Rhoetosaurus brownei and for the elephant.

We recognize, however, that our study only accounts for one type of tissue around bones,

namely cartilage. Future evaluations of ROM in extinct taxa ideally ought to incorporate

proxies of tissues that either presumably impede ROM (e.g., muscles, skin, ligaments) or

those that facilitate an increase of ROM (e.g., cartilage) (Hutson & Hutson, 2012).

4.3 Pedal postures

4.3.1 Rhoetosaurus brownei and other sauropodomorphs

This study demonstrates that a wide range of postural morphotypes can be reconstructed for

Rhoetosaurus brownei. Our preceding critique indicates the poor anatomical integrity of the plantigrade and unguligrade morphotypes (postures 1 and 5), and that such configurations as

This article is protected by copyright. All rights reserved. Jannel 39 specifically reconstructed in Rhoetosaurus brownei (Fig. 11) are also unknown among the feet of modern tetrapods. Instead, based on our aforementioned assessments about anatomical configurations, comparisons with modern taxa, ichnology, and recognised mechanical principles (see Results), we suggest that the most plausible postures for Rhoetosaurus brownei may have occurred within a range bracketed by mid-digitigrady and subunguligrady

(postures 2–4).

Postures 2–4 record a proper pes-tarsus articulation compared to plantigrady and unguligrady, with the metatarsals articulating to the craniodistal or distal margins of the astragalus. Unlike the feet-tarsus joint of large mammals, which typically involve five or more precise and close-fitting tarsals (e.g., in elephants), sauropods comprise mainly a single enlarged astragalus, sometimes with a reduced globular calcaneum. Therefore, the ankle articulation in sauropods effectively rests at the astragalometatarsal joint (generally between the astragalus and metatarsals I–III). The integrity of this articulation would therefore have been pivotal in bearing limb loads and facilitating transmission of forces into the more proximal parts of the limb. To our knowledge, no known living taxon records a similar articulation, preventing direct morphofunctional corroboration. This form of astragalometatarsal joint appears to be a unique configuration particular to sauropods.

While the relatively flat astragalus in Rhoetosaurus brownei indicates some incompleteness (Nair & Salisbury, 2012), better-preserved astragali in other sauropods (and even in many basal sauropodomorph) exhibit a distal articulation that forms a roller-like surface (Wilson & Sereno, 1998). Bonnan (2005) suggested that this roller-like surface might have permitted a change in posture by reorienting the metatarsals more horizontally to

This article is protected by copyright. All rights reserved. Jannel 40 maintain the advantage of plantarflexor muscles. While this idea is consistent with our mid- digitigrade morphotype (posture 2), our results do not rule out the other digitigrade and subunguligrade postures. Until these hypotheses are further quantified, it cannot be asserted that a singular adaptation ultimately governs postural change. Instead, we think it is more likely that a combination of multiple changes in morphology and structural configuration might have driven changes in pedal form during sauropod evolution.

Postures 2–4 in Rhoetosaurus brownei demonstrate an additional possible mechanical benefit relating to the proximally arched configuration of the metatarsus. The presence of arched feet is consistent with the condition in modern taxa, in which it might provide a propulsive lever by acting as shock absorber and behaving as a counterbend during locomotion (Morton, 1922; Brinkman, 1980a; b; Ward, Kimbel & Johanson, 2011; Haile-

Selassie et al., 2012). Such an arch, reconstructed in Rhoetosaurus brownei and other sauropods (Figs. 11, 15–16), would have brought peripheral elements closer to the centroid of weight transmission, hence presumably minimizing stresses to which the elements were subjected. Moreover, the arched configurations of postures 2–4 result in the most lateral digits (IV and V) being positioned caudally (Fig. 11). This configuration may be consistent with the notion that metatarsal V could have functioned as a ‘pseudocalcaneum’ for muscle attachment (Bonnan, 2005: p. 368). Unlike elephants, most sauropods lack a complex calcaneum, thus limiting muscle attachment. It is plausible that sauropods could have also evolved neomorphic structures in the pes (somewhat similar to predigits in elephants;

Hutchinson et al., 2011), functioning for muscle attachment.

Our results demonstrate that external and internal arch curvature angles largely vary

between each postural morphotype and between taxa (Fig. 16 and Table 2). Because this

study can infer at best only a range of postures rather than a specific morphotype, this renders

the attribution of a single metapodial configurations, as is often proposed in the literature,

problematic. Existing descriptions of metapodial configurations, and thus of postural

morphotypes, ought to be assessed cautiously. For example, the flat metatarsal arcades

figured for Shunosaurus lii and Omeisaurus tianfuensis (He, Li & Cai, 1988; Zhang, 1988; conveyed in Nair and Salisbury, 2012: fig. 15) lacks quantification using biomechanical insights. Without evaluating the potential autopodial posture that a taxon could have achieved, there is a limitation to determining with certainty the precise form of the corresponding metapodial arch.

Irrespective of any mechanical advantages theorized in postures 2–4, an important

consideration concerns stresses associated with fulfilling habitual locomotor tasks (Fig. 12C).

Repetitive heavy loading, as expected via the stance phase of each step cycle during

locomotion, may have resulted in chronic injuries (i.e., fatigue fractures), as observed in some

sauropod autopodia (Rothschild & Molnar, 2005; Tschopp et al., 2016). Thus, it seems

intuitive that, if considering only osteologically, the pedal posture in most sauropods was

unlikely to be fully biomechanically optimized, at least not enough to avoid incurring

consistent chronic issues.

4.3.2 Pedal pads in sauropods

To further comprehend how the sauropod pes may have functioned, we discuss insights from

the feet of large living animals, and morphological aspects from the track record. The

This article is protected by copyright. All rights reserved. Jannel 42 elevated bones of the feet of large-bodied mammals, such as elephants, are supported by fibrous ‘toe and sole cushions’ laying underneath and between the digits (Weissengruber &

Forstenpointner, 2004). From a biomechanical perspective, such fibrous tissues may function to absorb mechanical shock, store and return elastic strain energy (Weissengruber et al.,

2006). In turn, these tissues protect against localized stress and help maintain low loading pressure. Given that sauropods attained considerably greater body masses than extant elephants (10–90 tonnes [Benson et al., 2014], versus typically <7 tonnes for elephants, respectively), the loading on any sauropod pes would be considerable. From this perspective, sauropods like Rhoetosaurus brownei would have most likely had a soft tissue structure beneath the elevated bony elements to similarly reduce the significant pressure exerted on their pedes. Among extant archosaurs of the sauropod EPB approach, feet with caudally expanded soft tissue pads are unknown. The nearest corresponding soft tissue structure is the cushioning seated underneath the pedal phalanges of large ground birds, such as casuariiforms (e.g., Dromaius, Struthio or Casuarius). Recently, impressions of similar soft pads were described in the non-avian theropod Concavenator corcovatus (Cuesta et al.,

2015). Based on this feature, we can only speculate that, if basal sauropodomorphs possessed similar cushioning, such soft tissue may have provided the foundation for the development of caudally expanded pads as most sauropods attained larger sizes.

Based solely on ichnology, inferring foot anatomy might be problematic because distinct pedal skeletal and functional postures can result in comparatively similar track shapes (Fig.

18: postures A1 and A3). Nonetheless, sauropod tracks can offer further clues regarding the shape of their in-life autopodia. In the case of the plausible mid-digitigrade, absolute

digitigrade and subunguligrade postures, the proximal pedal region is elevated when the

osseous stance is considered alone. In this hypothetical scenario, we would expect the

trackmaker to leave impressions of only the cranial-most portion of the pes (i.e., of its phalanges; Fig. 18). However, few known sauropod pedal tracks reflect morphologies that corresponds with this scenario. For Rhoetosaurus brownei, there is a lack of spatiotemporal

proximate sauropod tracks, which might have otherwise provided inferences on its general

pedal morphology. Therefore, the virtual pseudo-tracks (Fig. 14) provide an alternative

means for comparison, within which inclusion of virtual pads expands the plantar surfaces

caudally (i.e., relative to the skeletal only condition; Figs. 11 and 18). Consistent with the

pseudo-tracks, actual sauropod tracks generally exhibit a rounded caudal margin, with a

distinct ‘heel pad’ impression often identified in better preserved instances (Farlow, Pittman

& Hawthorne, 1989; Thulborn, 1990; Santos, Moratalla & Royo-Torres, 2009; Mazin,

Hantzpergue & Olivier, 2017; Salisbury et al., 2017). Some tracks made by non-mammalian

therapsids exhibit comparable rounded caudal margins, and have been similarly inferred to

correlate to heel pads (Citton et al., 2018). Thus, the rounded caudal margin impressions

commonly identified in the majority of sauropod tracks provide some measure of

corroboration for the presence of a soft tissue padding in Rhoetosaurus brownei and other

sauropods.

Accordingly, we hypothesize that Rhoetosaurus brownei would have most likely

possessed a type of soft tissue pad that would have engaged the plantar region of the osseous

elevated metatarsus. The presence of a pad would have increased the surface area of contact,

thereby reducing the overall pressure exerted on the pedal elements. Such a structure would

This article is protected by copyright. All rights reserved. Jannel 44 influence the mechanical conditions of the pes by absorbing shock, through the stance and support phases, via its assumed elastic properties. In the context of repetitive loadings of the stance phase particularly, the role of the pad would have been vitally important. Above all, the pad would have made the pes of Rhoetosaurus brownei and other sauropods functionally plantigrade in life.

4.3.3 African bush elephant (Loxodonta africana)

Our study demonstrates that all pedal morphotypes may be reconstructed for a modern elephant pes, if purely considering the osseous assembly (Fig 17A). When compared with the known in vivo posture generated via CT data, the closest morphotype corresponds to a range between subunguligrady to unguligrady. However, based on the geometrical definitions of postures (Fig. 5), the in vivo elephant pes appears to be best termed unguligrade (contra to prior studies; e.g., Weissengruber & Forstenpointner, 2004; Hutchinson et al., 2011;

Panagiotopoulou et al., 2012) because all phalanges are elevated, with only the tips of the distalmost phalanges in contact with the substrate (Fig. 17B). Regardless, our results show that the digitigrade, subunguligrade, and unguligrade morphotypes (i.e., postures 2–5) appears anatomically comparable and structurally plausible (Fig. 17A). Based purely on osteological configuration, none of these morphotypes may be confidently ruled out.

However, we suggest that this range of postures may not be mutually exclusive. It is worth mentioning that our CT data correspond to the pes of a deceased animal, therefore it represents an unloaded model. As shown by Hutchinson et al. (2011), the configuration of the pedal elements are modified under varying loading conditions. As a result, different postural morphotypes may occur during a stance phase (i.e., an unloaded vs. loaded pes). For instance,

the presence of a compressed pad and distally spread digits observed in the mid-digitigrade

model (i.e., Fig 17, posture 2) may represent a loaded version of an unloaded unguligrade

model (i.e., Fig 17, posture 5), with subunguligrady perhaps being intermediate.

The idea that loading conditions alter osseous postures may be applicable to Rhoetosaurus

brownei and other fossil models represented herein. In Rhoetosaurus brownei, the mid-

digitigrade model may correspond to a loaded modification of the absolute digitigrade and/or

subunguligrade versions. In Plateosaurus engelhardti, the full range of osseous postures that

were reconstructed (Fig. 15) probably corresponds to different stance phases. The plantigrade

morphotype may correspond to a hypothetical resting pose, whereas a mid-digitigrade,

absolute digitigrade, subunguligrade and unguligrade morphotypes may represent successive

stages occurring through the stance and swing phases of Plateosaurus engelhardti. Hence, the

widespread attribution of a single morphotype to the feet of extinct and/or extant taxa should

be cautioned, which should integrate information regarding the gait cycle that may be

applicable. This reinforces the argument that an in-life autopodial posture may encompass a swathe of subtly different postures.

4.4 Hypothetical kinematics

Some considerations on the kinematics of the sauropod pes during a stance phase can be

speculated upon, with the assumption that they bore substantial plantar pads underneath

elevated metatarsi. The stance phase of an animal is considered as three, but continuous,

components: touch down, weight bearing, and propulsive (the latter has also been referred to

as kick-off by Thulborn & Wade, 1989). The scenario of kinematics that follows remains purely hypothetical, which we likened with those observed among extant graviportal animals,

such as elephants. First, during touch down, the caudal region of the pes (specifically the

caudal portion of the plantar pad) is speculated to interact with the substrate foremost, as

observed in elephants (Muybridge, 1985: Plate 36). We assume that substantial forces would have been exerted upon the pes at this point, although we posit that a soft tissue pad would have facilitated absorption of most of this stress. Second, at weight bearing, we suppose that

the plantar surface of the pes to be in complete interaction with the substrate. The joints

within each digit would have been adjusted to some extent in order to further stabilize the

pes, perhaps resulting in a close to mid-digitigrade morphotype. Such stabilization may be

contingent on the substrate properties upon which the animal moved. At this stage, the pad

would have helped in the distribution of forces, in storing or absorbing mechanical forces, as

in elephants (Weissengruber et al., 2006). Third, during the final propulsive stage, the caudal

region of the pes may have been initially elevated while the cranial portion retained contact

with the substrate. The final propulsive phase may also have involved lateral rotations of the

digits (specifically, lateral torque of the unguals), which provided some momentum to lift the

pes into the succeeding swing phase. We speculate that during this final phase the others autopodia to be in contact with the substrate, consequently reducing the pressure exerted on the uplifting pes. Finally, the release of loads during the swing phase potentially may alter the

postural configuration into a relatively passive resting pose, perhaps resulting in a more

absolute digitigrade and/or subunguligrade morphotypes.

4.5 Functionality

Varied ideas for the application of sauropod pedal capabilities (e.g., unguals mobility) have

been previously proposed, but some of these remain debated. The two main ideas currently

preferred are ‘substrate gripping’ and ‘scratch digging’. Substrate gripping involves the use

of unguals during the propulsive phase of locomotion, as preferred by Christiansen (1997) and Bonnan (2005). Scratch digging (sensu Gallup, 1989) involves the use of unguals for activities such as nest excavation, an idea favored by Fowler & Hall (2011), Hall, Fragomeni

& Fowler (2016) and Tschopp et al. (2016). It should be noted that ‘scratch digging’ sensu

Gallup (1989) was originally defined as ‘hind feet-first digging’ (scratch digging actually refers primarily to a forelimb-driven action [Hildebrand, 1985: p. 94–102]). Vila et al. (2010) further correlated the flattened unguals of sauropods with the elongated and shallow nature of sauropod nests.

Intriguingly, none of these ideas have yet been informed by biomechanical insights, and are commonly considered as independent, with one function preferred over the other (e.g.,

Bonnan, 2005; Hall, Fragomeni & Fowler, 2016). Based on our study, the wide osteological ranges of DIP joint motion recorded within the pes of Rhoetosaurus brownei does not rule out the use of unguals for ‘substrate gripping’ and/or ‘scratch-digging’ behaviors, or even in other ways. For instance, we suggest that these biomechanical abilities may have been vital to the equilibrium and minimization of stresses to help maintain a biomechanical safety factor of the autopodium. Pedal functionality is undoubtedly linked to the structural configuration of the pes, kinematics, and the properties of the substrate upon which the animal moved/behaved. Whilst our ROM evaluations do not discount prior ideas (in fact, it supports their potential), further investigations are certainly required (e.g., finite element analysis) to more fully inform on the morphofunctional understanding of the in-life sauropod pes.

4.6 Evolutionary implications

Various hypotheses for the origin of sauropod pedal postures have been proposed (Cooper,

1984; Wilson & Sereno, 1998; Apesteguía, 2005; Bonnan, 2005; Wright, 2005). These

generally record a single skeletal postural change associated with, in most instances, the

acquisition of a ‘heel’ pad. In comparison, early sauropodomorph precursors of sauropods have been regarded as exhibiting a digitigrade pedal posture without a heel pad, which is also considered to be the plesiomorphic condition in dinosaurs (Carrano, 2005). The majority of

these hypotheses, however, are rarely supported by pedal postures attributed to specific

specimens, and often incorporate little in the way of compelling biomechanical evidence.

Our evaluations of the ROM and postures of Rhoetosaurus brownei and others sauropods

imply a slightly different scenario. This study primarily supports the hypothesis that

Rhoetosaurus brownei bore a range of skeletal pedal posture around ‘digitigrady’ during

weight bearing, but would have been functionally plantigrade via possession of a theorised

soft tissue pad. The pes of Plateosaurus engelhardti, the outgroup exemplar herein, demonstrates a broad range of postures (Fig. 15), which we assume to be digitigrade during weight bearing. We propose that in sauropods, the skeletal pedal postural range was relatively narrower compared to Plateosaurus engelhardti (Fig. 15), but that early sauropods ultimately retained a postural range close to the plesiomorphic skeletal digitigrady. Among the chronologically younger sauropod exemplars (i.e., Diplodocus carnegii, Camarasaurus sp. and Giraffatitan brancai), this range appears to be conserved (Fig. 15), thus it is suggested that sauropods retained this skeletal postural range during their evolution. In this context, it is

plausible that early evolutionary changes in the sauropod pes may have been driven by the

acquisition of neomorphic soft tissues (i.e., a pad) rather than skeletal postural modification.

The permanent attainment of graviportalism, linked to columnar limbs, in the earliest

sauropods and their closest relatives (McPhee & Choiniere, 2018), was likely a key driver in the eventual evolution of a functionally plantigrade pedal posture. This idea contradicts the notion that large-bodied dinosaurs evolved non-plantigrade postures (sensu Kubo & Kubo,

2016). The upper Middle Jurassic chronostratigraphic position of Rhoetosaurus brownei suggests that contemporaneous sauropod lineages may have acquired the plantigrade functional postural morphotype by this time. However, near complete pedal skeletons of sauropods are rare prior to the Middle Jurassic, and are best known for the early sauropods

Tazoudasaurus naimi and Vulcanodon karibaensis from the upper Lower Jurassic (Cooper,

1984; Allain & Aquesbi, 2008). However, these materials appear to be osteologically intermediate between known Middle Jurassic forms (e.g., He et al., 1984; Zhang, 1988; Tang et al., 2001; Nair & Salisbury, 2012) and known non-sauropod sauropodomorph pedal

specimens (e.g., Galton & van Heerden, 1998; McPhee et al., 2014). Based on our

reconstructions of postural evolution, Tazoudasaurus naimi and Vulcanodon karibaensis

would have likely borne a skeletal postural range around digitigrady, presumably retained in

post-Lower Jurassic sauropods. However, whether these taxa (and presumably other Early

Jurassic sauropods) acquired a functionally plantigrade morphotype is unknown at present.

Regarding the acquisition of the functionally plantigrade pedal posture, eusauropods likely evolved this sometime in the late Early–Middle Jurassic.

5 CONCLUSIONS AND LIMITATIONS

Biomechanical and postural evaluations of the pes of Rhoetosaurus brownei, coupled with

additional sauropodomorphs, and compared with the foot of an African bush elephant, reveal

further insights into the abilities, paleobiology and evolution of the sauropod pes. We initially

evaluated the range of motion at each pedal joint, and subsequently attempted to determine

the likely range of pedal postures in Rhoetosaurus brownei using a constraint-based framework, a process that can be extended to other taxa. We show that the range of motion for these osteological joints were particularly large in all axes, appreciably more than previously inferred for sauropod pedes. In life, these large movements would most likely have been considerably restricted by soft tissue, as is observed in extant elephants.

Nonetheless, the resultant degrees of freedom for each joint are inferred to have played a crucial role in the stability, stress accommodation and possible kinematics of the autopodium during locomotion. These outcomes are consistent with results obtained and compared from an African bush elephant pes subjected to identical procedures.

Anatomical integrity, comparisons with modern taxa, ichnology, and accepted mechanical principles lead us to suggest that Rhoetosaurus brownei exhibited a range of likely skeletal postures, encompassing a mid-digitigrady, full digitigrady, to subunguligrady continuum.

When compared with evidence from extant large-bodied analogues and sauropod tracks, the presence of a soft tissue pad within the structural pes is inferred. Positioned beneath the elevated metatarsals, this pad would have helped reduce the pressure exerted on the pes during support and locomotion. Fundamentally, the presence of this structure would ultimately have resulted in a functionally plantigrade pedal morphology.

This study highlights that interpretations and studies made on the biomechanical abilities

of the sauropod pes appear more complex than previously assumed. It should be emphasized,

however, that our interpretations of the pes in many aspects rest largely on theoretical

grounds. Accordingly, further insights (e.g., myology, finite element analysis, further

comparative analyses with additional taxa and ichnological data) need to be integrated to

bolster paleobiological understanding of the sauropod pes. These further investigations will

ultimately be required to delimit the plausible range of pedal postures hypothesized here for

Rhoetosaurus brownei, and other sauropodomorphs.

Until a soft tissue pad structure is found preserved and/or further mechanical validations are undertaken, we theorise that the probable acquisition of a soft tissue pad is likely to have been a key innovation in sauropod evolution. This would have potentially been facilitated by antecedent shifts in the body plan and a trend towards larger body mass among earlier

precursors of sauropods (McPhee et al., 2018; McPhee & Choiniere, 2018). Such a soft tissue

pad in sauropods, and hence a functional plantigrade pedal posture, may have been

established by Middle Jurassic times.

ACKNOWLEDGEMENTS

For access to Queensland Museum specimens in their care, we thank S. Hocknull, A.

Rozefelds, C. Spring, and D. Lewis. The casts used in the present study were previously produced by JPN with the assistance of M. Herne (Nair & Salisbury, 2012), and L. Pollard,

with the assistance of AJ, produced additional casts. For access to the Plateosaurus

engelhardti materials from the centre de conservation et d’études, we thank S. Deschamps.

For access to specimens at Museum für Naturkunde and Musée des Confluences, we are

CT data, we are greatly indebted to J. Hutchinson. S. Bloomberg provided helpful advice and assistance with statistical analyses. We thank J. Carballido for providing the original phylogenetic dataset to JPN. We appreciate the constructive feedback provided on an earlier version of our manuscript by J. Whitlock and an anonymous reviewer, and we thank M.

Bonnan and two anonymous reviewers for their comments and suggestions. This research was made possible with a University of Queensland International Scholarship to AJ, and this study forms part of the doctoral thesis of AJ.

AUTHOR CONTRIBUTIONS

Study concept and design: AJ, OP, AR, SWS. Acquisition of data: AJ, OP. Analysis and interpretation of data: AJ, OP. Drafting of the manuscript: AJ, JPN, SWS. Generation of

Figures: AJ. Critical revision of the manuscript for important intellectual content: AJ, JPN,

OP, AR, SWS. Obtained funding: AJ. Administrative, technical, and material support: AJ,

JPN, OP, SWS. All authors gave final approval for publication.

ORCID

Andréas Jannel https://orcid.org/0000-0002-6625-5693

Jay P. Nair https://orcid.org/0000-0001-9218-8124

Olga Panagiotopoulou https://orcid.org/0000-0002-6457-448X

Anthony Romilio https://orcid.org/0000-0002-1218-3567

Steven S. Salisbury https://orcid.org/0000-0003-4097-8567

REFERENCES

Alexander, R. M. (1998). All-time giants: the largest animals and their problems. Palaeontology, 41, 1231–1245.

Allain, R., & Aquesbi, N. (2008). Anatomy and phylogenetic relationships of Tazoudasaurus naimi (Dinosauria, Sauropoda) from the late Early Jurassic of Morocco. Geodiversitas, 30, 345–424.

Apesteguía, S. (2005). Evolution of the titanosaur metacarpus. In V. Tidwell & K. Carpenter (Eds.), Thunder-lizards: the sauropodomorph dinosaurs (pp. 321–345). Bloomington & Indianapolis: Indiana University Press.

Benson, R. B. J., Campione, N. E., Carrano, M. T., Mannion, P. D., Sullivan, C. S., Upchurch, P., & Evans, D. C. (2014). Rates of dinosaur body mass evolution indicate 170 million years of sustained ecological innovation on the avian stem lineage. PLoS Biology, 12, e1001853. https://doi.org/10.1371/journal.pbio.1001853

Biewener, A. A. (1989a). Mammalian terrestrial locomotion and size: mechanical design principles define limits. BioScience, 39, 776–783. https://doi.org/10.2307/1311183

Biewener, A. A. (1989b). Scaling body support in mammals: limb posture and muscle mechanics. Science, 245, 45–48. https://doi.org/10.1126/science.2740914

Bonaparte, J. F., & Coria, R. A. (1993). Un nuevo y gigantesco sauropodo titanosaurio de la Formacion Rio Limay (Albiano-Cenomaniano) de la Provincia del Neuquén, Argentina. Ameghiniana, 30, 271–282.

Bonnan, M. F. (2003). The evolution of manus shape in sauropod dinosaurs: implications for functional morphology, forelimb orientation, and phylogeny. Journal of Vertebrate Paleontology, 23, 595–613. https://doi.org/10.1671/A1108

Bonnan, M. F. (2005). Pes anatomy in sauropod dinosaurs: implications for functional morphology, evolution, and phylogeny. In V. Tidwell & K. Carpenter (Eds.), Thunder-lizards: the sauropodomorph dinosaurs (pp. 346–380). Bloomington & Indianapolis: Indiana University Press.

Bonnan, M. F. (2016). The bare bones: an unconventional evolutionary history of the skeleton, Bloomington & Indianapolis: Indiana University Press.

Bonnan, M. F., Sandrik, J., Nishiwaki, T., Wilhite, D. R., Elsey, R. M., & Vittore, C. P. (2010). Calcified cartilage shape in archosaur long bones reflects overlying joint shape in stress-bearing elements: implications for nonavian dinosaur locomotion. The Anatomical Record, 293, 2044–2055. https://doi.org/10.1002/ar.21266.

Bonnan, M. F., & Senter, P. J. (2007). Were the basal sauropodomorph dinosaurs Plateosaurus and Massospondylus habitual quadrupeds? Special Papers in Palaeontology, 77, 139–155.

Brassey, C., Margetts, L., Kitchener, A. C., Withers, P. J., Manning, P. L., & Sellers, W. I. (2013). Finite element modelling versus classic beam theory: comparing methods for stress estimation in a morphologically diverse sample of vertebrate long bones. Journal of the Royal Society Interface. https://doi.org/10.1098/rsif.2012.0823

Brinkman, D. B. (1980a). The hind limb step cycle of Caiman sclerops and the mechanics of the crocodile tarsus and metatarsus. Canadian Journal of Zoology, 58, 2187–2200. https://doi.org/10.1139/z80-301

Brinkman, D. B. (1980b). Structural correlates of tarsal and metatarsal functioning in Iguana (Lacertilia; Iguanidae) and other lizards. Canadian Journal of Zoology, 58, 277–289. https://doi.org/10.1139/z80-033

Carballido, J. L., Pol, D., Otero, A., Cerda, I. A., Salgado, L., Garrido, A. C., Ramezani, J., Cúneo, N. R., & Krause, J. M. (2017). A new giant titanosaur sheds light on body mass evolution among sauropod dinosaurs. Proceedings of the Royal Society of London, B, 284, 20171219. https://doi.org/10.1098/rspb.2017.1219

Carpenter, K. (2002). Forelimb biomechanics of nonavian theropod dinosaurs in predation. Senckenbergiana lethaea, 82, 59–76. https://doi.org/10.1007/BF03043773

Carrano, M. T. (1997). Morphological indicators of foot posture in mammals: a statistical and biomechanical analysis. Zoological Journal of the Linnean Society, 121, 77–104. https://doi.org/10.1006/zjls.1996.0087

Carrano, M. T. (2005). The evolution of sauropod locomotion: morphological diversity of a secondarily quadrupedal radiation. In K. A. Curry Rogers & J. A. Wilson (Eds.), The Sauropods: Evolution and Paleobiology (pp. 229–251). Berkeley: University of California Press.

Charig, A. J. (1972). The evolution of the archosaur pelvis and hindlimb: an explanation in functional terms. In K. A. Joysey & T. S. Kemp (Eds.), Studies in vertebrate evolution: essays presented to F R Parrington (pp. 121–155). Edinburgh: Oliver and Boyd.

Christian, A., Heinrich, W.-D., & Golder, W. (1999). Posture and mechanics of the forelimbs of Brachiosaurus brancai (Dinosauria: Sauropoda). Mitteilungen aus dem Museum fur Naturkunde, Geowissenschaftliche Reihe, 2, 63–73. https://doi.org/10.1002/mmng.1999.4860020103

Christiansen, P. (1997). Locomotion in sauropod dinosaurs. Gaia, 14, 45–75.

Citton, P., Díaz-Martínez, I., de Valais, S., & Cónsole-Gonella, C. (2018). Triassic pentadactyl tracks from the Los Menucos Group (Río Negro province, Patagonia Argentina): possible constraints on the autopodial posture of Gondwanan trackmakers. PeerJ, 6:e5358. https://doi.org/10.7717/peerj.5358

Clifford, A. B. (2010). The evolution of the unguligrade manus in artiodactyls. Journal of Vertebrate Paleontology, 30, 1827–1839. https://doi.org/10.1080/02724634.2010.521216

Cooper, M. R. (1984). A reassessment of Vulcanodon karibaensis Raath (Dinosauria: Saurischia) and the origin of the Sauropoda. Palaeontologia Africana, 25, 203–231.

Cuesta, E., Díaz-Martínez, I., Ortega, F., & Sanz, J. L. (2015). Did all theropods have chicken-like feet? First evidence of a non-avian dinosaur podotheca. Cretaceous Research, 56, 53–59. https://doi.org/10.1016/j.cretres.2015.03.008

D'Emic, M. D. (2013). Revision of the sauropod dinosaurs of the Lower Cretaceous Trinity Group, southern USA, with the description of a new genus. Journal of Systematic Palaeontology, 11, 707–726. https://doi.org/10.1080/14772019.2012.667446

Davis, D. D. (1964). The giant Panda—a morphological study of Evolutionary mechanisms. In L. A. Ross, P. M. Williams & E. G. Nash (Eds.), Fieldiana: zoology memoirs. Chicago: Chicago Natural History Museum.

Egger, G. F., Witter, K., Weissengruber, G. E., & Forstenpointner, G. (2008). Articular cartilage in the knee joint of the African Elephant, Loxodonta africana, Blumenbach 1797. Journal of Morphology, 269, 118–127. https://doi.org/10.1002/jmor.10600

Falkingham, P. L. (2012). Acquisition of high resolution 3D models using free, open-source, photogrammetric software. Palaeontologia Electronica, 15, 1T. https://doi.org/10.26879/264

Farlow, J. O., Pittman, J. G., & Hawthorne, J. M. (1989). Brontopodus birdi, Lower Cretaceous sauropod footprints from the U.S. Gulf Coastal Plain. In D. D. Gillette & M. G. Lockley (Eds.), Dinosaur Tracks and Traces (pp. 371–394). Cambridge: Cambridge University Press.

Floyd, R. T. (2014). Manual of Structural Kinesiology, 19th Edition. New York: McGraw- Hill Education.

Fowler, D. W., & Hall, L. E. (2011). Scratch-digging sauropods, revisted: similarity of sauropod pes unguals to tortoises, and inferred nest-excavating bahaviour. Historical Biology, 23, 27–40. https://doi.org/10.1080/08912963.2010.504852

Fowler, M. E. (2001). An overview of foot conditions in Asian and African Elephants. In B. Csuti, E. L. Sargent & U. S. Bechert (Eds.), The elephant's foot: prevention and care of foot conditions in captive Asian and African elephants (pp. 3–7): Ames, Iowa: Iowa State University Press.

Gallup, M. R. (1989). Functional morphology of the hindfoot of the Texas sauropod Pleurocoelus sp. indet. Geological Society of America, Special Paper, 238, 71–74. https://doi.org/10.1130/SPE238-p71

Galton, P. M., & van Heerden, J. W. A. (1998). Anatomy of the prosauropod dinosaur Blikanasaurus cromptoni (Upper Triassic, South Africa), with notes on the other tetrapods from the lower Elliot Formation. Paläontologische Zeitschrift, 72, 163–177. https://doi.org/10.1007/BF02987824

Gatesy, S. M., Baker, M., & Hutchinson, J. R. (2009). Constraint-based exclusion of limb poses for reconstructing theropod dinosaur locomotion. Journal of Vertebrate Paleontology, 29, 535–544.

Haile-Selassie, Y., Saylor, B. Z., Deino, A., Levin, N. E., Alene, M., & Latimer, B. M. (2012). A new hominin foot from Ethiopia shows multiple Pliocene bipedal adaptations. Nature, 483, 565–569. https://doi.org/10.1038/nature10922

Hall, L. E., Fragomeni, A. E., & Fowler, D. W. (2016). The flexion of sauropod pedal unguals and testing the substrate grip hypothesis using the trackway fossil record. In P. L. Falkingham, D. Marty & A. Richter (Eds.), Dinosaur tracks: the next steps (pp. 138–151). Bloomington & Indianapolis: Indiana University Press.

Hamill, J., Knutzen, K. M., & Derrick, T. R. (2014). Biomechanical basis of human movement, 4th Edition. Baltimore: Williams & Wilkins.

Hatcher, J. B. (1901). Diplodocus (Marsh): Its osteology, taxonomy, and probable habits, with a restoration of the skeleton. Memoirs of the Carnegie Museum, 1, 1–63.

Hay, O. P. (1910). On the manner of locomotion of the dinosaurs especially Diplodocus, with remarks on the origin of birds. Proceedings of the Washington Academy of Sciences, 12, 1–25.

He, X., Li, K., & Cai, K. (1988). The Middle Jurassic dinosaurian fauna from Dashanpu, Zigong, Sichuan. IV. The sauropod dinosaurs 2. Omeisaurus tianfuensis, Chengdu, Sichuan: Sichuan Scientific and Technological Publishing House.

He, X., Li, K., Cai, K., & Gao, Y. (1984). [Omeisaurus tianfuensis - a new species of Omeisaurus from Dashanpu, Zigong, Sichuan]. Journal of the Chengdu College of Geology, Supplement, 2, 13–32.

Henderson, D. M. (2006). Burly gaits: centers of mass, stability, and the trackways of sauropod dinosaurs. Journal of Vertebrate Paleontology, 26, 907–921. https://doi.org/10.1671/0272-4634(2006)26[907:BGCOMS]2.0.CO;2

Hildebrand, M. (1985). Digging of quadrupeds. In M. Hildebrand, D. M. Bramble, K. F. Liem & D. B. Wake (Eds.), Functional Vertebrate Morphology (pp. 89–109). Cambridge and London: The Belknap Press of Harvard University Press.

Holland, W. J. (1910). A review of some recent criticisms of the restorations of sauropod dinosaurs existing in the museums of the United States, with special reference to that of Diplodocus carnegiei in the Carnegie Museum. American Naturalist, 44, 259–283.

Holliday, C. M., Ridgely, R. C., Sedlmayr, J. C., & Witmer, L. M. (2010). Cartilaginous epiphyses in extant archosaurs and their implications for reconstructing limb function in dinosaurs. PLoS ONE, 5, e13120. https://doi.org/10.1371/journal.pone.0013120

Hutchinson, J. R., Delmer, C., Miller, C. E., Hildebrandt, T., Pitsillides, A. A., & Boyde, A. (2011). From flat foot to fat foot: structure, ontogeny, function, and evolution of Elephant “sixth toes”. Science, 334, 1699–1703. https://doi.org/10.1126/science.1211437

Hutson, J. D. (2015). Quadrupedal dinosaurs did not evolve fully pronated forearms: new evidence from the ulna. Acta Palaeontologica Polonica, 60, 599–610. https://doi.org/10.4202/app.00063.2014

Hutson, J. D., & Hutson, K. N. (2012). A test of the validity of range of motion studies of fossil archosaur elbow mobility using repeated-measures analysis and the extant phylogenetic bracket. The Journal of Experimental Biology, 215, 2030–2038. https://doi.org/10.1242/jeb.069567

Kambic, R. E., Roberts, T. J., & Gatesy, S. M. (2014). Long-axis rotation: a missing degree of freedom in avian bipedal locomotion. Journal of Experimental Biology, 217, 2770– 2782. https://doi.org/10.1242/jeb.101428

Klinkhamer, A. J., Mallison, H., Poropat, S. F., Sloan, T., & Wroe, S. (2018). Comparative three-dimensional moment arm analysis of the sauropod forelimb: Implications for the transition to a wide-gauge stance in titanosaurs. The Anatomical Record. https://doi.org/10.1002/ar.23977

Kubo, T., & Kubo, M. O. (2016). Nonplantigrade foot posture: a constraint on dinosaur body size. PLoS ONE, 11, e0145716. https://doi.org/10.1371/journal.pone.0145716

Lacovara, K. J., Lamanna, M. C., Ibiricu, L. M., Poole, J. C., Schroeter, E. R., Ullmann, P. V., Voegele, K. K., Boles, Z. M., Carter, A. M., Fowler, E. K., Egerton, V. M., Moyer, A. E., Coughenour, C. L., Schein, J. P., Harris, J. D., Martínez, R. D., & Novas, F. E. (2014). A gigantic, exceptionally complete titanosaurian sauropod dinosaur from southern Patagonia, Argentina. Scientific Reports, 4, 6196. https://doi.org/10.1038/srep06196

Langer, M. C., de França, M. A. G., & Gabriel, S. (2007). The pectoral girdle and forelimb anatomy of the stem-sauropodomorph Saturnalia tupiniquim (Upper Triassic, Brazil). Special Papers in Palaeontology, 77, 113–137.

Longman, H. A. (1926). A giant dinosaur from Durham Downs, Queensland. Memoirs of the Queensland Museum, 8, 183–194.

Mahmoud, K. M. (Ed). (2009). Safety and reliability of bridge structures. London, UK: CRC Press/Balkema. 497 p.

Malda, J., Grauw, J. C., Benders, K. E. M., Kik, M. J. L., van de Lest, C. H. A., Creemers, L. B., Dhert, W. J. A., & van Weeren, P. R. (2012). Of mice, men and elephants: the relation between articular cartilage thickness and body mass. PLoS ONE, 8, e57683. https://doi.org/10.1371/journal.pone.0057683

Mallison, H. (2010). The digital Plateosaurus II: An assessment of the range of motion of the limbs and vertebral column and of previous reconstructions using a digital skeletal mount. Acta Palaeontologica Polonica, 55, 433–458. https://doi.org/10.4202/app.2009.0075

Mallison, H., & Wings, O. (2012). Photogrammetry in paleontology – a practical guide. Journal of Paleontological Techniques, 12, 1–31.

Marsh, O. C. (1878). Principal characters of American Jurassic dinosaurs. Part I. American Journal of Science (Series 3), 16, 411–416. https://doi.org/10.2475/ajs.s3-16.95.411

Mattheck, C., Bethge, K., & Schäfer, J. (1993). Safety factors in trees. Journal of Theoretical Biology, 165, 185–189. https://doi.org/10.1006/jtbi.1993.1184

Matthew, W. D. (1910). The pose of sauropodous dinosaurs. American Naturalist, 44, 547– 560.

Mazin, J.-M., Hantzpergue, P., & Olivier, N. (2017). The dinosaur tracksite of Plagne (early Tithonian, Late Jurassic; Jura Mountains, France): the longest known sauropod trackway. Geobios, 50, 279–301. https://doi.org/10.1016/j.geobios.2017.06.004

Mazzetta, G. V., Christiansen, P., & Fariña, R. A. (2004). Giants and bizarres: body size of some southern South American Cretaceous dinosaurs. Historical Biology, 16, 71–83. https://doi.org/10.1080/08912960410001715132

McClure, R. C. (1993). Functional anatomy of the horse foot, MU Guide G 2740. University of Missouri: Columbia, Missouri

McIntosh, J. S. (1990). Sauropoda. In D. B. Weishampel, P. Dodson & H. Osmolska (Eds.), The Dinosauria (pp. 345–401). Berkeley: University of California Press.

McPhee, B., Benson, R. B. J., Botha-Brink, J., Bordy, E. M., & Choiniere, J. N. (2018). A Giant Dinosaur from the Earliest Jurassic of South Africa and the Transition to Quadrupedality in Early Sauropodomorphs. Current Biology, 28, 1–9. https://doi.org/10.1016/j.cub.2018.07.063

McPhee, B. W., & Choiniere, J. N. (2018). The osteology of Pulanesaura eocollum: implications for the inclusivity of Sauropoda (Dinosauria). Zoological Journal of the Linnean Society, 182, 830–861. https://doi.org/10.1093/zoolinnean/zlx074

McPhee, B. W., Yates, A. M., Choiniere, J. N., & Abdala, F. (2014). The complete anatomy and phylogenetic relationships of Antetonitrus ingenipes (Sauropodiformes, Dinosauria): implications for the origins of Sauropoda. Zoological Journal of the Linnean Society, 171, 151–205. https://doi.org/10.1111/zoj.12127

Mills, N. J. (2007). Polymer foams handbook: engineering and biomechanics applications and design guide, Oxford: Butterworth-Heinemann.

Morton, D. J. (1922). Evolution of the Human Foot. American Journal of Physical Anthropology, 5, 305–336.

Muybridge, E. (1985). Horses and other animals in motion. 45 classic photographic sequences, New York, United States: Dover Publications Inc.

Nair, J. P., & Salisbury, S. W. (2012). New anatomical information on Rhoetosaurus brownei Longman, 1926, a gravisaurian sauropodomorph dinosaur from the Middle Jurassic of Queensland, Australia. Journal of Vertebrate Paleontology, 32, 369–394. https://doi.org/10.1080/02724634.2012.622324

Netti, P. A. (2014). Biomedical foams for tissue engineering applications, Cambridge: Woodhead Publishing.

Panagiotopoulou, O., Pataky, T. C., Day, M., Hensman, M. C., Hensman, S., Hutchinson, J. R., & Clemente, C. J. (2016). Foot pressure distributions during walking in African elephants (Loxodonta africana). Royal Society Open Access, 3, 160203. https://doi.org/10.1098/rsos.160203

Panagiotopoulou, O., Pataky, T. C., Hill, Z., & Hutchinson, J. R. (2012). Statistical parametric mapping of the regional distribution and ontogenetic scaling of foot pressures during walking in Asian elephants (Elephas maximus). Journal of Experimental Biology, 215, 1584–1593. https://doi.org/10.1242/jeb.065862.

Parrish, J. M. (1987). The origin of crocodilian locomotion. Paleobiology, 13, 396–414. https://doi.org/10.1017/S0094837300009003

Patel, M., Fransson, P. A., Lush, D., & Gomez, S. (2008). The effect of foam surface properties on postural stability assessment while standing. Gait & Posture, 28, 649– 656. https://doi.org/10.1016/j.gaitpost.2008.04.018

Paul, G. S. (1988). The brachiosaur giants of the Morrison and Tendaguru with a description of a new subgenus, Giraffatitan, and a comparison of the world's largest dinosaurs. Hunteria, 2, 1–14.

Perry, J. M. G., & Prufrock, K. A. (2018). Muscle functional morphology in paleobiology: the past, present, and future of “paleomyology”. The Anatomical Record, 301, 538– 555. https://doi.org/10.1002/ar.23772

Preuschoft, H., Hohn-Schulte, B., Stoinski, S., & Witzel, U. (2011). Why so huge? Biomechanical reasons for the acquisition of large size in sauropod and theropod dinosaurs. In N. Klein, K. Remes, C. T. Gee & P. M. Sander (Eds.), Biology of the sauropod dinosaurs: understanding the life of giants (pp. 197–218). Bloomington & Indianapolis: Indiana University Press.

R Development Core Team. 2017. R: A language and environment for statistical computing: R Foundation for Statistical Computing, Vienna, Austria.

Ramsay, E. C., & Henry, R. W. (2001). Anatomy of the elephant foot. In B. Csuti, E. L. Sargent & U. S. Bechert (Eds.), The elephant's foot: prevention and care of foot conditions in captive Asian and African elephants (pp. 9–12): Ames, Iowa: Iowa State University Press.

Reiss, S., & Mallison, H. (2014). Motion range of the manus of Plateosaurus engelhardti von Meyer, 1837. Palaeontologia Electronica, 17, 12A. https://doi.org/palaeo- electronica.org/content/2013/502-titanosaur-skeleton

Romilio, A., & Salisbury, S. W. (2014). Large dinosaurian tracks from the Upper Cretaceous (Cenomanian–Turonian) portion of the Winton Formation, Lark Quarry, central- western Queensland, Australia: 3D photogrammetric analysis renders the ‘stampede trigger’ scenario unlikely. Cretaceous Research, 51, 186–207. https://doi.org/10.1016/j.cretres.2014.06.003

Rothschild, B. M., & Molnar, R. E. (2005). Sauropod stress fractures as clues to activity. In V. Tidwell & K. Carpenter (Eds.), Thunder-lizards: the sauropodomorph dinosaurs (pp. 381–392). Bloomington & Indianapolis: Indiana University Press.

Salisbury, S. W., Romilio, A., Herne, M. C., Tucker, R. T., & Nair, J. P. (2017). The dinosaurian ichnofauna of the Lower Cretaceous (Valanginian–Barremian) Broome Sandstone of the Walmadany area (James Price Point), Dampier Peninsula, Western Australia. Society of Vertebrate Paleontology Memoir (Journal of Vertebrate Paleontology Vol 36, supplement to 6, November 2016), 16, 1–152. https://doi.org/10.1080/02724634.2016.1269539

Sander, P. M., Christian, A., Clauss, M., Fechner, R., Gee, C. T., Griebeler, E. M., Gunga, H.-C., Hummel, J., Mallison, H., Perry, S. F., Preuschoft, H., Rauhut, O. W. M., Remes, K., Tütken, T., Wings, O., & Witzel, U. (2011). Biology of the sauropod

dinosaurs: the evolution of gigantism. Biological Reviews, 86, 117–155. https://doi.org/10.1111/j.1469-185X.2010.00137.x

Santos, V. F. d., Moratalla, J. J., & Royo-Torres, R. (2009). New sauropod trackways from the Middle Jurassic of Portugal. Acta Palaeontologica Polonica, 54, 409–422. https://doi.org/10.4202/app.2008.0049

Schaller, N. U., D'Août, K., Villa, R., Herkner, B., & Aerts, P. (2011). Toe function and dynamic pressure distribution in ostrich locomotion. Journal of Experimental Biology, 214, 1123–1130. https://doi.org/10.1242/jeb.043596

Schwarz-Wings, D., Wings, O., & Meyer, C. A. (2007). Super sizing the giants: first cartilage preservation at a sauropod dinosaur limb joint. Journal of the Geological Society, London, 164, 61–65. https://doi.org/10.1144/0016-76492006-019

Sellers, W. I., Margetts, L., Coria, R. A., & Manning, P. L. (2013). March of the titans: the locomotor capabilities of sauropod dinosaurs. PLoS ONE, 8, e78572. https://doi.org/10.1371/journal.pone.0078572

Senter, P. J. (2005). Function in the stunted forelimbs of Mononykus olecranus (Theropoda), a dinosaurian anteater. Paleobiology, 31, 373–381. https://doi.org/10.1666/0094- 8373(2005)031[0373:FITSFO]2.0.CO;2

Senter, P. J. (2006). Comparison of forelimb function between Deinonychus and Bambiraptor. Journal of Vertebrate Paleontology, 26, 897–906. https://doi.org/10.1671/0272-4634(2006)26[897:COFFBD]2.0.CO;2

Senter, P. J., & Parrish, J. M. (2005). Functional analysis of the hands of the theropod dinosaur Chirostenotes pergracilis: evidence for an unusual paleoecological role. PaleoBios, 25, 9–19.

Senter, P. J., & Parrish, J. M. (2006). Forelimb function in the theropod dinosaur Carnotaurus sastrei, and its behavioral implications. PaleoBios, 26, 7–17.

Senter, P. J., & Robins, J. H. (2005). Range of motion in the forelimb of the theropod dinosaur Acrocanthosaurus atokensis, and implications for predatory behaviour. Journal of Zoology, 266, 307–318.

Sereno, P. C. (1991). Basal archosaurs: phylogenetic relationships and functional implications. Society of Vertebrate Paleontology Memoir, (Journal of Vertebrate Paleontology 11, Supplement to 4), 2 1–53. https://doi.org/10.2307/3889336

Simon, W. H. (1970). Scale effects in animal joints. I. Articular cartilage thickness and compressive stress. Arthritis and Rheumatism, 13, 244–256. https://doi.org/10.1002/art.1780130305

Stockwell, R. A. (1971). The interrelationship of cell density and cartilage thickness in mammalian articular cartilage. Journal of Anatomy, 109, 411–421.

Tang, F., Jing, X., Kang, X., & Zhang, G. (2001). Omeisaurus maoianus - a complete Sauropoda from Jingyan, Sichuan: China Ocean Press.

Thulborn, R. A., & Wade, M. J. (1989). A footprint as a history of movement. In D. D. Gillette & M. G. Lockley (Eds.), Dinosaur Tracks and Traces (pp. 51–56). Cambridge: Cambridge University Press.

Thulborn, T. (1990). Dinosaur tracks, London: Chapman and Hall.

Tornier, G. (1909). Wie war der Diplodocus carnegii wirklich gebaut? Sitzungsberichte der Gesellschaft Naturforschender Freunde zu Berlin, 1909, 193–209.

Tschopp, E., Wings, O., Frauenfelder, T., & Rothschild, B. M. (2016). Pathological phalanges in a camarasaurid sauropod dinosaur and implications on behaviour. Acta Palaeontologica Polonica, 61, 125–134. https://doi.org/10.4202/app.00119.2014

Upchurch, P., Barrett, P. M., & Dodson, P. (2004). Sauropoda. In D. B. Weishampel, P. Dodson & H. Osmólska (Eds.), The Dinosauria (2nd edition) (pp. 259–322). Berkeley: University of California Press.

Vila, B., Jackson, F. D., Fortuny, J., Sellés, A. G., & Galobart Lorente, À. (2010). 3-D modelling of megaloolithid clutches: insights about nest construction and dinosaur behaviour. PLoS ONE, 5, e10362. https://doi.org/10.1371/journal.pone.0010362

Ward, C. V., Kimbel, W. H., & Johanson, D. C. (2011). Complete fourth metatarsal and arches in the foot of Australopithecus afarensis. Science, 331, 750–753. https://doi.org/10.1126/science.1201463

Weems, R. E. (2003). Plateosaurus foot structure suggests a single trackmaker for Eubrontes and Gigandipus footprints. In P. M. LeTourneau & P. E. Olsen (Eds.), The Great Rift Valleys of Pangea in Eastern North America Volume 2: Sedimentology, Stratigraphy, and Paleontology (pp. 293–313). New York: Columbia University Press.

Weissengruber, G. E., Egger, G. F., Hutchinson, J. R., Groenewald, H. B., Elasser, L., Famini, D. J., & Forstenpointner, G. (2006). The structure of the cushions in the feet of African elephants (Loxodonta africana). Journal of Anatomy, 209, 781–792. https://doi.org/10.1111/j.1469-7580.2006.00648.x

Weissengruber, G. E., & Forstenpointner, G. (2004). Musculature of the crus and pes of the African elephant (Loxodonta africana): insight into semiplantigrade limb architecture. Anatomy and Embryology, 208, 451–461. https://doi.org/10.1007/s00429-004-0406-1

Wilson, J. A. (2005). Integrating ichnofossil and body fossil records to estimate locomotor posture and spatiotemporal distribution of early sauropod dinosaurs: a stratocladistic approach. Paleobiology, 31, 400–423. https://doi.org/10.1666/0094- 8373(2005)031[0400:IIABFR]2.0.CO;2

Wilson, J. A., & Carrano, M. T. (1999). Titanosaurs and the origin of "wide-gauge" trackways: a biomechanical and systematic perspective on sauropod locomotion. Paleobiology, 25, 252–267. https://doi.org/10.1017/S0094837300026543

Wilson, J. A., & Sereno, P. C. (1998). Early evolution and higher-level phylogeny of sauropod dinosaurs. Society of Vertebrate Paleontology Memoir, (Journal of Vertebrate Paleontology 18, supplement to 2), 5, 1–68. https://doi.org/10.1080/02724634.1998.10011115

Witmer, L. M. (1995). The extant phylogenetic bracket and the importance of reconstructing soft tissues in fossils. In J. J. Thomason (Ed.), Functional Morphology in Vertebrate Paleontology (pp. 19–33). Cambridge: Cambridge University Press.

Wright, J. L. (2005). Steps in understanding sauropod biology: the importance of sauropod tracks. In K. A. Curry Rogers & J. A. Wilson (Eds.), The Sauropods: Evolution and Paleobiology (pp. 252–284). Berkeley: University of California Press.

Yates, A. M. (2004). Anchisaurus polyzelus (Hitchcock): the smallest known sauropod dinosaur and the evolution of gigantism among sauropodomorph dinosaurs. Postilla, 230, 1–58.

Yates, A. M. (2012). Basal Sauropodomorpha: the “prosauropods”. In M. K. Brett-Surman, T. R. Holtz, Jr. & J. O. Farlow (Eds.), The Complete Dinosaur (2nd edition) 2 (pp. 425–444). Bloomington & Indianapolis: Indiana University Press.

Zhang, Y. (1988). The Middle Jurassic dinosaurian fauna from Dashanpu, Zigong, Sichuan. III. The sauropod dinosaurs 1. Shunosaurus, Chengdu, Sichuan: Sichuan Scientific and Technological Publishing House.

FIGURE CAPTIONS

FIGURE 1 Biomechanical coordinates systems. (A) Neutral coordinate system of a

sauropod skeleton illustrating the (non-natural) alignment of all the bones derived from an

Osteological Neutral Pose (ONP); (B) Joint coordinate system, illustrating two pedal joints in digit I of Rhoetosaurus brownei. The ‘locator’ (the virtual axes) denotes the center of rotation of the virtual joint, at the metatarsophalangeal (MTP) joint (bottom left), and at the most distal interphalangeal joint (IP) joint (bottom right). Regarding motions, lateral/medial transverse rotations occur along the x-axis, dorsiflexion/plantiflexion occur along the y-axis, and abduction/adduction occur along the z-axis. Scale bar: 10 cm.

<> Color online only

FIGURE 2 Cladogram showing the phylogenetic relationships of Rhoetosaurus brownei and four additional comparative sauropodomorphs, as indicated. The phylogeny of sauropodomorphs were hypothesized using the dataset of Carballido et al. (2017) under implied weighting methodology (see Supplemental Text). This particular cladogram is an abridgement of the maximum agreement subtree derived from the primary result of 80 most parsimonious trees each at 1350 steps long, showing selected taxa of interest (see

Supplemental Figs. S6 and S7, for full cladograms with all original operational taxonomic units). Schematic pes of each taxa used in this study illustrated on the right side, with (↔) indicating mirrored left pes for ease of comparisons. Illustrations not to scale.

<> Color online only

FIGURE 3 Definitions of angular measurements of range of motion (ROM) within each plane of rotation, illustrated using pedal digit I of Rhoetosaurus brownei. Top row: bones aligned in their Osteological Neutral Position (ONP); middle row: ROM movements measured at the metatarsophalangeal (MTP) joints from the ONP; bottom row: ROM movements measured at the distalmost interphalangeal (IP) joints from the ONP. The ROM movements measured in the sagittal plane as dorsiflexion/plantiflexion (left column), frontal plane as abduction/adduction (middle column), and transverse plane as lateromedial rotation

(right column).

<> Color online only

FIGURE 4 Schematic algorithm explaining the approaches used in this study. For each digit, the digital ROM was evaluated independently using traditional physical (A) and virtual

(B) approaches. Within both procedures, ROM values were then considered in direct bone-to- bone articulations (A1 or B1) and with the presence of hypothetical cartilaginous facets (A2 or B2). In each case, ROM data were measured at the MTP and at the IP joints of each digit in the sagittal, frontal, and transverse planes. Finally, for each component of movements (i.e., dorsiflexion/plantiflexion, abduction/adduction, lateral/medial rotations), ten repeated ROM measures were taken from the respective ONP of each joint.

<> Color online only

FIGURE 5 Autopodial postural morphotypes. (A) Autopodial posture terminology. Four main autopodial postural morphotypes commonly attributed among terrestrial tetrapods,

arbitrarily defined within the parasagittal plane according to how the elements interact with

the substrate (1 = mesopodium, 2 = metapodium, 3a & 3b = proximal and distal (generally

the ungual) phalanges. (B) Angular dimensions in autopodial posture. Proposed angular

dimensions attributed to each autopodial morphotype, where α corresponds to the angle at the

MTP joints relative to the substrate; β corresponds to the angle at the IP joints relative to the

substrate; and γ corresponds to the angle of the distalmost phalanx relative to the substrate.

These dimensions allow for quantification of each postural morphotype. A plantigrade

morphotype is defined as the alignment of all the pedal elements parallel to the substrate,

where α = β = γ = 0. A digitigrade morphotype occurs when α > 0. A subunguligrade

morphotype occurs when β > 0 and α is ≥ 0 (a negative α might define another, yet undefined,

morphotype). An unguligrade morphotype occurs when γ > 0 and α and β are ≥ 0 (negative α

and/or β values might define another, yet undefined, morphotype).

Hatching pattern indicates substrate.

<> Color online only

FIGURE 6 Pedal posture reconstructions in Rhoetosaurus brownei within a constraint-

based approach. (A) Angular dimensions for each pedal morphotype investigated in this

study following the range of morphotypes defined in Figure 4. Seven morphotypes

considered comprise: (1) plantigrady, where α1 = β1 = γ1 = 0; (2) mid-digitigrady, where α2 =

αmax*0.5 (i.e., 50% of the maximal dorsiflexion at the MTP joints); (3) absolute digitigrady,

where α3 = αmax (i.e., maximal dorsiflexion at the MTP joints); (4) prim-subunguligrady, where β4 = βmax (i.e., arbitrarily as the maximal dorsiflexion at the DIP joints); (5)

unguligrady, where α5 = αmax, β5 = βmax, and γ5 > 0 (i.e., maximal dorsiflexion at the MTP and

IP joints and proximal portion of the unguals off the ground). (B) Illustration of the procedure

behind a given virtual configuration, using posture 3 (absolute digitigrady) as example. Each

digit is initially articulated according to the angular dimensions of the morphotype (top row).

The proximal end of each metatarsal is then articulated to form a respective metatarsal

bridge, and finally positioned in relation to the vertically fixed crus and tarsus. The central

illustration represents the end results of posture 3 in mediolateral view. The pedal

reconstruction in each degree of freedom within the Cartesian system is now distinct from the

coordinate system of the ONP (Figure 2), as shown in the bottom row (including the sagittal

[left], transverse [middle], and frontal [right] planes; scale bars: 10 cm).

<> Color online only

FIGURE 7 External and Internal Metatarsal arc curvature angles (Φ). (A) Geometric measurements of a curvature angle. Considering an elliptical curve that represent a sketch of the overall metatarsal curvature, a line segment [AB] is initially drawn; where A corresponds

to the most medial (external or internal) point and B the most lateral (external or internal),

point of the proximocaudal surface of the defined inner metatarsal curvature. A perpendicular

bisector a of the segment [AB] is then traced. This bisector a intersects the line segment [AB]

at point C and the arc at point D, where the associated line segments [AD] and [BD] are then drawn. The perpendicular bisectors b and c of the segments [AD] and [BD] (respectively) are then traced. These bisectors b and c intersect the lines segment [AD] and [BD] at points E and

F (respectively). The intersection of the bisectors at point G represents the center of the

circle, of which the arc is a part. Finally the metatarsal bridge curvature angle (Φ) is found by

tracing the two radii AG and BG. (B–C) Illustration of the metatarsal bridge curvature angles measured: (B) at the external arc curvature; and (C) at the internal arc curvature.

<> Color online only

FIGURE 8 Digital ROM evaluated in the pes of Rhoetosaurus brownei. In both

procedures (analyses A and B), the ROM values are reported in the three planes of rotation

from their respective ONP, comprising the sagittal plane (involving dorsiflexion and

plantiflexion; top row), the frontal plane (involving abduction and adduction; middle row),

and the transverse plane (involving lateral and medial rotations; bottom row). For each digit,

mean ROM values with standard errors are reported at the MTP (left) and IP (right) joints.

<> Color online only

FIGURE 9 Digital ROM evaluated in the pes of an African elephant. The ROM values are

reported in the three planes of rotation from their respective ONP, comprising the sagittal

plane (involving dorsiflexion and plantiflexion; top row), the frontal plane (involving

abduction and adduction; middle row), and the transverse plane (involving lateral and medial

rotations; bottom row). For each digit, mean ROM values with standard errors are reported at

the MTP (left) and IP (right) joints.

<> Color online only

FIGURE 10 Boxplots of the digital ROM values within the three planes of rotation. Four measurements (A1, A2, B1, B2) are reported for each digital (rows) articulation (MTP and IP joints), within each plane of rotation (columns). Individual data points (dot) represent singular ROM measurements. A1 = analysis A with bone-to-bone; A2 = analysis A with cartilaginous facets; B1 = analysis B with bone-to-bone; B2 = analysis B with cartilaginous facets. For specific details of ANOVA and statistical significances, see Supplementary Table

S5.

<> Color online only

FIGURE 11 Reconstruction of the postural morphotypes of Rhoetosaurus brownei in 3D.

This includes the five postural morphotypes defined previously (numbered at top row). Top row: Proximal metatarsal bridge articular surfaces (as 3D reconstructions and schematics).

Three-dimensional reconstructions of the postural morphotypes in the Cartesian system in: the sagittal plane (second row); the transverse plane (third row); and the frontal plane (bottom row). Virtual pads are shown as voluminous render. Scale bar: 50 cm.

<> Color online only

FIGURE 12 Mechanical characteristics associated to the postural morphotypes. (A)

Relationship between loading forces and load vector, with combined stress plotted against the angle of the load direction (arrows) with the long axis of the bone (here, as Mt I). (B)

Relationship between loading forces and cross sectional area. For the same load (larger arrows), the amount of pressure exerted will depend of the cross sectional area under loading.

A larger surface area (A1) will result in a smaller amount of pressure (σ1) compared to a smaller surface area (A2). (C) Function of load magnitude and frequency of loading

(repetitions). Although large forces can be sustained with a single load, successive loadings might result in injuries. Circled numbers correspond to postural morphotypes.

<> Color online only

FIGURE 13 Reconstruction of the pedal flexibility within all postural morphotypes

(numbered at top row) of Rhoetosaurus brownei. Top row: dorsoplantar flexibilities

(illustrated in the sagittal plane). Middle row: abduction/adduction flexibilities (illustrated in

the transverse plane). Bottom row: lateromedial flexibilities (illustrated in the frontal plane).

Virtual pads are shown as voluminous render. Scale bar: 50 cm.

<> Color online only

FIGURE 14 Hypothetical static outlines of all postural morphotypes with inclusion of a

virtually restored fifth digit and theoretical soft tissue pad (numbered at top row) in plantar

view. Top row: morphotypes with maximal recorded abduction at each joint. Middle row:

morphotypes with each joint at their ONP. Bottom row: morphotypes with maximal recorded

adduction at each joint. Scale bar: 50 cm.

<> Color online only

FIGURE 15 3D reconstructions of the postural morphotypes of a range of

sauropodomorphs in craniomedial view. This includes the five postural morphotypes defined

previously (numbered at top row). From top to bottom, these include the: Late Triassic

Plateosaurus engelhardti, representing a non-sauropod sauropodomorph (outgroup)

exemplar; Middle Jurassic sauropod Rhoetosaurus brownei; Late Jurassic diplodocoid

Diplodocus carnegii; Late Jurassic macronarian Camarasaurus sp.; and Late Jurassic

macronarian Giraffatitan brancai. Virtual pads are shown as voluminous render. Not to scale.

<> Color online only

FIGURE 16 Relationship between external and internal metatarsus arc angles at distinct

postural morphotypes for sauropodomorphs (Plateosaurus engelhardti, Rhoetosaurus

brownei, Diplodocus carnegii, Camarasaurus sp., Giraffatitan brancai) and the African

elephant (Loxodonta africana). Respective postural morphotypes indicated above the graph,

with: 1, plantigrade; 2, mid-digitigrade; 3, digitigrade; 4, subunguligrade; 5, unguligrade.

Respective taxa indicated below the graph. Because the pes of Rhoetosaurus brownei is

missing DV, we provided a range of metatarsal bridge curvature angles between digits I–IV

(ΦI-IV), and with inclusion of a hypothetical fifth digit (ΦI-V), under an assumption that digit

V would likely have followed the same arc as the defined by digits I–IV. Hence, two set of values were constrained between the absence and presence of a hypothetical digit V.

Schematic illustration of the geometric measurements of a curvature angle, considering an elliptical curve that represent a sketch of the overall metatarsal curvature (see Fig. 7 for its estimation).

<> Color online only

FIGURE 17 Reconstruction of the postural morphotypes of the pes of an African elephant

(Loxodonta africana) in 3D. (A) Five postural morphotypes defined previously (numbered at

top row). Top row: Proximal metatarsal bridge articular surfaces (as 3D reconstructions and

schematics). Three-dimensional reconstructions of the postural morphotypes in the Cartesian

system in: the sagittal plane (second row); the transverse plane (third row); and the frontal

plane (bottom row). Virtual pads are shown as voluminous render. (B) End result of the 3D

model generated from CT scanned data, in: medial view (left), dorsal view (center left),

cranial view (center right), craniomedial view (right).

<> Color online only

FIGURE 18 Skeletal versus functional pedal morphotypes, with illustrations of the

resemblance in plantar outlines derived from distinct skeletal postures. Left column: the

skeletal pedal posture is plantigrade (top left), corresponding to a functional plantigrade plantar outline (bottom left). Middle column: the skeletal pedal posture is digitigrade (top middle), corresponding to a functional digitigrade plantar outline (bottom middle). Right column: the skeletal pedal posture is digitigrade (top right), corresponding to a functional plantigrade plantar outline with incorporation of a pad (bottom right).

<> Color online only

TABLE CAPTIONS

TABLE 1 Biomechanical terms used in this work.

TABLE 2 External and internal metatarsus curvature angles (Φ) for each of the five assessed postures for all taxa investigated in this study. Curvature angles for Rhoetosaurus brownei listed as preserved digits (I–IV) and with hypothetical (non-preserved) fifth digit included (I–V).

Term Definition Remarks and reference Range of motion Measurement of movement around a Floyd, 2014 (ROM) specific joint or body part Osteological See Methods Reiss and Mallison, 2014 Neutral Position Craniocaudal/caud Movement or direction between the ocranial front and back surfaces/and vice versa Dorsoplantar/plant Movement or direction between the odorsal dorsal and plantar surfaces/and vice versa Lateromedial/ Movement or direction between the mediolateral outer and inner surfaces/and vice versa Dorsiflexion Movement that results in the top of the Floyd, 2014; Hamill, bones moving toward the proximal 2014 one, decreasing the angle between two adjacent bones at a joint Plantiflexion Movement that results in the bones Although typically moving away from proximal one, referred to as ‘plantar increasing the angle between two flexion’ in most works adjacent bones at a joint (e.g., Floyd, 2014; Hamill, 2014), we use plantiflexion because it employs an inflexion consistent with ‘dorsiflexion’ Abduction Lateral movement away from the Floyd, 2014; Hamill, midline of the ONP 2014 Adduction Movement medially toward the Floyd, 2014; Hamill, midline of ONP 2014 Transverse medial Medial rotation toward the x axis In this work, application rotation of pronation/supination (e.g., Hutson, 2015) does not extend to the autopodia, which interact with the substrate and are therefore positioned differently to the long bones. We consider the use of medial and lateral rotations more consistent with regard to the pedal ROM from the ONP Transverse lateral Lateral rotation toward the x axis Remark as for transverse rotation medial rotation

Excursion angle Full swathe of motion in one plane, which combines the ranges of both components of the movement (i.e., dorsiflexion + plantiflexion; abduction + adduction; lateral + medial rotations TABLE 1. Biomechanical terms used in this work.

TABLE 2. External and internal metatarsus curvature angles (Φ) for each of the five assessed postures for all taxa investigated in this study. Curvature angles for Rhoetosaurus listed as preserved digits (I–IV) and with hypothetical (non-preserved) fifth digit included (I–

V).

Taxon Postures External arch (°) Internal arch (°)

Rhoetosaurus brownei (I- 1 39 24 IV) 2 82 56 3 133 116 4 103 88 5 136 115

Rhoetosaurus brownei (I- 1 46 30 V) 2 91 68 3 162 150 4 113 102 5 161 156 Plateosaurus engelhardti 1 34 10 2 47 28 3 83 64 4 72 51 5 65 45 Diplodocus carnegii 1 100 61 2 102 72 3 156 130 4 147 106 5 149 107 Camarasaurus sp. 1 32 11 2 89 66 3 156 129 4 125 100 5 114 86 Giraffatitan brancai 1 69 23 2 114 59 3 142 90

4 128 80 5 109 57 African Elephant 1 162 55 2 167 95 3 212 117 4 202 117 5 203 105 Postural Morphotypes: 1, Plantigrade; 2, Mid-digitigrade; 3, Digitigrade; 4, Subunguligrade; 5, Unguligrade

GRAPHICAL ABSTRACT

For the paper by Jannel, A., Nair, J. P., Panagiotopoulou, O., Romilio, A., & Salisbury, S. W. “Keep your feet on the ground”: Simulated range of motion and hind foot posture of sauropods — a case study of the Middle Jurassic Rhoetosaurus brownei

• Sauropods were iconic colossal terrestrial animals during Mesozoic times. In recent decades an emerging picture of how they operated has been developed, but many aspects of their paleobiology still remain unclear. Here, we quantify how they may have kept their hind feet on the ground.

• Using various conventional and computational methods, we assessed the range of motion in the digits of the hind foot of Rhoetosaurus brownei, a Middle Jurassic sauropod from Australia, to establish a gamut of potential foot postures. With additional input of biomechanical principles we ultimately delimit these to a predictable in-life posture.

• We found that the foot exhibited greater flexibility than previously inferred. The hindfoot of Rhoetosaurus brownei was determined to be ‘skeletally digitigrade’, implying the animal walked with its phalanges in a “high-heeled” fashion. However, the posture was almost certainly ‘functionally plantigrade’, meaning the animal actually impressed the full bottom surface of its foot, by virtue of an inferred cushioning pad.

• The soft tissue pad appears to be a key biomechanical innovation, suggested to have arisen in sauropods during the Early to Middle Jurassic.