The birth of a dinosaur footprint: Subsurface 3D motion reconstruction and discrete element simulation reveal track ontogeny Peter L. Falkinghama,b,1 and Stephen M. Gatesyb aStructure and Motion Laboratory, Department of Comparative Biomedical Sciences, Royal Veterinary College, Hatfield AL97TA, United Kingdom; and bDepartment of Ecology and Evolutionary Biology, Brown University, Providence, RI 02912 Edited by Neil H. Shubin, The University of Chicago, Chicago, IL, and approved October 30, 2014 (received for review August 22, 2014) Locomotion over deformable substrates is a common occurrence in is rarely straightforward. A track is not a simple mold of static nature. Footprints represent sedimentary distortions that provide pedal anatomy, but rather the end product of a dynamic sequence anatomical, functional, and behavioral insights into trackmaker of interactions between the moving foot and substrate (11–14). biology. The interpretation of such evidence can be challenging, Even data collected by direct observation of track formation is however, particularly for fossil tracks recovered at bedding planes incomplete and frustratingly elusive. Because sinking of the below the originally exposed surface. Even in living animals, the foot is intrinsic to the process, motions of the distal limb, the complex dynamics that give rise to footprint morphology are limb–substrate interface, and all subsurface sediment remain obscured by both foot and sediment opacity, which conceals hidden from view. animal–substrate and substrate–substrate interactions. We used Several previous studies have focused on describing de- X-ray reconstruction of moving morphology (XROMM) to image formation at deeper levels to interpret fossil tracks exposed at and animate the hind limb skeleton of a chicken-like bird travers- bedding planes beneath the surface on which the animal walked. ing a dry, granular material. Foot movement differed significantly Experimental tracks have been created predominantly in hori- from walking on solid ground; the longest toe penetrated to zontal layers of colored material that were later sectioned or split a depth of ∼5 cm, reaching an angle of 30° below horizontal be- apart (15–19). Such destructive methods lack a temporal com- EVOLUTION fore slipping backward on withdrawal. The 3D kinematic data ponent, precluding direct association of individual track features were integrated into a validated substrate simulation using the with specific anatomical structures and explicit events in the step discrete element method (DEM) to create a quantitative model cycle. Recent work using two X-ray systems (20) noninvasively of limb-induced substrate deformation. Simulation revealed that documented the 3D path of radiopaque sediment markers despite sediment collapse yielding poor quality tracks at the air– throughout track formation; however, this approach suffers from substrate interface, subsurface displacements maintain a high artificial indenter motion and limited spatial resolution. level of organization owing to grain–grain support. Splitting the Herein we describe results of a study in which we recorded live substrate volume along “virtual bedding planes” exposed prints birds traversing granular and solid substrates with biplanar X-ray that more closely resembled the foot and could easily be mistaken video. The 3D motion of the hind limb bones was reconstructed, for shallow tracks. DEM data elucidate how highly localized defor- both above and below the substrate surface. These skeletal ki- mations associated with foot entry and exit generate specific fea- nematics were then incorporated into a validated computer tures in the final tracks, a temporal sequence that we term “track simulation of the substrate to visualize subsurface sediment de- ontogeny.” This combination of methodologies fosters a synthesis formation. Exploration of the simulated volume over time offers between the surface/layer-based perspective prevalent in paleon- a dynamic glimpse into the previously invisible process of track tology and the particle/volume-based perspective essential for formation, which we term “track ontogeny.” a mechanistic understanding of sediment redistribution during track formation. Results Three-Dimensional Foot and Limb Kinematics. The guineafowl footprint | dinosaur | locomotion | discrete element method | XROMM walked steadily over both a solid platform and a trough of poppy errestrial locomotion is vital to the survival of many verte- Significance Tbrate animals, and is expressed in typical behaviors such as food acquisition, predator avoidance, mate finding, and pop- We reconstructed the 3D foot movements of guineafowl tra- ulation dispersal. Generalized models of legged movement are versing a granular substrate from biplanar X-rays, and then in- typically derived from laboratory studies of walking and running corporated those kinematics into a discrete element simulation. on stiff, solid surfaces; however, locomotion over compliant, Digital track models permitted visualization of in vivo track yielding substrates is also important, for two reasons. First, ani- formation at the surface and at virtual bedding planes for the mals frequently encounter such terrain—unconsolidated desert first time. Application of these volumetric data to fossil dinosaur sand, river banks, shorelines, snow, or simply soil after rain—in tracks uncovered the developmental origin of previously enig- their natural environments. Movement over such deformable matic features. A “track ontogeny” perspective helps integrate substrates is, accordingly, a major research area in biomechanics limb and substrate dynamics into the interpretation of track and robotics (1–3). Second, feet that deform malleable substrates morphology, from which foot anatomy cannot be read directly. leave tracks. Footprints can be a major source of information about an animal or group of animals (4–6), and this is particu- Author contributions: P.L.F. and S.M.G. designed research, performed research, analyzed larly true for extinct taxa that cannot be observed directly (7–9). data, and wrote the paper. Indeed, the only movements that have been recorded in the fossil The authors declare no conflict of interest. record were necessarily over/through suitably compliant sub- This article is a PNAS Direct Submission. strates (10). 1To whom correspondence should be addressed. email: [email protected]. Despite tracks being so common and holding so much potential, This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. extracting reliable inferences from a footprint’s final morphology 1073/pnas.1416252111/-/DCSupplemental. www.pnas.org/cgi/doi/10.1073/pnas.1416252111 PNAS Early Edition | 1of6 Downloaded by guest on September 28, 2021 Fig. 1. XROMM analysis of guineafowl limb movement through a compliant substrate (poppy seeds). (A) End view of the sediment trackway, showing the volume covered by the two X-ray beams (blue and yellow), and the two calibrated light cameras (red and green). The intersection of the X-ray beams continues below the sediment surface. (B) Perspective view of the Maya scene showing the four image planes, the reconstructed skeletal model, and the photogrammetric model of the tracks. (C) Frame of the X-ray video showing subsurface imaging and the registered bone models. (D) Comparison of steps on solid (Left) and dry, granular (Right) substrates for the same individual. (Scale bars: 20 cm in A; 5 cm in D.) seeds. The biplanar X-ray volume was large enough to allow are rendered invisible, a footprint is revealed that is analogous to reconstruction of 3D motions of the limb skeleton from the knee one exposed by splitting along a bedding plane in fossilized sedi- down throughout one step by X-ray reconstruction of moving ments. Transitional and final track contours are clearly shown on morphology (XROMM) (Fig. 1 A–C and Movie S1). We iden- color height maps. Warm-colored particles are elevated above each tified comparable stages of the locomotor cycle based on motion green “virtual bedding plane,” whereas cool-colored particles are of the tarsometatarsus (Fig. 1D), and distinguished major dif- exposed lying below these planes. ferences in kinematic patterns between substrates. Final track morphology varied dramatically with depth (Fig. – On the solid platform, the knee remained 16 18 cm above the 3E). The most striking difference was the definition of upper- surface during the stance phase. Soon after contact with the most and deeper surfaces. Unlike the collapsed air-sediment – platform, the three forward-facing toes (digits II IV) came to lie boundary, sediment–sediment interfaces below approximately 1 horizontally along most of their length. Later in stance phase, cm clearly recorded the passage of the foot. Toe impressions these toes rolled smoothly up and forward, sequentially lifting were preserved as deep, steep-walled incisions. Tracks at a depth from proximal to distal. The longest toe (digit III) pivoted about of 1–2 cm also showed obvious uplifted features formed from the tip of its claw, cleanly transitioning into swing phase without particles that rose well above their original starting depth. At slipping backward. 3 cm, a very faint trace of the hallux (digit I) was discerned, but On poppy seeds, the foot immediately plunged beneath the at this and deeper bedding planes, the tracks are reduced sediment on contact. Penetration lowered knee height, which – varied from only 12 cm to 15 cm above the level of the undisturbed